Indoor Positioning System Based on BP Neural Network Optimized by Genetic Algorithm Using ZigBee Wireless Sensor Network

Abstract

Accurate and efficient indoor positioning is increasingly essential for a wide range of applications. Traditional indoor positioning systems, such as those using RFID, ZigBee, or ultra-wideband technology, have been hindered by multipath propagation, signal reflection, and interference in complex environments. This paper proposes an indoor positioning system utilizing a BP (Back Propagation) neural network optimized by a genetic algorithm, based on the ZigBee wireless sensor network. The method addresses inaccuracies in received signal strength indicator (RSSI)-based non-ranging algorithms, improving signal acquisition and filtering for higher precision. Experimental results demonstrate that the proposed system achieves an average positioning error of just 0.22 meters for non-training points within a 2m × 2m range, marking a significant improvement in indoor positioning accuracy.