Datasets:

eswardivi

/

books

like 0

of a new Ethernet-based technology which ensures sub-nanosecond synchronization and deterministic data transfer. The project uses an open-source paradigm for the development of its hardware and software components (https://www.ohwr.org/projects/ white-rabbit/wiki/). Synchronous Ethernet is an ITU-T standard for computer networking that facilitates the transfer of clock sig- nals over the Ethernet physical layer. This signal can then be made traceable to an external clock. 5G Network Architecture 75 CPRI Transport The CPRI is an industry forum defining a publicly available specification for the interface between a radio equipment control (REC) and a radio equipment (RE) in wireless networks. CPRI specifies a digitized serial interface between a base station referred to as REC in CPRI terminology and an RRH or RE. The specifications cover the user- plane, the control-plane transport mechanisms, as well as the synchronization schemes. The specification supports both electrical and optical interfaces as well as point-to-point, star, ring, daisy-chain topologies. The CPRI interface provides a physical connection for I/Q samples transport as well as radio unit management, control signaling, and synchronization such as clock frequency and timing synchronization [75]. CPRI transports I/Q samples to/from a particular antenna port and RF carrier. This is called an antenna-carrier (AxC) and is the amount of digital baseband (I/Q) user-plane data neces- sary for either reception or transmission of only one carrier at one independent antenna ele- ment. An AxC group is an aggregation of multiple AxC streams with the same sample rate, the same sample width, and the same destination. An AxC container consists of a number of AxCs and is a part of a basic CPRI frame (see Fig. 1.28). Data is organized into basic frames of 16 words. The first word of each basic frame is the control word. Each word can be 8, 16, or 32 bits, depending on the width of the I/Q samples. The width of the word depends on the CPRI line rate. For example, in

an LTE system, if I = 16 bits and Q = 16 bits, then one AxC is 32 bits. Each 256 basic frames make up a hyperframe and 150 hyper- frames are needed to transport an LTE 10 ms frame. Data in a basic frame is encoded with 8B/10B encoding, that is, 8 bits of data are encoded in 10 bits. The extra bits are used to detect link failures. Some of the CPRI rates support 64B/66B encoding scheme and this CPRI option 1 (basic frame structure (128 bits/260.42 ns) 16 bits/260.42 ns 16 bits/260.42 ns- 8B/10B encoding CPRI line rate 614.4 Mbps W = 15 CPRI payload (I/Q data block 120 bits) 16 bits/260.42 ns 64 bits CPRI payload (I/Q data block 120 bits) 16 bits/260.42 ns 32 bits 16 bits/260.42 ns 16 bits 8 Bits Option (payload = 120 bits) Option 2 (payload = 240 bits) Option 3 (payload = 480 bits) Option 5 (payload = 960 bits) Radio Radio equipment control equipment (REC) LTE/NR upper LTE/NR PHY LTE/NR RF layers Control and management IP backhaul Timing and synchronization Figure 1.28 CPRI frame structure 60]. 76 Chapter 1 extension is used to detect sync header impairments and link failures. Note that 8B/10B and 64B/66B encodings incur 20% and 3% overhead, thus the latter would be a significant improvement in overhead reduction. The CPRI specification specifies the maximum allowed effect of the fronthaul jitter on the frequency accuracy of the clock recovered at the RRH relative to a master reference clock at the BBU. One of the CPRI technical requirements defines the clock frequency accuracy of RRH as + 0.002 ppm. This requirement states that the maximum impact of jitter from the CPRI fronthaul on the frequency accuracy of RRH should be less than + 0.002 ppm. In addition to jitter, there are other factors that may affect the accuracy of clock frequency. CPRI requires very high reliability with bit error rates (BERs) of V 10-12. The CPRI framing process is illustrated in Fig. 1.29. User data is transported as baseband digital I/Q stream in a data block of a CPRI basic frame. The RRH, upon receiving the data, converts it into an

analog signal, amplifies it, and then radiates the signal over the air. Control and management data and synchronization information are delivered through CPRI subchannels, more specifically through control words in the CPRI basic frames. This infor- mation is only used by the REC (on the BBU side) and the RE (on the RRH side). CPRI subchannels are created per CPRI hyper-frame, which is 66.67 us and a hyperframe consists of 256 basic frames (260.42 ns). Each basic frame has one byte of a control word and 15 bytes of payload. A group of 256 control words in one hyperframe collectively constitute 64 subchannels. Fig. 1.29 shows a CPRI subframe and how the control and management data and synchronization information are mapped and transported [75]. eCPRI Transport The concepts of CPRI-over-Ethernet and replacing the TDM-like CPRI format with Ethernet messaging both hold the promise of reducing the bandwidth require- ments of CPRI transport and making fronthaul affordable and available to all mobile opera- tors. Ethernet is a very cost-effective transport technology that is widely deployed in the 010001000100111 0100010001001117 LTE/NR LTE/NR RF User data LTE/NR IP backhaul Control and management Synchronization Optical fiber medium Fast C&M information (Ethernet) Fast C&M information (Ethernet) Slow C&M information (HDLC) Slow C&M information (HDLC) L1 Inband protocol 16 bits/260.42 ns L1 Inband protocol Vendor specific information Vendor specific information Timing and synchronization Timing and synchronization Radio equipment (RE)/RRH/RRU W= 15 Radio Equipment Control (REC)/BBU/DU CPRI payload (I/Q data block 120 bits) CPRI interface Figure 1.29 Illustration of CPRI framing [60]. 5G Network Architecture backhaul transport network. However, it is also an asynchronous best effort technology that has not been originally designed to meet the low latency, low jitter, and tight synchroniza- tion requirements of baseband signal transmission. The new specification, known as eCPRI, introduces improved transport efficiency t

o match the speed and bandwidth requirements of 5G fronthaul networks. The eCPRI specification was released in August 2017 that supports partitioning of base station functions. The main advantages of the eCPRI protocol include support of functional split option 7, flexible bandwidth scaling according to user-plane traf- fic, and the use of mainstream transport technologies, which makes it possible carrying eCPRI and other traffic simultaneously in the same switched network. The main difference between eCPRI and CPRI v7.0 can be summarized by looking at their respective characteristics [76]. CPRI characteristics It is intrinsically a point-to-point interface. There is a master port and a slave port connected directly by optical/electrical cable (s) as a hop. Networking functions are application layer functions and not supported by the CPRI interface itself. Supported topologies depend on REC/RE functions. Supported logical connections include point-to-point (one REC one RE) and point-to-multipoint (one REC several REs). Redundancy, QoS, security, etc. are REC/RE functions. eCPRI characteristics An eCPRI network consists of eCPRI nodes (eRECs and eREs), transport network, as well as other network elements including grand master for timing and EMS/NMS for management. There is no longer a master port/slave port classification at physical level. SAPs: master of PTP and synchronous Ethernet is not an eREC entity in general. SAPCM: some of management-plane entities may be managed by EMS/NMS. The eCPRI layer is above the transport networking layer. The eCPRI layer does not depend on a specific transport network layer (TNL) topology. The transport network may include local network and local switches provided by the eREC/eRE vendors. Supported logical connections include point-to-point (one eREC one eRE), point- to-multipoint (one eREC several eREs), multipoint-to-multipoint (eRECs eREs, eRECs eRECs, eREs eREs). Redundancy, QoS, security, etc. are mainly transport network functions; eCPRI nodes need to implement proper TNL

protocols to support these capabilities. 78 Chapter 1 Logical connection for synchronization (eREC<eRE) Logical connection for C&M data eREC eRE) connection for (eREC<>eRE) Transport network SAPCM User-plane User-plane SAPUL SAPCM SAPU. SAPCM eCPRI Standard eCPRI Standard specific protocls specific protocls Transport network layer Transport network layer eCPRI radio equipment controller eCPRI radio equipment Figure 1.30 eCPRI system and interface definition [76]. As shown in Fig. 1.30, in eCPRI, the radio base station is divided into two building blocks: eCPRI radio equipment control (eREC) and eCPRI radio equipment (eRE), which are physi- cally separated and are connected via a transport network. The eREC implements part of the physical layer functions and higher layer functions of the air interface, whereas the eRE con- tains the remaining part of the physical layer functions and the analog RF functions. User- plane data, control and management, and synchronization signals (i.e., synchronization data used for frame and timing alignment) are packetized, multiplexed, and transferred over the transport network which connects eREC(s) and eRE(s). The eCPRI does not rely on specific transport network and data-link-layer protocols, thus any type of network can be used for eCPRI provided that eCPRI requirements are fulfilled (see Fig. 1.31). Fig. 1.32 shows high-level protocol stack and physical layer processing of an LTE eNB or NR gNB. The eCPRI specification defines five functional splits identified as A to E splits. An additional set of intra-PHY functional divisions identified as I, II, and IU are also defined. It is understood that the CPRI specification supports only functional split E. The physical layer processing stages shown in Fig. 1.32 are consistent with those of the NR. The eCPRI specification focuses on three different reference splits, two splits in the down- link and one split in the uplink. Any combination of the different DL/UL splits is also possi- ble. Other functional splits within the physical l

Ethernet Ethernet PHY C&M: Control and management VLAN: Virtual LAN OAM: Operations, administration, and maintenance IPsec: Internet protocol security UDP: User datagram protocol MACsec: Media access control security ICMP: Internet control message protocol SNMP: Simple network management protocol Figure 1.33 eCPRI protocol stack over IP or Ethernet [76]. management plane; and synchronization plane. The control and management information exchanged between the control and management entities are within the eREC and the eRE. This information flow is provided to the higher protocol layers and is not considered to be time critical. An eCPRI protocol layer for the user plane is defined, as shown in Fig. 1.33. The eCPRI specification identifies Ethernet and IP as two transport options for the user plane. It further defines certain messages for the user plane, which include user data, real- time control data, and other services. The I/Q Data and Bit Sequence message types are defined for the user data whose selection depends on the functional split that is used. The Real-Time Control Data message type is defined for real-time control information. These message formats are shown in Fig. 1.34. The eCPRI specification does not provide the detailed description of the information fields of the above message types. In these message formats, the following fields are identified: PC_ID (an identifier of a series of I/Q Data Transfer messages; or an identifier of a series of Bit Sequence Transfer messages), RTC ID (an identifier of a series of Real-Time Control Data messages), SEQ_ID (an identifier of each message in a series of I/Q Data Transfer messages; or an identifier of each message in a series of Bit Sequence Transfer messages), I/Q samples of user data (a sequence of I/Q sample pairs (I,Q) in frequency domain or time domain and associated control information, if necessary), bit sequence of user data, real- time control data, whose interpretation is left to implementation. Therefore, the details of 5G Network Architecture P

C_ID PC_ID RTC ID SEQ_ID SEQ_ID SEQ_ID User data IQ samples (1st octet) User data bit sequence (1st octet) Real-time control data (1st octet) User data IQ samples (Lth octet) User data bit sequence (Lth octet) Real-time control data (Lth octet) Bit sequence transfer message Real time control data message IQ data transfer message format format format Figure 1.34 eCPRI user-plane message formats [76]. information flow are out of the scope of the eCPRI specification. This flexibility implies that there is additional work required to realize multi-vendor interoperable solutions. IEEE 1914.3 Radio-Over-Ethernet Transport As we mentioned earlier, the TSN has been designed to ensure timely transport of delay-sensitive packetized streams such as CPRI-over- Ethernet; however, it does not deal with the encapsulation of various fronthaul transport pay- loads. RoE is a standard for radio transport over Ethernet including specification of encapsula- tions and mappings developed by IEEE 1914.3 working group. 46 It enables transport data over Ethernet (i.e., native RoE packet mapper) as well as support of structure-aware and structure-agnostic mappers for CPRI and other data formats. The work targeted among others definition of a native RoE encapsulation and mapper transport format for both digitized radio payload (I/Q data) and management and control data. The Ethernet packet format itself is not changed and neither is the MAC protocol, as shown in Fig. 1.35. The RoE defines a native encapsulation header format for transporting time-sensitive radio data and control information. The definition of protocol primitives allows multiplexing of independent streams, for example, antenna and carriers; time stamping or sequence number- ing to enable time synchronization of RoE packets and timing alignment of the streams; control protocol for auxiliary non-data streams and for link and RoE endpoint management; and mapper(s) for existing CPRI framing standards to a native RoE encapsulation and trans- port. There is also focus on enabling

support for non-native radio data where the transport structure is simply a container for the data (see Fig. 1.35). IEEE 1914.3 has further IEEE 1914.3: Standard for radio-over-Ethernet encapsulations and mappings (http://sites.ieee.org/sagroups- 1914/p1914-3/). 82 Chapter 1 Logical connection for timing packets Logical connection for control packets Logical connection for data packets RoE node Pass-thru Termination Ethernet frame Ethernet frame Preamble + SFD Transport of a Preamble + SFD MAC DA combination of antenna MAC DA MAC SA carriers (2G/3G/4G/5G IQ MAC SA data can be pre- EthType processed, compressed, or Ethtype unmodified) OBSAI frames unchanged RoE header-> length, AxC RoE header-> length, ID, FrameNum, Type = IQ SeqNum, type = structure Data/VSD/C&M, sample size, etc. agnostic, etc. Transparent vendor- specific data (transport CPRI frames unchanged only) RoE payload RoE payload Inter-packet GAP Inter-packet GAP Transparent C&M data New transport frames (transport only) unchanged Figure 1.35 RoE encapsulation: RoE structure aware and agnostic mappers [77]. considered implementing a structure-aware mapper for CPRI with well-defined encapsula- tion procedure. One can therefore distinguish between a simple tunneling mapper for the former use case and the structure-aware mapper for CPRI. There is also a structure-agnostic mapper in the specs which offers a middle ground for efficiency and complexity relative to the other mappers. Several use cases of RoE can be considered including aggregation of multiple CPRI streams from a number of RRHs to a single RoE link to the BBU pool, or a native edge-to-edge RoE connection from the RRH directly to the BBU pool. The RoE will logically add a new switching/aggregation node between the baseband pool and the radio resources. IEEE 1914.1 Next-Generation Fronthaul Interface The fronthaul packet transport enables implementation of critical 4G/5G technologies such as massive MIMO, CoMP transmission and reception, and scalable centralized/virtual RAN functions. Current ne

twork deployment models and practices based on traditional backhaul or fronthaul requirements are likely to be unsustainable and expensive for 5G deployments as the need for integrated access and backhaul in heterogeneous networks becomes more compelling. The NGFI is the new 5G Network Architecture Leaf access network Metro network- Backhauk EPC/5GC ((1)")) Split option X Split option y NGFI-I NGFI-II- EPC/5GC Transport sub networks Figure 1.36 Example of RAN aggregation node with CU and DUs 78]. packet-based fronthaul interface specified by IEEE 1914.1 47 in order to provide bandwidth efficiency and to achieve scalability in transport networks as the CPRI standard cannot scale with the bandwidth requirements of 5G. IEEE 1914.1 standard simplifies network design and operation, increases network flexibility and resource utilization, and lowers cost by leveraging the existing and mature Ethernet-based solutions for critical functions such as QoS, synchronization, and data security. The fronthaul architecture provides unified man- agement and control mechanisms, common networking protocols, and universal network elements, thus facilitating migration to cloud-based mobile networks. As shown in Fig. 1.36, the high-level architecture for transport network for the next- generation mobile systems is typically hierarchical and rigorously follows the physical OTN topology. In the reference architecture, the core network is where the packet core gateways are located. The metro aggregation network aggregates one or more metro ANs, which again aggregates one or more cell site or leaf transport networks. Ring network topologies are common due to their resilience properties; however, other topologies are also conceiv- able. The leaf networks are often point-to-point topologies. There is no single location for central units and BBU pools. Each of the larger transport network domains may have their own CU/BBU pool sites. The same also applies to DUs. For instance, a DU may be located in an evolved RRH, in an aggregation node conn

ecting to multiple RRHs, or in a central IEEE 1914.1: Standard for packet-based fronthaul transport networks (http://sites.ieee.org/sagroups-1914/ p1914-1/). 84 Chapter 1 FH-I (CPRI) FH-II (RoE) FH-II (Ethernet) RoE mapper aggregator and aggregator Case 1 Subclass 1 (100 ust Subclass 2 (1 ms) Subclass 3 (10 ms) Case 2 Subclass 1 (100 us) Subclass 2 (1 ms) Transport connection L1 aggregation blocks Subclass 3 (10 ms) L2 aggregation blocks eNB/gNB blocks URLLC FH-III-> CPRI/eCPRI/ Ultra-low latency transport Low latency transport Subclass 0 (50 us) Medium latency transport Subclass 1 (100 us) Subclass 2/3 (1/10 ms) Subclass 1 (100 us) Subclass 2/3 (1/10 ms) Subclass 3 (10 ms) Figure 1.37 Next-generation fronthaul: CPRI-over-Ethernet for LTE and NR 78]. office. These networks must be able to transport 5G flows with heterogeneous traffic pro- files. The traffic profiles may consist of traditional backhaul IP traffic, several different 3GPP functional split options traffic profiles, application traffic with varying latency needs, or non-IP CPRI TDM traffic. The transport network has to be able to serve all traffic pro- files with vastly different service-level requirements in the same transport network infra- structure [78]. The transport network dedicated to 5G services is hierarchical, which means it comprises different domains and progressively aggregates signals from RRHs, at one end and up to the packet core at the other end. The support of multi-level functional split, results in a logi- cal partitioning where more fronthaul segments and a backhaul segments may be identified. Fig. 1.37 depicts a generic model for converged fronthaul/backhaul network, where the 5G Network Architecture 85 following segments are identified: Fronthaul-I (NGFI-I), connecting the RU to a DU; Fronthaul-II (NGFI-II), connecting the DU to a CU; and Backhaul (BH), connecting the CU to the packet core elements. The NGFI reference architecture assumes all network deploy- ment scenarios can leverage the same transport network infrastructure.

