Indoor location algorithm of back propagation Neural Network based on residual analysis

In indoor environment, there are gross errors in random measured values of base station, which has effect on generalization ability of BP neural network and then results in low location accuracy. In order to improve location accuracy, location algorithm of BP Neural Network based on residual analysis is proposed, namely conducting pretreatment on measured values separately in training phase and location phase of BP neural network with twice residual analysis and getting rid of measured value with bigger error. The simulation result shows that such algorithm is better than BP algorithm both in aspects of convergence rate and location effect.