Research on Computation Offloading Method of Digital Twin Assisted Internet of Vehicles

The rapid development of information has brought about the explosive growth of the amount of data computation, and the intelligent vehicle terminal is inevitably affected, and its local computing resources are difficult to cope with the complex computing needs. In the face of such a dilemma, computation unloading is undoubtedly an effective means to alleviate the shortage of vehicle computing resources. However, most computing tasks in vehicle networking scenarios have the characteristics of high delay sensitivity, so the delay requirements will be relatively higher. How to shorten the delay under the highly dynamic vehicle is an urgent problem to be solved. In this paper, digital twin technology is used to obtain network dynamics and business requirements, build a vehicle-connected architecture based on digital twin, measure task design priorities, model the computational offloading problem in dynamic environment as a Markov decision process, generate real-time resource allocation strategies, thereby reducing latency and maximizing task utility. In the unloading part, an unloading algorithm based on dual deep Q network is proposed to minimize the task processing delay.