Fig. 1.37 illustrates a deployment scenario where both new and legacy radio technologies coexist in the same cell site. In essence, the same transport infrastructure has to serve legacy backhaul, legacy non- packet CPRI fronthaul, and the highly versatile 5G fronthaul incorporating multiple func- tional split options with varying traffic profiles. It is also possible to deploy multi-level functional splits between an RRH and a BBU pool, resulting in multiple fronthaul transport domains in the network. However, practical limitations limit the number of splits that can be supported. For instance, an RRH to the aggregating DU node connection may implement 3GPP functional split option 7, and the DU to the CU connection may subsequently imple- ment 3GPP functional split option 2. Fig. 1.37 illustrates an example deployment with two functional splits between the RRH and the CU. Typically, the DU would be an aggregation node close to the radio access network edge, and possibly in charge of time- and latency- sensitive coordinated scheduling and MAC-level retransmission functions. IEEE 1914.1 standard specifies a number of service classes corresponding to the charac- teristics of traffic being transported over the backhaul/fronthaul links in a C-RAN, which includes control information or user traffic. In particular, it defines dedicated sub- classes providing requirements for maximum tolerable latency for mobile control, trans- port control, and synchronization signals. The data-plane (or user traffic) class is further divided into five different subclasses, each addressing a specific network application (e.g., 5G service transport or functional split options implemented between RRH, DU, and CU). More specifically, the very low-latency subclass addresses network segment supporting URLLC service; the low-latency subclass addresses network segment where 3GPP split options 6, 7, 8 are implemented, the medium-latency subclass addresses net- work segment where 3GPP split options 2, 3, 4, 5 are implemented, the high-latency s

ubclass applies to functional split options 2 and 3 with longer transport distances, and the very high-latency subclass is applicable to functional split option 1 or legacy trans- port that connects the access and core networks with much longer transport distance. Network slicing enables next-generation network to provide multiple type of services and applications with different service requirements under a common and shared physical net- work infrastructure. An NGFI-compliant transport network supports transport of the network slicing traffic and serves as a sub-network instance and resource manageable by the network slicing orchestrator. From architectural perspective, two operation modes may be defined depending on the level of transparency to the network slicing. In slicing-agnostic transport mode, the NGFI deployment scenarios may coexist across multiple instances in the same transport network segment, potentially owing to variations in RAN node (RRU, DU, CU) locations that resulted from network slicing operation. 86 Chapter 1 Therefore, it is possible that multiple classes of service be simultaneously enacted in a transport connection between two network nodes. The combined classes of service are logically presented in Fig. 1.38, which are supported by a slicing-agnostic transport net- work. The slicing-agnostic transport is able to provide multiple levels of transport ser- vice and to maintain different transport QoS based on the class of service requirement. It also performs transport key performance indicator (KPI) monitoring and reports the results to NGFI operation and maintenance (OAM), on the basis of classes of service. The transport network in this mode is not directly aware of the network slicing opera- tion. Instead, it simply makes the classes of service available via its interface to the other network entities, where a slicing-to-classes of service mapping needs to be per- formed by aggregating the slicing traffic with similar KPI requirements. If RAN nodes that support the sliced services are

not geographically co-located, this class of service may be routable to multiple destinations. Upon addition, deletion, or modification of the slicing operations, the slicing-agnostic transport is not expected to adapt to the change and to perform any reconfiguration process for optimization of the transport network. In slicing-aware transport mode, while the user plane remains class of service based and the transport QoS is maintained at class of service level, a slicing-aware transport network interfaces with the network slicing orchestrator which is aware of the network slicing operation. Thus, the transport operation can be controlled and managed via the NGFI OAM configuration and provisioning functions for the purpose of network slicing optimi- zation. As depicted in Fig. 1.38, the slice-to-class of service mapping is required to sup- port class of service-based transport operation and is realized within the transport network and controlled by OAM configuration function that communicates with the network slic- ing orchestrator. This slice-to-class of service mapping correspondence is flexible by nature and should be dynamically or semi-dynamically reconfigured without service inter- ruption. Furthermore, the transport OAM monitoring function reports the transport KPIs to the network slicing orchestrator, providing a means of feedback for integrated network optimization. Upon deletion, addition, or any modification of the network slicing services, the slice-aware transport adapts to the changes and performs seamless reconfiguration over the transport network for the purpose of overall network optimization. A slice-based transport that is fully optimized for network slicing should not be class of service-based where each slice traffic should be individually identified, labeled, and transported accord- ing to its own QoS requirements [78]. 1.1.6.3 Backhaul Transport Options The increasing demand for broadband wireless applications has a significant effect on the entire mobile infrastructure including the mobile

backhaul network. In an operator's Network slicing orchestrator NGFI OAM NGFI OAM configuration performance NGFI OAM RAN-transport interface monitoring provisioning KPI monitoring Class of service configuration KPI monitoring Slice 1 Class of service-Subclass 0 Slice 1 Slice 1 Class of service Subclass 0 Slice 1 Slice 2 Slice 2 Slice 2 Slice 2 Class of service Subclass 1 Class of service Subclass 1 Slice M-1 Slice M-1 Slice M-1 Slice M-1 Slice M Slice M Slice M Slice M Class of service Subclass N Class of service Subclass N Node A Node B Node A Node B (RRU or DU) (DU or CU) (RRU or DU) (DU or CU) Figure 1.38 Comparison of slice-agnostic and slice-aware transports 781 88 Chapter 1 network, the mobile backhaul connects small and macro-cell sites to the core network that is further connected to external data centers serving content and applications. The RAN is an increasingly critical part of the global network infrastructure and is the primary reason that network operators are extremely focused on the mobile backhaul network as a key ele- ment of their short to long-term business strategies. Therefore the capacity, latency, reliabil- ity, and availability of mobile backhaul networks must improve as the wireless access data rates increase to enable video-centric and other broadband user applications. In addition, mobile backhaul networks support specific technologies that together ensure an acceptable quality of experience, which include network timing and synchronization as well as operations, administration, maintenance, and provisioning. Recent technology advances and 4G/5G network deployments have created a new land- scape in the access networks through integration of wireless and fiber technologies. In the past, the use and deployment of fiber links in the backhaul was slow compared to that of wireless backhaul schemes due to somewhat low data rate requirements of the backhaul supporting typically large cell sites. With the decreasing cost of fiber deploy- ments and penetration of the fiber in the access netwo

rks as well as the demand of the latest wireless standards for smaller and higher bandwidth cells, the use of fiber connec- tivity has become more prevalent. Depending on the demarcation point between key net- work elements, one can decide whether fiber should be used only as a high data rate backhaul path or a transition to radio-over-fiber techniques can be afforded for the fronthaul links, as well. Backhaul traffic comprises a number of components in addition to the user-plane traffic. As the transport networks evolve, network operators are evalu- ating and, in some cases, have already started to deploy mobile fronthaul networks to support centralized RANs and ultimately C-RANs. In these networks, the BBU is moved from the cell site to a central location. This creates a new fronthaul network between the BBU and the cell site that has typically utilized CPRI protocol until just recently, which effectively carry digitized RF signals. These networks require fiber-based back- haul of Ethernet traffic from either the cell site or the BBU location, regardless of the last mile technology. The last mile could be the same Ethernet-over-fiber connection or Ethernet over some other media such as copper or microwave. As the cell topology changes from a traditional macro-cell to smaller cells, the requirements for the endpoint of the backhaul service typically become more stringent to meet certain deployment spe- cifications such as temperature range, space, and power consumption. The performance of the underlying transport network must substantially improve in order to support the considerably tighter transport performance requirements in terms of frequency synchroni- zation, phase synchronization, and latency. Since the performance of the backhaul net- works is becoming more critical to end-to-end 5G services/applications, performance monitoring capabilities are imperative to ensure that the QoS requirements and service- level agreements are met (see Fig. 1.39). 5G Network Architecture External networks Aggregation netwo

rk Core network Last mile (serves cells) IPsec overhead Transport protocol overhead OAM, sync, etc. X2 user-plane and controlplane Control plane S1 User-plane traffic Transport network Figure 1.39 Illustration of backhaul components in a typical cellular network [66]. Mobile backhaul is an example application area where packet optical technology48 enables the paradigm shift from traditional TDM-based backhaul to high performance, low cost, and scalable solutions that are currently required. Modern packet optical solutions enable the optimization of IP traffic between cell site gateways and core routers, avoiding unnecessary router hops. New generation of mobile backhaul supports Ethernet transport for all cell types and all locations regardless of last mile technology, whether CPRI-based fronthaul, DASs, fiber-connected small-cells or macro-cells, or fiber aggregation points supporting microwave backhaul in non-fiber environments are utilized. The use of frequency-division duplex scheme over the air-interface requires only simple frequency synchronization using SyncE or IEEE 1588v2 packet-based synchronization schemes. However, time-division duplex scheme requires phase synchronization, in which the network needs to track the phase of the synchronization signal and to receive accurate time-of-day timestamps. The more complex radio-access network features such as CoMP transmission and reception and eICIC further require phase synchronization. Frequency synchronization is provided through a number of methods, where the most com- mon solution is SyncE. It can also be provided in some regions using the global navigation Packet optical transport covers technologies and architectures that enable the transport of Internet protocol packets on both fixed and mobile optical networks. Converged products include the functional switching capability of wavelength division multiplexing schemes, Ethernet switching via various protocols, as well as time division multiplexing and optical transport network switching. The technologi

es include a combination of optical networking products that operate separately or within a single converged platform called packet optical transport system such as reconfigurable optical add-drop multiplexers, time division multiplexing, and carrier Ethernet switching products. These platforms reside in the metro edge, metro core, and the long- haul networks of major service providers. 90 Chapter 1 satellite system (GNSS) methods such as global positioning system (GPS). Phase synchroni- zation is provided using IEEE 1588v2 precision timing protocol. There are a number of ways in which good network performance can improve PTP quality within the base station. The use of network elements with low jitter has a positive impact on the quality of the received PTP by reducing errors. Furthermore, networks that support both SyncE and IEEE 1588v2 are able to operate in hybrid mode, with SyncE assisting the IEEE 1588v2 protocol for an improved overall performance. Inadequate synchronization has a negative impact on network performance, resulting in less efficient radio interface, poor performance for data traffic, and dropped calls. Mobile backhaul is provisioned throughout the cellular network to transport voice and data traffic between the access and core networks. Wireless equipment in the radio access net- work includes macro-cell base stations, small-cell access points, and DASs. Wired and wire- less transport mechanisms are the two types of mobile backhaul deployed across the RANs. With the emergence of heterogeneous networks, mobile backhaul has become a critical component in the 4G and 5G networks. The cellular networks are evolving toward a heterogeneous architecture where different classes of small-cell base stations and DAS installations are coordinated, cooperating with macro-site base stations. A HetNet topology improves cellular network capacity and cover- age to support the exponential growth in mobile traffic. Subsequently, HetNets can deliver ubiquitous connectivity to the mobile users with exceptional qua

lity of experience. Backhaul is the confluence of mobile broadband users, small cells, DAS, macro-cells, and the core network. The emerging HetNet topologies have created a need for diverse wired and wireless mobile backhaul solutions. Small cells are being deployed in indoor and out- door environments, on utility poles, and other urban structures. The sites can be located in private or public locations. Depending on the use case, each small-cell site has specific requirements for power sourcing, power budget, and backhaul transport. Meanwhile, the conventional macro-cell base stations will continue to increase network capacity, further driving demand for high-throughput backhaul. Wireless backhaul, emerged as a cost-effective connectivity option, has many advantages rela- tive to wired technologies, but wireless solutions also present unique design challenges among which are the need for spectrally efficient radio links, low operating power, small form factor, and environmentally resilient high-reliability equipment. RF analog integration also plays an important role. As an example, a typical microwave radio transmitter relies on RF analog inte- gration and RF building blocks to reduce the size, lower the power, and improve the dynamic performance. A wireless backhaul equipment requires four key components: antennas, radio transceiver, modem, and interface. The antenna transmits and receives electromagnetic waves. The radio transceiver handles RF carrier frequency transformation to and from the baseband. 5G Network Architecture 91 The modem performs channel coding/decoding and modulation/demodulation of the baseband signals, and the interface transports information between the radio and TDM/IP transport. As shown in Fig. 1.40, wired and wireless backhaul solutions employ broadband technologies that vary in terms of physical media and access method. Wired backhaul physical media include copper wire, hybrid fiber-coaxial (HFC)49 cable, and single-mode and multimode optical fiber. Transport access technologies are

fractional-T/E carrier (T1/E1), digital sub- scriber line (DSL), pseudo-wire, Ethernet, WDM, and gigabit passive optical network. The choice of wired backhaul is driven by the availability of physical media, cost, and capacity requirements. Wireless backhaul is needed when base stations do not have access to copper, HFC, or fiber transport. Wireless backhaul is also attractive when time to deployment is criti- cal or when leasing costs are prohibitive. It is estimated that nearly 70% of worldwide LTE base station installations use wireless backhaul [80]. The methods of delivering wireless back- haul transport are LoS microwave, LoS millimeter wave, non-LoS (NLoS) sub-6 GHz micro- wave, broadband satellite links, and in-band or out-of-band relay nodes. The HetNet architecture comprises four general classes of base station: (1) macro-cell, (2) metro-cell, (3) pico-cell, and (4) femto-cell. Table 1.2 compares the types of base station, deployment scenarios, and the set of possible wireless backhaul solutions. Wireless back- haul radios operate over a wide spectrum of licensed and unlicensed RF bands extending to 80 GHz (see Fig. 1.41). The RF spectrum for wireless backhaul ranges from sub-6 GHz NLoS to C/Ka/Ku-band microwave LoS, and Q/V/E-band mmWave LoS. Each RF band has spectrum restrictions, channel bandwidth limitations, and specific propagation character- istics. Channel bandwidth can vary from 5 to 160 MHz in NLoS systems; from 3.5 to 56 MHz in microwave LoS systems; or from 28 to 112 MHz and 250 MHz to 5 GHz in mmWave systems. All these specifications impact the type of modulation and carrier-to- noise ratio, and thereby mandating certain capacity trade-offs and maximum link distance. Table 1.2 shows that backhaul throughput for each base station class must match the respec- tive cell-site capacity. An optimal wireless backhaul solution further adapts the performance requirement with a particular deployment scenario. The method of wireless backhaul deter- mines frequency band operation, radio design specific

ations, and the radio architecture. A hybrid fiber coaxial network is a telecommunication technology in which optical fiber cable and coaxial cable are used in different portions of a network to carry broadband content. An advantage of hybrid fiber coaxial is that some of the characteristics of fiber-optic cable (high bandwidth and low noise and interference susceptibility) can be brought close to the user without having to replace the existing coaxial cable that is installed at home or business. Gigabit passive optical network is a point-to-multipoint access network. Its main characteristic is the use of passive splitters in the fiber distribution network, enabling one single feeding fiber from the provider to serve multiple homes and small businesses. Gigabit passive optical network has a downstream capacity of 2.488 Gbps and an upstream capacity of 1.244 Gbps that is shared among users. Encryption is used to keep each user's data secured and private from other users [ITU-T G.984]. Wired backhaul link Wireless backhaul link mmwave (())) Digital subscriber line Small-cell Copper access multiplexer Macro-cell FTTN (fiber to the node) EPC/5GC (core network) Passive optical network Fiber (optical line (())) Macro-cell termination) sub-6 GHz Metro Cable Edge router Edge router FTTN (fiber to Macro-cell the node) (())) FTTN (fiber to the Fiber node) Macro-cell Small-cell Passive optical (())) network Relay (optical line FTTN (fiber to the termination) LoS microwave Macro-cell node) Digital subscriber line Satellite access multiplexer Macro-cell Figure 1.40 Illustration of mobile backhaul network physical media and access layer with macro-cell and small-cell base station [79]. Table 1.2: Example of LTE-advanced base stations and wireless backhaul solutions [79]. Number of Power Amplifier Signal Number of Sectors/ Total Base Station Wireless Station Deployment Active Radius Output Power Bandwidth Number of Theoretical Capacity Backhaul Scenario Users (dBm) (MHz) Antennas (Gbps) Scheme Macro- Outdoor 200-1000 Microware,

4 X 4 mmWave, Metro- Outdoor Microware, 2 X 2 mmWave, Pico-cell Indoor/ 32-100 < 0.3 mmWave, outdoor 2 X 2 sub-6 GHz, relay Femto- Indoor < 0.1 Sub-6 GHz, 2 X 2 relay 94 Chapter 1 Narrowband channels Wider channels Ultra-wideband channels 3, 5, 40 MHz 3.5, 56, 112 MHz High-order modulation 64QAM, 250, 500, 5000 MHz Higher order modulation 256QAM Low-order modulation QPSK, 16QAM, 64QAM 64QAM, 256QAM, 1024QAM S, C band C, Ku, K, Ka band Q band V band E band Frequency (GHz) Sub-6 GHz Microwave mmWave Figure 1.41 Frequency spectrum, channel bandwidth, and modulation considerations for RF bands [79]. Table 1.3 summarizes different characteristics for each wireless backhaul method and fur- ther outlines the key factors to be considered in backhaul selection. Low-order modulation schemes such as QPSK can be used with wideband channels or for operation in poor atmospheric conditions with low SNR channel conditions. On the other hand, high-order modulation up to 2048 QAM can be used with narrowband channels or for operation in good channel quality and in clear atmospheric conditions. Depending on the link capacity requirements, deployment scenario, SNR, and atmospheric conditions, the data throughput can range from 100 Mbps to 10 Gbps. For macro-cell base station applications, the migration from indoor units to full outdoor units lowers the power, improves signal quality, and lowers OPEX. Macro-cells take advantage of full outdoor unit partition because RF loss in the waveguide or coaxial cable is minimized, or eliminated, which lowers RF power output and improves receiver input sensitivity. In small-cell base stations, the adop- tion of an integrated configuration means that systems can achieve a small footprint with lower equipment cost. Small cells benefit from an embedded partition because a single unit houses the wide and local area network connectivity, and wireless backhaul radio functions, resulting in reduced system size and simplified installation. Furthermore, since many small- cell deployments rely on E-band/

V-band backhaul operating at 60-80 GHz, the RF losses associated with radio and antenna connections are significantly reduced. Conventional point-to-point, LoS microwave systems operate in the licensed spectrum from C-band up to Ka-band. Common operating band frequencies are 6, 11, 18, 23, 26, and 38 GHz. These systems require unobstructed propagation. Fig. 1.42 illustrates a point-to-point micro- wave backhaul link connecting a macro-cell base station node to an aggregation node. For small-cell base station applications some microwave equipment vendors have demonstrated NLoS operation using conventional LoS microwave bands, leveraging high antenna gains with Table 1.3: Wireless backhaul options and associated parameters [79]. Single- Channel Channel Backhaul NLoS/ Frequency Licensed/ Bandwidth Range Latency Capacity Unlicensed (MHz) Modulation (Gbps) Application Notes Microwave C, Ku, Ka Licensed 3.5/7/14/ Macrocell High reliability and 28/56 small-cell high capacity links 10/20/30/ 16-2048QAM aggregation 40/50 mmWave Licensed 28/56/112 <0.05 0.3-10 Macrocell Narrow beams and 16-2048QAM small-cell oxygen absorption to Unlicensed > 250 aggregation improve frequency Lightly 250-5000 16QAM, 64QAM reuse in V band licensed C, Ku, Ka Licensed 26/33/50/ QPSK, 8PSK, Small cell Remote and rural 72/500 32ASK (medium areas with no earth infrastructure orbit) (GEO) NLoS/ Licensed/ 5/10/20/ 64QAM,256QAM Metro-cell, Fast deployment and 6 GHz unlicensed 40/80/160 (NLoS) picocell, unpredictable capacity femtocell (LoS) In-band LTE bands Licensed 5/10/15/ 64QAM > >10 Picocell, Occupies cellular relay 20/40 femtocell spectrum and adds latency 96 Chapter 1 Macro-cell site tail Macro-cell site aggregation node C, Ku, Ka band backhaul link fiber Macro-cell Microwave baseband baseband Core network Fiber backhaul Figure 1.42 A point-to-point LoS microwave link with a relay node and aggregation node [79]. operating guidelines for the electromagnetic wave propagation effects such as diffraction, reflection, and penetration to overcome t

he additional path loss as a result of NLoS operation. In clear weather conditions when radio-link SNR is high, the spectral efficiency and through- put are increased by employing higher order QAM constellations such as 256QAM, 1024QAM, or even 4096QAM. In poor weather conditions as SNR degrades, the modulation order can be lowered to 16QAM or QPSK to ensure operational link for high-priority data but at reduced throughput. However, shifting to higher order QAM constellations yields a point of diminishing returns in terms of throughput gained versus added cost, RF transmitter power utilized, required RF signal-chain linearity, and higher dynamic range. With each increase in modulation order, 3-4 dB increase in SNR or transmitter power is needed; however, with each increase in modulation order the throughput only improves by about 10%. Co-channel dual polarization (CCDP) utilizes cross-polarization interference cancelation (XPIC)51 to double the link capacity over the same channel. CCDP-XPIC allows simulta- neous transmission of two separate data streams on the same frequency. Data is transmitted Cross-polarization interference cancelation is an algorithm to suppress mutual interference between two received streams in a polarization-division multiplexing communication system. The cross-polarization inter- ference canceler is a signal processing technique implemented on the demodulated received signals at the baseband level. This technique is typically necessary in polarization-division multiplexing systems where the information to be transmitted is encoded and modulated at the system's symbol rate and upconverted to an RF carrier frequency, generating two (orthogonally polarized) radio streams radiated by a single dual-polarized antenna. A corresponding dual-polarized antenna is located at the remote receiver site and connected to two RF receivers, which down-convert and later combine the radio streams into the baseband signal. 5G Network Architecture Small-cell site ((< ) fiber Satellite VSAT network gateway rout

er Core network T/E copper fiber cable Figure 1.43 Illustration of an example satellite backhaul [79]. on orthogonal antenna polarizations (vertical and horizontal) and cross-polarization interfer- ence is canceled using digital signal processing. Spatial multiplexing significantly improves spectral efficiency and uses multi-antenna techniques to send multiple data streams over the same RF channel. A 2 X 2 MIMO link can ideally double the capacity. Spatial multiplexing has been used in many wireless access technologies including LTE/NR and IEEE 802.11n/ ac/ax, which relies on multipath interference and exploits spatial propagation paths caused by reflections. However, an LoS microwave link does not exhibit multipath phenomenon, thus a multipath condition is simulated by deliberate separation of the antennas, thereby cre- ating a pseudo multipath condition. While it is evident that high-density RF analog integra- tion is important to reduce the size and lower the component count, there are still many radio functions that rely on [discrete] RF building blocks. The IF circuitry and frequency up/down conversion require several key analog functions including phase-locked loop (PLL)5 frequency synthesizer, and the variable gain amplifier (VGA). Commercial satellite systems use very small aperture terminals (VSAT) for cost- effective delivery of telephony, broadband access, and video content. VSAT systems are deployed in enterprise-grade private networks, consumer broadband services, and cellular base station backhaul. In cellular base station backhaul applications, VSAT sys- tems are ideal for remotely located small-cell sites. Fig. 1.43 shows a typical VSAT system used in a base station backhaul application. Router and gateway VSAT A phase-locked loop is a control system that generates an output signal whose phase is related to the phase of an input reference signal. There are several different phase-locked loop types, but the simplest form is an electronic circuit consisting of a variable frequency oscillator and a p

