231. INTEROPERABILITY BETWEEN FEDERATED LEARNING AND WIRELESS POWERED COMMUNICATIONS: MINIMIZING TRAINING DELAYS (Интероперабилност помеѓу федеративно учењеи безжично напојувани комуникации:Минимзирање на времето на тренирање

Slavche Pejoski; Marija Poposka; Zoran Hadzi-Velkov

doi:10.51466/JEEIT25101231041p

Slavche Pejoski Faculty of Electrical Engineering and Information Technologies, Ss. Cyril and Methodius University in Skopje, 18 Rugjer Bošković Str., 1000 Skopje, Republic of Macedonia
Marija Poposka Faculty of Electrical Engineering and Information Technologies, Ss. Cyril and Methodius University in Skopje, 18 Rugjer Bošković Str., 1000 Skopje, Republic of Macedonia
Zoran Hadzi-Velkov Faculty of Electrical Engineering and Information Technologies, Ss. Cyril and Methodius University in Skopje, 18 Rugjer Bošković Str., 1000 Skopje, Republic of Macedonia

DOI: https://doi.org/10.51466/JEEIT25101231041p

Keywords: wireless powered communication networks, resource allocation, federated learning

Abstract

Abstract: This paper explores an interoperability scenario between wireless powered transfer and federated learning (FL) technologies. In this scenario, the base station does not only coordinate training of the global FL model, but it also charges energy harvesting (EH) clients, which are responsible for training the local models. These EH clients are equipped with rechargeable batteries, enabling them to perform local processing and energy harvesting concurrently. We propose an efficient resource allocation scheme that optimizes both computing and communication parameters to minimize training latency. Simulation results show a significant latency reduction compared to state-of-the-art FL system that operates with non-overlapping local processing and energy harvesting phases.