A New Indoor Positioning Technique Based on Neural Network

摘要

This paper presents a novel indoor positioning technique based on neural network. Each sensor was modeled by an independent neural network in order to catch the mapping between the object's position coordinate and the received signal strengths from different directions and distances. Unlike the traditional neural network application in this field, the position coordinate of the object was used as the input and the received signal strength was used as the output in the neural model. We assume that if the distance between sensor and object is fixed, then the sensed signal strengths from different directions should be the same. A new least-squares location estimation method was developed to calculate the object's position in accordance with the received signal strengths sensed by different sensors and the estimated coordinates given by all neural networks. From the simulation results shown, it is clearly found that the new indoor position technique developed is more accurate and more stable than the traditional neural network approaches.