hase detector in a feedback loop. The oscillator generates a periodic signal, and the phase detector compares the phase of that signal with the phase of the input reference signal, adjusting the oscillator frequency to maintain phase matching. 98 Chapter 1 terminals are ground-based units with a two-way communication link to C/Ka/Ku-band satellites. The satellites can be geostationary or geosynchronous equatorial orbit, medium earth orbit, or low earth orbit. Orbiting satellites are powered by a finite energy source; therefore they are energy-constrained and a satellite downlink channel has limited transmitter power. The link is also susceptible to atmospheric loss because geosynchronous satellites orbit at 35,786 km from ground-level terminals. As a result, the radio link operates with a very low SNR. To achieve the desired data throughput with acceptable BER at low SNR, VSAT systems use wide-channel bandwidths with relatively low-order modulation schemes such as QPSK or 8PSK and low coding rates. Microwave and broadband satellite backhaul are two common wireless technologies deployed across the RAN. As the base station capacity and throughput increase to sup- port growing mobile data demand, backhaul capacity must increase. Similarly, as base station size and power decrease, the backhaul solutions must become smaller and more efficient. As such, wireless backhaul systems will continue relying on RF analog inte- gration and RF building block solutions to achieve high spectral efficiency, smaller form factor, and lower operating power. 1.1.7 Mobile Edge Computing MEC or multi-access edge computing is an emerging 5G technology which enables provi- sioning of cloud-based network resources and services (e.g., processing, storage, and net- working) at the edge of the network and in the proximity of the users. The edge may refer to the base stations themselves and/or data centers close to the radio network possi- bly located at the aggregation points. The end-to-end latency perceived by the mobile user can be signific

antly reduced using the MEC platform, which is a key enabling factor for many 5G services such as the tactile Internet. MEC supports different deployment scenar- ios, and the MEC servers can be located at different locations within the radio access net- work depending on technical and business requirements. Applications and analytics at the MEC server will not be impacted by congestion in other parts of the network. By perform- ing analytics or caching content at the MEC server, the volume of data transmitted to the core network for processing is reduced, and the impact on the data traffic through the net- work is minimized, resulting in more efficient use of existing network bandwidth. This establishes an ultra-low-latency environment capable of providing mission-critical and real-time services. The user location information can be used by applications and services hosted by the MEC server to offer location/context-related services to the subscribers. Since these applications and services are found at the edge of the network instead of within a centralized cloud, responsiveness can be improved, resulting in an enriched qual- ity of experience for the user. 5G Network Architecture Routing control network capability exposure platform management Mobile core network Mobile edge computing Radio access network Mobile terminal Figure 1.44 Conceptual MEC architecture [48]. Established in December 2014, the ETSI Industry Specification Group (ISG) 53 on MEC has produced normative specifications that enable the hosting of third-party applications in a multi-vendor environment. The initial scope of the ISG MEC was to focus on use cases, and to specify the requirements and the reference architecture, including the components and functional elements that were the key enablers for MEC solutions. The work has contin- ued on platform services, APIs, and interfaces. The MEC platform API is application- agnostic and allows smooth porting of value-creating applications on every mobile-edge server with guaranteed service-level agree

ment. The main functions in the MEC platform include routing, network capability exposure, and management (see Fig. 1.44). The routing entity is responsible for packet forwarding among the MEC platform, radio access network, and the mobile core network, as well as within the MEC platform. The network capability exposure entity enables the authorized exposure of the radio network information service and the RRM. The management entity supports the authentication, authorization, and accounting and management of the third-party applica- tions in the MEC platform. This section presents the architectural description of the MEC platform as well as main applications and key functionalities enabling this technology. 1.1.7.1 Service and Deployment Scenarios The primary objective of multi-access edge computing is to reduce network congestion and to improve application performance by processing the corresponding tasks in the proximity of the users. Furthermore, it intends to improve the type and delivery of the content and applications to those users. The use cases already being realized include augmented reality and virtual reality, which both benefit from fast response times and low-latency ETSI multi-access edge computing (http://www.etsi.org/index.php/technologies-clusters/technologies/multi- access-edge-computing). 100 Chapter 1 communications; connected cars, which also thrive in high-bandwidth, low-latency, highly available settings; and industrial/residential IoT applications that rely on high performance and smart utilization of network resources. Large public and enterprise locations are also beneficiaries of MEC. In large-scale situations where localized venue services are impor- tant, content is delivered to onsite consumers from a MEC server located at the venue. The content is locally stored, processed, and delivered, not requiring information transport through a backhaul or centralized core network. Large enterprises are also increasingly motivated to process user traffic locally rather than backhaul traffic t

o a central network, using small-cell networks instead [48]. As shown in Fig. 1.45, edge computing use cases may be classified into two major categories, that is, third-party applications and operator applications. The MEC ecosystem is likely to bring significant benefit to the mobile operators and other industries as well as to the consumers that will be able to experience services which need to rely on very accurate localization or high performance in terms of latency and throughput. In other words, MEC will be able to support new IoT services that would not be technically or economically feasible without 5G networks. Several use cases are addressed by MEC and also currently considered as part of next-generation mobile networks. Security and safety have been among the most important verticals for IoT. The advances in technology with an ever-increasing amount of information collected from sensors and high-resolution video cameras create the need for a scalable, flexible, and cost-effective solution to analyze the content in real time. MEC can host the analytics applications close to the source and enable increased reliability, better performance, and significant savings by locally proces- sing large volume of data. Edge computing use cases Operator Third-party applications applications Analytics Compliance Real time Immersive Security Cost reduction Private Self-contained Figure 1.45 MEC use cases [45,48]. 5G Network Architecture 101 The automotive sector is another area where the new technologies are revolutionizing the industry. Self-driving cars have been already demonstrated by both traditional automotive and new Internet players, and it is anticipated to make the first autonomous cars commer- cially available in 2020 While the work on future 5G systems is currently being con- ducted by various organizations around the globe, the digitalization in the automotive industry is clearly reflected in the use cases and the requirements coming from this sector. The IoT is a key driver for the next-generation technol

ogy and most of the use cases appear to focus on the connected cars. With the next-generation system, the latency requirements are set to less than 1 ms to empower a wide range of use cases. MEC is the ideal solution and has been identified as a key component to support these ultra-low latency scenarios as it enables hosting applications close to the users at the edge cloud and therefore provides the shortest path between the applications and the servers. Computationally intensive applications running on mobile terminals may be offloaded to the cloud for various reasons, such as availability of more computing power or of specific hardware capabilities, reliability, joint use of the resources in collaborative applications, or saving network capacity. The computation offload is particularly suitable for IoT applica- tions and scenarios where terminals have limited computing capabilities, to guarantee longer battery life. Such offload may happen statically (server components are deployed by the ser- vice provider proactively in advance) or dynamically (server components are deployed on demand by request from the UE). In this case, applications benefit from the low delay pro- vided by the MEC. 1.1.7.2 Architectural Aspects MEC provides a highly distributed computing environment that can be used to deploy appli- cations and services as well as to store and process content in close proximity of the mobile users. Applications can also be exposed to real-time radio and network information and can offer a personalized and contextualized experience to the mobile subscriber. This translates into a mobile broadband experience that is not only more responsive but also opens up new business opportunities and creates an ecosystem where new services can be developed in and around the base station. Fig. 1.46 shows the high-level functional entities in the MEC framework, which are further grouped into the system-level, host-level, and network-level entities. The host-level group consists of the MEC host and the corresponding MEC h

ost-level management entity. The MEC host is further split to include the MEC platform, applications, and the virtualization infrastructure. The network-level group consists of the corresponding external entities that are 3GPP radio access networks, the local networks, and the external networks. This layer represents the connectivity to local area networks, cellular networks, and external networks such as the Internet. Above everything is the MEC system-level management that by 102 Chapter 1 Radio Traffic network offload Communication Service information function service service registry (TOF) (RNIS) MEC application platform services MEC virtualization manager, laaS MEC application platform MEC virtualization layer MEC hardware resources MEC hosting infrastructure 3GPP radio network element Figure 1.46 Overview of ETSI MEC 46]. definition has the overall visibility to the entire MEC system. The MEC system consists of the MEC hosts and the MEC management necessary to run MEC applications within an operator network or a subset of an operator network [53]. An in-depth understanding of the MEC systems can be attained from the reference architec- ture depicted in Fig. 1.47, which defines the functional entities and their relationships. The reference architecture follows the earlier described functional grouping of the general framework and includes system-level and host-level functions; however, the network-level functional group is not visible because there are no MEC-specific reference points needed to access those entities. The MEC host is an entity consisting of the MEC platform and the virtualization infrastructure that provides computing, storage, and network resources for the MEC applications. In addition, the MEC host can provide storage and ToD information for the applications. The virtualization infrastructure includes a data plane that executes the for- warding rules received by the MEC platform and routes the traffic between the applications, services, and networks. The MEC server provides computing resour

3GPP network Local network External network Figure 1.47 MEC general framework [46]. The MEC platform represents a collection of baseline functionalities that are required to run applications on a particular MEC host and to enable MEC applications to discover, adver- tise, offer, and exploit the MEC services. MEC services can be provided by the platform and by the applications, where both the platform and applications may utilize MEC services. The baseline functionalities of the MEC platform are needed to navigate the traffic between the applications, services, and networks. The MEC platform receives the traffic forwarding rules from the MEC platform manager, MEC applications, and MEC services, and based on those criteria, as well as policies, it provides the instructions to the forwarding plane. The reference architecture shows the functional elements that comprise the mobile edge (ME) system and the reference points between them. Fig. 1.48 depicts the ME system refer- ence architecture. There are three groups of reference points defined between the system entities as follows [46]: 1. Reference points related to ME platform functionality (Mp) 2. Management reference points (Mm) 3. Reference points interfacing with external entities (Mx) The MEC applications run as VMs on top of virtualization infrastructure provided by the MEC host. The applications interact with the MEC platform over an Mp1 reference point to utilize the services offered by the platform. The MEC platform manager is a host-level entity that is further split into MEC platform element management, MEC application life cycle management, and MEC application rules and requirements management functions. The application life cycle management consists of application instantiation and termination procedures as well as providing indication to the MEC orchestrator on application-related 104 Chapter 1 portal Operations support system User application life cycle UE app management proxy Mobile edge orchestrator Other platform service ME platform element managem

ent ME app rules and requirements Service registry Service management Traffic ME app life cycle management ME app ME app ME app rules control handling ME platform manager ME platform Other Data plane Virtualization infrastructure Virtualization infrastructure ME host Figure 1.48 MEC reference architecture [46]. events. The MEC orchestrator is the central function in the MEC system that has visibility over the resources and capabilities of the entire MEC network. The MEC orchestrator main- tains information on the entire MEC system, the services and resources available in each host, the applications that are instantiated, and the topology of the network. The orchestrator is also responsible for managing the MEC applications and the related procedures by inte- grating the applications, checking the integrity and authenticity of the application, validating the policies for the applications, and maintaining a catalog of the applications that are available. The MEC applications may indicate their requirements for the resources, services, location, and performance, such as maximum allowed latency, and it is the MEC orchestrator's responsibility to ensure that their requirements are satisfied. The orchestrator uses the requirements received from the applications in the selection process for the target MEC host. The reference point toward the VIM is used to manage the virtualized resources of the MEC host and to manage the application images that are provided for instantiation. It is fur- ther used for maintaining status information on available resources. The operations support system of an operator is a function that is widely used to manage various services and sub- systems in the operators' network. The reference point between the MEC orchestrator and VIM is used for management of the application images and the virtualized resources as well as for monitoring the availability of the resources. The customer-facing service (CFS) acts as an entry point for a third-party application. This portal can be used for operations

to manage the provisioning, selection, and ordering of the 5G Network Architecture MEC applications. The user application life cycle management proxy is a function that the MEC-related clients and applications use to request services related to onboarding, instantia- tion, and termination of the applications. This proxy can be used to request transfer of the application from the MEC system to the external cloud or to the MEC system from the external cloud. More specifically, the ME system-level management includes the ME orchestrator as its core component, which has an overall view of the complete ME system. The ME host-level management comprises the ME platform manager and the VIM, which handles the management of the ME-specific functionality of a particular ME host and the applications running on it. The ME host is an entity that contains the ME platform and a virtualization infrastructure which provides compute, storage, and network resources for the ME applications. The virtualization infrastructure includes a data plane that executes the traffic rules received by the ME platform, and routes the traffic among applications, ser- vices, DNS server/proxy, 54 3GPP network, local networks, and external networks. The ME platform is responsible for offering an environment where the ME applications can discover, advertise, consume, and offer ME services, including ME services available via other platforms; receiving traffic rules from the ME platform manager, applications, or ser- vices; and instructing the data plane. This includes the translation of tokens representing UEs in the traffic rules into specific IP addresses, receiving DNS records from the ME plat- form manager, configuring a DNS proxy/server, hosting ME services, and providing access to persistent storage and ToD information. The ME applications run as VMs on top of the virtualization infrastructure provided by the ME host, and can interact with the ME platform to utilize and provide ME services. Under certain conditions, the ME applications can also

interact with the ME platform to perform certain support procedures related to the life cycle of the application, such as indicating availability, preparing relocation of user state, etc. The ME applications can have a certain number of rules and requirements associated with them such as required resources, maximum latency, required or useful services, etc. These requirements are validated by the ME system-level management and can be assigned to default values if not provided. The ME orchestrator is the core functionality in ME system-level management. The ME orchestrator is responsible for maintaining an overall view of the ME system based on deployed ME hosts, available resources, available ME services, and topology; on-boarding Domain name system is a hierarchical decentralized naming system for computers, services, or other resources connected to the Internet or a private network. It associates various information with domain names assigned to each of the participating entities. It translates readily memorized domain names to the numerical IP addresses needed for locating and identifying computer services and devices with the underly- ing network protocols. A domain name system proxy improves domain lookup performance by caching pre- vious lookups. A typical domain name system proxy processes domain name system queries by issuing a new domain name system resolution query to each name server that it has detected until the hostname is resolved. 106 Chapter 1 of application packages, including checking the integrity and authenticity of the packages, validating application rules and requirements, and if necessary adjusting them to comply with operator policies, keeping a record of on-boarded packages, and preparing the VIM(s) to handle the applications; selecting appropriate ME host(s) for application instantiation based on constraints such as latency, available resources, and available services; triggering application instantiation and termination; and triggering application relocation as needed when supported.

The operations support system shown in Fig. 1.48 refers to the OSS of an operator. It receives requests via the CFS portal and from UE applications for instantiation or termina- tion of applications and decides whether to grant these requests. Granted requests are for- warded to the ME orchestrator for further processing. The OSS also receives requests from UE applications for relocating applications between external clouds and the ME system. A user application is an ME application that is instantiated in the ME system in response to a request of a user via an application running in the UE. The user application life cycle man- agement proxy allows UE applications to request on-boarding, instantiation, termination of user applications, and relocation of user applications in and out of the ME system. It also allows informing the UE applications about the state of the user applications. The user application life cycle management proxy authorizes requests from UE applications in the UE and interacts with the OSS and the ME orchestrator for further processing of these requests. The user application life cycle management proxy is only accessible from within the mobile network. It is only available when supported by the ME system. The VIM is responsible for managing the virtualized resources for the ME applications. The management tasks consist of allocating and releasing virtualized computing, storage, and network resources provided by the virtualization infrastructure. The VIM also prepares the virtualization infrastructure to run software images, which can also be stored by the VIM for a faster application instantiation. Since it is possible for virtualized resources to run out of capacity or to fail in operation, it is important to closely monitor them. The VIM pro- vides support for fault and performance monitoring by collecting and reporting information on virtualized resources and providing the information to server and system-level manage- ment entities. The VIM has a reference point toward the virtualization in

frastructure to man- age the virtualized resources. The ME reference architecture shown in Fig. 1.48 incorporates the following reference points [46]: Mp1 is a reference point between the ME platform and the ME applications that pro- vides service registration, service discovery, and communication support for services. It also enables other functionalities such as application availability, session state relocation support procedures, traffic rules and DNS rules activation, access to persistent storage and ToD information, etc. 5G Network Architecture 107 Mp2 is a reference point between the ME platform and the data plane of the virtualiza- tion infrastructure and is used to instruct the data plane on how to route traffic among applications, networks, services, etc. Mp3 is a reference point between the ME platforms and is used for controlling commu- nication between ME platforms. Reference points related to the ME management Mm1 is a reference point between the ME orchestrator and the OSS that is used for trig- gering the instantiation and the termination of ME applications in the ME system. Mm2 is a reference point between the OSS and the ME platform manager that is used for the ME platform configuration, fault detection, and performance management. Mm3 is a reference point between the ME orchestrator and the ME platform manager and is used for the management of the application life cycle, application rules and requirements, and keeping track of available ME services. Mm4 is a reference point between the ME orchestrator and the VIM which is used to manage virtualized resources of the ME host including maintaining track of available resource capacity and managing application images. Mm5 is a reference point between the ME platform manager and the ME platform and is used to perform platform configuration, configuration of the application rules and requirements, application life cycle support procedures, management of application relo- cation, etc. Mm6 is a reference point between the ME platform manager and the VIM

which is used to manage virtualized resources and to realize the application life cycle management. Mm7 is a reference point between the VIM and the virtualization infrastructure that is used to manage the virtualization infrastructure. Mm8 is a reference point between the user application life cycle management proxy and the OSS and is used to handle UE requests for running applications in the ME system. Mm9 is a reference point between the user application life cycle management proxy and the ME orchestrator of the ME system and is used to manage ME applications requested by UE application. Reference points related to external entities Mx1 is a reference point between the OSS and the CFS portal and is used by a third party to request the ME system to run applications in the ME system. Mx2 is a reference point between the user application life cycle management proxy and the UE application, which is used by a UE application to request the ME system to run an application in the ME system, or to move an application in or out of the ME system. This reference point is only accessible within the mobile network. It is only available when supported by the ME system. 108 Chapter 1 The ME computing and NFV are complementary concepts that can exist independently. The ME architecture has been designed in such a way that a number of different deploy- ment options of ME systems are possible. An ME system can be realized independent of an NFV environment in the same network, or can coexist with that environment. Since both MEC and NFV are based on the use of virtualization concept, the MEC applications and VNFs can be fully or partially instantiated over the same virtualization infrastructure. The MEC reference architecture reuses the concept of a VIM similar to that of the VIM of NFV framework. Multiple scenarios for deployments are possible, depending on operators' pre- ferences for their networks and their network migration plans. The relationship between MEC and NFV-MANO components is an important aspect of integrated MEC/N

FV deployments. In 5G networks, there are three types of session and service continuity (SSC) modes. Different SSC modes can guarantee different levels of service continuity. As shown in Fig. 1.49, SSC mode 1 maintains the same UPF. In SSC mode 2, the network may trigger the release of the PDU session and instruct the UE to establish a new PDU session to the same data network immediately. Upon establishment of the new PDU session, a new UPF acting as PDU session anchor can be selected. In SSC mode 3, the network allows the establishment of UE connectivity via a new UPF to the same application server before con- nectivity between the UE and the previous UPF is terminated. Different applications have different service and session continuity requirements. Therefore, in order to achieve effi- cient control of MEC APPs with different SSC mode requirements in 5G network, the coor- dination between ME APPs on the ME host and the 5G network, for example, how to indicate the SSC mode requirement of an ME APP to the 5G network, must be carefully considered. Servers Servers Servers Servers ((HH)) ((HH)) ((HII) ((HH) (((III)) SSC mode 1 SSC mode 2 SSC mode 1 Figure 1.49 SSC modes [3,4]. 5G Network Architecture 109 A large part of the functionality providing data connectivity is for supporting flexible deployment of application functions (AFs) in the network topology as needed for edge com- puting, which is supported via the three different SSC modes or via the functionality of uplink classifiers and branching points. The SSC modes include the traditional SSC 1 mode, where the IP anchor remains stable to provide continuous support of applications and to maintain the path toward the UE as its location is updated. The new modes allow relocating the IP anchor. There are two options, make-before-break (SSC mode 3) and break-before- make (SSC mode 2). The architecture enables applications to influence selection of suitable data service characteristics and SSC modes. Given that 5G network deployments are expected to serve extremely

large amount of mobile data traffic, an efficient user-plane path management is critical. The system architecture defines in addition to the SSC modes the functionality of uplink classifiers and branching points to allow breaking out and inject- ing traffic selectively to and from AFs on the user plane path before the IP anchor. Also, as permitted by policies, AFs may coordinate with the network by providing information rele- vant for optimizing the traffic route or may subscribe to 5G system events that may be rele- vant for applications [3,4]. 1.1.8 Network Sharing A network sharing architecture allows multiple participating operators to share resources of a single shared network according to agreed allocation terms. The shared network includes a RAN. The shared resources include radio resources of that network. The shared network operator allocates shared resources to the participating operators based on their plans and current needs and according to service-level agreements. A UE that has a sub- scription to a participating network operator can select the participating network operator while within the coverage area of the shared network and to receive subscribed services from the participating network operator. 3GPP laid out two approaches to sharing a RAN, which are illustrated in Fig. 1.50, where they primarily differ in the core network aspects. In multi-operator core network (MOCN) approach, each network operator has its own core network. Maintaining a strict separation between the core network and the radio network has a number of benefits related to service differentiation, interworking with legacy net- works, fall back to circuit-switched voice services, and the support of roaming. Alternatively, in the gateway core network (GWCN) approach, the network operators also share the mobility management entity of the core network, which is responsible for bearer management and connection management between the mobile terminal and the network. The GWCN approach enables additional cost savings compared to the

