Research On WSN Indoor Location Algorithm Based On RF Fingerprint

With the rapid development of electronic, computer and wireless communication technology, people’s demand for location-based services in various fields is getting higher and higher. For indoor complex environments, GPS systems and cellular mobile networks for outdoor positioning can not meet the demand of indoor users’ precisely position, so finding a high-precision indoor positioning algorithm which is suitable for complex environment, low complexity and high environmental adaptability has become a research hotspot in recent years. Wireless sensor networks have been widely used in indoor positioning with its flexible network topology, low cost, easy to deploy and capable of autonomous networking, etc.Indoor localization algorithm which based on adaptive fingerprinting and WSN uses the easy measurement node received signal strength indication(RSSI), the target RSSI signal received in real time is matched with the radio frequency fingerprint which measured in the offline stage to obtain the target position. This method won the academic attention in the advantage of simple, efficient and it neither adder additional hardware nor consideration of the detailed model of each building. RF-based wireless sensor network fingerprint indoor positioning technology was researched in this paper and the main work and contribution are as follows:(1) The spatial and temporal characteristics of RSSI values has been researched and analysis in depth which based on the IRIS node and Mote Works test platform. This paper proposed a new dynamic maps update mechanism and it use the wireless sensor networks to monitor the spatial and temporal changes in the environment, and the same time, it analyzed the effect of the number of samples to RSSI distribution characteristics.(2) Based on the above mechanism, this paper gave two algorithms to build adaptive fingerprint map which based on linear regression model and the GABP neural network. In the linear regression model, it assumes that the nodes at different locations of received RSSI value from the same beacon nodes obey the linear correlation approximately and update the fingerprint map based on this correlation. In the GABP neural network algorithm, it use the genetic algorithm to optimize BP neural network and estimate the nonlinear space correlation between the signal strength at different locations to achieve real-time RF updates. The simulation results show that the GABP neural network algorithm can further enhance the fit of nonlinear, with strong environment adaptability.(3) In real-time positioning stage, the complex localization algorithm of Bayes-MAP+FSM was proposed in this paper for the environmental interference issues to the indoor mobile node. In this method, the use of Bayes-MAP estimate the initial position of the target, FSM calculation model is used to restrict the removable grid space, and it can greatly improve the positioning accuracy because of reducing the impossible position of transition. The simulation results show that the average positioning accuracy can reach 1.2 meters, and the maximum positioning error can be limited in 3 meters. The simulation results show that the positioning error of the composite algorithm can rech 2m with 85% and within 3m with 100 %, the average positioning accuracy can reach 1.2 meters.