Research On Wireless Sensor Network Fingerprint Location Method Based On Extreme Learning Machine

Although GPS global positioning system(GPS) has been well developped,it is very poor in the indoor environment.Because of the blocked by the buildings and walls,some of them can not even received GPS signals.Therefore, It is pay extensive attention by scholars at home and abroad to indoor positioning technology.It has become a hotspot research for indoor positioning technology that based on the position of the WiFi fingerprint.A feedforward neural network named extreme learning machine(ELM) and kernel extreme learning machine(KELM) are applied to the probelm of indoor fingerprint positioning based on the nolinear location fingerprinting. The fast learning speed and tightness of KELM network can reduce algorithm offline learning time and improve its generalization.It can solve the problem of online fingerprint location by online sequential extreme learning Machine(OS-ELM).The result shows that ELM has good ability of location by comparing with the widely used algorithms,through experimental verification of positioning performance.The research of thsis includes as follows:(1) Through analyzing the basic characteristics and learning algorithm of ELM and KELM.This article is researching on the improvement effect of ELM kernel function and OS-ELM theory of learning algorithm.(2) Based on the position of the WiFi fingerprint research.It is Including the received signal strength indication(RSSI) of information collection,the mode of access point(AP),the number of AP and the influence of multidimensional orientation fingerprint information for positioning.(3) Research on positioning performance of ELM,KELM and OS-ELM.Through analyzing localization performance of KELM by changing the environment of offline data,WiFi access point and multidimensional orientation fingerprint information.Under the same condition,experimental results suggest that KELM has faster speed and higher precision than ELM,SVM,BP,kNN and WkNN.Experimental results show that the positioning precision of the OS-ELM meet the needs of indoor location for online positioning.