MOCN approach, but it is relatively less flexible, potentially reducing the level of differentiation among the participating operators. 3GPP Rel-15 supports MOCN network sharing architecture, in which only the RAN is shared in the 5G system. However, RAN and AMF support for operators that use more than one PLMN ID is required [3]. 110 Chapter 1 Operator A Operator B Operator A Operator B core network core network core network core network Shared network S1 (eNB-MME) or N2/N3 (gNB-AMF) S1 (eNB-MME) or N2/N3 (gNB-AMF) eNB/gNB eNB/gNB eNB/gNB eNB/gNB eNB/gNB eNB/gNB eNB/gNB eNB/gNB Shared radio Shared radio access network access network Multioperator core network (MOCN) RAN sharing Gateway core network (GWCN) RAN sharing Figure 1.50 3GPP-defined network sharing architecture options [34]. In each case, the network broadcasts system information and supports signaling exchanges that allow the UE to distinguish between up to six different sharing network operators, to obtain service or to perform handover, with no consideration of the underlying network sharing arrangement. As network sharing becomes a central feature of mobile network oper- ation, there is a need to address a wide variety of technical, commercial, and regulatory requirements. Among other things, there is an interest in pooling spectrum, sharing resources asymmetrically and dynamically based on financial considerations and load, and the ability for participating operators to manage and control the use of resources indepen- dently. If a shared NG-RAN is configured to indicate the available core network operators to the UEs for selection, each cell in the shared RAN would then include the PLMNs related to the available core network operators in the shared network in the broadcast sys- tem information. The broadcast system information provides a set of PLMN IDs and option- ally one or more additional set of parameters per PLMN such as cell-ID, tracking areas (TAs), etc. All UEs compliant with the 5G system attempting to connect to NG-RAN must support rece

ption of the basic and additional set of PLMN IDs. The available core network operators must be the same for all cells of a TA in a shared NG-RAN network. The UE decodes the broadcast system information and takes the information concerning the avail- able PLMN IDs into account in the network and cell (re-)selection procedures. A UE that has a subscription to a participating operator in a network sharing scenario must be able to select this participating network operator while present within the cover- age area of the shared network and to receive subscribed services from the participating network operator. Each cell in the shared NG-RAN must include the PLMN-IDs corre- sponding to the available core network operators in the shared network in the broadcast 5G Network Architecture 111 system information. When a UE performs an initial access to a shared network, one of the available PLMNs is selected to serve the UE. The UE uses all received broadcast PLMN- IDs in its PLMN (re) selection processes and informs the NG-RAN of the selected PLMN SO that the NG-RAN can properly route its traffic. After initial access to the shared net- work, the UE does not switch to another available PLMN as long as the selected PLMN is available to serve the UE at its present location. The network does not move the UE to another available PLMN by handover as long as the selected PLMN is available to serve the UE. The NG-RAN uses the selected PLMN information, which is provided by the UE at RRC establishment or provided by the AMF/source NG-RAN at N2/Xn handover, to select target cells for future handovers and allocation of radio resources. In case of hand- over to a shared network, the NG-RAN selects a target PLMN based on either PLMN in use, preset configuration, or the equivalent PLMN list in the handover restriction list pro- vided by the AMF. For Xn-based handover procedure, the source NG-RAN indicates the selected PLMN ID to the target NG-RAN by using target cell ID. For N2-based hand- over procedure, the NG-RAN indicates the selec

ted PLMN ID to the AMF as part of the tracking area identity (TAI) sent in the handover required message. The source AMF uses the TAI information supplied by the source NG-RAN to select the target AMF/ MME and to forward the selected PLMN ID to the target AMF/MME. The selected PLMN ID is signaled to the target NG-RAN/eNB SO that it can select target cells for future handovers. In a network slicing scenario, a network slice is defined within a PLMN. Network sharing is performed among different PLMNs and each PLMN sharing the NG-RAN defines and supports its PLMN-specific set of slices that are supported by the common NG-RAN [3]. 1.2 Reference Architectures 5G systems have been designed to support seamless user connectivity and to render new services/applications which would require deployment of networks that exploit innovative techniques such as NFV and SDN. A distinct feature of 5G system architecture is network slicing. The previous generation supported certain aspects of this with the functionality for dedicated core networks. In the context of 3GPP 5G system architecture, a network slice refers to the set of 3GPP defined features and functionalities that together form a complete PLMN for providing services to UEs. Network slicing allows controlled composition of a PLMN from the specified NFs with their specifics and provides services that are required for a specific usage scenario. The need for these new techniques is increasing due to the versatility of data services that are supported by 5G networks. Mobile networks were tradi- tionally designed as voice-centric and later data-centric systems; however, with 5G this design philosophy has changed as a result of proliferation of new use cases and applications. 112 Chapter 1 Having such requirements in mind, 3GPP attempted to maintain the premises of flat archi- tecture where the control-plane functions are separated from the user plane in order to make them scale independently, allowing the operators to exploit logical functional split for dimensioning in deplo

ying and adapting their networks. Another central idea in the design of 5G network was to minimize dependencies between the access network and the core net- work with a converged access-agnostic core network with a common interface which inte- grates different 3GPP and non-3GPP access types. 5G is a service-centric architecture which strives to deliver the entire network as a service. 3GPP system architecture group took the approach to rearchitect the LTE core network based on a service-oriented framework. This involves breaking everything down to more functional granularity. The MME no longer exists and its functionality has been redistribu- ted between mobility management and session management NFs. As such, registration, reachability, mobility management, and connection management are now considered new services offered by a new general NF labeled as AMF. Session establishment and session management, also formerly part of the MME, are now new services offered by a new NF called session management function (SMF). Furthermore, packet routing and forwarding functions, currently performed by the SGW and PGW in 4G, will now be realized as ser- vices rendered through a new NF called UPF. The main reason for this new architectural approach is to enable a flexible network as a service solution. By standardizing a modular- ized set of services, various deployment options such as centralized, distributed, or mixed configurations will be enabled for different users or applications. The dynamic service chaining lay the groundwork for network slicing which is an important concept in 5G to sat- isfy the diverse user and application demands, shifting the design emphasis on software rather than hardware. The physical boxes where these software services are instantiated could be in the cloud or on any targeted general-purpose hardware in the system. In the following sections, we will discuss the reference architecture, network entities, and interfaces of the 3GPP 5G access and the core networks. The user-plane and control-plan

e protocols will be further discussed. 1.2.1 Access Network 1.2.1.1 Reference Architecture: Network Entities and Interfaces The overall reference architecture of next-generation network comprises the entities associ- ated with the radio access network (NG-RAN) and the core network (5GC) and their corre- sponding interfaces that terminate the protocols. In this section, we will describe the access network reference architecture. The NG-RAN architecture consists of a set of gNBs con- nected to the 5GC through the NG interface (see Fig. 1.51). The interface between the NG-RAN and 5GC is referred to as N2 and N3 depending on the termination point in 3GPP 5G Network Architecture NG-RAN Central entity gNB-CU gNB-DU gNB-DU Distributed entity Logical NG-RAN node Figure 1.51 Overall NG-RAN architecture [15]. system architecture specifications. Furthermore, gNBs are interconnected through the Xn interface. In a C-RAN architecture, a gNB may be further disaggregated such that some lower layer protocol functions are implemented in the DUs and the remaining upper layer protocol functions are implemented in the edge cloud and as part of the CU(s). In that case, the gNB would consist of a gNB-CU and gNB-DUs as shown in Fig. 1.51. The gNB-CU and gNB-DU entities are connected via F1 logical interface. Note that one gNB-DU is con- nected to only one gNB-CU. In some deployment scenarios, one gNB-DU may be con- nected to multiple gNB-CUs. In NG-RAN reference architecture, NG, Xn and F1 are logical interfaces [16]. The traditional architecture of a base station where all protocol layers and functionalities were concentrated in a single logical RAN entity has evolved into a disaggregated model in 5G network architecture. In a disaggregated gNB architecture, the NG and Xn-C interfaces are terminated at the gNB-CU. In an LTE-NR dual connectivity (EN-DC) scenario, the S1- U and X2-C interfaces for a disaggregated gNB terminate at the gNB-CU. It must be noted that the gNB-CU and its associated gNB-DUs are seen as a gNB to other gNBs and t

he 5GC. The NG-RAN comprises a radio network layer (RNL) and a TNL. The NG-RAN architecture, that is, the NG-RAN logical nodes and the corresponding interfaces, is defined as part of the RNL, whereas the NG-RAN interfaces (NG, Xn, F1) are specified as part of TNL protocols and functionalities. The TNL provides services for user-plane and signaling transport. The protocols over Uu (i.e., the radio air-interface) and NG interfaces are divided into two classes: user-plane protocols, which are the protocols implementing the actual PDU carrying user data through the AS; and control-plane protocols, which are the proto- cols for controlling the PDU sessions and the connection between the UE and the network from different aspects including requesting the service, controlling different transmission resources, handover, etc. as well as a mechanism for transparent transfer of NAS messages 114 Chapter 1 Central control-plane entity Central user-plane entity -Xn-C Xn-U- CU-CP CU-UP Distributed entity Figure 1.52 Disaggregated gNB model [15]. via encapsulation in RRC messages. The NG interface, comprising a user-plane interface (NG-U) and a control-plane interface (NG-C), connects the 5GC and the NG-RAN. In this architecture, the NG-RAN termination is an NG-RAN node which can be either an ng-eNB or a gNB and the 5GC termination is either the control-plane AMF logical node or the user- plane UPF logical node. There may be multiple NG-C logical interfaces toward 5GC from any NG-RAN nodes which are selected by the NAS node selection function. Likewise, there may be multiple NG-U logical interfaces toward 5GC from any NG-RAN node which can be selected within the 5GC and signaled to the NG-RAN node by the AMF [15,17] (see Fig. 1.52). The NG interface supports procedures to establish, maintain, and release NG-RAN part of PDU sessions; to perform intra-RAT handover and inter-RAT handover; the separation of each UE on the protocol level for user-specific signaling management; the transfer of NAS signaling messages between UE and AMF;

and mechanisms for resource reservation for packet data streams. The functions supported over the NG interface include the follow- ing [17]: Paging function which supports transmission of paging requests to the NG-RAN nodes that are part of the paging area to which the UE is registered. UE context management function which allows the AMF to establish, modify, or release the UE context in the AMF and the NG-RAN node in order to support user-specific sig- naling on NG interface. Mobility function for UEs in ECM-CONNECTED which includes the intra-system handover function to support mobility within NG-RAN and inter-system handover func- tion to support mobility from/to EPS system including the preparation, execution, and completion of handover via the NG interface. 5G Network Architecture 115 PDU session function that is responsible for establishing, modifying, and releasing the involved PDU sessions NG-RAN resources for user data transport once the UE context is available in the NG-RAN node. NAS signaling transport function provides means to transport or re-route a NAS mes- sage for a specific UE over the NG interface. NG-interface management functions which provide mechanisms to ensure a default start of NG-interface operation and handling different versions of application part implemen- tations and protocol errors. The interconnection of NG-RAN nodes to multiple AMFs is supported in the 5G system archi- tecture. Therefore, a NAS node selection function is located in the NG-RAN node to determine the AMF association of the UE, based on the UE's temporary identifier, which is assigned to the UE by the AMF. If the UE's temporary identifier has not been assigned or is no longer valid, the NG-RAN node may consider the slicing information to determine the AMF. This functionality is located in the NG-RAN node and enables proper routing via the NG interface. As shown in Fig. 1.53, each NG-RAN node is either a gNB, that is, an NR base station, pro- viding NR user-plane and control-plane protocol terminations toward the UE

or an ng-eNB, that is, a Rel-15/16 LTE base station, terminating LTE user-plane and control-plane proto- cols to/from the UE. The gNBs and ng-eNBs are interconnected through Xn interface. The gNBs and ng-eNBs are also connected via NG interfaces to the 5GC. More specifically, NG-C interfaces NG-RAN nodes to the AMF and NG-U interfaces the NG-RAN nodes to the UPF [16]. In 5G network architecture, the gNB entity is responsible for performing functions corre- sponding to RRM including radio bearer control, radio admission control, connection mobil- ity control, dynamic allocation of resources to UEs in both uplink and downlink (scheduling). The gNB further performs IP header compression and encryption and integrity protection of data as well as selection of an AMF upon UE attachment when no routing to an AMF can be determined from the information provided by the UE. It also provides rout- ing of user-plane data toward UPF(s) along with routing of control-plane information toward AMF; the connection setup and release; scheduling and transmission of paging mes- sages originated from the AMF; scheduling and transmission of system information origi- nated from the AMF or network operation and management; measurement and reporting configuration for mobility and scheduling; session management; support of network slicing; QoS flow management and mapping to data radio bearers (DRBs); support of UEs in RRC_INACTIVE state; as well as RAN sharing and DC [16]. In NG-RAN architecture, the role of AMF is similar to that of MME in the EPC which hosts core network control-plane functions such as terminating NAS signaling and security; AS security control; internetwork signaling for mobility between 3GPP access networks; UE IP address allocation NAS security Inter cell RRM PDU session control RB control Idle state mobility handling AMF/UPF/SMF AMF/UPF/SMF Connection mobility control Radio admission control Measurement configuration Mobility anchoring and provision Dynamic resource PDU handling Internet allocation (scheduler) ((H)) (

(III) ) ) ((H)) ng-eNB ng-eNB Figure 1.53 Overall access network architecture [16]. 5G Network Architecture 117 idle-mode UE reachability which includes paging control, registration area management; support of intra-system and inter-system mobility; access authentication and authorization; mobility management control; support of network slicing; and SMF selection. The role of SMF in the new architecture is similar to that of EPC PGW entity hosting user- plane functions such as session management; allocating and managing UE IP address; selecting and controlling of UPFs; configuring traffic steering at UPF in order to route traf- fic toward proper destination; controlling part of policy enforcement; and the QoS. The new architecture further features a UPF entity which is similar to the EPC SGW user- plane entity. This entity hosts the UPFs such as anchor point for intra- or inter-RAT mobil- ity, external PDU session point of interconnect to data network, packet routing and forward- ing, packet inspection and user plane part of policy rule enforcement, traffic usage reporting, uplink classifier to support routing traffic flows to a data network, QoS handling for user plane which includes packet filtering, gating, UL/DL rate enforcement, uplink traf- fic verification, that is, service data flow (SDF)55 to QoS flow mapping, and downlink packet buffering and downlink data notification triggering [16]. 1.2.1.1.1 Xn Control-Plane/User-Plane Functions and Procedures Xn is a reference point connecting the NG-RAN access nodes, supporting the transfer of signaling information and forwarding of user traffic between those nodes. Xn is a logical point-to-point interface between two NG-RAN access nodes which establishes a logical con- nection between the two nodes even when a direct physical connection does not exist between the two nodes (which is often the case). It comprises both user-plane and control-- plane protocols. The user-plane protocol stack of Xn interface is shown in Fig. 1.54. The TNL uses GTP-U 56 over UDP/IP to

transfer the user-plane PDUs. As a result, Xn-U interface can User traffic using different services or applications has different quality of service classes. A service data flow is an IP flow or an aggregation of IP flows of user traffic classified by the type of the service in use. Different service data flows have different quality of service class attributes, thereby a service data flow serves as a unit by which quality of service rules are applied in accordance with network policy and charging rules. GPRS tunneling protocol is a protocol that is used over various interfaces within the packet core in 3GPP networks to allow the user terminals to maintain a connection to a packet data network while on the move. The protocol uses tunnels to allow two GPRS support nodes to communicate over a GPRS tunneling protocol-based interface and separates traffic into different communication flows. GPRS tunneling protocol creates, modifies, and deletes tunnels for transporting IP payloads between the user equipment, the GPRS support node in the core network and the Internet. GPRS tunneling protocol comprises three types of traffic, namely control-plane (GTP-C), user-plane (GTP-U), and charging. GTP-C protocol supports exchange of control information for creation, modification, and termination of GPRS tunneling protocol tunnels. It creates data forwarding tunnels in case of handover. GTP-U protocol is used to forward user IP packets over certain network interfaces. When a GTP tunnel is established for data forwarding during LTE handover, an End Marker packet is transferred as the last packet over the GPRS tunneling protocol tunnel. 118 Chapter 1 User-plane PDUs Xn-AP GTP-U Data link layer Data link layer Physical layer Physical layer User-plane Control-plane Figure 1.54 Xn-U and Xn-C protocol stacks [16]. only provide non-guaranteed delivery of user-plane PDUs, but it supports data forwarding and flow control. The Xn interface specifications facilitate interconnection of NG-RAN nodes manufactured by different vendors as well

as separation of Xn interface radio net- work and transport network functionalities in order to allow future extensions. The Xn inter- face supports intra-NG-RAN mobility and DC procedures. The control-plane protocol stack of Xn interface is also illustrated in Fig. 1.54. The TNL uses stream control transmission protocol (SCTP)57 57 over IP. The application layer signaling protocol is referred to as Xn application protocol (Xn-AP). The SCTP layer provides guaranteed delivery of application layer messages. In the transport IP layer, point-to-point transmission is used to deliver the signaling PDUs. The Xn-C interface supports Xn inter- face management and error handling (the functionality to manage the Xn-C interface), mobility support for UE in CM-CONNECTED (the functionality to manage the UE mobility for connected mode between nodes in the NG-RAN), context transfer from the current serv- ing NG-RAN node to the new serving NG-RAN node, and control of user-plane tunnels between the existing serving NG-RAN node and the new serving NG-RAN node. It further supports DC (the functionality to enable usage of additional resources in a secondary node in the NG-RAN) as well as support of paging (sending of paging messages via the last serv- ing NG-RAN node toward other nodes in the RAN-based notification area to a UE in RRC_INACTIVE state). Stream control transmission protocol is a transport layer protocol for transmitting multiple streams of data at the same time between two end points that have established a connection in a network. It is message- oriented protocol like user data protocol and ensures reliable, in-sequence transport of messages with conges- tion control like transmission control protocol. Stream control transmission protocol supports multihoming and redundant paths to increase resilience and reliability. 5G Network Architecture 119 1.2.1.1.2 F1 Control-Plane/User-Plane Functions and Procedures As we described earlier, the NG-RAN gNB functions may be implemented in one gNB-CU and one or more gNB-DUs. The g

