hash
stringlengths 32
32
| doc_id
stringlengths 7
13
| section
stringlengths 3
121
| content
stringlengths 0
3.82M
|
|---|---|---|---|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.0 Introduction
|
This use case provides the background, objectives, solution, and requirements for the NSSI resource optimization, an rApp implemented in Non-RT RIC, which leverages AI/ML inference on slice performance measurement data to determine the actions to automatically optimize the resource allocation for network slice instances.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.1 Background and goal of the use case
|
Network slicing is essential to 5G, as it enables many new services across manufacturing, autonomous driving, gaming, and many more via the provision of ultra-low latency in URLLC and huge data volume in eMBB features that require different or contrasting QoS requirements exploiting a shared RAN node. The goal of this use case is to ensure the resources are allocated dynamically and efficiently among multiple network slices sharing the RAN node. Time PRB usage Slice #1 Allocated PRB For slice #1 (S-NSSAI=1) Slice #n Time PRB usage Slice #1 Allocated PRB For slice #1 (S-NSSAI=1) Slice #n Decrease PRBs for Slice #1 via O1 (rRMPolicyRatio) Increase PRBs for Slice #1 via O1 (rRMPolicyRatio) Time PRB usage Slice #1 Allocated PRB For slice #1 (S-NSSAI=1) Slice #n Monitoring and Reconfiguration in Non-RT RIC Initial portion is calculated with “Wideband CQI distribution” and “Data volume in UL” Throughput deterioration Low PRB usage High PRB usage ETSI ETSI TS 104 226 V10.1.0 (2025-08) 48 As the new 5G services have different characteristics, the network traffic tends to be sporadic, where there can be different usage pattern in terms of time, location, UE distribution, and types of applications. For example, most IoT sensor applications can run during off-peak hours or weekends. Special events, such as sport games, concerts, can cause traffic demand to shoot up at certain time and locations. Cars with autonomous driving capability tend to require more URLLC services in the morning or afternoon rush hours in major freeways in big cities, while subscribers tend to consume eMBB services to watch video streaming at night in residential areas. Therefore, NSSI resource optimization rApp trains the AI/ML model, based on the huge volume of performance data collected over days, weeks, months from O-RAN nodes. It then performs inference function on the model with input measurements to predict the traffic demand patterns of 5G networks in different times and locations for each network slice, and automatically optimize the resource allocation for network slice instances accordingly.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.2 Entities/resources involved in the use case
|
1) Non-RT RIC: a) Receive measurements to monitor the usage of RRM resources (e.g. PRB, RRC, DRB) identified by S-NSSAI from E2 nodes via the O1 interface. b) Perform the model training with input measurements data received from E2 nodes to create the model. c) Perform the inference function on the model with the input measurements data to determine if any actions shall be executed to update the resources on the E2 nodes. d) Configure the resources at the E2 nodes via O1 interface. e) Receive notifications from E2 nodes indicating the resource re-configuration was done. f) Update the O-Cloud resources via the O2 interface. g) Receive notifications from O-Cloud indicating the resource was updated. 2) E2 nodes (O-CU-CP, O-CU-UP, D-DU): a) Support the collections and reporting of measurements that are used to monitor the resource usage on per network slice basis via the O1 interface. b) Support the re-configuration of attributes to update the resources allocated to each network slice via the O1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.3 Solutions
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.3.1 NSSI resource optimization
|
The context of the NSSI resource optimization is captured in table 4.6.3.1-1. Table 4.6.3.1-1: NSSI resource optimization Use Case Stage Evolution / Specification <<Uses>> Related use Goal The goal is to ensure the resources (e.g. PRB, RRC, DRB) are allocated dynamically and efficiently among multiple network slices sharing the E2 nodes. Actors and Roles • SMO functions. • Non-RT RIC framework. • rApp: NSSI resource optimization. • E2 nodes (i.e. O-CU-CP, O-CU-UP, O-DU). Assumptions • All relevant functions and components are instantiated. • O1 interface connectivity is established. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 49 Use Case Stage Evolution / Specification <<Uses>> Related use Pre-conditions • O1 interfaces have been established to enable SMO to receive measurements from E2 nodes and configure the E2 nodes. • R1 interface has been established to enable the rApp to receive measurements form E2 nodes and configure the E2 nodes via Non-RT RIC framework. • E2 nodes have been configured to collect the measurements and send them to Non-RT RIC framework. Begins when The rApp utilizes the model to perform the inference function. Step 1 (M) The rApp performs the offline model training with input measurements data received from E2 nodes to create the model. Step 2 (M) Non-RT RIC framework receives the measurements from O-CU-CP via O1 to monitor the usage of RRM resources (e.g. RRC connected user). Step 3 (M) Non-RT RIC framework sends the measurements to rApp via R1 interface. Step 4 (M) Non-RT RIC framework receives the measurements from O-CU-UP via O1 to monitor the usage of RRM resources (e.g. the number of DRB allocated, and the number of PDU sessions). Step 5 (M) Non-RT RIC framework sends the measurements to rApp via R1 interface. Step 6 (M) Non-RT RIC framework receives the measurements from O-CU-UP via O1 to monitor the usage of RRM resources (e.g. the number of PRBs used in the downlink and uplink data traffic). Step 7 (M) Non-RT RIC framework sends the measurements to rApp via R1 interface. Step 8 (M) The rApp performs the inference function based on the model with input measurements data received to determine the actions to update the resources allocated to slices on the E2 nodes if needed. If the rApp decides the RRM resources (e.g. RRC) in O-CU-CP need to be updated, then steps 9 to 12 are executed. Step 9 (O) rApp requests Non-RT RIC framework via R1 interface to update the RRM resources for slices in O-CU-CP. Step 10 (O) Non-RT RIC framework uses the modify MOI (Managed Object Instance) operation to configure the MOI(s) associated with the RRM resources at O- CU-CP via O1 interface. Step 11 (O) Non-RT RIC framework receives a notification from O-CU-CP via O1 interface indicating the resource re-configuration was successful. Step 12 (O) Non-RT RIC framework notifies rApp via R1 interface indicating the RRM resources in O-CU-CP have been successfully updated. If the rApp decides the RRM resources (e.g. DRB) in O-CU-UP need to be updated, then steps 13 to 16 are executed. Step 13 (O) rApp requests Non-RT RIC framework via R1 interface to update the RRM resources for slices in O-CU-UP. Step 14 (O) Non-RT RIC framework uses the modify MOI operation to configure the MOI(s) associated with the RRM resource at O-CU-UP via O1 interface. Step 15 (O) Non-RT RIC framework receives a notification from O-CU-UP via O1 interface indicating the resource re-configuration was successful. Step 16 (O) Non-RT RIC framework notifies rApp via R1 interface indicating the RRM resources in O-CU-UP have been successfully updated. If the rApp decides the RRM resources (e.g. PRB) in O-DU need to be updated, then steps 17 to 20 are executed. Step 17 (O) rApp requests Non-RT RIC framework via R1 interface to update the RRM resources for slices in O-DU. Step 18 (O) Non-RT RIC framework uses the modify MOI operation to configure the MOI(s) associated with the RRM resource at O-DU via O1 interface. Step 19 (O) Non-RT RIC framework receives a notification from O-DU via O1 interface indicating the resource re-configuration was successful. Step 20 (O) Non-RT RIC framework notifies rApp via R1 interface indicating the RRM resources in O-DU have been successfully updated. If the rApp decides the O-Cloud resources need to be updated, then steps 21 to 24 are executed: Step 21 (O) rApp requests Non-RT RIC framework via R1 interface to update the O-Cloud resources. Step 22 (O) Non-RT RIC framework re-configures the O-Cloud resources via O2 interface. Step 23 (O) Non-RT RIC framework receives a notification via O2 interface indicating the resource re-configuration was successful. Step 24 (O) Non-RT RIC framework notifies rApp via R1 interface indicating the O-Cloud resources have been successfully updated. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 50 Use Case Stage Evolution / Specification <<Uses>> Related use Ends when The resources have been optimized. Exceptions None identified. Post Conditions None. Traceability REQ-R1-FUN9, REQ-R1-FUN10. NOTE: How the O-Cloud resources are to be monitored and updated is not defined in the present document. The flow diagram of the NSSI resource optimization is given in figure 4.6.3.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 51 Figure 4.6.3.1-1: NSSI resource optimization flow diagram ETSI ETSI TS 104 226 V10.1.0 (2025-08) 52
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.4 Required data
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.4.0 Introduction
|
This clause contains the input and output data of model training and inference.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.4.1 Input data
|
The measurement input data are used in model training and inference. They include the following measurements to monitor the resource usage for network slices in E2 nodes: 1) Measurements used to monitor the usage of RRC related resources in O-CU-CP include: - Mean number of RRC connections - provides the mean number of RRC connections with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.4.1). - Peak number of RRC connections - provides the peak number of RRC connections with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.4.2). 2) Measurements used to monitor the usage of DRB related resources in O-CU-UP include: - Mean number of DRBs being allocated - provides the mean number of DRBs being allocated in the PDU sessions with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.10.10). - Peak number of DRBs being allocated - provides the peak number of DRBs being allocated in the PDU sessions with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.10.9). 3) Measurements used to monitor the usage of PRB related resources in O-DU include: - Mean DL PRB used for data traffic - provides the mean number of PRBs used in downlink for data traffic with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.2.5). - Peak DL PRB used for data traffic - provides the peak number of PRBs used in downlink for data traffic with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.2.9). - Mean UL PRB used for data traffic – provides the mean number of PRBs used in uplink for data traffic with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.2.7). - Peak UL PRB used for data traffic - provides the peak number of PRBs used in uplink for data traffic with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.2.10). - Mean number of PDU sessions being allocated - provides the mean number of PDU sessions being allocated with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.5.4). - Peak number of PDU sessions being allocated - provides the peak number of PDU sessions being allocated with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.5.5). - Mean number of active UEs in the DL per cell - provides the mean number of active UEs in downlink with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.23.1). - Maximum number of active UEs in the DL per cell - provides the maximum number of active UEs in downlink with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.23.2). - Mean number of active UEs in the UL per cell - provides the mean number of active UEs in uplink with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.23.3). - Maximum number of active UEs in the UL per cell - provides the maximum number of active UEs in uplink with sub-counters per S-NSSAI (as specified in 3GPP TS 28.552 [5], clause 5.1.1.23.4). ETSI ETSI TS 104 226 V10.1.0 (2025-08) 53
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.4.2 Output data
|
The output data, including NRCellCU IOC, NRCellDU IOC, GNBDUFunction IOC, GNBCUCPFunction IOC, GNBCUUPFunction IOC and RRMPolicyRatio IOC with RRMPolicy abstract class (as specified in 3GPP TS 28.541 [4]), are needed to enable NSSI resource optimization rApp to re-configure the resources via O1 and O2 interfaces.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.6.5 O1 usage example
|
An example of two NSSIs, where NSSI#1 groups E2 nodes (i.e. O-DU, O-CU-CP, and O-CU-UP), and NSSI#2 groups 5GC NFs is shown in figure 4.6.5-1. It also shows that two network slices, identified by S-NSSAI#1 supporting URLLC, and S-NSSAI#2 supporting eMBB. The goal of this use case is to optimize the resources associated with RAN network slices. O-DU O-CU-UP O-CU-CP 5GC S-NSSAI#1 S-NSSAI#2 S-NSSAI#1 S-NSSAI#1 S-NSSAI#2 S-NSSAI#2 NSSI#1 NSSI#2 Control plane User plane Figure 4.6.5-1: NSSI resource optimization example NSSI resources optimization rApp runs model inference with input measurement data collected from E2 nodes for S-NSSAI#1 and S-NSSAI#2, and detects a traffic pattern for O-DU serving an area with high density of business and residential users at the time on a given day. An example of PRB resource allocation for S-NSSAI#1 and S-NSSAI#2 at the O-DU is shown in figure 4.6.5-2: • At 15:00, the dedicated resources and prioritized resources for S-NSSAI#1 were increased to 20 % and 50 % respectively for as more cars demand more URLLC services at the start of rush hours. • At 17:00, the dedicated resources, prioritized resources, and shared resources for S-NSSAI#1 were further increased as the rush hours traffic getting worse. • At 19:00, the dedicated resources, prioritized resources, and shared resources for S-NSSAI#2 were increased to 20 %, 60 %, and 75 % respectively as more residential users demand more eMBB services for home video streaming. • At 20:00, the dedicated resources, prioritized resources, and shared resources for S-NSSAI#1 were decreased as the rush hours traffic coming to end. • At 21:00, the dedicated resources, prioritized resources, and shared resources for S-NSSAI#2 were further increased as the demand for eMBB services increased. • At 22:00, the dedicated resources, prioritized resources, and shared resources for S-NSSAI#1were decreased as the demand for URLLC services further reduced. • At 24:00, the dedicated resources, prioritized resources, and shared resources for S-NSSAI#2 were decreased to 10 %, 25 %, and 60 % respectively as the demand for eMBB services further reduced. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 54 0 100 0 100 13:00 Shared resources 0 100 MaxRatio 0 100 0 100 0 100 0 100 0 100 MinRatio DedicatedRatio 0 100 15:00 17:00 19:00 20:00 21:00 22:00 24:00 S-NSSAI#1 S-NSSAI#2 10 20 25 75 50 75 80 60 30 45 15 60 30 10 45 10 25 60 20 60 75 25 65 80 10 25 60 Time MaxRatio MinRatio DedicatedRatio Prioritized resources Dedicated resources Figure 4.6.5-2: Example of network slice resource allocations for O-DU
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7 Use case 7: Massive MIMO optimization use cases
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.0 Introduction
|
Massive MIMO optimization is one of the top priority use cases in O-RAN. Several massive MIMO sub-features have been proposed and studied during the massive MIMO pre-normative study, which is documented in the O-RAN.WG1.MMIMO-USE-CASES-TR-v00.13 [i.3], including the potential data requirements for each of the sub-use cases. The following clauses provide the background, objectives, solutions, deployment options, and identified WG2 requirements for massive MIMO sub-features
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.1 Massive MIMO Grid-of-Beams Beamforming (GoB BF) optimization
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.1.1 Background and goal of the use case
|
Massive MIMO (mMIMO) is among the key methods to increase performance and QoS in 5G networks. Capacity enhancement is obtained by means of beamforming of the transmitted signals, and by spatially multiplexing data streams. Beamforming can increase the received signal power and simultaneously decrease the interference generated for other users, hence resulting in higher SINR and higher user throughputs. Grid-of-Beams (GoB) with the corresponding beam sweeping has been introduced to allow beamforming of the control channels used during initial access as well as for data transmission and reception, mainly for high frequency (but can be used also for the sub-6 GHz band) MIMO operation. The physical properties of the antenna array and its possible configurations characterize the span of the beams, namely the horizontal and vertical aperture in which beamforming is supported, and therefore the coverage area and the shape of the cell. mMIMO can be deployed in 5G macrocell clusters as well as in heterogeneous networks, where macrocells and small cells co-exist and complement each other for better aggregated capacity and coverage. In order to obtain optimal beamforming and cell resources (Tx power, PRB) configuration, one will have to look at a multi-cell environment instead of a single cell. Moreover, different vendors can have different implementations in terms of the number of beams, the horizontal/vertical beam widths, azimuth and elevation range, to achieve the desired coverage. In a multi-node/multi-vendor scenario, centralized monitoring and control is required to offer optimal coverage, capacity and mobility performance as well as control over electromagnetic emissions in order to comply with regulatory requirements. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 55 The problem associated with traditional mMIMO BF is that its performance is highly dependent on the choice of the Beam Forming (BF) pattern. Manual configuration is usually based on the empirical knowledge and manual test results of the domain expert(s) and is performed in a semi-static way. That is, (near-)real time contextual, per-site information (such as cell geometry change, user/traffic distribution, mobility patterns, seasonalities, etc.) is taken into account in a suboptimal and non-real-time way. This can cause one or more of the following problems: 1) High inter-cell interference. 2) Unbalanced traffic between neighbouring cells. 3) Low performance at the cell edges or throughout the cell. 4) Poor handover performance. This solution proposes a framework that allows the operator to flexibly configure the mMIMO BF parameters in a cell or in a cluster of cells by means of policies and configuration assisted by Machine Learning (ML) techniques. The configuration optimization relies on contextual information and patterns such as the user distribution, traffic demand distribution, cell geometries, and mobility.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.1.2 Entities/resources involved in the use case
