Distributed Beamforming for Cooperative Multi-cell ISAC: A Federated Learning Approach

Date

2024-01-01

Author

Jiang, Lai
Meng, Kaitao
Temiz, Murat
Hu, Jiaming
Masouros, Christos

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

33
views

0
downloads

Cite This

In this work, we propose a distributed framework based on the federated learning (FL) for beamforming design in multi-cell integrated sensing and communications (ISAC) systems. Our aim is to address the following dilemma: 1) Beamforming strategies based on solely local information may cause severe inter-cell interference (ICI) affecting both communication users and sensing receivers in the adjacent cells, leading to degraded network-level performance in communication and sensing, 2) Centralized beamforming strategies require the knowledge of global communication and sensing channel information, which incurs additional transmission overhead and latency. In the proposed framework, multiple base stations (BSs) jointly train a deep neural network (DNN) to cooperatively design the optimal beamforming matrices, aiming at maximizing the weighted sum of communication rate and radar information rate. To implement a fully decentralized design without channel information exchange among BSs, we develop a novel loss function to manage the interference leakage, which can be computed by only using local channel information. Numerical results demonstrate that the proposed method achieves performance comparable to optimization-based algorithms and surpasses closed-form solutions in terms of both communication rate and radar information rate.

URI

https://hdl.handle.net/11511/117818

Collections

Department of Electrical and Electronics Engineering, Conference / Seminar