Intelligent data offloading learning model for autonomous vehicular networks

Abstract

AbstractWhen individuals are accustomed to receiving information in automobiles, mobile data offloading is becoming more common. However, the effects of movement of the vehicle in correspondence to relative speed and direction between vehicles, have a significant impact on mobile data offloading. An innovative deep learning algorithm namely, intelligent deep neural network (IDNN) is proposed for vehicle data offloading and an optimal algorithm namely, quasi opposition based C‐hen swarm optimization (QOCSO) is proposed for efficient vehicle resource allocation. Initially, the vehicles in the vehicular networks are clustered with the help of the cosine similarity‐based K‐means algorithm for transmitting the data in an energy‐aware manner. Then cluster heads (CHs) are optimally selected for the generated clusters using Boltzmann selection probability‐based earth worm algorithm. The selected CHs are responsible for collecting the data from the cluster members and that is forwarded to the roadside unit. Then the suitable mobile edge servers are selected according to the IDNN algorithm that offloads the data from the CHs to the appropriate server. These received tasks of the vehicles could be stored as a blockchain for providing security to the vehicular network and finally, the resource allocation of the incoming tasks to the vehicles is performed using the QOCSO algorithm. Experiment findings reveal that both offloading and resource scheduling techniques outperform existing state‐of‐the‐art vehicular network algorithms.