|
1) SMO & Non-RT RIC Framework (FW): a) Collect the necessary configurations, performance indicators, and measurement reports from the E2 nodes (O-DU), triggered by Non-RT RIC FW if required. b) Transfer collected data towards rApp. c) Provide optimized mMIMO GoB parameters via O1 (to O-DU) or open FH M-plane (to O-RU) interface. d) Optional: Retrieve necessary enrichment information (UE location related information, e.g. GPS coordinates) for the purpose of i) training relevant rApps and ii) execution of relevant rApps. NOTE 1: Exposure of enrichment information to rApps is not defined in the present document. e) Monitor the performance of the respective cells; when the optimization objective fails, initiate fallback procedure and/or trigger the rApp model retraining and re-optimization. f) Execute the inference/control loop periodically or event-triggered. g) Optional: The ML model training can be done by the Non-RT RIC FW. 2) rApps: a) Retrieve the necessary configurations, performance indicators, and measurement reports from the E2 nodes and necessary enrichment information via the SMO, for the purpose of training and execution of relevant AI/ML models. b) Infer an optimized GoB BF configuration. 3) E2 nodes & O-RU: a) Collect and provide necessary measurements and KPIs to the SMO (see Required data clause). b) Apply mMIMO GoB configuration received from the SMO. NOTE 2: Both aggregated and disaggregated gNB architecture are supported.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.1.3 Solutions
|
The context of the creation and deployment of mMIMO GoB BF optimization applications is captured in table 4.7.1.3-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 56 Table 4.7.1.3-1: Creation and deployment of mMIMO GoB BF optimization applications Use Case Stage Evolution / Specification <<Uses>> Related use Goal Optimized beamforming configuration with the Grid-of-Beams method. Actors and Roles SMO, Non-RT RIC, E2 nodes, O-RU. Assumptions All relevant functions and components are instantiated. O1 and OFH-MP interface connectivity is established. Pre-conditions Near-RT RIC and Non-RT RIC are instantiated. O1 interface is established between SMO and Near-RT RIC and E2 nodes, OFH-MP is established between O-DU(s) and O-RU(s). Begins when GoB BF optimization rApp with initial ML model is deployed. Step 1 (M) SMO/Non-RT RIC FW collects the necessary configurations, performance indicators, and measurement reports from E2 nodes (O-DU). Step 2 (O) SMO/Non-RT RIC FW collects input data from external apps. Step 3-6 (M) Collected data is transferred to rApp from the SMO/Non-RT RIC FW and rApp trains the necessary ML model(s). Step 7-10 (M) A new optimization trigger is applied or re-optimization of the GoB BF is necessary due to low performance. ML model assisted rApp infers optimized GoB BF configuration and transfers it to the SMO/Non-RT RIC. Step 11-13 (M) SMO/Non-RT RIC FW applies the optimized GoB BF configuration via O1 or via O1 and OFH-MP. Step 14-20 (O) SMO/Non-RT RIC FW continuously monitors GoB BF performance in respective cells. Optionally, it initiates fallback in case performance is unsatisfactory and requests ML model retraining/update. Then, rApp retrains/updates the respective ML model(s). Ends when On operator request of rApp is disabled. Exceptions None identified. Post Conditions GoB BF configuration is active. Traceability REQ-Non-RT-RIC-FUN1, REQ-Non-RT-RIC-FUN5, REQ-Non-RT-RIC- FUN6, REQ-Non-RT-RIC-FUN8, REQ-Non-RT-RIC-FUN9 The flow diagram of GoB BF optimization is given in figure 4.7.1.3-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 57 Figure 4.7.1.3-1: Flow diagram of GoB BF optimization ETSI ETSI TS 104 226 V10.1.0 (2025-08) 58
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.1.4 Required data
|
The specification of the data communicated over O1 is outside the scope of WG2. There are no data that are relevant for the A1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2 Massive MIMO Non-GoB Beamforming (Non-GoB BF) optimization
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.0 Introduction
|
This use case provides the background and motivation for the O-RAN architecture to support Non-Grid of Beams beamforming optimization. Non-RT RIC could be used to train AI/ML models for Non-GoB BF selection xApps, which intelligently recommend best Non-GoB BF modes to a O-gNB or O-DU. Note that non-AI/ML based solutions for Non-GoB BF optimization is not precluded. The AI/ML model training, deployment, and inference described below are not applicable to non-AI/ML based solutions.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.1 Background and goal of the use case
|
Non-GoB BF approaches are an important class of beamforming algorithms for 5G massive MIMO deployments, especially for implementations in sub-6 GHz frequency bands. For example, beam weights can be computed at the O-gNB or O-DU based on channel measurements of Sounding Reference Signals (SRS) without predefined beam sets, assuming uplink and downlink correspondence. Noted that Non-GoB BF modes are not standardized, instead they are vendor-specific proprietary algorithms. Multiple Non-GoB BF modes can be implemented, as some modes perform better than others under particular wireless conditions. Non-GoB BF modes can differ in the following aspects: • MIMO modes (i.e. SU-MIMO or MU-MIMO) • Channel estimation algorithms • Beam weight calculation approaches (e.g. Matched Filter (MF), Zero-Forcing (ZF), eigen-beamforming, etc.) • Time and frequency granularity of beamforming, etc. To select the best Non-GoB BF modes, the SMO/Non-RT RIC and the Near-RT RIC are not required to understand the details of a specific Non-GoB BF modes, e.g. how the beamforming weights are computed under a mode. They only need to know how many Non-GoB BF modes are supported in the O-DU and the performance of each mode. The goal of this use case is therefore to provide an intelligent control over multiple supported Non-GoB BF modes in order to recommend a preferred mode to a BS as a function of wireless conditions, such as channel quality, UE location and mobility, interference condition, PHY layer configuration, etc.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.2 Entities/resources involved in the use case
|
1) SMO/Non-RT RIC: a) Retrieve the number of supported Non-GoB BF modes in O-DU via the O1 interface. b) Retrieve performance measurement data and UE context information (e.g. SRS periodicity) from O-DU via the O1 interface, for each Non-GoB BF mode. c) Collect enrichment information from external sources such as application servers. d) Associate enrichment information with collected measurements and configurations. e) Perform model training. f) Perform model deployment. g) Perform model performance monitoring and model re-training. h) Send enrichment information to the Near-RT RIC for inference via the A1 interface. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 59 Note that the model can be trained in the Non-RT RIC framework or in the Non-GoB BF optimization rApp. 2) Near-RT RIC: a) Receive enrichment information via the A1 interface. b) Support AI/ML model deployment from the SMO/Non-RT RIC. c) Receive performance measurement data and UE context information (e.g. SRS periodicity) from O-DU via the E2 interface. d) Associate enrichment information with collected measurements and configurations. e) Select Non-GoB modes, e.g. by performing model inference. f) Send Non-GoB BF control/policy message to O-DU via the E2 interface. Note that the requirements of Near-RT RIC are under the scope of WG3. 3) O-DU: a) Send the number of supported Non-GoB BF modes to SMO/Non-RT RIC via the O1 interface. b) Send measurement data and UE context information (e.g. SRS periodicity) to SMO/Non-RT RIC via the O1 interface. c) Send measurement data and UE context information (e.g. SRS periodicity) to the Near-RT RIC via the E2 interface. d) Support Non-GoB control/policy message received from the Near-RT RIC via the E2 interface. Note that the requirements of O1 interface for O-DU are under the scope of WG5.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.3 Solutions
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.3.1 AI/ML-assisted Non-GoB BF mode selection
|
Note that data collection over E2 interface and E2 control/policy procedures shown in table 4.7.2.3.1-2 and in figure 4.7.2.3.1-2 are under the scope of WG3. Note that external interface between the Non-RT RIC and the external sources (e.g. application servers) is not specified by O-RAN. The context of the AI/ML-assisted Non-GoB BF mode selection - model training, deployment, and update is captured in table 4.7.2.3.1-1. Table 4.7.2.3.1-1: AI/ML-assisted Non-GoB BF mode selection - model training, deployment, and update Use Case Stage Evolution / Specification <<Uses>> Related use Goal To train an AI/ML model to select the best Non-GoB BF modes given wireless conditions and per-UE configurations. Actors and Roles SMO, Non-RT RIC, Near-RT RIC, O-DU, external sources, e.g. application server. Assumptions • All relevant functions and components are instantiated. Pre-conditions • O1 interface is established between SMO and O-DU to enable SMO/Non-RT RIC to obtain the number of supported Non-GoB BF modes and to collect performance measurement data and associated per-UE configuration. • A1 interface is established between Non-RT RIC and Near-RT RIC to enable enrichment information transfer. • O-DU supports Non-GoB BF. Begins when Operator specified trigger condition or event is satisfied. Step 1 (M) SMO requests the number of supported Non-GoB BF modes in O-DU via the O1 interface. Step 2 (M) SMO collects the number of supported Non-GoB BF modes in O-DU via the O1 interface. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 60 Use Case Stage Evolution / Specification <<Uses>> Related use Step 3 (M) Non-RT RIC retrieves collected information. Step 4 (M) For each Non-GoB BF mode, SMO requests performance measurement data and associated UE context information (e.g. SRS periodicity) from O-DU for model training via the O1 interface. Step 5 (M) SMO collects required performance measurement data and UE context information (e.g. SRS periodicity) from O-DU via the O1 interface. Step 6 (O) SMO collects enrichment information (e.g. UE mobility and location info) from external sources, e.g. application server. Step 7 (O) Non-RT RIC retrieves collected data and enrichment information. Step 8 (O) For each Non-GoB BF mode, Non-RT RIC associates received enrichment information with measurement data and UE context information. Step 9 (M) Non-RT RIC performs model training. Step 10 (M) Non-RT RIC requests to deploy the trained AI/ML model. Step 11 (M) SMO/Non-RT RIC deploys trained model to the Near-RT RIC via O1 or O2 interface. Step 12 (M) SMO requests performance measurement data, including the active Non- GoB BF mode index, from O-DU for performance monitoring via the O1 interface. Step 13 (M) SMO collects performance measurement data, including the active Non- GoB BF mode index, from O-DU for performance monitoring via the O1 interface. Step 14 (O) SMO collects enrichment information (e.g. UE mobility and location info) from external sources, e.g. application server. Step 15 (O) Non-RT RIC retrieves collected performance monitoring data and enrichment information. Step 16 (M) Non-RT RIC evaluates the performance of deployed AI/ML model. Step 17 (M) Non-RT RIC re-trains the model. Step 18 (M) Non-RT RIC requests to deploy the updated AI/ML model. Step 19 (M) SMO/Non-RT RIC updates model in the Near-RT RIC via O1 or O2 interface. Ends when Operator specified trigger condition or event is satisfied. Exceptions None identified. Post Conditions Near-RT RIC executes the deployed model for AI/ML-assisted Non-GoB BF. Traceability REQ-Non-RT-RIC-FUN1, REQ-Non-RT-RIC-FUN4, REQ-Non-RT-RIC- FUN5, REQ-Non-RT-RIC-FUN9, REQ-A1-FUN2 The flow diagram of the AI/ML-assisted Non-GoB BF mode selection - model training, deployment, and performance monitoring is given in figure 4.7.2.3.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 61 Figure 4.7.2.3.1-1: AI/ML-assisted Non-GoB BF mode selection - model training, deployment, and performance monitoring The context of the AI/ML-assisted Non-GoB BF mode selection - model inference is captured in table 4.7.2.3.1-2. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 62 Table 4.7.2.3.1-2: AI/ML-assisted Non-GoB BF mode selection – model inference Use Case Stage Evolution / Specification <<Uses>> Related use Goal To generate Non-GoB control/policy message. Actors and Roles SMO, Non-RT RIC, Near-RT RIC, O-DU, external sources, e.g. application server. Assumptions • All relevant functions and components are instantiated. Pre-conditions • A1 interface is established between Non-RT RIC and Near-RT RIC to enable enrichment information transfer. • E2 interface is established between Near-RT RIC and O-DU to enable Non-GoB BF mode selection via E2 control/policy message. • O-DU supports Non-GoB BF. Begins when Operator specified trigger condition or event is satisfied. Step 1 (O) The Near-RT RIC queries available EI type identifiers. Step 2 (O) The Non-RT RIC responds an array of identifiers of all available EI types. Step 3 (O) The Near-RT RIC queries the EI type to support Non-GoB BF inference (e.g. UE mobility/location info). Step 4 (O) The Non-RT RIC responds detailed information related to the queried EI type. Step 5 (O) The Near-RT RIC creates an EI job. Step 6 (O) The Non-RT RIC responds to the EI job creation request. Step 7 (O) SMO collects enrichment information (e.g. UE mobility/location info) from external sources, e.g. application server. Step 8 (O) Non-RT RIC retrieves collected enrichment information. Step 9 (O) Non-RT RIC delivers collected enrichment information as EI job results to the Near-RT RIC via the A1 interface. Step 9 (M) Near-RT RIC requests measurement data and UE context information (e.g. SRS periodicity) from O-DU via the E2 interface. Step 9 (M) Near-RT RIC collects measurement data and UE context information (e.g. SRS periodicity) from O-DU via the E2 interface. Step 9 (M) Near-RT RIC associates received enrichment information with collected measurement data and UE context information. Step 9 (M) Near-RT RIC selects the best Non-GoB BF mode, e.g. by performing model inference. Step 9 (M) Near-RT RIC generates Non-GoB control/policy message based on inferred Non-GoB BF mode selection. Step 9 (M) Near-RT RIC sends Non-GoB control/policy message to O-DU via the E2 interface. Ends when Operator specified trigger condition or event is satisfied. Exceptions None identified. Post Conditions Non-RT RIC monitors the performance of AI/ML-assisted Non-GoB BF mode selection in the Near-RT RIC. Traceability REQ-Non-RT-RIC-FUN9, REQ-A1-FUN3 The flow diagram of the AI/ML-assisted Non-GoB BF mode selection - inference is given in figure 4.7.2.3.1-2. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 63 Figure 4.7.2.3.1-2: AI/ML-assisted Non-GoB BF mode selection - inference
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.4 Required data
|
The specification of the data communicated over O1 is outside the scope of WG2. The following enrichment information from external sources (e.g. application server) are used in model training and inference: • UE location • UE mobility • Time granularity of the enrichment information reports (e.g. integer multiple of a second) Note that for model inference, above EI is sent from Non-RT RIC to Near-RT RIC via the A1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.2.5 A1 enrichment information example
|
In training phase, the retrieved enrichment information (e.g. UE mobility and location information) needs to be associated with collected per-UE L1/L2 measurement reporting (e.g. L1-RSRP and/or L1-SINR, etc.) and UE context information (e.g. UE-specific SRS periodicity) by the Non-RT RIC. In the inference phase, such data association is performed by the Near-RT RIC. Therefore, the EI delivered over the A1 interface should contain necessary UE identification to facilitate the data association at the Near-RT RIC. The Near-RT RIC shall be able to recognize the UE identification and be able to map it to the UE identification used over the E2 interface. For example, the A1 enrichment information contains the following information elements: • UE identifier • Position of the UE • Height of the UE ETSI ETSI TS 104 226 V10.1.0 (2025-08) 64 • Time stamp when the position and height was recorded 4.7.3 MIMO optimization via MIMO DL Tx power optimization, MU-MIMO pairing, and MIMO mode selection
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.0 Introduction
|
This use case will provide the objective, solutions, and data requirements related to MIMO optimization based on three key sub-features involving downlink transmit power, MIMO pairing enhancement (user separability), and user MIMO mode selection (MU-MIMO or SU-MIMO) that are described in detail in the O-RAN.WG1.MMIMO-USE-CASES- TR-v00.13 [i.3]. The use-case leverages Non-RT RIC to train and host the relevant models and applications that rely on O1 interface services to intelligently optimize MIMO capacity and user experience.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.1 Background and goal of the use case