NB-CU and gNB-DU are connected via an F1 logical interface, which is a standardized interface that supports interchange of signaling informa- tion and data transmission between the termination points. From a logical point of view, F1 is a point-to-point interface between the termination points, which may be established irre- spective of the existence of a physical connectivity between the two points. F1 interface supports control-plane and user-plane separation and separates radio network and TNLs. This interface enables exchange of UE and non-UE associated information. Other nodes in the network view the gNB-CU and a set of gNB-DUs as a gNB. The gNB terminates X2, Xn, NG, and S1-U interfaces [28]. The RRM functions ensure the efficient use of the available network resources. In a gNB- CU/gNB-DU disaggregated model, different RRM functions may be located at different locations. As an example, radio bearer control function (for establishment, maintenance, and release of radio bearers) can be located either at gNB-CU or gNB-DUs, whereas inter- RAT RRM function (for management of radio resources in conjunction with inter-RAT mobility) and dynamic resource allocation and packet scheduling functions (for allocation and de-allocation of resources to user and control-plane packets) are exclusively located at gNB-CU and gNB-DU(s), respectively. Fig. 1.55 shows F1 control-plane and user-plane protocol structures. In the control plane, the TNL is based on IP transport using SCTP over IP for transfer of control messages. The application layer signaling protocol is referred to as F1 application protocol (F1AP). In the user plane, the IP-based TNL uses GTP-U over UDP/IP for transfer of data packets. User-plane PDUs GTP-U Data link layer Data link layer Physical layer Physical layer User plane Control plane Figure 1.55 User-plane and control-plane protocol structure of F1 interface [28]. 120 Chapter 1 F1AP provides a signaling conduit between gNB-DU and the gNB-CU whose services are divided into non-UE-associated (related to t

he entire F1 interface instance between the gNB-DU and gNB-CU utilizing a non-UE-associated signaling connection); and UE- associated (F1AP functions that provide these services are associated with a UE-associated signaling connection that is maintained for a specific UE) services [31]. F1AP consists of elementary procedures (EPs), where an EP is defined as the unit of interaction between gNB-CU and gNB-DU over F1. These EPs are defined separately and are used to create complete sequences in a flexible manner. Unless otherwise stated by the restrictions, these EPs may be invoked independent of each other as stand-alone procedures, which can be active in parallel. An EP consists of an initiating message and possibly its response mes- sage. Two types of EPs, referred to as class 1 and class 2, are used, where the former con- sists of elementary procedures with response (success and/or failure) and the latter comprises EPs without response. A gNB-CU UE F1AP ID is allocated to uniquely identify the UE over the F1 interface within a gNB-CU. The gNB-DU stores the received gNB-CU UE F1AP ID for the duration of time that the UE-associated logical F1-connection is valid. A gNB-DU UE F1AP ID is assigned to uniquely identify the UE over the F1 interface within a gNB-DU. When a gNB- CU receives a gNB-DU UE F1AP ID, it stores it for the period of time that the UE- associated logical F1-connection for the UE remains valid. The UE-associated signaling is used when F1AP messages corresponding to a UE utilize the UE-associated logical F1- connection for association of the message to the UE in gNB-DU and gNB-CU. A UE- associated logical F1-connection uses unique identities that are used to identify UE F1AP messages by the gNB-CU and gNB-DU. A UE-associated logical F1-connection may exist before the F1 UE context58 is setup in the gNB-DU. The F1 interface management procedures include reset, error indication, F1 setup, gNB-DU configuration update, and gNB-CU configuration update on the control plane. The F1-C context management pro

cedures include UE context setup, UE context release request (gNB-DU/gNB-CU initiated), UE context modification (gNB-CU/gNB-DU initiated), and UE mobility command. The F1-C RRC message transfer procedures include initial uplink RRC message transfer as well as UL/DL RRC message transfer. The F1 control plane fur- ther includes system information procedure and paging procedures [28]. The error indication function is used by the gNB-DU or gNB-CU to indicate occurrence of an error. The reset A gNB UE context is a block of information in the gNB associated with an active user equipment. The block of information contains the necessary information in order to provide NG-RAN services to the active user equipment. The gNB user equipment context is established when the transition to active state for a user equipment is completed or in target gNB after completion of handover resource allocation during handover preparation, where in each case the user equipment state information, security information, user equipment capability information, and the identities of the user equipment-associated logical NG connection are stored in the gNB user equipment context. 5G Network Architecture 121 function is used to initialize the peer entity after node setup and after a failure event. This procedure can be used by both the gNB-DU and gNB-CU. The F1 setup function, initiated by gNB-DU, allows exchange of application-level data between gNB-DU and gNB-CU while ensuring proper interoperability over the F1 interface. The gNB-CU and gNB-DU configuration update functions facilitate updating application-level configuration data between gNB-CU and gNB-DU to properly interoperate over the F1 interface, and activate or deactivate the cells. Scheduling of broadcast system information is performed in the gNB-DU. The gNB-DU is responsible for encoding of NR-MIB (master information block portion of NR system information). In case broadcast of RMSI (remaining system informa- tion) or other SI messages is needed, the gNB-DU will be responsible for enco

ding of RMSI and the gNB-CU is responsible for encoding of other SI messages. The gNB-DU and gNB-CU measurement reporting functions are used to report the measurements of gNB-DU and gNB-CU, respectively. The gNB-DU is further responsible for transmitting paging information according to the scheduling parameters. The gNB-CU provides paging information to enable gNB-DU to cal- culate the exact paging occasion and paging frame. The gNB-CU is responsible for calculat- ing paging area. The gNB-DU combines the paging records for a particular paging occasion, frame, and area and further encodes the RRC message and broadcasts the paging message. The F1 UE context management function supports the establishment and modifi- cation of the necessary overall initial UE context. The mapping between QoS flows and radio bearers is performed by gNB-CU where the granularity of bearer related management over F1 is at radio-bearer level. To support PDCP duplication for intra-DU carrier aggrega- tion, one DRB should be configured with two GTP-U tunnels between gNB-CU and gNB-DU. 1.2.1.1.3 E1 Control-Plane Functions and Procedures As we discussed earlier, the disaggregated gNB model was introduced to enable separation of control-plane and user-plane functions in addition to the functional split of the protocol stack in gNB which facilitates design and development of new-generation gNBs based on the SDN/NFV concepts. As such, a new interface between the gNB-CU-CP and gNB-CU- UP components has been defined by 3GPP to support exchange of signaling information between these entities (see Fig. 1.52). E1 is an open interface which would allow multi- vendor implementation of the control-plane and data-plane components of the CU. E1 estab- lishes a logical point-to-point interface between a gNB-CU-CP and a gNB-CU-UP even in the absence of a direct physical connection between the endpoints. The E1 interface sepa- rates radio network and TNLs and enables exchange of UE associated/non-UE associated information. The E1 interface is a control inter

face and is not used for user data forwarding. As shown in Fig. 1.52, a gNB may consist of one gNB-CU-CP, multiple gNB-CU-UPs, and multiple gNB-DUs. One gNB-DU can be connected to multiple gNB-CU-UPs under the 122 Chapter 1 Data link layer Physical layer Figure 1.56 Protocol stack for E1 interface [36]. control of the same gNB-CU-CP and one NB-CU-UP can be connected to multiple DUs under the control of the same gNB-CU-CP. The connectivity between a gNB-CU-UP and a gNB-DU is established by the gNB-CU-CP using bearer or UE context management func- tions. The gNB-CU-CP further selects the appropriate gNB-CU-UP(s) for the requested UE services. The E1 interface would support independent virtualization of the control and UPFs. It would also enable more flexible allocation of the functions of the central unit. It allows for energy and cost-efficient central processing and resource pooling for the user plane. Several functions such as security, packet inspection, header compression, and data mining could benefit from centralization. Furthermore, it would provide optimum routing of packets in case of multi-connectivity and interworking with other systems. The protocol structure for E1 is shown in Fig. 1.56. The TNL uses IP transport with SCTP over IP. The application layer signaling protocol is referred to as E1 application protocol (E1AP). 1.2.1.1.4 NG Control-Plane/User-Plane Functions and Procedures The NG control-plane interface (NG-C) is defined between the gNB/ng-eNB and the AMF. The control-plane protocol stack of NG interface is shown in Fig. 1.57. The TNL relies on IP transport; however, for more reliable transmission of signaling mes- sages, the SCTP protocol is used over IP. The application layer signaling protocol is referred to as NG application protocol (NGAP). The SCTP layer provides guaranteed delivery of application layer messages. IP layer point-to-point transmission is used to deliver the signaling PDUs. The NG-C interface further enables interface management (the functionality to manage the NG-C inter

face), UE context management (the functionality to manage the UE context between NG-RAN and 5GC), UE mobility 5G Network Architecture 123 User-plane PDUs GTP-U Data link layer Data link layer Physical layer Physical layer User plane Control plane Figure 1.57 NG-U and NG-C protocol stack [16]. management (the functionality to manage the UE mobility in connected mode between NG-RAN and 5GC), transport of NAS messages (procedures to transfer NAS messages between 5GC and UE), and paging (the functionality to enable 5GC to generate paging messages sent to NG-RAN to allow NG-RAN to page the UE in RRC_IDLE state). It further enables PDU session management (the functionality to establish, manage, and remove PDU sessions and respective NG-RAN resources that are made of data flows carrying user-plane traffic) as well as configuration transfer (the functionality to trans- fer the NG-RAN configuration information, e.g., transport layer addresses for establish- ment of Xn interface, between two NG-RAN nodes via 5GC). The NGAP protocol consists of elementary procedures where an elementary procedure is a unit of interac- tion between the NG-RAN node and the AMF. These elementary procedures are defined separately and are used to create complete sequences in a flexible manner. Unless otherwise stated by the restrictions, the EPs may be invoked independent of each other as stand-alone procedures, which can be active in parallel. An EP consists of an initiating message and possibly the corresponding response message. Two types of EPs referred to as class 1 and class 2 are defined, where class 1 EP consists of mentary procedures with response (success and/or failure) and class 2 EP comprises elementary procedures without response. The NG user-plane (NG-U) interface is defined between a NG-RAN node and a UPF. The NG-U interface provides non-guaranteed delivery of user-plane PDUs between the two nodes. The user-plane protocol stack of the NG interface is also shown in Fig. 1.57. The TNL relies on IP transport and use of GTP-U over UDP

/IP to carry the user-plane PDUs between the NG-RAN node and the UPF. 124 Chapter 1 1.2.1.2 Bearers and Identifiers 1.2.1.2.1 Radio Bearers and Packet Data Unit Sessions In LTE, IP connection between a UE and a PDN is established via a PDN connection or EPS session. An EPS session delivers the IP packets that are labeled with UE IP address through logical paths between the UE and the PDN (UE-PGW-PDN). An EPS bearer is a logical pipe through which IP packets are delivered over the LTE network, that is, between a UE and a PGW (UE-eNB-SGW-PGW) A UE can have multiple EPS bearers concur- rently. Thus different EPS bearers are identified by their EPS bearer IDs, which are allo- cated by the MME. An EPS bearer in reality is the concatenation of three underlying bearers: (1) A DRB between the UE and the eNB, which is set up upon user-plane establish- ment between the UE and the access node; (2) an S1 bearer between the eNB and SGW, which is set up upon establishment of EPS connection between the UE and the EPC; and (3) an S5/S8 bearer between the SGW and PGW, which is set up upon establishment of PDN connection between the UE and the PGW. An E-URAN random access bearer (E- RAB) is a bearer with two endpoints at the UE and at the SGW, which consists of a DRB and an S1 bearer. As such, E-RAB is a concatenation of a DRB and an S1 bearer, which logically connects the UE to the SGW. LTE QoS architecture describes how a network operator could create and configure different bearer types in order to map various IP traffic categories (user services) to the appropriate bearers according to the service QoS require- ments. In that context, a bearer is an encapsulation mechanism or a tunnel that is created per user for transporting various traffic flows with specific QoS requirements within the LTE access and core networks. In LTE and NR, there are two types of radio bearers, namely signaling radio bearers (SRB) and DRB. The SRBs are radio bearers that are only used for the transmission of RRC and NAS messages, whereas the DRBs are r

adio bearers that are used to transport user-plane traffic. The RRC messages are used as signaling between UE and the access node (i.e., eNB or gNB). The NAS messages are used for signaling between the UE and the core network. The RRC messages are used to encapsulate the NAS messages for transfer between the UE and the core network through the access node (transparent to the access node). The SRBs are further classified into three types: SRBO, SRB1, and SRB2, where SRBO is used to transfer RRC messages which use common control channel, SRB1 is used to transfer RRC messages which use dedicated control channel, and SRB2 is used to transfer RRC messages which use dedicated control channel and encapsulate NAS messages. The SRB1 can be used to encapsulate NAS message, if SRB2 has not been configured. The SRB2 has lower priority than SRB1 and it is always configured after security activation. The SRBO uses RLC transparent mode, while SRB1 and SRB2 use RLC acknowledged mode. In LTE, upon connection establishment and at the beginning of an EPS session, a default EPS bearer, with no guaranteed bit rate and best effort QoS characteristics, is established. 5G Network Architecture 125 Access node Radio Interface Encapsulated IP IP flow IP flow 2 Radio bearer User-plane tunnel IP flow N - QoS flow(s) PDU session Figure 1.58 High-level illustration of 3GPP bearer architecture [16]. Dedicated EPS bearers with guaranteed bit rate and specific QoS attributes for various ser- vices can be established through negotiation and connection reconfiguration for active users. In LTE, the QoS control is performed per EPS bearer such that EPS bearers and radio bearers have a one-to-one relationship. There are two types of end-to-end bearers. The default bearer is established during the attach procedure and after an IP address is allocated to the UE and has best effort QoS characteristics. The dedicated bearer is typically estab- lished during the call setup after transition from the idle mode but can also be established during the attach pr

ocedure and supports various QoS characteristics with guaranteed bit rate. Bearer establishment negotiations are performed between the UE and the access point name (APN) 59 in the core network which maps the bearers to an external network such as the Internet or an IP multimedia subsystem (IMS). However, it is important to note that the availability and provisioning of bearers is strictly controlled by operator configuration, as well as the association between the UE and the PDN that provides PDU connectivity ser- vice. The PDU sessions can be based on IPv4, IPv6, Ethernet or unstructured. In NR and as shown in Fig. 1.58, the UE receives services through a PDU session, which is a logical connection between the UE and the data network. Various PDU session types are supported, for example, IPv4, IPv6, and Ethernet. Unlike the EPS, where at least one default bearer is always created when the UE attaches to the network, 5GS can establish a session Access point name is a gateway or anchor node between a mobile network and another IP network such as the Internet. A mobile device attempting a data connection must be configured with an access point name to present to the carrier. The carrier will then examine this identifier in order to determine what type of network connection should be created, which IP addresses should be assigned to the wireless device, and which secu- rity methods should be used. Therefore, the access point name identifies the packet data network with which a mobile data user communicates. The IP multimedia subsystem is an architectural framework developed by 3GPP for delivering IP multimedia services. Mobile networks originally provided voice services through circuit-switched networks and later migrated to all-IP network architectures. IP multimedia subsystem provides real-time multimedia sessions (voice over IP, video, teleconferencing, etc.) and nonreal-time multimedia sessions (push to talk, presence, and instant messaging) over an all-IP network. IP multimedia subsystem enables convergence of s

ervices provided by different types of networks which include fixed, mobile, and Internet. 126 Chapter 1 when service is needed and independent of the attachment procedure of UE, that is, attach- ment without any PDU session is possible. 5GS also supports the UE to establish multiple PDU sessions to the same data network or to different data networks over a single or multi- ple access networks including 3GPP and non-3GPP. The number of UPFs for a PDU session is not specified. The deployment with at least one UPF is essential to serve a given PDU session. For a UE with multiple PDU sessions, there is no need for a convergence point such as SGW in the EPC. In other words, the user-plane paths of different PDU sessions are completely disjoint. This implies that there is a distinct buffering node per PDU session for the UE in the RRC_IDLE state. 5G QoS framework has been designed to allow detection and differentiation of sub-service flows in order to provide improved quality of experience relative to the previous generations of 3GPP radio interface standards. Since LTE bearer framework was insufficient to address certain 5G service requirements, a refined QoS model based on the concept of QoS flow was introduced in 5G. The QoS flow is the finest granularity for QoS enforcement in 3GPP 5G systems. All traffic mapped to the same 5G QoS flow receives the same forwarding treatment. Providing different QoS forwarding treatment requires the use of different 5G QoS flows. Fig. 1.59 illustrates the comparison between 4G and 5G QoS models. It is shown that the 5G concept allows flexible mapping of the 5G QoS flows to radio bearers, for example, the first 5G QoS flow is transported over the first 5G radio bearer while the second and third 5G QoS flows are transported together in the second 5G radio bearer. In order to support 5G QoS flows, either the existing protocols (e.g., PDCP) had to be enhanced or a new (layer-2) sublayer known as service data adaptation protocol (SDAP) had to be introduced. The main services and functio

ns of SDAP include mapping between a Access node UPF (PDN gateway) Radio interface Encapsulated IP IP flow 1 IP flow 2 LTE QoS Radio bearer S1 bearer + S5/S8 bearer IP flow N model EPS bearer IP flow IP flow 2 NR QoS Radio bearer User-plane tunnel IP flow N - model QoS flow(s) PDU session End-to-end service Figure 1.59 Comparison of QoS models in 4G LTE and 5G NR. 5G Network Architecture 127 QoS flow and a data radio bearer, marking QoS flow identifier (QFI) in the downlink and uplink packets. For each DRB, the UE monitors the QoS flow ID(s) of the downlink packets and applies the same mapping in the uplink, that is, for a DRB, the UE maps the uplink packets belonging to the QoS flows(s) corresponding to the QoS flow ID(s) and PDU ses- sion in the downlink packets for that DRB. In order to establish a PDU session, a PDU session establishment message is sent by 5GC to the gNB serving the UE, which includes the NAS message to be transferred to the UE containing the QoS-related information. The gNB sends a DRB setup request message to the UE and includes the DRB parameters and the NAS message that it received earlier. The UE establishes at least a default DRB associated with the new PDU session. It further cre- ates the QFI to DRB mapping and sends an RRC DRB setup complete message to the gNB. The gNB sends PDU session establishment acknowledgment message to 5GC, indicating successful establishment of the PDU session. Data is sent over the N3 tunnel to the gNB and then over the DRB to the UE. The data packets may optionally include a QoS marking (same as or corresponding to QFI) in their SDAP header. The UE sends uplink packets over the DRB to the gNB. The uplink data packets include a QoS marking (same as or corre- sponding to QFI) in the SDAP header [16]. Dual-connectivity was introduced as part of LTE Rel-12. In LTE DC, the radio protocol architecture that a particular bearer uses depends on how the bearer is setup. Three bearer types have been defined: master cell group (MCG), secondary cell group (SCG), and spl

MACLTE MeNB (LTE) SgNB (NR) MeNB (NR) SgNB (LTE) Figure 1.62 Illustration of split bearer via SCG [13]. the S1-U/NG-U interface terminates in the master node, whereas for SCG bearer, the S1- U/NG-U interface terminates at the secondary node [13]. 3GPP NR supports DC in which a UE in RRC_CONNECTED is configured to utilize radio resources provided by two distinct schedulers located in two gNBs connected via a non- ideal backhaul. The gNBs involved in DC operation for a certain UE may assume two dif- ferent roles, that is, a gNB may either act as an MgNB or as SgNB. In DC operation, a UE connected to one MgNB and one SgNB. There are four bearer types in NR DC, namely MCG bearer, MCG split bearer, SCG bearer, and SCG split bearer. The dual-connectivity between LTE and NR supports similar bearer types. Split bearer via MCG, SCG bearer (a bearer whose radio protocols are split at the SgNB and belongs to both SCG and MCG), and MCG split bearer (a bearer whose radio protocols are split at the MgNB and belongs to both MCG and SCG). The MCG bearer and one SCG bearer are used for two different QoS flows [16]. 5G Network Architecture 129 In the downlink, the incoming data packets are classified by the UPF based on SDF6 tem- plates according to their precedence (without initiating additional N4 signaling). The UPF conveys the classification of user-plane traffic associated with a QoS flow through an N3 (and N9) user-plane marking using a QFI. The AN binds QoS flows to AN resources (i.e., data radio bearers). There is no one-to-one relationship between QoS flows and AN resources and it is the responsibility of the AN to establish the necessary resources for map- ping to the QoS flows [13]. In 3GPP NR, the DRB defines the packet treatment on the radio interface. A DRB serves packets with the same packet forwarding treatment. Separate DRBs may be established for QoS flows requiring different packet handling. In the downlink, the RAN maps QoS flows to DRBs based on QoS flow ID and the associated QoS profiles. In the uplink, the U

