Energy and Bandwidth-Aware Federated Learning in Wireless Internet of Things

Abstract

In this thesis, we examine the challenges associated with implementing Federated Learning (FL) in wireless systems that have constraints on energy and bandwidth. First, we investigate decentralized FL in a wireless system, taking into account the transmission energy budget of each client which fundamentally limits the range and bandwidth of data communications. To do so, we propose a model partitioning method that highlights the design choices available within the energy constraint -- sharing of a larger partition of the model among clients requires transmission range to shrink and therefore sharing with fewer neighboring nodes. It is non-obvious what the best setting of partition size/transmission range is, and we demonstrate that such a setting can be found for particular deployments.

This thesis delves further into integrating Deep Reinforcement Learning (DRL) into the FL system, offering a smarter approach to bandwidth allocation. We propose a DRL-empowered FL framework for wireless clients that utilizes a Deep Deterministic Policy Gradient (DDPG) agent at the central server to allocate communication bandwidth to each client. The DRL aims to reduce each clients' transmission energy by considering their respective channels and distances to the server. Along with the DRL agent, we perform model partitioning to trade-off the amount of information transmitted by each client and the accuracy of the central model.

Ultimately, this study applies these findings to wireless Internet of Things (IoT) systems, demonstrating the practicality of our FL solutions in distributed systems and how innovative Machine Learning (ML) techniques can foster a balanced and efficient wireless IoT under energy constraints. The main contribution is a novel method for bandwidth management and model partitioning, which is vital for fully realizing the potential of FL in environments limited by resource availability.