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.1.1 MIMO downlink transmit power optimization
|
For general downlink precoding, the downlink transmit power is usually evenly distributed across the UEs. However, depending on the UE separability and path loss deltas, this can result in good cell capacity at the expense of individual UE quality. This can be due to several issues such as cell edge UEs having general downlink SINR issues (even without MU-MIMO), poor UE separability between cell edge UEs, and poor uplink SINR resulting in degraded SRS which are a few example issues. The result of these issues can be manifested by observations of very poor individual UE SINRs (either downlink, uplink, or both) when running in a MU-MIMO mode. Therefore, although the capacity of the cell has been significantly increased, certain customer experiences can become unacceptable in this MU-MIMO mode. The solution to the problem described above is to simply provide observations of UE performance in the form of periodic histograms of UE channel quality as well as the overall cell capacity in order to compute an optimal solution via AI/ML with control of the downlink minimum required SINR threshold to achieve a minimal UE quality requirement that is set by the operator. The minimum required SINR is a threshold recommendation and thus does not require real time AI/ML adjustment of transmit power directly but rather leaves this to the scheduler to adjust and optimize consistent with its numerous other priorities and requirements. The value of this observability and adjustability allows the operator to optimize the trade-off between cell capacity and individual user/customer quality which is essential to provide the best customer experience. The trade-off, for example, can reduce a very high cell centre data rate (which would likely be unnoticeable for the user) to allow more power to be allocated to the cell edge user (who is noticing low throughput and large latencies) to improve the cell edge data rate situation.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.1.2 MU-MIMO pairing enhancement (user separability)
|
Existing channel orthogonality between multiple users is critical to create user separability and allow for the opportunity to share radio frequency resources simultaneously. Failing this, residual interference will be too high to maintain adequate post pairing radio link signal quality levels required to sustain MU-MIMO mode assignments. With mobility there is an added demand to adjust beamforming weight assignments to not only maintain signal power levels at the user end (beam quality), but also to continuously limit the inter user interference experienced between users assigned with the same radio resource allocations. If these challenges are left unaddressed, a 5G massive MIMO deployment will fail to utilize the full capability of large antenna arrays powered by transceivers designed to transmit data channel signals towards a spatially confined direction. Further, the network will also fail to realize potential multiplexing gains as fewer radio resource blocks are shared between users within the same cell, reducing spectral efficiency. Another important aspect is the need to efficiently identify users with low demand for radio resources - sources of bursty traffic. An intelligent assessment of how best such users can be effectively paired, if at all, with other users, needs to be pre-determined by the RIC. In summary, this use case suggests various measurement objects that are recommended as input into the AI/ML analytics Apps to optimally determine the outputs required to optimize the MU-MIMO feature operation. The AI/ML assisted modelling and training output, along with the Non-RT RIC based enhancement/inference, will strive to deliver end goal solution selections and system configuration options that upon adoption within the respective domains where they reside, realize an optimization framework that maximizes the potential of a MU-MIMO feature. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 65 Capacity augmentation will be realized by successfully assigning MU-MIMO layers to a greater number of users simultaneously, more often, and more uniformly across the serving area of each gNB.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.1.3 MIMO mode selection optimization (MU-MIMO vs SU-MIMO selection)
|
A successful MU-MIMO operation involves the realization of as many orthogonal radio frequency channel links between multiple spatially separated users as possibly as supported by the implementation software at the digital domain. Key to such realization is the successful beamforming weight determination that enables not only the phase addition of multipath signals at the user receiver, but also the choice of precoding algorithms which limit the residual interference between the paired users. It can make sense for the scheduler to prioritize the assignment of radio resources to a MU-MIMO mode of operation during periods of congestion or when high latency requiring applications are supported (to free up other resources that can be assigned sooner). However, doing so at the expense of undesirably lower spectral efficiency on these assigned radio resources will reduce overall sector throughput levels and create poor user experience. It is important to find a means through the AI/ML agent to distinguish users and identify sectors where optimal operation means a greater assignment of SU-MIMO modes independently to users, especially those requiring higher throughput, using devices that are capable of supporting higher layer SU-MIMO count, and operating in an environment that sustains a greater channel rank. With increased loading massive MIMO systems will incur rising levels of interference on the uplink from connected users and on the downlink from the gNB. In addition to normal SINR measurements, the diagnosis of interference from all spatial directions uniformly (white spatial noise) versus specific directions (spatially correlated noise) will be of interest and will require MIMO modes (SU-MIMO vs MU-MIMO) to be properly selected for assignment on a user basis. Such implementation will optimize the per user and per cell throughputs, taking into consideration channel orthogonality conditions rank realizable, and per user effective bandwidth requirement.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.2 Entities/resources involved in the use case
|
1) SMO/Non-RT RIC: a) Retrieve relevant performance measurement data and RAN configurations from O-DU via the O1 interface. b) Perform model training and model deployment based on identified measurement data. c) Perform model performance monitoring and model re-training as required. d) Provide RAN configuration recommendations based on identified parameters to O-DU over O1 interface. e) Allow rApps to access the measurement data and to provide configuration recommendations via relevant R1 interface services. 2) O-DU: a) Send measurement data and RAN configurations to SMO/Non-RT RIC via the O1 interface. b) Support implementation of MIMO configuration parameters received from the SMO/Non-RT RIC via the O1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.3 Solutions
|
4.7.3.3.1 MIMO optimization via DL SINR threshold, MU-MIMO pairing, and MIMO mode selection The context of the MIMO optimization via DL Tx power, pairing enhancement, and mode selection is captured in table 4.7.3.3.1-1. Table 4.7.3.3.1-1: MIMO optimization via DL Tx power, pairing enhancement, and mode selection Use Case Stage Evolution / Specification <<Uses>> Related use Goal To train and deploy AI/ML models for MIMO optimization that given wireless conditions and RAN configuration information as input will generate configuration recommendations for DL SINR threshold, MU-MIMO user pairing, and MIMO mode selection. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 66 Use Case Stage Evolution / Specification <<Uses>> Related use Actors and Roles SMO, Non-RT RIC, O-DU Assumptions • All relevant functions and components are instantiated. • O1 interface connectivity is established. Pre-conditions • O1 interface is established between SMO and O-DU to enable SMO/Non-RT RIC to collect performance measurement data and associated RAN configurations. • O-DU supports the implementation of identified configuration parameters when configuration recommendation is received via O1 interface. Begins when Operator specified trigger condition or event is satisfied. Step 1 (M) SMO requests performance measurement data and associated RAN configurations from O-DU for model training via the O1 interface. Step 2 (M) SMO collects required performance measurement data and RAN configurations from O-DU via the O1 interface. Step 3 (M) Non-RT RIC FW retrieves collected information. Step 4 (O) Non-RT RIC performs model training/update. Step 5 (O) Non-RT RIC deploys trained model for inference. Step 6 (M) SMO requests performance measurement data from O-DU for performance monitoring via the O1 interface for rApp execution and optionally model inference. Step 7 (M) SMO collects performance measurement data from O-DU for performance monitoring via the O1 interface for rApp execution and optionally model inference. Step 8 (M) Non-RT RIC FW retrieves the collected data. Step 9 (M) rApp accesses the collected data via R1 interface services. Step 10 (M) rApp performance monitoring and evaluation and optional model inference. Step 11 (M) rApp generates configuration recommendation. Step 12 (M) Non-RT RIC FW retrieves the configuration recommendation via R1 interface services. Step 13 (M) Non-RT RIC provides configuration output to SMO O1 termination. Step 14 (M) SMO communicates MIMO configuration recommendation to O-DU via O1 interface. Ends when Operator specified trigger condition or event is satisfied. Exceptions None identified. Post Conditions O-DU implements the configuration recommendations provided by MIMO optimization App. Traceability REQ-Non-RT-RIC-FUN1, REQ-Non-RT-RIC-FUN2, REQ-Non-RT-RIC- FUN5, REQ-Non-RT-RIC-FUN8 REQ-R1-FUN9 The call flow for MIMO optimization use case is given in figure 4.7.3.3.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 67 Figure 4.7.3.3.1-1: Call flow for MIMO optimization use case
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.3.4 Required data
|
The specification of the data communicated over O1 is outside the scope of WG2. There are no data that are relevant for the A1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4 AI/ML-based initial access (SS Burst Set) configuration optimization
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4.1 Background and goal of the use case
|
3GPP NR based wireless cellular networks promises to provide leaner system design compared to its predecessors in a bid to improve spectral efficiency, power consumption performance and reduce interferences. Ultra-lean design aims to minimize "always on" reference signal transmissions in the downlink. Synchronization Signals Burst (SSB) sets are one of the high-overhead "always on" reference signals. In large scale NR networks with thousands of gNB/Transmission- Reception Points (TRPs) deployed, system configurations derived statically/manually aiming to accommodate worst case scenario which may arise only for a small window of time can impact on the followings: 1) Increased overhead i.e. degraded utilization of time-frequency resources affecting Spectral Efficiency (SE). 2) Increased interferences among the cells. 3) Increased power consumptions in both network and UEs leading increased network CAPEX and reduced UE battery life respectively. In this context, this sub-use case proposes an AI/ML assisted optimization framework wherein AI/ML agent/engine running at Non-RT RIC can infer optimal SSB set configuration (i.e. number of SS blocks, SS beam directions and SS burst periodicity) based on Configuration (CM) parameters, Performance Measurements (PM), Key Performance Indicators (KPI), and measurement report traces obtained from the E2 nodes (O-CU-CP, O-CU-UP, O-DU, O-eNB) and O-RU. At high-level, the goal of the optimization problem is to minimize SS signal transmissions overhead i.e. determine the minimum number of SSB beams required, their directions and periodicity subject to KPI (integrity, mobility, etc.) targets. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 68 The overall scheme can be summarized as follows. Based on the history/trace/distribution of UE-specific beams (e.g. P2 beams), the AI/ML engine determines the minimum number of SSB beams, their directions and the maximum periodicity of the SS burst required to achieve KPI targets as per history/trace/distribution of UE-specific beams. Furthermore, in order to handle the lower probability occurrences in the statistical models e.g. UEs appearing in a completely new direction that has not been considered in the training data, the AI/ML engine (if required) updates the optimal set (e.g. adds beam directions that compliments the optimal directions, updates the SS burst periodicity, etc.). Finally, the AI/ML engine shares the optimized SSB set configuration with gNB.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4.2 Entities/resources involved in the use case
|
1) SMO & Non-RT RIC framework: a) Collect the necessary Configuration (CM) parameters, Performance Measurements (PM), Key Performance Indicators (KPI), and measurement report traces from the E2 nodes (O-CU-CP, O-CU-UP, O-DU, O-eNB) and O-RU. b) Allow the rApp to receive the CMs, PMs, KPIs measurement data (collected via O1) to perform the AI/ML model training and provide inference on the SSB set configuration parameters (number of SS blocks, SS beam directions and SS burst periodicity). c) Write/update the optimized SSB set configuration via O1 (to O-DU) interface. 2) rApps: a) Retrieve the necessary Configuration (CM) parameters, Performance Measurements (PM), Key Performance Indicators (KPI), and measurement report traces pertaining to the E2 nodes and O-RU from the SMO/Non-RT RIC framework via R1 for the purpose of optimizing SSB set configuration. b) Train AI/ML model to optimize the SSB set configurations. c) Modify/update the SSB set configurations, optimized by the inference engine of the rApp, and write the configuration output to the SMO/Non-RT RIC framework via R1. d) Monitor the performance of the respective cells. Upon KPI degradation, initiate rollback to the previous version of the AI/ML model, if any, and/or trigger the AI/ML model retraining. e) Execute the inference/control loop periodically or on an event-triggered based. 3) E2 nodes and O-RU: a) Report the desired performance measurements and KPIs, configuration parameters and CM changes, trace reports and measurements to the SMO via O1. b) Receive the optimized SSB set configurations from the SMO via O1 and apply the configuration on the O-DU which may further exercise the configuration update on the O-RU. NOTE: Both aggregated and disaggregated gNB architectures are supported.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4.3 Solutions
|
The context of the creation and deployment of mMIMO SSB set optimization applications is captured in table 4.7.4.3-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 69 Table 4.7.4.3-1: Creation and deployment of mMIMO SSB set optimization applications Use Case Stage Evolution / Specification <<Uses>> Related use Goal SSB set optimization. Actors and Roles SMO/Non-RT RIC framework, SS burst set optimization rApp, E2 nodes, O-RU. Assumptions All relevant functions and components are instantiated. O1 interface connectivity is established. Pre-conditions SMO/Non-RT RIC framework is instantiated. O1 interface is established between SMO and E2 nodes. Begins when SSB optimization rApp with initial ML model is deployed. Step 1-3(M) SMO/Non-RT RIC framework collects the necessary configurations, performance indicators, and measurement reports from E2 nodes. Step 4-5 (M) rApp retrieves the necessary configurations, performance indicators, and measurement data from SMO/Non-RT RIC framework via R1 and trains AI/ML model for the purpose of optimizing SSB set configuration. Step 6 (M) SMO/Non-RT RIC framework collects observation data from E2 nodes. Step 7-9 (M) rApp retrieves the observation data from SMO/Non-RT RIC framework via R1, infers SSB set configuration, and shares the configuration to SMO/Non-RT RIC framework. Step 10-11 (M) SMO/Non-RT RIC framework writes/updates SSB set configuration at E2 nodes. Step 12-18(M) rApp continuously monitors KPIs in respective cells. In case of unsatisfactory performance, it initiates fallback and retrains/updates the respective AI/ML model(s). Ends when On operator request for rApp to be disabled. Exceptions None identified. Post Conditions SSB set configuration is active. Traceability REQ-Non-RT-RIC-FUN1, REQ-Non-RT-RIC-FUN2, REQ-Non-RT-RIC- FUN5, REQ-Non-RT-RIC-FUN6 The flow diagram of SSB set optimization is given in figure 4.7.4.3-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 70 Figure 4.7.4.3-1: Flow diagram of SSB set optimization
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4.4 Required data
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4.4.1 Input data
|
1) Supported SSB configurations per cell (as specified in 3GPP TS 38.331 [20], clause 6.3.2 and in 3GPP TS 28.541 [4], clause 4.3.5). 2) Key Performance Indicator (KPIs) such as integrity and cell/beam mobility KPIs, etc., for service level assurance (as specified in 3GPP TS 28.552 [5] and in 3GPP TS 28.554 [17]). 3) CSI-RS beam configuration and CSI-RS beam-specific UE measurement reporting from tracing of RRC messages (as specified in 3GPP TS 38.331 [20], clause 5.5.5.2). 4) PMs, such as the distribution of SS-RSRP, SS-RSRQ across of UEs measured per cell, number of RRC connected UEs (mean, max) measured per cell, intra-NRCell SSB beam switch measurement and received random access preambles on a per SSB/per cell basis (as specified in 3GPP TS 28.552 [5], clauses 5.1.1.22, 5.1.1.31, 5.1.1.4, 5.1.1.20 and 5.1.1.21). 5) Radio Link Failure (RLF) tracing across UEs across SSBs per cell (as specified in 3GPP TS 37.320 [19], clause 5.4.1.2, 3GPP TS 32.422 [18], clause 4.3 and in 3GPP TS 38.331 [20], clause 5.3.10). 6) DL/UL throughput/spectral efficiency per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.3 and in 3GPP TR 38.913 [i.6], clause 7.13).
