Learning-Based Demand-Aware Communication Computing and Caching in Vehicular Networks

Abstract

With the development of communication technologies, more and more in-vehicle applications have emerged, which require complex computing and mass storage. In addition, different types of in-vehicle applications have different demands for communication and computing. This paper studies a demand-aware joint communication, computing and caching optimization problem by making full use of computing and caching resources in vehicular networks to meet demands of different services. We propose a vehicle-network cooperation learning framework that uses a deep reinforcement learning approach to enable dynamic allocation of communication, computing and caching resources, which can perform different resource allocation strategies based on different demands of different services for communication rate and computing rate. Simulation results with different schemes are presented to show that the proposed scheme improves the system performance and meets the demands of different services.