Energy-efficient Federated Edge Learning for Internet of Vehicles via Rate-Splitting Multiple Access

Abstract

Rate-Splitting Multiple Access (RSMA) has recently found favor in the multi-antenna-aided wireless network. Considering the heterogeneous demands and the qualities of Channel State Information at the Transmitter (CSIT), RSMA is crucial for advancing the quality of Internet of Vehicles (IoV) operations. However, it is challenging to incorporate RSMA into IoV operations under realistic autonomous driving constraints. To tackle this problem, we propose an RSMA-based IoV system to achieve energy-efficient Federated Edge Learning (FEEL) downlink broadcasting for autonomous driving. Specifically, the proposed framework is designed for transmitting the unicast control messages to the IoV platoon, as well as for broadcasting the global FEEL model to each vehicle. Thus, Non-Orthogonal Unicasting and Multicasting (NOUM) transmission is considered, where the unicast control message for vehicular platoons and broadcast FEEL model for autonomous driving can be transmitted simultaneously. Given the non-convexity of the formulated problems, a Successive Convex Approximation (SCA) approach is developed for solving the FEEL-based downlink problem. The simulation results show that our proposed RSMA-based IoV system can outperform the Multi-User Linear Precoding (MU-LP) by means of the NOUM and conventional Non-Orthogonal Multiple Access (NOMA) system that only supports unicast. In addition, the SCA method is shown to generate near-optimal solutions in reduced computation time.