E marks uplink packets over the radio air-interface with the QoS flow ID for the purpose of marking forwarded packets to the core network. Downlink traffic is marked to enable priori- tization in the IP network and in the access node. Similar traffic marking may be used for uplink traffic, according to the operator's configuration. Standardized packet marking informs the QoS enforcement functions of what QoS to provide without any QoS signaling, although the option with QoS signaling offers more flexibility and QoS granularity. 1.2.1.2.2 Radio Network Identifiers Each entity and bearer in NG-RAN and 5GC are identified with a unique identifier. The identifiers are either permanently provisioned, such as the International Mobile Subscriber Identity (IMSI) and International Mobile Equipment Identity (IMEI) that are assigned by an operator to the UE, or they are assigned during the lifetime of UE operation in an operator's network. Fig. 1.56 illustrates the identities that are either provisioned or dynamically/semi- statically assigned to the bearers and the entities in 5GS. This section describes the bearers and identifiers that have been introduced in 3GPP 5G specifications. The direction of the arrows in the figure indicates the entity which assigns the identifier and the entity to which the identifier is assigned. The bearers and various bearer identifiers depending on the net- work interface and the flow direction are also shown in the figure. Note that the protocols over NR-Uu and NG interfaces are divided into two categories: user-plane protocols, which Service data flow is a fundamental concept in the 3GPP definition of QoS and policy management. Service data flows represent the IP packets related to a user service (web browsing, e-mail, etc.). Service data flows are bound to specific bearers based on policies defined by the network operator. The traffic detection filters, for example, IP packet filter, required in the user-plane function can be configured either in the SMF and provided to the UPF, as service

data flow filter(s), or be configured in the UPF, as the application detection filter identified by an application identifier. In the latter case, the application identifier has to be configured in the SMF and the UPF. In this context, service data flow filter is a set of packet flow header parameter values/ranges used to identify one or more of the packet (IP or Ethernet) flows constituting a service data flow. service data flow template is the set of service data flow filters in a policy rule or an application identi- fier in a policy rule referring to an application detection filter, required for defining a service data flow. 130 Chapter 1 are the protocols implementing the actual PDU session carrying user data through the AS; and the control-plane protocols that are the protocols for controlling the PDU session and the connection between the UE and the network from different aspects including requesting the service, controlling different transmission resources, handover, etc. In LTE, several radio network temporary identifiers (RNTIs) were used to identify a connected-mode UE within a cell, a specific physical channel, a group of UEs in case of paging, a group of UEs for which power control command is issued by the eNB, system information transmitted for all UEs by the eNB, etc. In general, the RNTIs are used to scramble the CRC part of the radio channel messages. This implies that if the UE does not know the exact RNTI values for each of the cases, it cannot decode the radio channel mes- sages even though the message reaches to the UE. The radio network and UE identifiers in NR, while similar to those of LTE, have been adapted to support new features and function- alities of NR and the 5GC such as multi-connectivity and network slicing. When an NR- compliant UE is connected to 5GC, the following identities are used at cell level to uniquely identify the UE (Fig. 1.63): Cell Radio Network Temporary Identifier (C-RNTI) is a unique identification, which is used as an identifier of the RRC connection and for sc

heduling purposes. In DC scenarios, two C-RNTIs are independently allocated to the UE, one for MCG and one for SCG. Temporary C-RNTI, which is used during the random-access procedure, is a random value for contention resolution which during some transient states, the UE is temporar- ily identified with a random value used for contention resolution purposes. Inactive RNTI (I-RNTI) is used to identify the UE context for RRC_INACTIVE The following identities are used in NG-RAN for identifying a specific network entity: AMF Name is used to identify an AMF. The AMF Name fully-qualified domain name (FQDN) uniquely identifies an AMF, where FQDN consists of one or more labels. Each label is coded as a one octet-length field followed by that number of octets coded as 8 bit ASCII characters. An AMF Set within an operator's network is identified by its AMF Set ID, AMF Region ID, mobile country code (MCC), and mobile network code (MNC). NR cell global identifier (NCGI) is used to globally identify the NR cells. The NCGI is constructed from the PLMN identity to which the cell belongs and the NR cell identity (NCI) of the cell. It can be assumed that it is equivalent to CGI in LTE system. gNB Identifier (gNB ID) is used to identify gNBs within a PLMN. The gNB ID is con- tained within the NCI of its cells. Global gNB ID is used to globally identify the gNBs, which is constructed from the PLMN identity to which the gNB belongs and the gNB ID. The MCC and MNC are the same as included in the NCGI. TAI is used to identify TAs. The TAI is constructed from the PLMN identity to which the TA belongs and the tracking area code (TAC) of the TA. S-NSSAI is used to identify a network slice. 5G Network Architecture PLMN ID Provisioned identities (Dynamically/semistatically) Assigned identities UE IP (static IP address) UPF IP address (FQDN) GUAMI AMF NAME TAI list allocation UPF IP address (FQDN) policy/rule TAI list 5G-GUTI AMF UE NGAP ID AMF/SEAF AMF NAME SMF/PCF gNB UE NGAP ID gNB ID 5G-GUTI 5G-S-TMSI AMF NAME UPF IP address (FQDN) UPF I

P address (FQDN) TAI 1 UPF IP address (FQDN) TAI 1 SUPI/SUPI C-RNTI PEI/IMEI I-RNT 5G-TMSI 5G UE PDU session ID/DNN, Data network TAI 1 Source: 3GPP TS 23.003 (Release 15) TAI 2 Figure 1.63 NG-RAN/5GC entities and bearers and identifiers. An application protocol identity (AP ID) is assigned when a new UE-associated logical con- nection is created in either a gNB or an AMF. An AP ID uniquely identifies a logical con- nection with a UE over the NG interface or Xn interface within a node (gNB or AMF). Upon receipt of a message that has a new AP ID from the originating node, the receiving node stores the corresponding AP ID for the duration of the logical connection. The defini- tion of AP IDs used over the NG, Xn, or F1 interface are as follows [15]: gNB UE NG application protocol (NGAP) ID is used to uniquely identify a UE over the NG interface within a gNB. This identifier is stored by the AMF for the duration of the UE association through the logical NG connection. This identifier is included in all UE associated NGAP signaling. AMF UE NGAP ID is allocated to uniquely identify the UE over the NG interface within an AMF. This identifier is stored for the duration of the UE-associated logical NG connection by the gNB. This identifier is included in all UE-associated NGAP sig- naling once known to the gNB. 132 Chapter 1 Old gNB UE XnAP ID is used to uniquely identify a UE over the Xn interface within a source gNB and it is stored by the target gNB for the duration of the UE association over the logical Xn connection. This identifier is included in all UE-associated XnAP signaling. New gNB UE XnAP ID is used to uniquely identify a UE over the Xn interface within a target gNB. When a source gNB receives a new gNB UE XnAP ID, it stores it for the duration of the UE association over logical Xn connection. This identifier is included in all UE-associated XnAP signaling. MgNB UE XnAP ID is allocated to uniquely identify the UE over Xn interface within an MgNB for dual-connectivity. The SgNB stores this identity for the du

ration of the UE association via logical Xn connection. This identifier is included in all UE- associated XnAP signaling. SgNB UE XnAP ID is used to uniquely identify the UE over Xn interface within a SgNB for dual-connectivity. The MgNB stores this identity for the duration of the UE-associated logical Xn connection. This identifier is included in all UE-associated XnAP signaling. gNB-CU UE F1AP ID uniquely identifies the UE association over the F1 interface within the gNB-CU. gNB-DU UE F1AP ID uniquely identifies the UE association over the F1 interface within the gNB-DU. gNB-DU ID is configured at the gNB-DU and is used to uniquely identify the gNB-DU within a gNB-CU. The gNB-DU informs the gNB-CU of its gNB-DU ID during F1 setup procedure. The gNB-DU ID is used over F1AP procedures. 1.2.1.3 User-Plane and Control-Plane Protocol Stacks This section describes an overview of the NG-RAN protocol structure, protocol layer termi- nations at various access and core network nodes, as well as the functional split between the NG-RAN and the 5GC. The NG-RAN radio protocols can be divided into control-plane and user-plane categories, where the user-plane protocols are typically responsible for carrying user data and the control-plane protocols are used to transfer signaling and control information. In general, a communication protocol is a set of rules for message exchange and/or sending blocks of data known as PDUs between network nodes. A protocol may define the packet structure of the data transmitted and/or the control commands that manage the session. A protocol suite consists of several levels of functionality. This modularity facilitates the design and evaluation of protocols. Since each protocol layer usually (logically or physi- cally) communicates with its peer entity across a communication link, they are commonly seen as layers in a stack of protocols, where the lowest protocol layer always deals with physical interaction of the hardware across the communication link. Each higher layer 5G Network Architecture

133 protocol adds more features or functionalities. User applications usually deal with the top- most layers. In the context of protocol structure, we will frequently use the terms service and protocol. It must be noted that services and protocols are distinct concepts. A service is a set of primi- tives or operations that a layer provides to the layer(s) to which it is logically/physically connected. The service defines what operations a layer performs without specifying how the operations are implemented. It is further related to the interface between two adjacent layers. A protocol, in contrast, is a set of rules presiding over the format and interpretation of the information/messages that are exchanged by peer entities within a layer. The entities use protocols to implement their service definitions. Thus a protocol is related to the imple- mentation of a service. The protocols and functional elements defined by 3GPP standards correspond to all layers of the open system interconnection (OSI), that is, the seven-layer network reference model.6 62 As shown in Fig. 1.64, what 3GPP considers as layer-2 and layer-3 protocols is mapped to the OSI data link layer. The higher layer protocols in the 3GPP stack are the application and transport layers. The presentation and session layers are often abstracted in practice. The NG-RAN protocol structure is depicted in Fig. 1.65 for UEs and gNBs in the user plane and control plane. In the control plane, the NAS functional block is used for network attach- ment, authentication, setting up bearers, and mobility management. All NAS messages are ciphered and integrity protected by the AMF and the UE. There is also a mechanism for transparent transfer of NAS messages. As shown in Fig. 1.66, the layer-2 of NR is divided into MAC, RLC, PDCP, and SDAP sublayers. The SAP or the interface between two adja- cent protocol layers is marked with a circle at the interface between the sublayers in the fig- ure. The SAP between the physical layer and the MAC sublayer provides the transport

channels. The SAP between the MAC sublayer and the RLC sublayer provides the logical channels. The physical layer provides transport channels to the MAC sublayer. From the physical layer perspective, the MAC sublayer provides and receives services in the form of transport channels. The data in a transport channel is organized into transport blocks. By Open systems interconnection is a standard description or a reference model for the computer networks which describes how messages should be transmitted between any two nodes in the network. Its original pur- pose was to guide product implementations to ensure consistency and interoperability between products from different vendors. This reference model defines seven layers of functions that take place at each end of a communication link. Although open systems interconnection is not always strictly adhered to in terms of grouping, the related functions together in a well-defined layer, many if not most products involved in tele- communication make an attempt to describe themselves in relation to the open systems interconnection model. Open systems interconnection was officially adopted as an international standard by the International Organization of Standards and it is presently known as Recommendation X.200 from ITU-T. The layers of open systems interconnection model are classified into two groups. The upper four layers are used whenever a message passes from or to a user. The lower three layers (up to the network layer) are used when any mes- sage passes through the host computer. Chapter 1 NR protocol structure Non-access stratum OSI Seven-layer network (NAS) Internet protocol (IP) model Application layer Radio resource control (RRC) Presentation layer QoS flows Session layer Service data adaptation protocol Control/ (SDAP) configuration Transport layer Radio bearers RRC PDUs Packet data convergence Network layer Control/configuration protocol (PDCP) Data-link layer RLC Channels Radio link control (RLC) Physical layer Control/configuration Logical channels Medi

um access Control/configuration control (MAC) Transport channels| Physical layer (PHY) Control/configuration/ physical layer measurements Physical channels Figure 1.64 Mapping of lower NR protocol layers to OSI network reference model. varying the transmission format of the transport blocks, the MAC sublayer can realize dif- ferent data rates and reliability levels. The MAC sublayer receives the RLC SDUs mapped to various logical channels in the downlink, and generates the MAC PDUs that further become the transport blocks in the physical layer. The RLC sublayer provides RLC chan- nels to the PDCP sublayer and the latter provides radio bearers to the SDAP sublayer. As we mentioned earlier, radio bearers are classified into data radio bearers for user-plane data and SRBs for control-plane information. The SDAP sublayer is configured by RRC and maps QoS flows to DRBs where one or more QoS flows may be mapped into one DRB in the downlink; however, one QoS flow is mapped into only one DRB at a time in the uplink 5G Network Architecture User-plane Access and mobility function management (UPF) function (AMF) User-plane ((HII) protocol Control-plane stack protocol stack Control/ User data signaling Figure 1.65 NG-RAN protocol stack [16]. [16]. The introduction of a new sublayer in NR layer-2 was meant to support the improved flow-based QoS model in NR as opposed to bearer-based QoS model in LTE. MAC sublayer is responsible for mapping between logical channels and transport channels and multiplexing/de-multiplexing of MAC SDUs belonging to one or different logical chan- nels to transport blocks which are delivered to or received from the physical layer through the transport channels. The MAC sublayer further handles scheduling and UE measurements/ reporting procedures as well as error correction through HARQ (one HARQ entity per carrier in case of CA). It further manages user prioritization via dynamic scheduling, priority han- dling among logical channels of one UE through logical channel prioritization. A single MAC ins

tantiation can support one or more OFDM numerologies, transmission timings as well as mapping restrictions on logical channels. In case of CA, the multi-carrier properties of the physical layer are only exposed to the MAC sublayer. In that case, one HARQ entity is required per serving cell. In both uplink and downlink, there is one independent HARQ entity per serving cell and one transport block is generated per transmission time interval per serv- ing cell in the absence of spatial multiplexing. Each transport block and the associated HARQ retransmissions are mapped to a single serving cell [16]. QoS flows QoS flows QoS flow QoS flow QoS flow handling handling handling Radio bearers Radio bearers Security Security Security Security Security Security RLC channels RLC channels Segmentation Segmentation Segmentation Segmentation Segmentation Segmentation Logical channels Logical channels Scheduling Schedulingand priority handling Multiplexing Multiplexing UE1 Multiplexing UEn Transport channels Transport channels Uplink layer 2 structure Downlink layer 2 structure Figure 1.66 3GPP NR DL/UL layer-2 protocol structure [ 16] 5G Network Architecture 137 MAC sublayer provides different type of data transfer services through mapping of logical channels to transport channels. Each logical channel type is defined by the type of informa- tion being transferred. Logical channels are classified into two groups of control and traffic channels. In NR, the control channels are used for transport of control-plane information and are of the following types (see Fig. 1.67): Broadcast control channel (BCCH) is a downlink channel for broadcasting system con- trol information. Paging control channel (PCCH) is a downlink channel that transports paging and system information change notifications. Common control channel (CCCH) is a logical channel for transmitting control information between UEs and the network when the UEs have no RRC connection with the network. Dedicated control channel (DCCH) is a point-to-point bidirectional channel

that trans- mits UE-specific control information between a UE and the network and is used by the UEs that have established RRC connection with the network. Traffic channels are used for the transfer of user-plane information and are of the following type: Dedicated traffic channel (DTCH) is a UE-specific point-to-point channel for transport of user information which can exist in both uplink and downlink. The physical layer provides information transfer services to the MAC and higher layers. The physical layer transport services are described by how and with what characteristics data is transferred over the radio interface. This should be clearly distinguished from the classification of what is transported which relates to the concept of logical channels in the Control Traffic channels channels Control channels Traffic channels Uplink Downlink logical logical channels channels Uplink Downlink transport transport UL-SCH channels DL-SCH channels Figure 1.67 Mapping of logical and transport channels [16]. 138 Chapter 1 MAC sublayer. In the downlink, the logical channels and transport channels are mapped as follows (see Fig. 1.67): BCCH can be mapped to broadcast channel (BCH), which is characterized by fixed, predefined transport format, and is required to be broadcast in the entire coverage area of the cell. A DL-SCH may support receptions using different numerologies and/or TTI duration within the MAC entity. An UL-SCH may also support transmissions using dif- ferent numerologies and/or TTI duration within the MAC entity. BCCH can be mapped to downlink shared channel(s) (DL-SCH), which is characterized by support for HARQ protocol, dynamic link adaptation by varying modulation, coding, and transmit power, possibility for broadcast in the entire cell, possibility to use beam- forming, dynamic and semistatic resource allocation, and UE discontinuous reception (DRX) to enable power saving. PCCH can be mapped to paging channel (PCH), which is characterized by support for UE DRX in order to enable power saving, require

ment for broadcast in the entire cover- age area of the cell, and is mapped to physical resources which can also be used dynam- ically for traffic or other control channels. This channel is used for paging when the network does not know the location of the UE. Common control channel (CCCH) can be mapped to DL-SCH and represents a logical channel for transmitting control information between UEs and gNBs. This channel is used for UEs that have no RRC connection with the network. Dedicated control channel (DCCH) can be mapped to DL-SCH and is a point-to-point bidirectional channel that transmits dedicated control information between a UE and the network. It is used by UEs that have already established RRC connection. Dedicated traffic channel (DTCH) can be mapped to DL-SCH and represents a point-to- point bidirectional channel dedicated to a single UE for the transfer of user information. In the uplink, the logical channels and the transport channels are mapped as follows: CCCH can be mapped to uplink shared channel(s) (UL-SCH), which is characterized by possibility to use beamforming, support for dynamic link adaptation by varying the transmit power and modulation and coding schemes, support for HARQ, support for both dynamic and semistatic resource allocation. DCCH can be mapped to UL-SCH. DTCH can be mapped to UL-SCH. Random access channel(s) (RACH), which is characterized by limited control informa- tion and collision risk. The RLC sublayer is used to format and transport traffic between the UE and the gNB. The RLC sublayer provides three different reliability modes for data transport: acknowledged mode (AM), unacknowledged mode (UM), and transparent mode (TM). The UM is suitable for transport of real-time services since such services are delay-sensitive and cannot 5G Network Architecture tolerate delay due to ARQ retransmissions. The acknowledged mode is appropriate for non- real-time services such as file transfers. The transparent mode is used when the size of SDUs are known in advance such as for broadcastin

g system information. The RLC sub- layer also provides sequential delivery of SDUs to the upper layers and eliminates duplicate packets from being delivered to the upper layers. It may also segment the SDUs. The RLC configuration is defined per logical channel with no dependency on numerologies and/or TTI durations, and ARQ can operate with any of the numerologies and/or TTI durations for which the logical channel is configured. For SRBO, paging and broadcast system informa- tion, RLC TM mode is used. For other SRBs, RLC AM mode is used. For DRBs, either RLC UM or AM mode is used. The services and functions provided by the PDCP sublayer in the user plane include header compression/decompression of IP packets; transfer of user data between NAS and RLC sublayer; sequential delivery of upper layer PDUs and duplicate detection of lower layer SDUs following a handover in RLC acknowledged mode; retransmission of PDCP SDUs following a handover in RLC acknowledged mode; and ciphering/deciphering and integrity protection. The services and functions provided by the PDCP for the control plane include ciphering and integrity protection and transfer of control-plane data where PDCP receives PDCP SDUs from RRC and forwards them to the RLC sublayer and vice versa. The main services and functions provided by SDAP sublayer include mapping between a QoS flow and a DRB and marking QoS flow IDs in downlink and uplink packets. A single instantiation of SDAP protocol is configured for each individual PDU session, with the exception of dual-connectivity mode, where two entities can be configured. The RRC sublayer in the gNB makes handover decisions based on neighbor cell mea- surements reported by the UE; performs paging of the users over the air-interface; broadcasts system information; controls UE measurement and reporting functions such as the periodicity of channel quality indicator reports; and further allocates cell-level temporary identifiers to the active users. It also executes transfer of UE context from the serving gNB to th

e target-gNB during handover and performs integrity protection of RRC messages. The RRC sublayer is responsible for setting up and maintenance of radio bearers. Note that the RRC sublayer in 3GPP protocol hierarchy is considered as layer-3 protocol [16]. Fig. 1.68 shows an example of layer-2 data flow and packet processing, where a transport block is generated by MAC sublayer by concatenating two RLC PDUs from the radio bearer RBX and one RLC PDU from the radio bearer RBy. In this figure, H denotes the layer-specific headers or subheaders of each sublayer. The two RLC PDUs from RBX each corresponds to one of the IP packets n and n + 1, while the RLC PDU from RBy is a seg- ment of the IP packet m. IP packet n IP packet n +1 IP packet m Radio bearerx Radio bearery SDAP SDU SDAP SDU SDAP SDU PDCP SDU PDCP SDU PDCP SDU RLC SDU RLC SDU SDU segment SDU segment MAC SDU MAC SDU MAC SDU MAC SDU MAC PDU - -transport block Figure 1.68 Example of layer-2 packet processing in NR [ 16]. 5G Network Architecture 1.2.2 Core Network The design of 5G core network architecture in 3GPP was based on the following design principles in order to allow efficient support of new service networks such as information- centric networking (ICN)63 [81]: Control- and user-plane separation (CUPS): This was a departure from LTE's vertically integrated control/user-plane network design to the one adopting NFV framework with modular NFs decoupled from the hardware for service centricity, flexibility, and programmability. In doing so, NFs are going to be implemented both physically and vir- tually, while allowing each to be customized and scaled based on their individual requirements, also allowing the realization of multi-slice coexistence. This further allows the introduction of UPF with new control functions, or reusing/extending the existing ones, to manage the new user-plane realizations. Decoupling of RAT and core network: Unlike LTE's unified control plane for access and the core networks, 5GC offers control-plane separation of RAN from the cor

e network. This allows introduction of new radio access technologies and mapping of multiple heteroge- neous RAN sessions to arbitrary core network slices based on service requirements. Non-IP PDU session support: A PDU session is defined as the logical connection between the UE and the data network. The PDU session establishment in 5GC supports both IP and non-IP PDUs (known as unstructured payloads), and this feature can poten- tially allow the support for ICN PDUs by extending or reusing the existing control functions. Service-centric design: 5GC service orchestration and control functions, such as naming, addressing, registration/authentication, and mobility, will utilize cloud-based service APIs. This enables open interfaces for authorized service function interaction and creat- ing service-level extensions to support new network architectures. These APIs include widely used approaches, while not precluding the use of procedural approach between functional units. Compared to LTE core network, where PDU session states in RAN and core were synchro- nized from session management perspective, 5GC decouples those states by allowing PDU Information-centric networking is an approach to evolve the Internet infrastructure away from a host-centric paradigm based on perpetual connectivity and the end-to-end principle, to a network architecture in which the focus is on content or data. In other words, information-centric networking is an approach to evolve the Internet infrastructure to directly support information distribution by introducing uniquely named data as a core Internet principle. Data becomes independent from location, application, storage, and means of transpor- tation, enabling or enhancing a number of desirable features, such as security, user mobility, multicast, and in-network caching. Mechanisms for realizing these benefits is the subject of ongoing research in Internet Engineering Task Force and elsewhere. Current research challenges in information-centric networking includes naming, security, routing

