Using a Backpropagation Neural Network in WSN for NOx Detection

Abstract

Abstract- This study is aim to design and simulate a backpropagation neural network in wireless sensor network (WSN) in order to process incoming data from number of sensors and training the network with using a training function to optimize the result and make a proper decision according to the purpose of design.

The backpropagation neural network considers as one of the most powerful artificial neural networks using an optimize method to upgrade the biases and weights on the ‘hidden and output layers.

Particle Swarm Optimization (PSO) had been used in this paper as a training function to optimize the result in neural network. and introduce the best result with mean square error (MSE) equal to zero.

The system will be simulated and design using MATLAB software package, and the test of processes given a results matching to desired results of output with mean square error equal to zero with 46 iterations that improve the previous work done in this field.