Multi-Access Edge Computing Based Vehicular Network: Joint Task Scheduling and Resource Allocation Strategy

Abstract

With the progress of 5G technology, Internet of vehicles (IoV) has entered a stage of rapid development. The deficiencies of cloud computing, such as centralized deployment, network congestion and the long distance from the terminal, make it hard to satisfy vehicular service demands. As a promising computing paradigm, Multi-access Edge Computing (MEC) can be applied to IoV scenarios. The resource management in the changeable MEC environment has become a challenging problem to be addressed. In particular, it should be noticed that IoV tasks have different sensitivities to delay and need to be treated differently. In this article, we focus on the Joint Task Scheduling and Resource Allocation problem in the MEC based vehicular system, proposing a JTSRA algorithm. Firstly, we classify computing tasks into different priorities according to their delay tolerances. Based on this, in order to improve processing efficiency, we design an approach to adjust the order of tasks. Moreover, we further transform the computing resource allocation problem into the Markov decision process and solve it with reinforcement learning. Finally, the feasibility and effectiveness of the proposed scheme are verified by simulation.