An Indoor Localization Algorithm Based on RBF Neural Network Optimized by the Improved PSO

Abstract

Aimed at the problem of large localization error based on indoor received signal strength indication (RSSI), a RBF neural network (RBFNN) localization algorithm is proposed optimized by improved particle swarm optimization (PSO). Combined with resource allocation network (RAN), the number of nodes in hidden layer increase dynamically to determine the center of RBFNN, the number of nodes in hidden layer and spread constant. The inertia weight of PSO is improved to advance the global search ability of PSO and optimize the output weight of RBFNN. Finally, the optimized RBFNN is used for indoor RSSI positioning. Simulation and experimental results show that the improved localization algorithm has higher positioning accuracy.