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.7.4.4.2 Output data
|
1) Inferred SSB set (number of SS blocks, SS beam directions and SS burst periodicity) configuration per cell (as specified in 3GPP TS 28.541 [4], clauses 4.3.39 and 4.3.40). ETSI ETSI TS 104 226 V10.1.0 (2025-08) 71
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8 Use case 8: Network energy saving use cases
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.0 Introduction
|
This clause contains the set of energy saving use cases.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1 Carrier and cell switch off/on
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.1 Background and goal of the use case
|
Mobile networks often utilize multiple frequency layers (carriers) to cover the same service area. At low load, i.e. when the expected traffic volume is lower than a fixed threshold, ES can be achieved by switching off one or more carriers or entire cells without impairing the user experience. UEs previously served by the carrier or cell will be offloaded by the E2 node(s) to a new target carrier or cell prior to the switch off. However, the switch off/on decisions are not a trivial task. There are conflicting targets between system performance and energy savings. Other carriers / cells will have to serve the additional traffic and traffic is changing over time. E2 node(s) and O-RU(s) support several techniques effecting energy consumption which might also be load dependent. While energy savings for the switched off carrier/cell is maximized, the overall energy consumption of the network might even increase. Carrier and cell switch off/on control by the Non-RT or Near-RT RIC can consider overall network energy efficiency instead of local optimization. The switch off/on decision can optionally be made by an AI/ML model within the inference host, deployed at the Non-RT RIC to further improve decision making. Among others, the AI/ML models' functionality can include prediction of future traffic, user mobility, and resource usage and can also predict expected energy efficiency enhancements, resource usage, and network performance for different ES optimization states. Also, with addition of per-UE geographical location information such as trace record for immediate MDT measurement (as specified in 3GPP TS 32.423 [11], clause 4.34.1) and trace record for UE location information (as specified in 3GPP TS 32.423 [11], clause 4.34.2) as input data, the increased accuracy for UE location/trajectory prediction could be expected for more efficient solution so that it could prevent switched off cell(s) from being switched on even though meaningful number of UEs generating/receiving traffic do not exist in that cell(s). In that sense these collections could be conditionally activated during some cells being switched off and be deactivated once all cells switched on in terms of UE energy saving. Possible differences among collected types of geographical location information such as between MDT and LMF are expected to be absorbed and exposed to rApp(s) based on R1 data type definition. Before switching off/on carrier(s) or cell(s), there is a possibility the E2 node(s) and O-RU(s) of performing some preparation actions for off switching (e.g. check ongoing emergency calls and warning messages, to enable, disable, modify carrier aggregation and/or dual connectivity, to trigger HO traffic and UEs from cells/carriers to other cells or carriers, informing neighbour nodes via X2/Xn interface, etc.) as well as for on switching (e.g. cell probing, informing neighbour nodes via X2/Xn interface, etc.).This solution proposes a framework that allows the operator to flexibly configure carrier and cell switch off/on parameters in a cell or in a cluster of cells through O1 configuration formulated by rApp towards E2 node(s) and O-RU(s) or A1 policies formulated by rApp towards Near-RT RIC through SMO/Non- RT RIC framework assisted by machine learning (ML) techniques.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.2 Entities/resources involved in the use case
|
1) SMO/Non-RT RIC framework function: a) Collect the configurations, performance indicators and measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) from E2 node(s) and trace records (e.g. per-UE measurement metrics and location information) through O1 Interface, for the purpose of decision making, optionally using training and inference of AI/ML models that assist such EE/ES functions. It is assumed that configurations, performance indicators and measurement reports collected from the O-DU contain the related information for the corresponding O-RU(s). b) Transfer collected data towards rApp. c) Signal updated configurations for EE/ES optimization towards E2 node(s) (O-CU) through O1 Interface. d) (Optionally) Retrain, update, configure EE/ES AI/ML models in Non-RT RIC. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 72 e) Provide A1 policies to Near-RT RIC via A1 interface based on the request from energy saving rApp in the case of A1 policy-based solution. f) Send enrichment information to the Near-RT RIC for calculation of coverage overlap via the A1 interface in the case of A1 policy-based solution. 2) rApps: Energy saving rApp a) Collect the necessary configurations, performance indicators, and measurement reports from SMO/Non-RT RIC framework, for the purpose of training and execution of relevant AI/ML models. b) (Optionally) Retrain, update, configure EE/ES AI/ML model. c) Infer an optimized O1 configuration for EE/ES based on the data collected using R1 interface. d) Infer an optimized A1 policy for EE/ES based on the data collected using R1 interface in the case of A1 policy-based solution. EI producer rApp ( For A1-EI solution) a) Support to produce enrichment information data requested by Near-RT RIC to ascertain overlapping carriers/cells and the coverage of those carriers/cells (e.g. geo location information of carriers/cells , coverage samples mapped geo location, etc.) b) Send enrichment information to the Near-RT RIC through SMO/Non-RT RIC framework functions for calculation of coverage overlap via the A1 interface. 3) E2 nodes: a) Report cell configuration, performance indicators and measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports, trace records such as per-UE measurement metrics and location information) to SMO via O1 interface. b) Report measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) to Near-RT RIC via E2 interface in the case of A1 policy-based solution. c) Perform actions required for EE/ES optimization: e.g. check ongoing emergency calls and warning messages, perform some preparation actions for off switching (e.g. to enable, disable, modify carrier aggregation and/or dual connectivity, to trigger HO traffic and UEs from cells/carriers to other cells or carriers, informing neighbour nodes via X2/Xn interface, etc.) as well as for on switching (e.g. cell probing, informing neighbour nodes via X2/Xn interface, etc.) and make final decision on switch off/on and notify SMO via O1 interface about performed actions in case of O1 based solution or notify Near-RT RIC via E2 interface about performed actions in case of A1 policy-based solution. 4) O-RU: a) Report EC and EE related information via open FH M-plane interface to O-DU. b) Support actions required to perform EE/ES optimization. Updated carrier configuration (i.e. activation, deactivation or sleep). 5) Near-RT RIC (For A1 policy-based solution): a) Collect measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) from E2 nodes. b) (Optionally) Receive EE/ES AI/ML model for deployment via O1 or O2 interface. c) Receive EE/ES related policies via A1 interface for consideration during optimization. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 73 d) Analyse the received data from E2 nodes and perform AI/ML model inference to determine EE/ES. Optimization (i.e. if carriers or cells are recommended to be switched off/on) considering the optimization targets/policies. e) Provide policies or required information to E2 node (O-CU) via E2 to trigger actions for EE/ES optimization. f) Receive enrichment information via the A1 interface. g) Associate enrichment information with collected measurements.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.3 Solution
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.3.1 O1 interface based carrier and cell switch off/on optimization for energy saving
|
In this solution, decision making, potentially including AI/ML model training and inference, is done at the Non-RT RIC. The context of the O1 interface based carrier and cell switch off/on optimization for energy saving is captured in table 4.8.1.3.1-1. Table 4.8.1.3.1-1: O1 interface based carrier and cell switch off/on optimization for energy saving Use Case Stage Evolution / Specification <<Uses>> Related use Goal Enable carrier and cell switch off/on energy saving functions in the network by means of configuration parameter change and actions controlled by Non-RT RIC and allow for AI/ML-based solutions. Actors and Roles SMO/Non-RT RIC framework function: O1 termination. rApp: Carrier and cell switch off/on optimization. E2 node(s), O-RU: Enforces carrier and cell switch off/on optimization configurations. Assumptions O1 interface connectivity is established. Open FH M-plane interface is established between E2 node(s) and O- RU. Network is operational. Non-RT RIC has knowledge about overlapping carriers/cells and the coverage of those carriers/cells (e.g. which carrier/cell is a coverage layer, and which is a capacity layer). Pre-conditions Operator has set the targets for energy saving functions in the Non-RT RIC. Begins when Operator enables the optimization rApp along with initial ML model for carrier and cell switch off/on energy saving functions and E2 node(s) and O-RU(s) become operational. Step 1 (M) rApp requests to collect necessary configurations, performance indicators, and measurement data (e.g. cell load related information and traffic information, EE/EC measurement reports, cell level configurations, per-UE measurement metrics and location information) towards SMO/Non-RT RIC framework function over R1 interface. Step 2 (M) SMO/Non-RT RIC framework function requests data collection towards E2 node(s) and O-RU(s) (via E2 node(s)) over O1 Interface. Step 3 (M) E2 node(s) upon receiving request from SMO/Non-RT RIC framework function requests and collect necessary data form O-RU(s) over open FH M-plane interface. Step 4 (M) E2 node(s) sends the configuration data, configured measurement data to SMO/Non-RT RIC framework function periodically or event based. Step 5 (M) rApp collects the configuration data, collected measurement data for processing. Step 6 (O) AI/ML models can be retrained either on Non-RT RIC framework or on rApp. If Non-RT RIC framework is hosting retraining, then rApp to select AI/ML model and initiate retraining on Non-RT RIC framework. See note 1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 74 Use Case Stage Evolution / Specification <<Uses>> Related use Step 7 (O) Upon receiving retraining request from rApp. Non-RT RIC framework initiates AI/ML model retraining. Step 8 (O) rApp monitors retrained AI/ML model and retrieves retrained AI/ML model. See note 2. Step 9 (O) Upon receiving retrieval request from rApp, Non-RT RIC framework to transfer AI/ML model to rApp. See note 3. Step 10 (O) If AI/ML model retraining is hosted by rApp then AI/ML model to be retrained on rApp itself. Step 11 (O) Once AI/ML model retraining is performed, AI/ML models are deployed and activated for inferencing for which rApp constantly monitors, performance and energy consumption of the E2 node(s) energy consumption of O-RU(s). rApp monitors performance and energy consumption for evaluation of necessary O1 configurations required to perform cell and carrier shutdown. Step 12 (M) rApp generates O1 configurations to prepare and execute cell(s) and carrier(s) off/on SMO/Non-RT RIC framework function. Step 13 (M) SMO/Non-RT RIC framework function instructs E2 node(s) via O1 interface to perform the received request(s) from the rApp. Step 14 (M) O-RU(s) is informed about the updated O-RU(s) configuration via open FH M-plane interface by E2 node. O-RU(s) to notify E2 node(s) once O- RU(s) configuration is implemented. Step 15 (M) E2 node(s) will inform SMO/Non-RT RIC framework function once cell or carrier switch off/on is completed. Step 16 (M) SMO/Non-RT RIC framework inform rApp about completion of cell or carrier switch off/on over R1 interface. Step 17 (M) rApp monitors energy saving objectives and if energy saving objectives are not achieved, it can decide to initiate fallback mechanism for example, reverting changes over O1 interface for carrier and cell switch off/on optimization, and/or AI/ML model update or retraining. Ends when E2 node(s) becomes non-operational or when the operator disables the optimization functions or ML model for energy saving. Exceptions TBD Post Conditions rApp continues monitoring of energy saving function at E2 node(s) and O-RU. E2 node(s) and O-RU(s) operate using the newly deployed parameters/models and state (off/on). NOTE 1: AI/ML procedures mentioned here can be subject to change based on future work on AI/ML workflow in WG2. NOTE 2: AI/ML model retrieval procedure over R1 interface is for illustration purpose, can be subject to change based on future work on AI/ML workflow in WG2. NOTE 3: AI/ML model transfer procedure over R1 interface is for illustration purpose, can be subject to change based on future work on AI/ML workflow in WG2. The flow diagram of O1 interface based carrier and cell switch off/on optimization for energy saving is given in figure 4.8.1.3.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 75 Figure 4.8.1.3.1-1: Flow diagram of O1 interface based carrier and cell switch off/on optimization for energy saving
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.3.2 A1 policy based carrier and cell switch off/on optimization for energy saving
|
In this solution, decision making, potentially using AI/ML model inference, is done at Near-RT RIC. While AI/ML model training might be hosted in Non-RT or Near-RT RIC, the description below is based on AI/ML model training in the Non-RT RIC. The context of the A1 policy based carrier and cell switch off/on optimization for energy saving is captured in table 4.8.1.3.2-1 ETSI ETSI TS 104 226 V10.1.0 (2025-08) 76 Table 4.8.1.3.2-1: A1 policy based carrier and cell switch off/on optimization for energy saving Use Case Stage Evolution / Specification <<Uses>> Related use Goal Enable A1 policy based carrier and cell switch off/on energy saving functions in the network by means of configuration parameter change and actions controlled by Near-RT RIC and allow for AI/ML-based solutions. Actors and Roles SMO/Non-RT RIC framework function: A1 & O1 termination. rApp: Carrier and cell switch off/on optimization. E2 node(s), O-RU: Enforces carrier and cell switch off/on optimization configurations. Near -RT RIC: Energy savings decision making function. Assumptions O1 interface connectivity is established between SMO and E2 nodes. R1 interface connectivity is established. Open FH M-plane interface is established between E2 node(s) and O-RU. E2 interface connectivity is established between E2 node and Near-RT RIC. A1 interface is established between Non-RT RIC and Near-RT RIC. Network is operational. Pre-conditions Operator has set the targets for energy saving functions in the Non-RT RIC. Begins when Operator enables the optimization rApp along with initial ML model for carrier and cell switch off/on energy saving functions and E2 node(s) and O-RU(s) become operational. See note 1. Step 1 (M) rApp requests to collect necessary configurations, performance indicators, and measurement data (e.g. cell load related information and traffic information, EE/EC measurement reports, cell level configurations) towards SMO/Non-RT RIC framework function over R1 interface. Step 2 (M) SMO/Non-RT RIC framework function requests data collection towards E2 node(s) and O-RU(s) (via E2 node(s)) over O1 Interface. Step 3 (M) E2 node(s) upon receiving request from SMO/Non-RT RIC framework function requests and collect necessary data form O-RU(s) over open FH M-plane interface. Step 4 (M) E2 node(s) sends the configuration data, configured measurement data to SMO/Non-RT RIC framework function periodically or event based. Step 5 (M) rApp collects the configuration data, collected measurement data for processing. Step 6 (O) AI/ML models can be retrained either on SMO/Non-RT RIC framework or on rApp. If SMO/Non-RT RIC framework is hosting retraining, then rApp to select AI/ML model and initiate retraining on Non-RT RIC framework. See note 2. Step 7 (O) Upon receiving retraining request from rApp. SMO/Non-RT RIC framework initiates AI/ML model retraining. Step 8 (O) rApp monitors retrained AI/ML model before requesting to deploy towards Near- RT RIC. Step 9 (O) rApp request SMO/Non-RT RIC framework to deploy AI/ML model in Near-RT RIC over R1 Interface. See note 3. Step 10 (O) If AI/ML model retraining is hosted by rApp then AI/ML model to be retrained on rApp itself. Step 11 (O) Once AI/ML model retraining is performed rApp to transfer AI/ML model to Non- RT RIC framework for deployment to Near-RT RIC. Step 12 (O) Upon receiving request to deploy AI/ML model, Non-RT RIC framework to deploy AI/ML model in Near-RT RIC. Step 13 (M) rApp to trigger EE/ES optimization through A1 policy to prepare and execute cell(s) and carrier(s) off/on. Step 14 (M) Non-RT RIC framework receives policy from rApp and forwards it towards Near- RT RIC via A1 Interface. Step 15 (M) Near-RT RIC provides A1 policy response to SMO/Non-RT RIC framework. Step 16 (M) Non-RT RIC framework informs rApp about A1 policy feedback. Step 17 (M) rApp monitors energy saving objectives as per A1 policy. Step 18 (M) Near-RT RIC receives the policy from the Non-RT RIC over the A1 interface and inferences the EE/ES related AI/ML models and converts policy to specific E2 control or policy commands. