Joint Resource Allocation and Interference Management for Vehicular Networks

Abstract

Vehicular communication is a promising technology for the intelligent transportation system. With the growth of data requirement, the rapid evolvement of vehicular communication has been hindered by the constraints imposed by limited network resources and the resource sharing is benefit for this issue. Vehicle-to-vehicle (V2V) pairs, by sharing the spectrum of vehicle-to-infrastructure (V2I) links, can improve the network resource utilization significantly. However, they also bring out lots of interference to the V2I links, with the aim of enhancing the user experience quality and driving safety of the vehicular networks and then there needs suitable methods to solve it. In our study, we put forward an algorithm that combines one-to-many resource allocation and interference management to improve the spectrum efficiency for vehicular networks. Considering complex network environment with high mobility of vehicular networks, it is difficult to accurately model such requirements mathematically. We resolve the joint optimization problem with the assistance of deep reinforcement learning (DRL). Simulation results show that our proposed method exhibits significant advantages over the baseline algorithm in terms of network capacity and system interference. These findings indicate our proposed method can effectively reduce the cumulative interference between vehicular users and increase the total capacity of vehicular networks.