wintin0826/RF-only_channel_dataset_for_WiFo

Pretraining channel datasets for WiFo

📌 Overview

To fully unleash WiFo’s prediction capabilities across various CSI configurations, we have constructed a series of diverse 3D CSI datasets, generated through channel generator QuaDRiGa compliant with the 3GPP standards. We consider MISO-OFDM systems, where the BS is equipped with a UPA and the user has a single antenna. The adjacent antenna spacing is half the wavelength at the central frequency. A total of 19 datasets are simulated, indexed from D1 to D19, covering various STF CSI configurations, cell scenarios, and user speed. Among them, the first 16 datasets are used for pre-training, and the last three datasets are used for generalization testing. The detailed simulation configurations of each dataset are shown in Table 1, where fC represents the center frequency. There are seven 5G new radio (NR) frequency bands, eight 3GPP scenarios, and seven user speed ranges considered. For each CSI sample, the user has a random initial position and a straight-line motion trajectory, where the speed is uniformly selected within the corresponding speed range. Each dataset contains 12000 samples, which are randomly split into 9000, 1000, and 2000 samples for training, validation, and inference, respectively.

---

📜 Related paper

B. Liu, S. Gao, X. Liu, X. Cheng, L. Yang. “WiFo: Wireless Foundation Model for Channel Prediction. ” SCIENCE CHINA Information Sciences, June 2025, 68(6): 162302， https://arxiv.org/abs/2412.08908