, system scalability, mobility management, wireless networking, transport services, in-network caching, and network management. 142 Chapter 1 Nnssf Nausf Figure 1.69 5G system service-based reference architecture [3]. sessions to be defined prior to a PDU session request by a UE. This de-coupling allows dynamic and policy-based interconnection of UE flows with slices provisioned in the core net- work. The SMF is used to handle IP anchor point selection and addressing functionality, man- agement of the user-plane state in the UPFs such as in uplink classifier and branching point functions during PDU session establishment, modification and termination, and interaction with RAN to allow PDU session forwarding in UL/DL to the respective data networks. In the user plane, UE's PDUs are tunneled to the RAN using the 5G RAN protocols. From the RAN perspective, the PDU's five-tuple header information (IP source/destination, port, protocol, etc.) is used to map the flow to an appropriate tunnel from RAN to UPF. 1.2.2.1 Reference Architecture: Network Entities and Interfaces The 5G core network architecture comprises a number of NFs, some of which have newly been introduced. In this section, we provide a brief functional description of these NFs. Fig. 1.69 illustrates the 5GC service-based reference architecture. Service-based interfaces are used within the control plane. The main 5GC network entities include the following [3]: Fig. 1.70 depicts the 5G system architecture in the non-roaming case, using the reference- point representation showing how various NFs interact with each other. Note that the NF is a 3GPP-adopted processing function in next generation network that has both functional behavior and interface. An NF can be implemented either as a network element on a dedicated hardware, as a software instance running on a dedicated hardware, or as a virtualized function instantiated on an appropriate platform such as cloud infrastructure. 5G Network Architecture (R)AN PDN (Internet) Figure 1.70 Non-roaming reference-po

int representation of 5G system architectures [3]. service-based and reference-point representations are two different representations of the 5GC which make it distinguished from LTE EPC. Service-based interfaces and reference points are two different ways to model interactions between architectural entities. A reference point is a conceptual point at the conjunction of two non-overlapping functional groups. A reference point can be replaced by one or more service-based interfaces that provide equiva- lent functionality. A unique reference point exists between two NFs, which means even if the functionality of two reference points is the same with different NFs, different reference point names must be assigned. However, when service-based interface representation is used, the same service-based interface is assigned if the functionality is equal on each interface. The functional description of the network functions is as follows [3]: Authentication server function (AUSF) is responsible for performing authentication pro- cess with the user terminals. AMF is responsible for termination of RAN control-plane interface (N2) and NAS (N1); ciphering and integrity protection of NAS messages; registration management; connection management; mobility management; lawful interception; transport of session management messages between UE and SMF; transparent proxy for routing session management messages; access authentication and authorization; and security anchor function (SEAF) and security context management (SCM). In addition to the functional- ities described earlier, the AMF may support functions associated with non-3GPP ANs. Data network (DN) comprises operator's services, Internet access, or other services. 144 Chapter 1 Unstructured data storage network function (UDSF) is an optional function that supports storage and retrieval of information as unstructured data by any NF and the deploy- ments can choose to collocate UDSF with other NFs such as UDR. Network exposure function (NEF) to securely expose the services and ca

pabilities pro- vided by 3GPP NFs, internal exposure/reexposure, AFs, and the edge computing. In addition, it provides a means for the AFs to securely provide information to 3GPP net- work, for example, mobility pattern. In that case, the NEF may authenticate, authorize, and regulate the AFs. It translates information exchanged with the AF and information exchanged with the internal NF. For example, it translates between an AF-service- identifier and internal 5G core information. The NEF receives information from other NFs based on exposed capabilities of other NFs. It may implement a frontend entity to store the received information as structured data using a standardized interface to a uni- fied data repository (UDR). NF repository function (NRF) supports service discovery function; receives NF discov- ery request from NF instance; and provides the information of the discovered NF instances to the NF instance. It further maintains the NF profile of available NF instances and their supported services. NSSF selects the set of NSIs to serve a UE and to determine the allowed NSSAI and to determine the AMF set to serve the UE or depending on the configuration, a list of can- didate AMF(s). Policy control function (PCF) supports interactions with the access and mobility policy enforcement in the AMF through service-based interfaces and further provides access and mobility management-related policies to the AMF. SMF handles session management (session establishment, modification, and release); UE IP address allocation and management; selection and control of UPF; traffic steering configuration at UPF to route traffic to the proper destination; termination of interfaces toward PCFs; control part of policy enforcement and QoS; and lawful interception among other functions. Unified data management (UDM) supports generation of 3GPP authentication and key agreement (AKA)65 authentication credentials; user identification handling; access authorization based on subscription data; UE's serving NF registration management; serv

ice/session continuity; lawful interception functionality; subscription management; and SMS management. To provide these functions, the UDM uses subscription data (including authentication data) that may be stored in the UDR, in that case the UDM implements the application logic and does not require an internal user data storage, thus Authentication and key agreement is a mechanism which performs authentication and session key distribution in UMTS networks. Authentication and key agreement is a challenge-response-based mechanism that uses symmetric cryptography. Authentication and key agreement is typically run on a UMTS IP multimedia ser- vices identity module, which resides on a smart card device that also provides tamper-resistant storage of shared secrets. Authentication and key agreement is defined in IETF RFC 3310. 5G Network Architecture 145 different UDMs may serve the same user in different transactions. The UDM is located in the home PLMN of the subscribers which it serves and accesses the information of the UDR located in the same PLMN. UDR supports storage and retrieval of subscription data by the UDM; storage and retrieval of policy data by the PCF; storage and retrieval of structured data for exposure; and application data by the NEF, including packet flow descriptions for application detection, and application request information for multiple UEs. During the deploy- ments, the operators can opt to co-locate UDR with UDSF. Non-3GPP interworking function (N3IWF) supports untrusted non-3GPP access to 5GC. It further supports IPsec tunnel establishment with the UE; terminates the IKEv260 or IPsec 67 protocols with the UE over NWu and relays over N2 the informa- tion needed to authenticate the UE and authorize its access to the 5G core network as well as termination of N2 and N3 interfaces to 5G core network for control plane and user plane, respectively; relaying uplink and downlink control-plane NAS (N1) signal- ing between the UE and AMF; handling of N2 signaling from SMF related to PDU ses- sions an

d QoS; and establishment of IPsec security association (IPsec SA) to support PDU session traffic. UPF acts as the anchor point for intra-RAT or inter-RAT mobility; external PDU ses- sion point of interconnect to the data network; packet routing and forwarding; packet inspection and user-plane part of policy rule enforcement; lawful interception; traffic usage reporting; uplink classifier to support routing traffic flows to a data network; branching point to support multi-homed PDU session; QoS handling for user plane (packet filtering, gating, and UL/DL rate enforcement); uplink traffic verification (SDF to QoS flow mapping); transport-level packet marking in the uplink and downlink; and downlink packet buffering and downlink data notification triggering. AF is responsible for interacting with the 3GPP core network in order to support appli- cation influence on traffic routing; accessing network exposure function; and interacting with the policy framework for policy control. Based on operator deployment, the AF is considered to be trusted by the operator and can be allowed to interact directly with rel- evant NFs. IKE or IKEv2 is the protocol used to set up a security association in the IPsec protocol suite. The IKE proto- col is based on a key-agreement protocol and the Internet security association and key management protocol which is a protocol defined by IETF RFC 2408 for establishing security associations and cryptographic keys in an Internet environment. The IKE uses X.509 certificates for authentication which are either pre-shared or distributed to set up a shared session secret from which cryptographic keys are derived. In addition, a secu- rity policy for every peer which will connect must be manually maintained. IPsec is a set of protocols for securing IP-based communications by authenticating and encrypting each IP packet of a data stream. IPsec also includes protocols for establishing mutual authentication between agents at the beginning of the session and negotiation of cryptographic keys to be used

during the session. IPsec can be used to protect data flows between a pair of hosts, between a pair of security gateways, or between a security gateway and a host. 146 Chapter 1 Security edge protection proxy is a non-transparent proxy which supports message fil- tering and policing inter-PLMN control-plane interfaces and topology hiding. User equipment Access network Reference-point representation of the architecture can be used to develop detailed call flows in the normative standardization. N1 is defined to carry signaling between UE and AMF. The reference points for connecting AN and AMF and AN and UPF are defined as N2 and N3, respectively. There is no reference point between AN and SMF, but there is a reference point, N11, between AMF and SMF. Therefore, the SMF is controlled by AMF. N4 is used by SMF and UPF SO that the UPF can be configured using the control information generated by the SMF, and the UPF can report its state to the SMF. N9 is the reference point for the connection between different UPFs, and N14 is the reference point connecting different AMFs. N15 and N7 are defined for the PCF to apply policies to AMF and SMF, respec- tively. N12 is required for the AMF to perform authentication of the UE. N8 and N10 are defined to provide the UE subscription data to AMF and SMF. The 5G core network supports UE connectivity via untrusted non-3GPP access networks such as Wi-Fi. Non-3GPP access networks can connect to the 5G core network via a non- 3GPP interworking function (N3IWF). The N3IWF interfaces the 5G core network control plane and UPFs via N2 and N3 interfaces, respectively, with the external network. The N2 and N3 reference points are used to connect stand-alone non-3GPP access networks to 5G core network control plane function and UPF, correspondingly. A UE that attempts to access the 5G core network over a stand-alone non-3GPP access, after UE attachment, must support NAS signaling with 5G core network control-plane functions using N1 reference point. When the UE is connected via a NG-RAN and

via a stand-alone non-3GPP access, multiple N1 instances could exist for the UE, that is, there is one N1 instance over NG- RAN and one N1 instance over non-3GPP access [3]. 1.2.2.2 PDN Sessions and 5GC Identifiers In an LTE network, once a UE connects to a PDN using the IP address assigned to it upon successful initial attach to the network, the IP connection remains in place after a default EPS bearer is established over the LTE network and until the UE detaches from the LTE network (i.e., the PDN connection is terminated). Even when there is no user traffic to send, the default EPS bearer always stays activated and ready for possible incoming user traffic. Additional EPS bearers can be established, if the best effort QoS attributes of the default EPS bearer do not satisfy the service requirements. The additional EPS bearer is called a dedicated EPS bearer, where multiple dedicated bearers can be created, if required by the user or the network. When there is no user traffic, the dedicated EPS bearers can be removed. Dedicated EPS bearers are linked to a default EPS bearer. Therefore, IP traffic from or to a UE is delivered through an EPS bearer depending on the required QoS class 5G Network Architecture 147 over the LTE network. Uplink IP traffic is mapped from a UE to the EPS bearer while downlink IP traffic is mapped from a PGW to the EPS bearer. Each E-RAB is associated with a QCI and an allocation and retention priority (ARP), where each QCI is characterized by priority, packet delay budget (PDB), and acceptable packet loss rate. The 5G core network supports PDU connectivity service, that is, a service that provides exchange of PDUs between a UE and a data network. The PDU connectivity service is sup- ported via PDU sessions that are established upon request from the UE. The PDU sessions are established (upon UE request), modified (upon UE and 5GC request), and released (upon UE and 5GC request) using NAS session management signaling over N1 between the UE and the SMF. Upon request from an application serv

er, 5GC is able to trigger a specific application in the UE. The UE conveys the message to the application upon receiving the trigger message. Note that unlike LTE, 3GPP NR Rel-15 does not support dual-stack PDU session. The 5GC supports dual-stack UEs using separate PDU sessions for IPv4 and IPv6. In 3GPP NR, the QoS granularity is refined further to the flow level. In a typical case, mul- tiple applications will be running on a UE; however, in LTE eNB, each E-RAB does not have an associated QCI or an ARP, whereas in NR, the SDAP sublayer can be configured by RRC sublayer to map QoS flows to DRBs. One or more QoS flows may be mapped to one DRB. Thus, QFI is used to identify a QoS flow within the 5G system. User-plane traffic with the same QFI within a PDU session receives the same traffic forwarding treatment (e.g., scheduling, admission threshold, etc.). The QFI is carried in an encapsulation header on N3 and is unique within a PDU session. The 5GC is access-agnostic and allows running the N1 reference point on non-3GPP radio access schemes such as Wi-Fi. The UE can also send NAS messages for session and mobil- ity management to the 5GC via a non-3GPP access, which was not possible in the previous 3GPP radio access standards. Non-access stratum is the signaling protocol of the UE for mobility and session-related control messages, which requires a new security procedure in order to authenticate the UE over the non-3GPP access with the AMF in the 3GPP network. Non-3GPP access networks must be connected to the 5G core network via N3IWF entity. The N3IWF interfaces the 5G core network control-plane function and UPF via N2 and N3 interfaces, respectively. When a UE is connected via an NG-RAN and via a stand-alone non-3GPP access, multiple N1 instances, that is, one N1 instance over NG-RAN and one N1 instance over non-3GPP access, will be created. The UE is simultaneously connected to the same 5G core network of a PLMN over a 3GPP access and a non-3GPP access and is served by a single AMF provided that the selected N

3IWF is located in the same PLMN as the 3GPP access. However, if the UE is connected to the 3GPP access network of a PLMN and if it selects an N3IWF which is located in a different PLMN, then the UE will be served separately by two PLMNs. The UE is registered with two separate AMFs. The PDU ses- sions over 3GPP access are served by the visiting SMFs which are different from the ones serving the PDU sessions over the non-3GPP access. The UE establishes an IPsec tunnel 148 Chapter 1 with N3IWF in order to attach to the 5G core network over the untrusted non-3GPP access and is authenticated by and attached to the 5G core network during the IPsec tunnel estab- lishment procedure [3]. The network identifiers in 5G system are divided into subscriber identifiers and UE identi- fiers. Each subscriber in the 5G system is assigned a 5G subscription permanent identifier (SUPI) to use within the 3GPP system. The 5G system treats subscription identification independent of the UE identification. In that sense, each UE accessing the 5G system is assigned a permanent equipment identifier (PEI). The 5G system assigns a temporary identi- fier (5G-GUTI) to the UE in order to protect user confidentiality. The 5G network identi- fiers can be summarized as follows [3]: 5G Subscription Permanent Identifier is a global unique identifier that is assigned to each subscriber in the 5G system, which is provisioned in the UDM/UDR. The SUPI is used only within 3GPP system. The previous generations' IMSI68 and network access identifier (NAI)69 can still be used in 3GPP Rel-15 as SUPI. The use of generic NAI makes the use of non-IMSI-based SUPIs possible. The SUPI must contain the address of the home network in order to enable roaming scenarios. For interworking with the EPC, the SUPI allocated to the 3GPP UE is based on the IMSI. Furthermore, 5GS defines a subscription concealed identifier (SUCI) which is a privacy preserving identi- fier containing the concealed SUPI. Permanent Equipment Identifier is defined for a 3GPP UE accessing the 5G sy

stem. The PEI can assume different formats for different UE types and use cases. The UE pre- sents the PEI to the network along with an indication of the PEI format being used. If the UE supports at least one 3GPP access technology, the UE must be allocated a PEI in the IMEI format. In 3GPP Rel-15, the only format supported for the PEI parameter is an IMEI. 5G Globally Unique Temporary Identifier is allocated by an AMF to the UE that is common in both 3GPP and non-3GPP access. A UE can use the same 5G-GUTI for International Mobile Subscriber Identity (IMSI) is used as a unique identification of mobile subscriber in the 3GPP networks. The IMSI consists of three parts: (1) mobile country code consisting of three digits. The mobile country code uniquely identifies the country of residence of the mobile subscriber; (2) mobile net- work code consisting of two or three digits for 3GPP applications. The mobile network code identifies the home public land mobile network of the mobile subscriber. The length of the mobile network code (two or three digits) depends on the value of the mobile country code; and (3) mobile subscriber identification num- ber identifying the mobile subscriber within a public land mobile network. Network access identifier defined by IETF RFC 7542 is a common format for user identifiers submitted by a client during authentication. The purpose of the network access identifier is to allow a user to be associated with an account name, as well as to assist in the routing of the authentication request across multiple domains. Note that the network access identifier may not necessarily be the same as the user's email address or the user identifier submitted in an application-layer authentication. 5G Network Architecture 149 accessing 3GPP access and non-3GPP access security context within the AMF. The AMF may assign a new 5G-GUTI to the UE at any time. The AMF may delay updating the UE with its new 5G-GUTI until the next NAS signaling exchange. The 5G-GUTI comprises a GUAMI and a 5G-TMSI, where GUAMI id

entifies the assigned AMF and 5G-TMSI identifies the UE uniquely within the AMF. The 5G-S-TMSI is the shortened form of the GUTI to enable more efficient radio signaling procedures. AMF Name is used to identify an AMF. It can be configured with one or more GUAMIs. At any given time, the GUAMI value is exclusively associated to one AMF name. Data Network Name (DNN) is equivalent to an APN which may be used to select an SMF and UPF(s) for a PDU session, to select N6 interface(s) for a PDU session, or to determine policies that are applied to a PDU session. Internal-Group Identifier is used to identify a group as the subscription data for a UE in UDM may associate the subscriber with different groups. A UE can belong to a limited number of groups. The group identifiers corresponding to a UE are provided by the UDM to the SMF and when PCC applies to a PDU session by the SMF to the PCF. The SMF may use this information to apply local policies and to store this information in charging data record. Generic Public Subscription Identifier (GPSI) is used for addressing a 3GPP subscrip- tion in different data networks outside of the 3GPP system. The 3GPP system stores within the subscription data the association between the GPSI and the corresponding SUPI. GPSIs are public identifiers used both inside and outside of the 3GPP system. The GPSI is either a mobile subscriber ISDN number (MSISDN) or an external identi- fier. If MSISDN is included in the subscription data, it will be possible that the same MSISDN value is supported in both 5GS and EPS. There is no one-to-one relationship between GPSI and SUPI. 1.2.2.3 User-Plane and Control-Plane Protocol Stacks 1.2.2.3.1 Control-Plane Protocol Stacks The 5GC supports PDU connectivity service which provides exchange of PDUs between a UE and a data network. The PDU connectivity service is supported through PDU sessions that are established upon request from the UE. In order to establish a PDU session and access to the PDN, the UE must establish user plane and control plane over th

e NG-RAN and the 5GC network interfaces to the PDN. Connection management comprises establish- ing and releasing a signaling connection between a UE and the AMF over N1. This signal- ing connection is used to enable NAS signaling exchange between the UE and the core network, which includes both access network signaling connection between the UE and the access node (RRC connection over 3GPP access or UE-N3IWF connection over non-3GPP access) and the N2 connection for this UE between the access node and the AMF. A NAS connection over N1 is used to connect a UE to the AMF. This NAS connection is used for 150 Chapter 1 registration management and connection management functions as well as for transport of session management messages and procedures for the UE. The NAS protocol over N1 com- prises NAS mobility and session management (NAS-MM and NAS-SM) components. There are several protocol information that need to be transported over N1 using NAS-MM protocol between a UE and a core NF besides the AMF (e.g., session management signaling). Note that in 5G systems, registration/connection management NAS messages and other types of NAS messages as well as the corresponding procedures are decoupled. The NAS-MM sup- ports NAS procedures that terminate at the AMF such as handling registration and connection management state machines and procedures of the UE, including NAS transport. There is a single NAS protocol that applies to both 3GPP and non-3GPP access. When a UE is served by a single AMF while it is connected through multiple (3GPP and/or non-3GPP) access schemes, there would be one N1 NAS connection per access link. The security for the NAS messages is provided based on the security context established between the UE and the AMF. It is possible to transmit the other types of NAS messages (e.g., NAS SM) along with RM/ CM NAS messages by supporting NAS transport of different types of payload or messages that do not terminate at the AMF. This includes information about the payload type, informa- tion for forwarding purp