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 77 Use Case Stage Evolution / Specification <<Uses>> Related use Step 19 (O) The Near-RT RIC creates an EI job to ascertain overlapping carriers/cells and the coverage of those carriers/cells (e.g. geo location information of carriers/cells, coverage samples mapped geo location, etc.). Step 20 (O) SMO/Non-RT RIC framework functions requests EI rApp to deliver EI data as per details mentioned by Near-RT RIC. Step 21 (O) EI rApp delivers requested EI data to SMO/Non-RT RIC framework functions. Step 22 (O) Non-RT RIC delivers collected enrichment information as EI job results to the Near-RT RIC via the A1 interface. Step 23 (M) Near-RT RIC requests data collection from E2 node(s) and O-RU(s) via E2 interface. Step 24 (M) Upon receiving data collection request E2 nodes requests and collect measurement data. Step 25 (M) Near-RT RIC collects measurement data from E2 node(s) and O-RU(s) via E2 interface. Step 26 (M) Inferencing AI/ML model to evaluate & generate E2 control or policies for cell and carrier switch off/on. Step 27 (M) Near-RT RIC generates and sends cell and carrier switch off/on control/policy message based on inferred AI/ML model to E2 nodes and O-RU(s) via E2 interface. Step 28 (M) O-RU(s) node to update configurations to execute cell or carrier switch off/on. Step 29 (M) E2 nodes feedbacks E2 control/policy to Near-RT RIC. Step 30 (M) Near-RT RIC to notify Non-RT RIC if any change in the state of policy. Step 31 (M) SMO/Non-RT RIC framework forwards notification received from Near-RT RIC to rApp. Step 32 (O) If energy saving objectives are not achieved rApp can decide to initiate fallback mechanism for example, updating or deleting A1 policy for carrier and cell switch off/on optimization. Step 33 (O) SMO/Non-RT RIC framework send update or delete A1 policies to Near-RT RIC. Step 34 (O) rApp can update or retrain AI/ML model based on evaluation of energy saving objectives. Step 35 (O) SMO/Non-RT RIC framework send update or retrain AI/ML model to Near-RT RIC. Ends when E2 node(s) becomes non-operational or when the operator disables the rApp or ML model for energy saving. Exceptions N/A. Post Conditions rApp continues monitoring of energy saving function at E2 node(s) and O-RU. E2 node(s) and O-RU(s) operate using the newly deployed parameters/models and state (off/on). NOTE 1: Operator can set policies through rApp or allow AI/ML model in rApp to infer policies. NOTE 2: AI/ML procedures mentioned here can be subject to change based on future work on AI/ML workflow in WG2. NOTE 3: AI/ML model deployment procedure over R1 interface is for illustration purpose, can be subject to change based on future work on AI/ML workflow in WG2. The flow diagram of A1 interface based carrier and cell switch off/on optimization for energy saving is given in figure 4.8.1.3.2-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 78 Figure 4.8.1.3.2-1: Flow diagram of A1 interface based carrier and cell switch off/on optimization for energy saving NOTE: Above mentioned AI/ML procedures are illustration purpose and details are not defined in the present document. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 79
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.4 Required data
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.4.0 Introduction
|
This clause contains the input and output data of model training and inference for energy saving cell and carrier shutdown.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.4.1 Input data
|
The measurement input data are used in model training and inference. They can include the following measurements to monitor energy consumption and energy efficiency of E2 node(s) and O-RU(s): a) DL PDCP SDU data volume per interface (data volume in DL delivered from O-CU-UP to O-DU, per PLMN, per QoS level, per slice, per interface (F1-U, Xn-U, X2-U)) (as specified in 3GPP TS 28.552 [5], clause 5.1.3.6.2.3). b) UL PDCP SDU data volume per interface (data volume in UL delivered to O-CU-UP from O-DU, per PLMN, per QoS level, per slice, per interface (F1-U, Xn-U, X2-U)) (as specified in 3GPP TS 28.552 [5], clause 5.1.3.6.2.4). c) RSRQ measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.31). d) RSRP measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.22). e) SINR measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.32). f) Energy consumption (as specified in 3GPP TS 28.552 [5], clause 5.1.1.19.3). g) Power consumed by physical network function & its components (as specified in 3GPP TS 28.552 [5], clause 5.1.1.19.2 and in O-RAN.WG4.MP.0 [10], clauses B.1, B.5). h) Transmit power (as specified in O-RAN.WG4.MP.0 [10], clauses B.1, B.2.1). i) M1 MDT measurement in trace record for immediate MDT measurements (as specified in 3GPP TS 32.423 [11], clause 4.34.1). j) UE location in trace record for UE location information (as specified in 3GPP TS 32.423 [11], clause 4.34.2).
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.1.4.2 Output data
|
rApps to deliver energy saving & energy efficiency policies for cell/carrier switch off/on optimization through R1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2 RF channel reconfiguration
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.1 Background and goal of the use case
|
In mobile networks m-MIMO antennas are used for beamforming techniques to enhance cell capacity and throughput. To achieve beamforming, O-RU(s) need to concentrate the power amplifiers at the radome by combining radiating elements. At low load, i.e. when the expected traffic volume or number of connected users are lower than the configured threshold, ES can be achieved by reducing the power consumption of O-RU(s) by switching off certain Tx/Rx arrays. For example, 32 out of 64 Tx/Rx arrays of an O-RU(s) can be switched off in a digital m-MIMO architecture and correspondingly the number of spatial layers and SSBs can be reduced. The procedure (involvement of respective O-RAN interfaces) of the RF channel reconfiguration depends on the management architecture model (hybrid or hierarchical) and the deployment option. The switch off/on decision can be made by an AI/ML model within the inference host deployed at the Non-RT RIC. Among others the AI/ML models can include prediction of future traffic, user mobility, and resource usage and can also predict expected energy efficiency enhancements, resource usage, and network performance for different ES optimization states. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 80 This solution proposes a framework that allows the operator to flexibly configure RF channel reconfiguration parameters in a cell or in a cluster of cells through O1 configuration formulated by rApp towards E2 node(s) and O-RU(s) through SMO/Non-RT RIC or A1 policies formulated by rApp towards Near-RT RIC through SMO/Non-RT RIC framework assisted by Machine Learning (ML) techniques.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.2 Entities/resources involved in the use case
|
1) SMO/Non-RT RIC framework function: a) Collect the configurations, performance indicators and measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) from E2 node(s), for the purpose of decision making, optionally using training and inference of AI/ML models that assist such EE/ES functions. b) Transfer collected data towards rApp. It is assumed that configurations, performance indicators and measurement reports collected from the O-DU contain the related information for the corresponding O-RU(s). c) Signal updated configurations for EE/ES optimization towards E2 node(s) (O-CU) through O1 interface. d) (Optionally) Retrain, update, configure EE/ES AI/ML models in Non-RT RIC. e) Provide A1 policies to Near-RT RIC via A1 interface based on the request from energy saving rApp in the case of A1 policy-based solution. 2) rApps: a) Collect the necessary configurations, performance indicators, and measurement reports from SMO/Non-RT RIC framework function, for the purpose of training and execution of relevant AI/ML models. b) (Optionally) Retrain, update, configure EE/ES AI/ML model. c) Infer an optimized RF channel configuration for EE/ES based on the data collected using R1 interface. d) Provide optimized A1 policy for EE/ES based on the data collected using R1 interface in the case of A1 policy-based solution. 3) E2 nodes: a) Report cell configuration, performance indicators and measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) to SMO via O1 interface. b) Report measurement reports to Near-RT RIC via E2 interface in the case of A1 policy-based solution. c) Perform actions required to perform RF channel reconfiguration (i.e. O-RU Tx/Rx array selection, modification of the number of SSB beams, modification of the O-RU antenna transmit power, modification of the number of SU/MU MIMO data layers or spatial streams) as part of EE/ES optimization. 4) O-RU: a) Report EC and EE related information via open FH M-plane interface to O-DU. b) Perform actions required to be performed due to RF channel reconfiguration (i.e. O-RU Tx/Rx array selection, modification of the number of SSB beams, modification of the O-RU antenna transmit power, modification of the number of SU/MU MIMO spatial streams or data layers) as part of EE/ES optimization. 5) Near-RT RIC (For A1 policy-based solution): a) Collect measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) from E2 nodes. b) (Optionally) Receive EE/ES AI/ML model for deployment via O1 or O2 interface. c) Receive EE/ES related policies via A1 interface for consideration during optimization. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 81 d) Analyse the received data from E2 nodes and perform AI/ML model inference to determine EE/ES. Optimization (i.e. if carriers or cells are recommended to be switched off/on) considering the optimization targets/policies. e) Provide policies or required information to E2 node via E2 to trigger actions for EE/ES optimization.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.3 Solution
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.3.1 O1 policy based RF channel reconfiguration optimization for energy saving
|
In this solution, decision making, potentially including AI/ML model training and inference, is done at the Non-RT RIC. The context of the O1 interface based RF channel reconfiguration optimization for energy saving is captured in table 4.8.2.3.1-1. Table 4.8.2.3.1-1: O1 interface based RF channel reconfiguration optimization for energy saving Use Case Stage Evolution / Specification <<Uses>> Related use Goal Enable RF channel reconfiguration energy saving functions in the network by means of configuration parameter change and actions controlled by Non-RT RIC and allow for AI/ML-based solutions. Actors and Roles SMO/Non-RT RIC framework function: Termination point for O1 interface. E2 node(s), O-RU(s): Enforces optimized RF channel configuration. rApp: RF channel reconfiguration optimization. Assumptions O1 interface connectivity is established. Open FH M-plane interface is established between E2 node(s) and O-RU. Network is operational. Pre-conditions Operator has set the targets for energy saving functions in the Non-RT RIC. Begins when Operator enables the optimization rApp along with initial ML model for RF channel reconfiguration saving functions and E2 node(s) and O-RU(s) become operational. Step 1 (M) rApp requests to collect necessary configurations, performance indicators, and measurement data towards SMO/Non-RT RIC framework function over R1 interface. Step 2 (M) SMO/Non-RT RIC framework function requests data collection towards E2 node(s) and O-RU(s) (via E2 node(s)) over O1 interface. Step 3 (M) E2 node(s) upon receiving request from SMO/Non-RT RIC framework function requests and collect necessary data from O-RU(s) over open FH M-plane interface. Step 4 (M) E2 node(s) send the configuration data, configured measurement data to SMO/Non-RT RIC framework periodically or event based. Step 5 (M) rApp collects the configuration data, collected measurement data for processing. Step 6 (O) AI/ML models can be retrained either on Non-RT RIC framework or on rApp. If Non-RT RIC framework is hosting retraining, then rApp to select AI/ML model and initiate retraining on Non-RT RIC framework. See note 1. Step 7 (O) Upon receiving retraining request from rApp. Non-RT RIC framework initiates AI/ML model retraining. Step 8 (O) rApp monitors retrained AI/ML model and retrieves retrained AI/ML model. See note 2. Step 9 (O) Upon receiving retrieval request from rApp, Non-RT RIC framework to transfer AI/ML model to rApp. See note 3. Step 10 (O) If AI/ML model retraining is hosted by rApp then AI/ML model to be retrained on rApp itself. Step 11 (O) Once AI/ML model retraining is performed, AI/ML models are deployed and activated for inferencing for which rApp constantly monitors performance and energy consumption to evaluate necessary O1 configurations to perform RF channel reconfiguration. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 82 Use Case Stage Evolution / Specification <<Uses>> Related use Step 12 (M) rApp to request the SMO/Non-RT RIC framework to configure E2 node(s) to prepare and execute RF channel reconfigurations as below: i) O-RU Tx/Rx array selection. ii) Modify the number of SU/MU MIMO spatial streams or data layers. iii) Modify the number of SSB beams. iv) Modify O-RU antenna transmit power. Step 13 (M) SMO/Non-RT RIC framework function to requests to configure E2 node(s) for RF channel reconfiguration through O1 interface. Step 14 (M) O-RU(s) is informed about the updated O-RU configuration via open FH M- plane interface by E2 node(s). O-RU(s) to inform E2 node(s) once O-RU(s) configuration is implemented. Step 15 (M) E2 node(s) will inform SMO/Non-RT RIC framework function once RF channel reconfiguration is completed. Step 16 (M) SMO/Non-RT RIC framework function to inform rApp about completion of RF channel reconfiguration over R1 interface. Step 17 (M) rApp monitors energy saving objectives. If energy saving objectives are not achieved, it can decide to initiate fallback mechanism for example, reverting changes over O1 interface for RF channel reconfigurations, and/or AI/ML model update or retraining. Ends when E2 node(s) becomes non-operational or when the operator disables the optimization functions or ML model for energy saving. Exceptions TBD. Post Conditions rApp continues monitoring of energy saving function at E2 node(s) and O-RU(s). E2 node(s) and O-RU(s) operate using the newly deployed parameters/models and state. NOTE 1: AI/ML procedures mentioned here can be subject to change based on future work on AI/ML workflow in WG2. NOTE 2: AI/ML model retrieval procedure over R1 interface is for illustration purpose, can be subject to change based on future work on AI/ML workflow in WG2. NOTE 3: AI/ML model transfer procedure over R1 interface is for illustration purpose, can be subject to change based on future work on AI/ML workflow in WG2. The flow diagram of O1 interface based RF channel reconfiguration optimization for energy saving is shown in figure 4.8.2.3.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 83 Figure 4.8.2.3.1-1: Flow diagram of O1 interface based RF channel reconfiguration optimization for energy saving
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.3.2 A1 policy based RF channel reconfiguration optimization for energy saving
|
