Task Offloading Decision Algorithm for Vehicular Edge Network Based on Multi-dimensional Information Deep Learning

Traditional vehicle edge network task offloading decision is based on the nature of tasks and network status, and each vehicular node makes a distributed independent decision. Nevertheless, the network state considered in the decision is single and lacks global information, which is not conducive to the overall optimization of the system. Therefore, this paper proposes a task offloading decision algorithm for vehicular edge network based on deep learning of multi-dimensional information. With the optimization goal ofminimizing system overhead, the algorithm uses hybrid neural networks to deeply learn the state information of multi-dimensional networks and constructs the central task offloading decision model. A large number of simulation experiments show that the task offloading decision model trained by the hybrid neural network in this paper has high validity and accuracy when making the offloading decision and can significantly reduce system overhead and task computing delay.