Joint Training and Resource Allocation Optimization for Federated Learning in UAV Swarm
Yun Shen,Yuben Qu,Chao Dong,Fuhui Zhou,Qihui Wu
DOI: https://doi.org/10.1109/jiot.2022.3152829
IF: 10.6
2022-01-01
IEEE Internet of Things Journal
Abstract:Unmanned aerial vehicles (UAVs) have been widely used to perform search and tracking tasks in military and civil fields. To perform these tasks autonomously, a swarm of multiple UAVs need to be endowed with intelligence through machine learning (ML). However, the traditional centralized ML cannot be directly applied in UAV networks, since it is challenging to transmit raw data with limited bandwidth and energy budget. As a distributed manner, federated learning (FL) is more suitable for UAV networks than traditional ML schemes in order to boost edge intelligence for UAVs. Considering the limited energy supply of UAVs, we study how to minimize UAVs’ overall training energy consumption by jointly optimizing the local convergence threshold, local iterations, computation resource allocation, and bandwidth allocation, subject to the FL global accuracy guarantee and maximum training latency constraint. The formulated nonconvex mixed-integer programming problem is solved by a joint training and resource allocation optimization algorithm. In addition, we also study how to solve the problem considering fairness among different UAVs by changing the objective to minimizing the maximum energy consumption of UAVs, and extend the aforementioned approach to this problem. Our simulation results show that while satisfying both the training accuracy and latency constraints, the proposed algorithm can reduce more UAVs’ overall training energy consumption and the maximum energy consumption in the UAV swarm than four baseline schemes.
computer science, information systems,telecommunications,engineering, electrical & electronic