In this solution, decision making, potentially using AI/ML model inference, is done at Near-RT RIC. While AI/ML model training might be hosted in Non-RT or Near-RT RIC, the description below is based on AI/ML model inferencing in the Near-RT RIC. The context of the A1 policy based optimization for energy saving is captured in table 4.8.2.3.2-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 84 Table 4.8.2.3.2-1: A1 policy based optimization for energy saving Use Case Stage Evolution / Specification <<Uses>> Related use Goal Enable A1 policy based RF channel reconfiguration energy saving functions in the network by means of configuration parameter change and actions controlled by Near-RT RIC and allow for AI/ML-based solutions. Actors and Roles SMO/Non-RT RIC framework function: A1 & O1 termination rApp: RF channel reconfiguration optimization E2 node(s), O-RU: Enforces RF channel reconfiguration Near-RT RIC: RF channel reconfiguration energy savings decision making function Assumptions O1 interface connectivity is established between SMO and E2 nodes. R1 interface connectivity is established. Open FH M-plane interface is established between O-DU and O-RU. E2 interface connectivity is established between E2 nodes and Near-RT RIC. A1 interface is established between Non-RT RIC and Near-RT RIC. Network is operational. Pre-conditions Operator has set the targets for energy saving functions in the Non-RT RIC. Begins when Operator enables the optimization rApp along with initial ML model for RF channel reconfiguration energy saving functions and E2 node(s) and O-RU(s) become operational. See note. Step 1 (M) rApp requests for necessary configurations, performance indicators, and measurement data towards SMO/Non-RT RIC framework function over R1 interface. Configurations may contain data related to node capability such as O-RU RF channel configuration/TRx control capability information which rApp needs to know before formulating A1 policies for ASM optimization. Step 2 (M) SMO/Non-RT RIC framework function requests data collection towards E2 node(s) and O-RU(s) (via E2 node(s)) over O1 Interface. Step 3 (M) E2 node(s) upon receiving request from SMO/Non-RT RIC framework function requests and collect necessary data form O-RU(s) over open FH M-plane interface. Step 4 (M) E2 node(s) sends the configuration data, configured measurement data to SMO/Non-RT RIC framework function periodically or event based. Step 5 (M) rApp collects the configuration data, collected measurement data for processing. Step 6 (M) rApp to trigger EE/ES optimization through A1 policy to prepare and execute RF channel configuration/Trx control. Step 7 (M) Non-RT RIC framework receives policy from rApp and forwards it towards Near-RT RIC via A1 interface. Step 8 (M) Near-RT RIC provides A1 policy response to SMO/Non-RT RIC framework. Step 9 (M) Non-RT RIC framework informs rApp about A1 policy feedback. Step 10 (M) rApp monitors energy saving objectives as per A1 policy. Step 11 (M) The Near-RT RIC receives the policy from the Non-RT RIC over the A1 interface. It then interprets the policy to determine the required data collection and E2 control/policies. Step 12 (M) Inferencing AI/ML model to evaluate & generate E2 control or policies for RF channel configuration/Trx Control optimization. Step 13 (M) Near-RT RIC to notify Non-RT RIC if any change in the state of policy. Step 14 (M) SMO/Non-RT RIC framework forwards notification received from Near-RT RIC to rApp. Step 15 (O) If energy saving objectives are not achieved rApp may decide to initiate fallback mechanism for example, updating or deleting A1 policy for RF channel configuration/Trx control optimization. Step 16 (O) SMO/Non-RT RIC framework send update or delete A1 policies to Near-RT RIC. Step 17 (O) rApp can update or retrain AI/ML model based on evaluation of energy saving objectives. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 85 Use Case Stage Evolution / Specification <<Uses>> Related use Step 18 (O) SMO/Non-RT RIC framework send update or retrained AI/ML model to Near-RT RIC. Ends when E2 node(s) becomes non-operational. Exceptions N/A. Post Conditions rApp continues monitoring of energy saving function at E2 node(s) and O-RU. E2 node(s) and O-RU(s) operate using the newly deployed parameters/models and state (off/on). NOTE: Operator can set policies through rApp or allow AI/ML model in rApp to infer policies. The flow diagram of A1 interface based RF channel reconfiguration optimization for energy saving is shown in figure 4.8.2.3.2-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 86 Figure 4.8.2.3.2-1: Flow diagram of A1 interface based RF channel reconfiguration optimization for energy saving
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.4 Required data
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.4.0 Introduction
|
This clause contains the input and output data of model training and inference for energy saving using RF channel reconfiguration. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 87
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.4.1 Input data
|
The measurement input data are used in model training and inference. They can include the following measurements to monitor energy consumption and energy efficiency of E2 node(s) and O-RU(s): a) DL PDCP SDU data volume per interface (data volume in DL delivered from O-CU-UP to O-DU, per PLMN, per QoS level, per slice, per interface (F1-U, Xn-U, X2-U)) (as specified in 3GPP TS 28.552 [5], clause 5.1.3.6.2.3). b) UL PDCP SDU data volume per interface (data volume in UL delivered to O-CU-UP from O-DU, per PLMN, per QoS level, per slice, per interface (F1-U, Xn-U, X2-U)) (as specified in 3GPP TS 28.552 [5], clause 5.1.3.6.2.4). c) RSRQ measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.31). d) RSRP measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.22). e) SINR measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.32). f) Energy consumption (as specified in 3GPP TS 28.552 [5], clause 5.1.1.19.3). g) Power consumed by physical network function & its components (as specified in 3GPP TS 28.552 [5], clause 5.1.1.19.2 and in O-RAN.WG4.MP.0 [10], clauses B.1, B.5). h) Transmit power (as specified in O-RAN.WG4.MP.0 [10], clauses B.1, B.2.1).
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.2.4.2 Output data
|
rApps to deliver energy saving & energy efficiency A1 policies for RF channel reconfiguration optimization through R1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3 Advanced sleep mode
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.0 Introduction
|
This use case describes a method to achieve intelligent energy saving by optimizing the sleep mode via Non-RT RIC-based guidance.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.1 Background and goal of the use case
|
Mobile networks were often designed to provide higher data rates, better coverage, and ubiquitous connectivity. They have to be always available and well-dimensioned in order to ensure the best Quality of Service (QoS) even in peak hours and emergency or mass event scenarios. This may lead to an over-dimensioned and under-utilized network particularly when the traffic demand is low, such is the case during night hours. The energy consumption of a network is composed of two components: • A fixed component, which is mainly due to the system architecture and includes the power consumption of control signals, backhaul infrastructure, and load-independent consumption of baseband processors. • A variable, load-dependent component, which depends on the transported traffic. Over-provisioning of the network as well as low load periods translate into significant, and unnecessary, energy consumption, due to the fixed component. Sleep modes, which consist in shutting down the O-RU for a certain period of time, are an efficient way to handle this component. It consists in a progressive deactivation of the O-RU's components according to the time needed by each of them to shut down then reactivate again. According to this transition time, four levels of sleep modes have been specified in O-RAN. WG4.CUS.0 [12], clause 16. Deeper sleep levels allow more energy saving but induce larger delays for the users who arrive to the network and who need to wait longer for the components to be reactivated. Hence, Non-RT RIC can provide policies in such cases where reactivation time may not be concern or proactively reactivates. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 88 O-RU and E2 nodes (O-CU, O-DU) may implement various sleep modes. The sleep modes could be enabled by SMO/Non-RT RIC through A1 policy. When enabled, the Near-RT RIC selects among the sleep modes considering their capabilities, the actual traffic situation, and the network conditions. Different SM operate at different time scales (e.g. symbol, slot, frame). This solution proposes a framework that allows the operator to flexibly select various sleep modes parameters through A1 policy formulated by rApp towards Near-RT RIC through SMO/Non-RT RIC framework assisted by Machine Learning (ML) techniques.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.2 Entities/resources involved in the use case
|
1) SMO/Non-RT RIC framework function: a) Collect the configurations, performance indicators and measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) from E2 node(s) and trace records (e.g. per-UE measurement metrics and location information) through O1 interface, for the purpose of decision making, optionally using training and inference of AI/ML models that assist such EE/ES functions. b) Transfer collected data towards rApp. It is assumed that configurations, performance indicators and measurement reports collected from the O-DU contain the related information for the corresponding O-RU(s). c) Signal A1 policies for Near-RT RIC for ASM optimization A1 interface. d) (Optionally) Retrain, update, configure AI/ML models in Non-RT RIC. 2) rApps: a) Collect the necessary configurations, performance indicators, and measurement reports from SMO/Non-RT RIC framework function, for the purpose of training and execution of relevant AI/ML models. b) (Optionally) Retrain, update, configure EE/ES AI/ML model. c) Infer an optimized A1 policy for ASM based on the data collected using R1 interface. 3) E2 nodes: a) Report cell configuration, performance indicators and measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports, Trace records such as per-UE measurement metrics and location information) to SMO via O1 interface. b) Report measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) to Near-RT RIC via E2 interface. c) Support actions required to perform ASM and Trx control for EE/ES optimization. 4) O-RU: a) Report EC and EE related information via open FH M-plane interface to O-DU . b) Support actions required to perform ASM and Trx control for EE/ES optimization. 5) Near-RT RIC (For A1 policy based solution): a) Collect measurement reports (e.g. cell load related information and traffic information, EE/EC measurement reports) from E2 nodes. b) (Optionally) Receive EE/ES AI/ML model for deployment via O1 or O2 interface. c) Receive EE/ES related policies via A1 interface for consideration during optimization. d) Analyse the received data from E2 nodes and perform AI/ML model inference to determine ASM & Trx control EE/ES. Optimization (i.e. ASM to be activated for certain time and cells) considering the optimization targets/policies. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 89 e) Provide policies or required information to E2 node (O-CU) via E2 to trigger actions for EE/ES optimization.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.3 Solution
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.3.1 A1 policy based ASM optimization for energy saving
|
In this solution, decision making, potentially using AI/ML model inference, is done at Near-RT RIC. While AI/ML model training might be hosted in Non-RT or Near-RT RIC, the description below is based on AI/ML model inferencing in the Near-RT RIC. The context of the A1 policy based ASM optimization for energy saving is captured in table 4.8.3.3.1-1. Table 4.8.3.3.1-1: A1 policy based ASM optimization for energy saving Use Case Stage Evolution / Specification <<Uses>> Related use Goal Enable A1 policy based ASM energy saving functions in the network by means of configuration parameter change and actions controlled by Near-RT RIC and allow for AI/ML-based solutions. Actors and Roles SMO/Non-RT RIC framework function: A1 & O1 termination. rApp: ASM optimization. E2 node(s), O-RU: Enforces ASM E2 controls or policies. Near-RT RIC: ASM energy savings decision making function. Assumptions O1 interface connectivity is established between SMO and E2 nodes. R1 interface connectivity is established. Open FH M-plane interface is established between E2 node(s) and O-RU. E2 interface connectivity is established between E2 node and Near-RT RIC. A1 interface is established between Non-RT RIC and Near-RT RIC. Network is operational. Pre-conditions Operator has set the targets for energy saving functions in the Non-RT RIC. Begins when Operator enables the optimization rApp along with initial ML model for ASM energy saving functions and E2 node(s) and O-RU(s) become operational. See note. Step 1 (M) rApp requests to collect necessary configurations, performance indicators, and measurement data towards SMO/Non-RT RIC framework function over R1 interface. Configurations may contain data related to node capability such as O-RU ASM capability information which rApp needs to know before formulating A1 policies for ASM optimization. Step 2 (M) SMO/Non-RT RIC framework function requests data collection towards E2 node(s) and O-RU(s) (via E2 node(s)) over O1 interface. Step 3 (M) E2 node(s) upon receiving request from SMO/Non-RT RIC framework function requests and collect necessary data form O-RU(s) over open FH M-plane interface. Step 4 (M) E2 node(s) sends the configuration data, configured measurement data to SMO/Non-RT RIC framework function periodically or event based. Step 5 (M) rApp collects the configuration data, collected measurement data for processing. Step 6 (M) rApp to trigger EE/ES optimization through A1 policy to prepare and execute ASM & Trx control. Step 7 (M) Non-RT RIC framework receives policy from rApp and forwards it towards Near- RT RIC via A1 interface. Step 8 (M) Near-RT RIC provides A1 policy response to SMO/Non-RT RIC framework. Step 9 (M) Non-RT RIC framework informs rApp about A1 policy feedback. Step 10 (M) rApp monitors energy saving objectives as per A1 policy. Step 11 (M) The Near-RT RIC receives the policy from the Non-RT RIC over the A1 interface. It then interprets the policy to determine the required data collection and E2 control/policies. Step 12 (M) Inferencing AI/ML model to evaluate & generate E2 control or policies for ASM optimization. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 90 Use Case Stage Evolution / Specification <<Uses>> Related use Step 13 (M) Near-RT RIC to notify Non-RT RIC if any change in the state of policy. Step 14 (M) SMO/Non-RT RIC framework forwards notification received from Near-RT RIC to rApp. Step 15 (O) If energy saving objectives are not achieved rApp may decide to initiate fallback mechanism for example, updating or deleting A1 policy for ASM optimization. Step 16 (O) SMO/Non-RT RIC framework send update or delete A1 policies to Near-RT RIC. Step 17 (O) rApp can update or retrain AI/ML model based on evaluation of energy saving objectives. Step 18 (O) SMO/Non-RT RIC framework send update or retrain AI/ML model to Near-RT RIC. Ends when E2 node(s) becomes non-operational or when the operator disables the rApp or ML model for energy saving. Exceptions N/A. Post Conditions rApp continues monitoring of energy saving function at E2 node(s) and O-RU. E2 node(s) and O-RU(s) operate using the newly deployed parameters/models and state (off/on). NOTE: Operator can set policies through rApp or allow AI/ML model in rApp to infer policies. The flow diagram of A1 interface based ASM optimization for energy saving is shown in figure 4.8.3.3.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 91 Figure 4.8.3.3.1-1: Flow diagram of A1 interface based ASM optimization for energy saving
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.4 Required data
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.4.0 Introduction
|
This clause contains the input and output data of model training and inference for energy saving using ASM. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 92
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.4.1 Input data
|
The measurement input data are used in model training and inference. They can include the following measurements to monitor energy consumption and energy efficiency of E2 node(s) and O-RU(s): a) DL PDCP SDU data volume per interface (data volume in DL delivered per PLMN, per QoS level, per slice, per interface ((from O-CU-UP to O-DU over F1-U, to external O-CU-UP over Xn-U and to external O-eNB over X2-U)) (as specified in 3GPP TS 28.552 [5], clause 5.1.3.6.2.3). b) UL PDCP SDU data volume per interface (data volume in UL delivered per PLMN, per QoS level, per slice, per interface ((from O-DU to O-CU-UP over F1-U, to external O-CU-UP over Xn-U and to external O-eNB over X2-U)) (as specified in 3GPP TS 28.552 [5], clause 5.1.3.6.2.4). c) RSRQ measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.31). d) RSRP measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.22). e) SINR measurement per SSB per cell (as specified in 3GPP TS 28.552 [5], clause 5.1.1.32). f) Energy consumption (as specified in 3GPP TS 28.552 [5], clause 5.1.1.19.3). g) Power consumed by physical network function & its components (as specified in 3GPP TS 28.552 [5], clause 5.1.1.19.2 and in O-RAN.WG4.MP.0 [10], clauses B.1, B.5). h) Transmit power (as specified in O-RAN.WG4.MP.0 [10], clauses B.1, B.2.1). i) M1 MDT measurement in trace record for immediate MDT measurements (as specified in O-RAN. WG4.CUS.0 [12], clause 4.34.1). j) UE location in trace record for UE location information (as specified in O-RAN. WG4.CUS.0 [12], clause 4.34.2). k) O-RU ASM capability information (as specified in O-RAN.WG4.MP.0 [10], clause 20.4).
