Grant-free random access in massive machine-type communications enables low-latency connectivity with minimal signaling. However, sporadic device activation requires efficient device activity detection. We propose a federated learning-based device activity detection approach, leveraging distributed training to enhance security and privacy while maintaining low computational complexity. Compared to existing methods, our solution achieves competitive detection performance, addressing scalability and security challenges in mMTC networks.
@article{arxiv.2503.13513,
title = {Federated Learning for Secure and Efficient Device Activity Detection in mMTC Networks},
author = {Ali Elkeshawy and Ibrahim Al Ghosh and Haifa Fares and Amor Nafkha},
journal= {arXiv preprint arXiv:2503.13513},
year = {2025}
}