Distributed Computation Offloading Based on Deep Reinforcement Learning and Blockchain in Internet of Vehicles

With the rapid development of Internet of Vehicles (IoV), the traditional centralized cloud computing cannot efficiently satisfy the computation requirements of vehicles. The deployment of mobile edge server (MEC) on road side unit (RSU) can shorten the distance between vehicles and computing servers so as to reduce the service delay. In order to provide wide coverage and more computation resources, we introduce high altitude platform system (HAPS) as a central node for computation controlling in the IoV system. To improve data security and user privacy, blockchain technology is applied to verify and determine the offloading strategy between RSUs. Aiming at minimizing the sum of the transmission and computation time, an optimization problem is formulated under the power and offloading constraints. The deep reinforcement learning (DRL) with the soft actor-critic (SAC) algorithm is utilized to solve the problem. Simulation results demonstrate the time reduction by the application of HAPS and significant superiority of the proposed scheme compared to the existing approaches.