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.8.3.4.2 Output data
|
rApps to deliver energy saving & energy efficiency policies for ASM optimization through R1 interface.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9 Use case 9: O-Cloud resource optimization use cases
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.0 Introduction
|
This clause contains the set of O-Cloud resource optimization use cases.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1 Use case: O-Cloud node draining use case
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.0 Introduction
|
This use case describes the procedure for the SMO/Non-RT RIC to perform draining of specific O-Cloud node [O-Cloud node description based on O-RAN.WG6.O2-GA&P [13] recommendation by rApp through SMO, which can result in relocation of network functions or its components to another O-Cloud node, thereby restoring network function i.e. network healing.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.1 Background and goal of the use case
|
When one or more NF(s) is (are) experiencing some performance degradation, and there is possibility that issue could not be fixed, or root cause could not be identified just by analysing O1 (FCAPS) data. There can be a requirement of co- relating O1 and O2 (FCAPS) data optionally with the help of AI/ML, which can result in identification of root cause. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 93 Examples for reasons to perform O-Cloud node draining: • There is possibility of faulty or misconfiguration of underneath O-Cloud node. By co-relating RAN OAM and O2 related data rApp will help in identifying issue in O-Cloud node which is found to be the root cause of NF's performance degradation, then O-Cloud can be drained, and relocation of the NF(s) on another O-Cloud node can be performed. This example explains a scenario where an action can be invoked when NF(s) performance degradation happens due to O-Cloud node(s), which can be resolved by draining the O-Cloud node. NOTE: The O-Cloud node(s) is set to maintenance mode (as specified in O-RAN.WG6.O2-GA&P [13], clause 3.10.2) by default, when this use case is called.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.2 Entities/resources involved in the use case
|
1) SMO/Non-RT RIC framework: - To collect necessary performance, configuration, and other data for rApp to define and update policies which guides the SMO for O-Cloud resource management through O2 related functions over O2 interface. - Non-RT RIC framework should support rApp for managing data to and from O2 related functions related to resource management. - Train AI/ML model with data from O1 and O2, which supports rApp to predict possible requirement of O-Cloud node draining. 2) rApps: - Collect the necessary configurations, performance indicators, and measurement reports from SMO/Non-RT RIC framework. - (Optionally) Retrain, update, configure AI/ML model. - Infer an optimized policies/recommendations for O-Cloud node draining based on the data collected using R1 interface. 3) O2 related functions (NFO/FOCOM): - To support discovery and delivery of O-Cloud IMS/DMS FCAPS data. - To support reception of policy/recommendations from rApp and enforcing these policies/ recommendations towards O-Cloud over O2 interface. 4) RAN OAM functions: - Retrieve relevant PM, CM, FM data from E2 nodes via the O1 interface. - Allow rApps to access the PM, CM, FM data over R1 interface. 5) O-Cloud (IMS and DMS): - To support delivery of O-Cloud (IMS/DMS) resource performance, configuration, and other data to O2 related functions. - To provide feedback post completion or non-completion of recommendations to Non-RT RIC through O2 related functions. - To support to relocate network function and drain O-Cloud node. 6) E2 nodes: - Support to send fault and measurement data, RAN configurations to SMO/Non-RT RIC via the O1 interface. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 94
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.3 Solutions
|
In this solution, decision making, potentially using AI/ML model inference, is done at rApp. While AI/ML model training might be hosted in Non-RT RIC, the description below is based on AI/ML model training in the Non-RT RIC. The context of Non-RT RIC based O-Cloud node draining is captured in table 4.9.1.3-1. Table 4.9.1.3-1: Non-RT RIC based O-Cloud node draining Use Case Stage Evolution / Specification Goal Draining O-Cloud node through O2 related functions. Actors and Roles SMO/Non-RT RIC framework. rApp: O-Cloud policy/recommendations triggering function. O2 related functions (NFO/FOCOM): Orchestration and management related functions. RAN OAM functions: O1 FCAPS functions. O-Cloud IMS and DMS: O-Cloud policy enforcement and resource data reporting. E2 node: To report network function data to SMO/Non-RT RIC. Assumptions • All relevant functions and components are instantiated. • O1 and O2 interface connectivity is established with SMO including RAN OAM functions /O2 related functions. • R1 interface connectivity is established between rApp and Non-RT RIC framework. Pre-conditions • Network is operational. Begins when Operator specified trigger condition or event is detected. Step 1 (M) rApp sent discovery request to SMO/Non-RT RIC framework to discover O2 related services. Step 2 (M) Non-RT RIC framework resolves the request and sent service discovery result to rApp. Step 3 (M) rApp requests O2 related data from O2 related functions (NFO/FOCOM). rApp can request data such as compute utilization, memory usage, availability of network function, performance of API responses from NF deployments, status of AAL logical processing unit, etc. Step 4 (M) Non-RT RIC framework processes the data request and request O2 related function (NFO and FOCOM) to collect O2ims and O2dms related data. Step 5 (M) O2 related function (FOCOM) performs data request and collection from O-Cloud (IMS). Step 6 (M) O2 related function (NFO) performs data request and collection from O-Cloud (DMS). Step 7 (M) SMO/Non-RT RIC framework collect and store O2ims and O2dms related data. Step 8 (M) SMO/Non-RT RIC framework delivers O2 related data towards rApp over R1 Interface. Step 9 (M) rApp requests RAN OAM related data to be collected from E2 nodes such as availability of E2 node, accessibility KPIs, UEs connected, user traffic and alarms reported on interface level, etc., to understand performance of E2 nodes. Step 10 (M) Non-RT RIC framework forwards request to E2 nodes through RAN OAM functions. Step 11 (M) RAN OAM functions request and collect data from E2 nodes. Step 12 (M) SMO/Non-RT RIC framework to collect and store RAN OAM related PM data. Step 13 (M) SMO/Non-RT RIC framework delivers RAN OAM related O1 data towards rApp. Step 14 (O) AI/ML models can be retrained either on Non-RT RIC framework or on rApp. If Non-RT RIC framework is hosting retraining, then rApp selects AI/ML model and initiate retraining on Non-RT RIC framework. See note 1. Step 15 (O) Upon receiving retraining request from rApp. Non-RT RIC framework initiates AI/ML model retraining. Step 16 (O) rApp monitors retrained AI/ML model and retrieves retrained AI/ML model. See note 2. Step 17 (O) If AI/ML model retraining is hosted by rApp then AI/ML model to be retrained on rApp itself. Step 18 (O) Once AI/ML model retraining is performed, AI/ML models are deployed and activated for inferencing for which rApp constantly monitors O1 and O2 related data. Step 19 (M) rApp identifies the requirement of fault recovery or maintenance of O-Cloud node and formulate policy/recommendations to drain, which can relocate network function to another O-Cloud node. rApp can sent policy or recommendations to O2 related function (FOCOM) for O-Cloud node draining. rApp to include the necessary identifiers of the O-Clouds to be drained. See note 3. Step 20 (M) Post receiving policy/recommendations for draining of O-Cloud node, O-Cloud to perform NF relocation (optionally) and drain O-Cloud node as specified in O-RAN.WG6.ORCH-USE-CASES [14], clause 3.12.2, then O2 related function (FOCOM) Informs rApp about NF relocation (optionally) and drain O-Cloud node R1 Interface. Step 21 (M) rApp monitors O1 and O2 PM, FM data from NF for any undesirable behaviour. Ends when If network function/E2 node becomes non-operational or when the operator disables the rApp. Exceptions N/A. Post Conditions rApp continues close loop monitoring of O-Cloud node telemetry. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 95 Use Case Stage Evolution / Specification NOTE 1: AI/ML procedures mentioned here can be subject to change based on future work on AI/ML workflow in WG2. NOTE 2: AI/ML model retrieval procedure over R1 interface is for illustration purpose, can be subject to change based on future work on AI/ML workflow in WG2. NOTE 3: Identifiers mentioned above are not specified in the present document. The workflow for Non-RT RIC based O-Cloud node draining is shown in figure 4.9.1.3-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 96 Figure 4.9.1.3-1: Workflow for Non-RT RIC based O-Cloud node draining ETSI ETSI TS 104 226 V10.1.0 (2025-08) 97
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.4 Required data
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.4.0 Introduction
|
This clause contains the input and output data required for O-Cloud node drain.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.4.1 Input data
|
O2 related data (O-Cloud FCAPS data): 1) DMS telemetry data to understand state and health of network functions deployed on O-Cloud node. 2) IMS telemetry data to understand the state and health of O-Cloud nodes. 3) IMS/DMS inventory data to understand configuration of nodes and network functions deployments on O-Cloud nodes. 4) Monitoring data such as compute utilization, memory usage, availability of network function, performance of API responses from NF deployments, status of AAL logical processing unit, etc. RAN OAM related data (E2 node and O-RU/network function data): 1) The measurement counters and KPIs (as defined by 3GPP and will be extended for O-RAN use cases) should be appropriately aggregated by cell, slice, etc. 2) E2 node KPIs such as availability of E2 node, accessibility KPIs, UEs connected, user traffic and alarms reported on interface level, etc. to understand whether E2 nodes behaving as usual.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
4.9.1.4.2 Output data
|
O2 related data: 1) rApp to provide policy based guidance or trigger recommendations to drain O-Cloud node towards O2 related function (NFO/ FOCOM).
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5 Requirements
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.1 Functional requirements
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.1.1 Non-RT RIC functional requirements
|
The Non-RT RIC functional requirements are captured in table 5.1.1-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 98 Table 5.1.1-1: Non-RT RIC functional requirements REQ Description Note REQ-Non-RT-RIC-FUN1 Non-RT RIC shall support data retrieval and analysis; the data can include performance, configuration or other data related to the application (recommended data shown in required data clause for different use cases). REQ-Non-RT-RIC-FUN2 Non-RT RIC shall support relevant AI/ML model training based on the data in [REQ-Non-RT-RIC-FUN1] for non-real-time optimization of configuration parameters in RAN or Near-RT RIC, as applicable for the use case. REQ-Non-RT-RIC-FUN3 Non-RT RIC shall support relevant AI/ML model training based on the data in [REQ-Non-RT-RIC-FUN1] for generating/optimizing policies and intents to guide the behaviour of applications in Near-RT RIC or RAN, as applicable for the use case. REQ-Non-RT-RIC-FUN4 Non-RT RIC shall support training of relevant AI/ML models based on the data in [REQ-Non-RT-RIC-FUN1] to be deployed/updated in Near-RT RIC as required by the applications. REQ-Non-RT-RIC-FUN5 Non-RT RIC shall support performance monitoring and evaluation. REQ-Non-RT-RIC-FUN6 Non-RT RIC shall support a fallback mechanism to prevent drastic degradation/fluctuation of performance, e.g. to restore to the previous policy or configuration. REQ-Non-RT-RIC-FUN7 Non-RT RIC shall be able to produce enrichment information through data analysis. REQ-Non-RT-RIC-FUN8 Non-RT RIC shall be able to request O1 reconfiguration for non-real- time optimization of configuration parameters in E2 nodes and/or Near-RT RIC, as applicable for the use case. REQ-Non-RT-RIC-FUN9 Non-RT RIC shall support retrieval of external information as applicable for the use case.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.1.2 A1 interface functional requirements
|
The A1 interface functional requirements are captured in table 5.1.2-1. Table 5.1.2-1: A1 interface functional requirements REQ Description Note REQ-A1-FUN1 A1 interface shall support communication of policies from Non-RT RIC to Near-RT RIC. REQ-A1-FUN2 A1 interface shall support AI/ML model deployment and update from Non-RT RIC to Near-RT RIC. REQ-A1-FUN3 A1 interface shall support communication of enrichment information from Non-RT RIC to Near-RT RIC. REQ-A1-FUN4 A1 interface shall support feedback from Near-RT RIC for monitoring AI/ML model performance. REQ-A1-FUN5 A1 interface shall support the policy feedback from Near-RT RIC to Non-RT RIC.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.1.3 R1 interface functional requirements
|
The R1 interface functional requirements are captured in table 5.1.3-1. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 99 Table 5.1.3-1: R1 interface functional requirements REQ Description Note REQ-R1-FUN1 R1 interface shall support registration of services. Based on REQ- nRTRfW-R1r-10 REQ-R1-FUN2 R1 interface shall support discovery of registered services. Based on REQ- nRTRApp-R1r-30 REQ-R1-FUN3 R1 interface shall support authentication of rApp. Based on REQ- nRTRfW-R1r-10 REQ-R1-FUN4 R1 interface shall support authorization of service request. Based on REQ- nRTRfW-R1r-10 REQ-R1-FUN5 R1 interface shall support subscription and unsubscription of notifications for added/updated/removed registered services. Based on REQ- nRTRfW-R1r-120 REQ-R1-FUN6 R1 interface shall support registration of data types. Based on REQ- nRTRfW-R1r-30 REQ-R1-FUN7 R1 interface shall support subscription of data types. Based on REQ- nRTRfW-R1r-30 REQ-R1-FUN8 R1 interface shall support A1 related services. REQ-R1-FUN9 R1 interface shall support O1 related services. REQ-R1-FUN10 R1 interface shall support O2 related services. REQ-R1-FUN11 R1 interface shall support AI/ML workflow services. REQ-R1-FUN12 R1 interface shall support services related to network slice subnets. Refer to O-RAN.WG1.Slicin g-Architecture [15] for details.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.2 Non-functional requirements
| |
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.2.1 Non-RT RIC non-functional requirements
|
The Non-RT RIC non-functional requirements are captured in table 5.2.1-1. Table 5.2.1-1: Non-RT RIC non-functional requirements REQ Description Note REQ-Non-RT-RIC-NON-FUN1 Non-RT RIC shall not update the same policy or configuration parameter for a given Near-RT RIC or RAN function more often than once per second. REQ-Non-RT-RIC-NON-FUN2 Non-RT RIC shall be able to update policies in several Near-RT RICs.
