Deep neural networks in low energy algorithms for wireless sensor networks

Cluster routing protocols are one of the ways to effectively reduce the energy consumption, but most of the cluster protocols have high dependence on probability functions, the cluster head distribution method is random and poorly balanced, and there is a large amount of redundant data at nodes, which accelerates node energy consumption and affects the over-all network information transmission smoothness and effectiveness. Neural networks have good data processing capability, self-adaptive capability and learning capability, which can make up for the lack of performance of wireless sensor networks. APTEEN is a hybrid protocol. The periodicity and related thresholds of the teen protocol can be set according to user needs and application types. It can not only collect data periodically but also make rapid response to emergencies. Therefore, this paper incorporates deep neural networks in the APTEEN routing protocol and constructs a model and data fusion algorithm. Experimental results show that a more balanced distribution on the basis of effective clustering, maintain the number of cluster heads per round fluctuating in a small range, maintain the stability of model performance, enhance load balancing, and reduce node energy consumption. In addition, the convolutional self-coding model can help the cluster heads to deal with redundant data effectively, improve data classification accuracy, delay the generation of the first dead node, and extend the network life cycle. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.