A Hibernation Scheduling Method for Opportunistic Network Nodes Based on Convolutional Neural Network

Abstract

The conventional scheduling method is mainly based on the path scheduling of source nodes and destination nodes, and the failed nodes and redundant nodes affect the final scheduling effect. Therefore, a hibernation scheduling method of opportunistic network nodes based on convolutional neural network is designed. The dormant scheduling model of opportunistic network nodes is constructed, and the dormant network nodes in the overlapping areas of perception are fused to reduce the energy consumption of data transmission and ensure the efficiency of network message transmission. Based on convolutional neural network, the cross-layer scheduling residual of dormant opportunity network nodes is eliminated, the message characteristics of nodes scheduled in the previous layer are mapped to the current scheduling layer, and messages are propagated back to the previous layer during the scheduling process, thus alleviating the problem of network data gradient disappearing. The simulation results show that the scheduling effect of this method is better and can be applied to real life.