A Deployment Optimization for Wireless Sensor Networks Based on Stacked Auto Encoder and Probabilistic Neural Network
Hui Feng, Xu Chen, Bo Jin, Min Zhang
Abstract
In the field of consumer electronics, Wireless Sensor Networks (WSNs) can be used to build large-scale IoT systems and achieve intelligent data collection and processing. This paper proposes a deep neural network model SAE-PNN by Stacked Auto Encoder (SAE) and Probabilistic Neural Network (PNN) in a stack manner. A comprehensive training algorithm of SAE-PNN based on a group of orthogonal function bases is established by introducing a time-varying input and connection weight function, based on the conventional algorithm of unsupervised layer-by-layer initialization and gradient descent of deep neural network. Our proposed SAE-PNN model is used to predict the distance from unknown nodes to known nodes. Based on the collaborative work of nodes and network coverage, an optimal set of working nodes is established to reduce the energy loss of network nodes. A balance relationship between network coverage and energy consumption has been established to solve the problem of optimizing the balance in location coverage and the contradiction between individual nodes and the overall performance of the network. The experimental results show that the proposed method can improve the coverage of the monitored area and the perceived quality of service, and uses a distributed node synchronous scheduling mechanism to reduce the overall energy consumption of the network.