|
7e4409c8abd9ce1b6236b46207489f8c
|
104 226
|
5.2.2 A1 interface non-functional requirements
|
The A1 interface non-functional requirements are captured in table 5.2.2-1. Table 5.2.2-1: A1 interface non-functional requirements REQ Description Note 5.2.3 R1 interface non-functional requirements The R1 interface non-functional requirements are captured in table 5.2.3-1. Table 5.2.3-1: R1 interface non-functional requirements REQ Description Note ETSI ETSI TS 104 226 V10.1.0 (2025-08) 100 Annex A (informative): Change history Date Version Information about changes 2019.05.06 01.00 • First version with initial set of use cases. Describes the set of use cases that have been approved within O-RAN WG2. 2020.03.01 02.00 • Updates to traffic steering use case. • Updates to QoE use case. • Addition of a new use case (QoS based resource optimization). • Addition of a new use case (Context-based dynamic handover management for V2X). • Removal of 3D-MIMO beamforming optimization use case from WG2 and moving it to UCTG as it does not have WG2 specific impact. 2020.11.01 02.01 • Updates to traffic steering use case with multi-access network scenario enhancements. 2021.03.10 03.00 • Addition of a new use case (RAN slice SLA assurance). 2021.07.19 04.00 • Updates to Non-RT RIC related definitions and abbreviations. • Additions of R1 functional requirements. 2021.11.24 05.00 • Addition of NSSI resource optimization use case. • Updates to RAN slice SLA assurance use case. 2022.04.15 06.00 • Load balancing related updates to RAN slice SLA assurance use case. • Addition of a section for multiple massive MIMO optimization use cases. • Addition of three massive MIMO sub-use cases to "Massive MIMO optimization use cases" section; 1) Massive MIMO Grid-of-Beams Beamforming (GoB BF) optimization use case, 2) Massive MIMO Non-GoB Beamforming (Non-GoB BF) optimization use case and 3) MIMO optimization via MIMO DL Tx power optimization, MU-MIMO pairing, and MIMO mode selection use case. 2023.03.24 07.00 • Alignment with latest O-RAN TS template. 2023.07.27 08.00 • Addition of a new use case (O-Cloud resource optimization). • Enhancements to one use case (Energy savings – Carrier switch off/on). • Minor addition to an A1 policy (frequency preference). 2023.11.16 09.00 • Alignment to the latest ODR template. • Addition of NES use case: - MDT/Trace measurement metrics for cell & carrier switched off/on - Advanced sleep mode - Policy-based RF channel reconfiguration 2024.03.30 10.00 • Addition of a massive MIMO sub-use case (initial access). • Updates for O-RAN Drafting Rules (ODR) compliancy. 2024.07.12 10.01 • Editorial changes for O-RAN Drafting Rules (ODR) compliancy. ETSI ETSI TS 104 226 V10.1.0 (2025-08) 101 History Document history V10.1.0 August 2025 Publication
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
1 Scope
|
This Technical Specification (TS) specifies the technical realization of the handling of calls originated by a GSM mobile subscriber and calls directed to a GSM mobile subscriber, up to the point where the call is established. Normal release of the call after establishment is also specified. The handling of DTMF signalling and Off-Air Call setup (OACSU) are not described in this specification. The details of the effects of GSM supplementary services on the handling of a call are described in the relevant GSM 03.8x and GSM 03.9x series of specifications. The specification of the handling of a request from the HLR for subscriber information is not part of basic call handling, but is required for both CAMEL (GSM 03.78 [5]) and optimal routeing (GSM 03.79 [6]). The use of the Provide Subscriber Information message flow is shown in GSM 03.78 [5] and GSM 03.79 [6]. The specification of the handling of data calls re-routed to a SIWFS is described in GSM 03.54 [4]. The logical separation of the MSC and VLR (shown in clauses 4, 5 & 7), and the messages transferred between them (described in clause 8) are the basis of a model used to define the externally visible behaviour of the MSC/VLR, which is a single physical entity. They do not impose any requirement except the definition of the externally visible behaviour. If there is any conflict between this specification and the corresponding stage 3 specifications (ETS 300 557 [14], ETS 300 590 [16] and ETS 300 599 [17]), the stage 3 specification shall prevail.
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
2 Normative references
|
The following documents contain provisions which, through reference in this text, constitute provisions of the present document. - References are either specific (identified by date of publication, edition number, version number, etc.) or non-specific. - For a specific reference, subsequent revisions do not apply. - For a non-specific reference, the latest version applies. - A non-specific reference to an ETS shall also be taken to refer to later versions published as an EN with the same number. [1] ETS 300 500 (1994): " Digital cellular telecommunications system (Phase 2); Principles of telecommunication services supported by a GSM Public Land Mobile Network (PLMN) (GSM 02.01)”. [2] ETS 300 523 (1994): " Digital cellular telecommunications system (Phase 2); Numbering, addressing & identification (GSM 03.03)”. [3] ETS 300 534 (1994): " Digital cellular telecommunications system (Phase 2); Security related network functions (GSM 03.20)”. [4] GSM 03.54 (TS 101 252) ”Digital cellular telecommunications system (Phase 2+);Description for the use of a Shared Inter Working Function /SIWF) in a GSM PLMN Stage 2 ”. [5] TS 101 044 (GSM 03.78): "Digital cellular telecommunications system (Phase 2+); Customized Applications for Mobile network Enhanced Logic (CAMEL) - Stage 2. [6] TS 101 045 (GSM 03.79): "Digital cellular telecommunications system (Phase 2+); Support of Optimal Routeing (SOR); Technical Realization". [7] ETS 300 542 (1994): " Digital cellular telecommunications system (Phase 2); Line identification supplementary services - Stage 2 (GSM 03.81)”. ETSI TS 101 043 V5.6.0 (1998-11) 9 GSM 03.18 version 5.6.0 Release 1996 [8] ETS 300 543 (1994): "Digital cellular telecommunications system (Phase 2); Call Forwarding (CF) supplementary services - Stage 2 (GSM 03.82)”. [9] ETS 300 544 (1994): "Digital cellular telecommunications system (Phase 2); Call Waiting (CW) and Call Hold (HOLD) supplementary services - Stage 2 (GSM 03.83)”. [10] ETS 300 545 (1994): "Digital cellular telecommunications system (Phase 2); Multi Party (MPTY) supplementary services - Stage 2 (GSM 03.84)”. [11] ETS 300 546 (1994): "Digital cellular telecommunications system (Phase 2); Closed User Group (CUG) supplementary services - Stage 2 (GSM 03.85)”. [12] ETS 300 547 (1994): "Digital cellular telecommunications system (Phase 2); Advice of Charge (AoC) supplementary services - Stage 2 (GSM 03.86)”. [13] ETS 300 548 (1994): "Digital cellular telecommunications system (Phase 2); Call Barring (CB) supplementary services - Stage 2 (GSM 03.88)”. [14] ETS 300 557 (1995): "Digital cellular telecommunications system (Phase 2); Mobile radio interface layer 3 specification (GSM 04.08)”. [15] ETS 300 582 (1994): "Digital cellular telecommunications system (Phase 2); General on Terminal Adaptation Functions (TAF) for Mobile Stations (MS) (GSM 07.01)”. [16] ETS 300 590 (1995): "Digital cellular telecommunications system (Phase 2); Mobile-services Switching Centre - Base Station System (MSC - BSS) interface Layer 3 specification (GSM 08.08)”. [17] ETS 300 599 Fourth Edition (1997): "Digital cellular telecommunications system (Phase 2); Mobile Application Part (MAP) specification (GSM 09.02)”. [18] ETS 300 604 (1994): "Digital cellular telecommunications system (Phase 2); General requirements on interworking between the Public Land Mobile Network (PLMN) and the Integrated Services Digital Network (ISDN) or Public Switched Telephone Network (PSTN) (GSM 09.07)”. [19] ETS 300 605 (1995): "Digital cellular telecommunications system (Phase 2); Information element mapping between Mobile Station - Base Station System (MS - BSS) and Base Station System - Mobile-services Switching Centre (BSS - MSC) Signalling procedures and the Mobile Application Part (MAP) (GSM 09.10)”. [20] ETS 300 627 (1996): "Digital cellular telecommunications system (Phase 2); Subscriber and equipment trace (GSM 12.08)”. [21] ETS 300 356-1 (1995): "Integrated Services Digital Network (ISDN); Signalling System No. 7; ISDN User Part (ISUP) version 2 for the international interface; Part 1: Basic services”. [22] ITU-T Recommendation Q.850 (1996): "Usage of cause and location in the Digital Subscriber Signalling System No. 1 and the Signalling System No. 7 ISDN User Part”.
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
3 Definitions and abbreviations
| |
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
3.1 Definitions
|
For the purposes of the present document, the following definitions apply: A subscriber: The calling mobile subscriber. B subscriber: The mobile subscriber originally called by the A subscriber. C subscriber: The subscriber to whom the B subscriber has requested that calls be forwarded. The C subscriber may be fixed or mobile. ETSI TS 101 043 V5.6.0 (1998-11) 10 GSM 03.18 version 5.6.0 Release 1996 Location Information: Information to define the whereabouts of the MS, and the age of the information defining the whereabouts.
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
3.2 Abbreviations
|
For the purposes of the present document, the following abbreviations apply: A&O Active & Operative ACM Address Complete Message ANM ANswer Message AoC Advice of Charge BC Bearer Capability BOIC-exHC&BOIZC Barring of Outgoing International Calls except those directed to the HPLMN Country & Barring of Outgoing InterZonal Calls BOIZC Barring of Outgoing InterZonal Calls BOIZC-exHC Barring of Outgoing InterZonal Calls except those directed to the HPLMN Country CFB Call Forwarding on Busy CFNRc Call Forwarding on mobile subscriber Not Reachable CFNRy Call Forwarding on No Reply CFU Call Forwarding Unconditional CLIP Calling Line Identity Presentation CLIR Calling Line Identity Restriction COLP COnnected Line identity Presentation COLR COnnected Line identity Restriction CUG Closed User Group CW Call Waiting FTN Forwarded-To Number FTNW Forwarded-To NetWork GMSCB Gateway MSC of the B subscriber HLC Higher Layer Compatibility HLRB The HLR of the B subscriber HPLMNB The HPLMN of the B subscriber IAM Initial Address Message IPLMN Interrogating PLMN - the PLMN containing GMSCB IWU Inter Working Unit LLC Lower Layer Compatibility MO Mobile Originated MPTY MultiParTY MT Mobile Terminated NDUB Network Determined User Busy NRCT No Reply Call Timer PRN Provide Roaming Number SIFIC Send Information For Incoming Call SIFOC Send Information For Outgoing Call SIWF Shared Inter Working Function SIWFS SIWF Server. SIWFS is the entity where the used IWU is located. SRI Send Routeing Information UDUB User Determined User Busy VLRA The VLR of the A subscriber VLRB The VLR of the B subscriber VMSCA The Visited MSC of the A subscriber VMSCB The Visited MSC of the B subscriber VPLMNA The Visited PLMN of the A subscriber VPLMNB The Visited PLMN of the B subscriber ETSI TS 101 043 V5.6.0 (1998-11) 11 GSM 03.18 version 5.6.0 Release 1996
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
4 Architecture
|
Subclauses 4.1 and 4.2 show the architecture for handling a basic MO call and a basic MT call. A basic mobile-to- mobile call is treated as the concatenation of an MO call and an MT call.
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
4.1 Architecture for an MO call
|
A basic mobile originated call involves signalling between the MS and its VMSC via the BSS, between the VMSC and the VLR and between the VMSC and the destination exchange, as indicated in figure 1. MS VMSCA VLRA VPLMNA Radio I/F signalling SIFOC Complete call IAM (ISUP) BSSA 'A' I/F signalling Figure 1: Architecture for a basic mobile originated call In figure 1 and throughout this specification, the term ISUP is used to denote the telephony signalling system used between exchanges. In a given network, any telephony signalling system may be used. When the user of an MS wishes to originate a call, the MS establishes communication with the network using radio interface signalling, and sends a message containing the address of the called party. VMSCA requests information to handle the outgoing call (SIFOC) from VLRA, over an internal interface of the MSC/VLR. If VLRA determines that the outgoing call is allowed, it responds with a Complete Call. VMSCA: - establishes a traffic channel to the MS; and - constructs an ISUP IAM using the called party address and sends it to the destination exchange. NOTE: When the non-loop method is used for data calls, the IAM is sent to the SIWFS.
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
4.2 Architecture for an MT call
|
A basic mobile terminated call involves signalling as indicated in figure 2. Communication between VMSCB and the MS is via the BSS, as for the mobile originated case. The IPLMN, containing GMSCB, is in principle distinct from HPLMNB, containing HLRB, but the practice for at least the majority of current GSM networks is that a call to a GSM MS will be routed to a GMSC in HPLMNB. ETSI TS 101 043 V5.6.0 (1998-11) 12 GSM 03.18 version 5.6.0 Release 1996 IPLMN GMSCB VPLMNB HLRB HPLMNB IAM (ISUP) IAM (ISUP) Send Routeing Info/ack Provide Roaming Number/ack Radio I/F signalling MS VLRB VMSCB SIFIC Page/ack Complete call BSSB Figure 2: Architecture for a basic mobile terminated call When GMSCB receives an ISUP IAM, it requests routeing information from HLRB using the MAP protocol. HLRB requests a roaming number from VLRB, also using the MAP protocol, and VLRB returns a roaming number in the Provide Roaming Number Ack. HLRB returns the roaming number to GMSCB in the Send Routeing Info ack. GMSCB uses the roaming number to construct an ISUP IAM, which it sends to VMSCB. When VMSCB receives the IAM, it requests information to handle the incoming call (SIFIC) from VLRB, over an internal interface of the MSC/VLR. If VLRB determines that the incoming call is allowed, it requests VMSCB to page the MS. VMSCB pages the MS using radio interface signalling. When the MS responds, VMSCB informs VLRB in the Page ack message. VLRB instructs VMSCB to connect the call in the Complete call, and VMSCB establishes a traffic channel to the MS.
|
4adb5594a028990749bbb6b19a2cf30e
|
101 043
|
5 Information flows
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.