Over-the-Air Federated Edge Learning with Integrated Sensing, Communication, and Computation
Dingzhu Wen, Sijing Xie, Cao Xiaowen, Yuanhao Cui, Weijie Yuan, Yang Zhaohui, Jie Xu, Yuanming Shi, Shuguang Cui · 2024
This paper studies an over-the-air federated edge learning (Air-FEEL) system with integrated sensing, communication, and computation (ISCC), in which one edge server coordinates multiple edge devices to wirelessly sense the objects and use the sensing data to collaboratively train a machine learning model for conducting tasks like human motion recognition. In this system, over-the-air computation (AirComp) is employed for edge devices to efficiently aggregate their local models. Under this setup, we analyze the convergence of the ISCC-enabled Air-FEEL, by particularly taking into account the wireless sensing noise and AirComp distortions. It is shown that a higher convergence rate is achieved with higher sensing and AirComp signal-to-noise ratios (SNRs), and a larger sum batch size of all devices. Particularly, a larger sum batch size can enhance the benefit of high sensing SNR as well as suppressing the stochastic gradient variance. Based on the convergence analysis, we design the ISCC parameters via jointly optimizing batch size control and resource allocation to maximize the loss function degradation while ensuring the latency in each round. Experimental simulations are conducted based on human motion recognition to evaluate the performance.