Federated Learning and Control at the Wireless Network Edge

Abstract

We are at the cusp of two transformational technologies, namely the fifth generation of wireless communication systems, known as 5G, and machine learning (ML). On the one hand, while the evolutionary part of 5G, enhanced mobile broadband (eMBB), focusing mainly on millimeter-wave transmissions has made significant progress, fundamentals of ultra-reliable and low-latency communication (URLLC), one of the major tenets of the 5G revolution, are yet to be fully understood. In essence, URLLC warrants a departure from average-based system design toward a clean slate design centered on tail, risk, and scale [1]. While risk is encountered when dealing with decision making under uncertainty, scale is driven by the sheer amount of devices, antennas, sensors, and actuators, all of which pose unprecedented challenges in network design, optimization, and scalability.