oses, and the SM message in case of SM signaling. The NAS-SM messages control the session management functions between the UE and the SMF. The session management message is created and processed in the NAS-SM layer of UE and the SMF (see Fig. 1.71). The content of the NAS-SM message is transparent to the AMF. The NAS-MM layer creates a NAS-MM message, including security header, indicat- ing NAS transport of SM signaling, as well as additional information for the receiving NAS mobility management (NAS-MM) entity to determine how and where to forward the SM signaling message. The receiving NAS-MM layer performs integrity check and interpreta- tion of NAS message content. Fig. 1.71 further depicts the NAS-MM layer, which is a NAS protocol for mobility management; support of registration management; connection NAS-SM NAS-SM Relay NAS-MM NAS-MM Relay NG-AP NG-AP 5G-AN 5G-AN protocol protocol layer layer 5G-AN Figure 1.71 Control-plane protocol stack between the UE and the AMF/SMF [3]. 5G Network Architecture 151 management and user-plane connection activation and deactivation functions. It is also responsible for ciphering and integrity protection of NAS signaling. The UE and AMF support NAS signaling connection setup function, which is used to estab- lish a NAS signaling connection for a UE in CM-IDLE state. It will be explained in the next chapter that two connection management states are used to reflect the NAS signaling connectivity between the UE and the AMF, that is, CM-IDLE and CM-CONNECTED The CM state for 3GPP access and non-3GPP access are independent of each other, that is, one can be in CM-IDLE state while the other is in CM-CONNECTED state [3]. A UE must register with the network to be authorized to use network services, to assist mobility tracking, and to be reachable. The registration procedure is used when the UE needs to perform initial registration with the 5GS; location update upon entering a new tracking area outside of the UE's registration area in CM-CONNECTED and CM-IDLE modes; when the UE perfo

rms a periodic registration update (due to a predefined inac- tivity time interval); and additionally when the UE needs to update its capabilities or protocol parameters that were negotiated during registration procedure. The AMF pro- vides a list of recommended cells/TAs/NG-RAN node identifiers for paging. The N2 interface supports management procedures which are not UE-specific, rather for configuration or reset of the N2 interface. These procedures are applicable to any access scheme and access-specific messages that carry some information for a particular access scheme such as information on the default paging DRX cycle that is used only for 3GPP access. The N2 interface further supports UE-specific and NAS-transport procedures. These procedures are in general access-agnostic, but they may also correspond to uplink NAS transport messages that carry some access-dependent information such as user location information. The N2 interface also supports procedures related to UE context management and resources for PDU sessions. These messages carry information on N3 addressing and QoS requirements that should be transparently forwarded by the AMF between the 5G access node and the SMF. The N2 interface further enables procedures related to handover management for 3GPP access. The control-plane interface between a 5G access node and the 5G core supports connection of different types of 5G access nodes to the 5GC via a unique control-plane protocol. A sin- gle NGAP protocol is used for 3GPP and non-3GPP access schemes. There is a unique N2 termination point at the AMF for a given UE (for each access node used by the UE) regard- less of the number of PDU sessions of the UE. The N2 control plane supports separation of AMF and other functions such as SMF that may need to control the services supported by 5G access nodes, where in this case, AMF transparently forwards NGAP messages between the 5G access node and the SMF. The N2 session management information (i.e., a subset of NGAP information that AMF transparently relay

s between an access node and SMF) is exchanged between the SMF and the 5G access node which is transparent to the AMF. The NG application protocol enables message exchange between a 5G access node and the 152 Chapter 1 AMF over N2 interface using SCTP protocol, which guarantees delivery of signaling mes- sages between AMF and 5G access node. 1.2.2.3.2 User-Plane Protocol Stacks The protocol stack for the user-plane transport related to a PDU session is illustrated in Fig. 1.72. The PDU layer corresponds to the PDU that is transported between the UE and the PDN during a PDU session. The PDU session type can be IPv6 or Ethernet for trans- porting IP packets or Ethernet frames. The GPRS tunneling protocol for the user plane (GTP-U) supports multiplexing of the traffic from different PDU sessions by tunneling user data over N3 interface (i.e., between the 5G access node and the UPF) in the core network. GTP encapsulates all end-user PDUs and provides encapsulation per-PDU-session. This layer also transports the marking associated with a QoS flow [3]. The 5G encapsulation layer supports multiplexing the traffic from different PDU sessions over N9 interface (i.e., an interface between different UPFs). It provides encapsulation per PDU session and carries the marking associated with the QoS flows. The 5G access node protocol stack is a set of protocols/layers which are related to the access network as described in the previous sections. The number of UPF entities in the data path is not con- strained by the 3GPP specifications; therefore, there could be none or more than one UPF entities in the data path of a PDU session that may not support PDU session anchor func- tionality for that PDU session. In certain cases, there is an uplink classifier or a branching point in the data path of a PDU session, which does not act as the non-PDU session anchor UPF. In that case, there are multiple N9 interfaces branching out of the uplink classifier/ branching point, each leading to different PDU session anchors. Application PDU lay

er PDU layer Relay Relay 5G user-plane GTP-U GTP-U 5G user-plane encapsulation encapsulation 5G-AN 5G-AN UDP/IP UDP/IP UDP/IP UDP/IP protocol protocol layers layers 5G-AN (PDU session anchor) Figure 1.72 User-plane protocol stack between UE and UPF [3]. 5G Network Architecture Carrier 1 (F1) ((HH)) User-plane data Carrier 2 (F2) User-plane data Macro- Small cell gNB cell gNB Non-ideal backhaul (Xn) Figure 1.73 Illustration of inter-node radio resource aggregation (dual connectivity concept). 1.3 Dual Connectivity and Multi-connectivity Schemes Dual connectivity (or multi-connectivity) is a term that is used to refer to an operation where a given UE is allocated radio resources provided by at least two different network nodes connected with non-ideal backhaul (see Fig. 1.73). Each access node involved in dual connectivity for a UE may assume different roles. Those roles do not necessarily depend on the access node's power class and can vary for different UEs. To support tight interworking between LTE and NR, where both LTE eNB and NR gNB can act as a master node. It is assumed that the dual connectivity between LTE and NR supports the deployment scenario where LTE eNB is not required to be synchronized with NR gNB. 3GPP NR supports dual connectivity operation in which a UE in the connected mode is con- figured to utilize radio resources provided by two distinct schedulers, located in two gNBs connected via a non-ideal backhaul as shown in Fig. 1.73. The gNBs involved in the dual connectivity for a certain UE may assume two different roles, that is, a gNB may either act as a master gNB (MgNB) or as a secondary gNB (SgNB). Under this condition, a UE is con- nected to one MgNB and one SgNB. Under dual connectivity framework, the radio protocol stack that a radio bearer uses depends on how the radio bearer is setup. There are four bearer types in dual connectivity framework, namely MCG70 bearer, MCG split bearer, SCG7¹ bearer, and SCG split bearer as depicted in Fig. 1.74. Under dual connectivity framework, the UE is

Split bearer in multi-RAT dual connectivity is a bearer whose radio protocols are split either at the master node or at the secondary node and belongs to both secondary cell group and master cell group. 154 Chapter 1 split split bearer bearer bearer bearer Figure 1.74 MgNB and SgNB bearers for dual connectivity [16]. links (or both) that it transmits uplink PDCP PDUs. The RLC sublayer only transmits corre- sponding feedback for the downlink data over the link which is not used for transmitting PDCP PDUs. Multi-RAT DC (MR-DC) is a generalization of the intra-LTE dual connectivity, where a UE with multiple RF transceivers may be configured to utilize radio resources provided by two distinct schedulers located in two different nodes and connected via non-ideal backhaul, one providing LTE access and the other one providing NR connectivity. One scheduler is located in the master node (MN) 73 and the other in the secondary node (SN). 74 The MN and SN entities are connected via a network interface and the MN is typically connected to the core network. LTE supports MR-DC via E-UTRA-NR dual connectivity (EN-DC), in which a UE is connected to one eNB that acts as the MN and one gNB that acts as the SN. The eNB is connected to the EPC and the gNB is connected to the eNB via the X2 interface (i.e., the logical interface between LTE eNBs). In MR-DC scenarios, the UE has a single RRC state, based on the MN RRC state and a sin- gle control-plane connection toward the core network. Fig. 1.75 shows the control-plane architecture for MR-DC. Each radio node has its own RRC entity, which can generate RRC PDUs to be sent to the UE. The RRC PDUs generated by the SN can be transported via the MN to the UE. The MN always sends the initial SN RRC configuration via MCG SRB, for example, SRB1, but subsequent reconfigurations may be sent via the MN or the SN entities. When transporting RRC PDU from the SN, the MN does not modify the UE configuration Master node in a multi-RAT dual connectivity architecture is either a master eNB or a master

gNB. Secondary node in multi-RAT dual connectivity architecture is either a secondary eNB or a secondary gNB. 5G Network Architecture Master node Secondary NR RRC RRC (MeNB RRC (master state) node state) EN-DC MR-DC Figure 1.75 Control-plane architecture for EN-DC and MR-DC with 5GC [13]. provided by the SN. In EN-DC and NG-RAN E-UTRA-NR dual connectivity (NGEN- DC75) scenarios, during initial connection establishment SRB1 uses LTE PDCP; however, after initial connection establishment MCG SRB (SRB1 and SRB2) can be configured by the network to use either LTE PDCP or NR PDCP. The PDCP version change (release of old PDCP and establishment of new PDCP) of SRBs can be supported via a handover pro- cedure (reconfiguration with mobility) or through a reconfiguration without mobility, when the network is aware that there is no uplink data in the UE buffer. For EN-DC capable UEs, NR PDCP can be configured for DRBs and SRBs before EN-DC is configured. If the SN is a gNB (i.e., for EN-DC and NGEN-DC), the UE can be configured to establish an SRB with the SN (e.g., SRB376 to enable RRC PDUs for the SN to be sent directly between the UE and the SN. The RRC PDUs for the SN can only be sent directly to the UE for SN RRC reconfiguration without any coordination with the MN. Measurement reporting for mobility within the SN can be conducted directly from the UE to the SN, if SRB3 is config- ured. The MCG split SRB is supported for all MR-DC cases, allowing duplication of RRC PDUs generated by the MN, via the direct path and through the SN. The MCG split SRB uses NR PDCP. The SCG split SRB is not currently supported in 3GPP specifications [13]. NG-RAN supports NGEN-DC, in which a UE is connected to one ng-eNB that acts as a master node and one gNB that acts as a secondary node. The ng-eNB is connected to the 5GC and the gNB is connected to the ng-eNB via the Xn interface. SRB3 in EN-DC and NGEN-DC represents a direct signaling radio bearer between the secondary node and the user equipment. 156 Chapter 1 bearer split split bearer

bearer bearer NR PDCP NR PDCP NR PDCP NR PDCP LTE RLC LTE RLC LTE RLC NR RLC NR RLC NR RLC LTE MAC NR MAC Figure 1.76 Radio protocol stack for MCG, MCG split, SCG, and SCG split bearers in MR-DC with EPC (EN-DC) [13]. MCG split bearer bearer split bearer bearer NR PDCP NR PDCP NR PDCP NR PDCP MN RLC MN RLC MN RLC SN RLC SN RLC SN RLC MN MAC SN MAC Figure 1.77 Radio protocol stack for MGC, MCG Split, SCG, and SCG split bearers in MR-DC with 5GC (NGEN-DC, NE-DC) [13]. As we mentioned earlier, there are four bearer types identified as MCG bearer, MCG split bearer, SCG bearer, and SCG split bearer in MR-DC scenarios. These four bearer types are depicted in Fig. 1.76 for MR-DC with EPC (EN-DC) and in Fig. 1.77 for MR-DC with 5GC (NGEN-DC, NE-DC). For EN-DC, the network can configure either LTE PDCP or NR PDCP for MCG bearers while NR PDCP is always used for SCG bearers. For split bearers, NR PDCP is always used and from the UE perspective there is no difference between MCG and SCG split bearers. In MR-DC with 5GC, NR PDCP is always used for all bearer types. From system architecture point of view, in MR-DC, there is an interface between the MN and the SN entities to facilitate control-plane signaling and coordination. For each MR-DC 5G Network Architecture S1-MME ((H)I) ((III) )) (III)) Figure 1.78 Control-plane and user-plane architecture for EN-DC and MR-DC with 5GC [13]. capable UE, there is also one control-plane connection between the MN and a correspond- ing core network entity. The MN and the SN entities involved in MR-DC operation for a UE control their own radio resources and are primarily responsible for allocating radio resources of their cells. Fig. 1.78 shows control-plane connectivity of an MN and SN involved in MR-DC with a UE. In MR-DC with EPC (EN-DC) scenario, the involved core network entity is the MME. S1-MME is terminated in MeNB and the MeNB and the SgNB are interconnected via X2-C. In MR-DC with 5GC (NGEN-DC, NE-DC) scenario, the ter- minating core network entity is the AMF. The NG-C interface i

s terminated at the MN and the MN and the SN are interconnected via Xn-C interface [13]. There are different user-plane connectivity options for the MN and SN involved in MR-DC operation with a certain UE, as shown in Fig. 1.78. The user-plane connectivity depends on the configured bearer type. For MCG bearers, the user-plane connection to the core network entity is terminated at the MN. The SN is not involved in the transport of user-plane data for this type of bearer over the Uu interface (i.e., the radio interface). For MCG split bearers, the user-plane connection to the core network entity is terminated in the MN. 158 Chapter 1 PDCP data is transferred between the MN and the SN via MN-SN user-plane interface. The SN and MN participate in transmitting data of this bearer type over Uu interface. For SCG bearers, the SN is directly connected to the core network entity via a user-plane interface. The MN is not involved in the transport of user-plane data for this type of bearer over Uu interface. For SCG split bearers, the user-plane connection to the core network entity is ter- minated at the SN. The PDCP packets are transferred between the SN and the MN via MN/ SN user-plane interface. The SN and MN transmit data of this bearer type over the Uu inter- face. For MR-DC with EPC (EN-DC), X2-U interface is the user-plane interface between MeNB and SgNB, and S1-U is the user-plane interface between MeNB and SGW. For MR- DC with 5GC (NGEN-DC, NE-DC), Xn-U interface is the user-plane interface between MN and SN, and NG-U is the user-plane interface between MN and UPF [13]. 1.4 LTE-NR Interworking and Deployment Scenarios 1.4.1 RAN-Level and CN-Level Interworking To provide full 5G services to the users, 5G cells will have to be deployed with full cover- age and all UEs will have to be able to connect to 5G network everywhere. However, in the early phases of 5G deployments, the 5G cells will be partially deployed and there will be 5G coverage gaps. Large-scale 5G commercial services and deployments are expected in 2020

+ and the initial investments for 5G service are expected to be limited due to lack of 5G user equipment. As a result, the 5G networks need to be able to interwork with the exist- ing LTE networks. The interworking solution can provide seamless service to the users. In this section, we discuss the solutions for LTE-NR interworking and we will compare the solutions in terms of performance, features and ease of migration to a full 5G network. If 5G cells are not deployed with full coverage, a seamless service can be provided to the users by interworking with the existing LTE networks, which are already deployed with full coverage. For LTE-NR interworking, two types of solutions namely access-network-level interworking and core-network-level interworking, have been studied in 3GPP. In RAN-level interworking solutions, the interworking service between LTE and 5G is made possible by using a direct interface between LTE eNB and NR gNB. The RRC mes- sages are transmitted over the LTE radio interface, thus the connection and the mobility of UE are controlled by the LTE eNB. The user traffic is simultaneously transmitted through LTE eNB and NR gNB either by PDCP aggregation or using NR gNB split bearer. Although the RRC messages can be processed by the LTE eNBs that provide larger cover- age than NR gNBs, LTE radio interface always remains connected, even though user traffic is transmitted over the NR. RAN-level interworking is necessary in non-stand-alone archi- tecture, where the NR cannot be used without overlaid LTE network. Note that when the LTE EPC is used, only EPC-based services can be provided, even though 5G radio 5G Network Architecture 159 technology is used. Two different core networks can be used for RAN-level interworking, as shown in Fig. 1.79. LTE and NR interworking can be achieved by upgrading some LTE eNBs connected to NR gNBs and by increasing the gateway capacity in EPC. The new 5G core network, 5GC, has been designed to support RAN-level interworking. In this solution, the new 5GC network slicing

feature can be used to separate 5G services from the LTE ser- vices. In this case, all LTE eNBs will have to be upgraded to ng-eNBs SO that they can be connected to 5GC [16]. In core-network-level interworking, a direct interface between the LTE eNB and the NR gNB is not required and the EPC SGW is connected to the 5GC UPF. The UE manages LTE and NR interface connection independently, and can be connected to a single network, either LTE or 5G. When the UE is located in 5G coverage, it can only connect to the 5G network and receive 5G service. When the UE moves out of 5G coverage, it releases NR interface connection and establishes LTE radio interface connection. Although the network to which the UE connects changes, the IP address assigned to the UE stays the same and seamless service can be provided to the user. The core-network-level interworking is neces- sary in stand-alone architecture models, where NR can be used without relying on LTE net- work. In this case, single registration or dual registration is possible, as shown in Fig. 1.79. With the single registration, the UE registers with either LTE EPC or 5G networks at any time, and the UE context can be transferred via the control interface between the EPC MME and 5GC AMF when the UE's network association changes. In order to support the single registration, the MME will have to be upgraded in order to support the MME-AMF interface and the SGW needs to be connected to UPF in 5GC. LTE eNB must also be upgraded to support the mobility between the LTE and the 5G networks. N26 interface is an inter-CN interface between the MME and AMF in order to enable interworking between EPC and the 5G core. Support of N26 interface in the network is optional for interworking. N26 supports a subset of the functionalities that are essential for interworking. Networks that support interworking with EPC may support interworking procedures that use the N26 interface or interworking procedures that do not use the N26 interface. Interworking proce- dures with N26 support provide

IP address continuity during inter-system mobility to UEs that support 5GC NAS and EPC NAS. Networks that support interworking procedures with- out N26 must support procedures to provide IP address continuity during inter-system mobility to the UEs that operate in both single-registration and dual-registration modes [3]. Interworking procedures using N26 interface enable the exchange of MM and SM states between the source and target networks. Handover procedures are supported through N26 interface. When interworking procedures with N26 is used, the UE operates in single- registration mode. The network retains only one valid MM state for the UE, either in the AMF or MME. Either the AMF or the MME is registered in the HSS + UDM. The support for N26 interface between AMF in 5GC and MME in EPC is required to enable seamless session continuity (e.g., for voice services) for inter-system handover. When the UE moves ((<(H))) (foy) (<(H)) Control plane User plane 4G/5G NSA UE 4G/5G NSA UE 4G/5G SAUE 4G/5G RAN-level interworking 4G/5G SAUE 4G/5G RAN-level interworking 4G/5G CN-level interworking with connected to EPC 4G/5G CN-level interworking with connected to 5GC single registration dual registration Figure 1.79 RAN-level and core-network-level interworking models [82]. 5G Network Architecture 161 from 5GC to EPC, the SMF determines which PDU sessions can be relocated to the target EPS. The SMF can release the PDU sessions that cannot be transferred as part of the hand- over. However, the target EPS determines if the PDU session can be successfully moved to the target network. The dual registration approach requires the UE to register separately with the EPC and the 5GC. Thus it does not need to forward the UE context between MME and AMF, and the interface between MME and AMF is not required. The handover between LTE and 5G systems is decided by the UE. The UE performs normal access procedures after moving to the other network. Therefore, the solution can be supported by LTE eNBs with minimal changes. Furthermore, the

impact on EPC to support the dual registration is minimized. However, in order to improve the mobility performance between LTE and 5G, it is necessary to temporarily support dual radio transmission when moving to the other network, although the UE can support dual registration solution even with single radio transmission capability. Deployments based on different 3GPP architecture options (i.e., EPC based or 5GC based) and UEs with different capabilities (EPC NAS and 5GC NAS) may coexist at the same time within one PLMN. In order to interwork with EPC, the UE that supports both 5GC and EPC NAS can operate in single- or dual-registration mode. In single-registration mode, the UE has only one active MM state (either RM state in 5GC or EMM state in EPC) and it is either in 5GC NAS mode or in EPC NAS mode (when connected to 5GC or EPC, respectively). The UE main- tains a single coordinated registration for 5GC and EPC. Accordingly, the UE maps the EPS-GUTI to 5G-GUTI during mobility between EPC and 5GC and vice versa. In dual- registration mode, the UE can handle independent registrations for 5GC and EPC. In this mode, UE maintains 5G-GUTI and EPS-GUTI independently. In this mode, the UE pro- vides native 5G-GUTI, if previously allocated by 5GC, for registration with 5GC and it pro- vides native EPS-GUTI, if previously allocated by EPC, for Attach/TAU with EPC. In this mode, the UE may be registered to 5GC only, EPC only, or to both 5GC and EPC. The sup- port of single registration mode is mandatory for UEs that support both 5GC and EPC NAS. During LTE initial attach procedure, the UE supporting both 5GC and EPC NAS indi- cates its support of 5G NAS in UE network capability [3]. The Ethernet and unstructured PDU session types are transferred to EPC as non-IP PDN type (when supported by UE and network). The UE sets the PDN type to non-IP when it moves from 5GS to EPS. After the transfer to EPS, the UE and the SMF maintain informa- tion about the PDU session type used in 5GS, that is, the information indicating that th