FEEL-enhanced Edge Computing in Energy Constrained UAV-aided IoT Networks

Abstract

In this work, we investigate the performance of a federated edge learning (FEEL)-enhanced edge computing in unmanned aerial vehicles (UAV)-aided internet of things (IoT) system under the consideration of limited energy at each UAV. It consists of multiple UAVs which apply FEEL for local training and, then, transmit the required parameters to a centralized IoT-server. We use a new cost metric obtained by a linear combination of latency and energy consumption, and formulate an optimization problem to jointly optimize the central processing unit (CPU)-frequency during FEEL and allotted bandwidth under the consideration of the limited overall system bandwidth and energy available at each UAV. Due to the non-convex nature of the formulated problem, we propose a twin delayed deep deterministic policy gradient (TD3)-based algorithm that solves the problem and provides the optimum CPU frequency and allotted bandwidth to each user. We validate the accuracy and convergence of the proposed algorithm via exhaustive simulations and highlight its effectiveness by comparing its performance with that of deep deterministic policy gradient (DDPG) and deep Qnetwork (DQN)-based solutions.