Research And Implementation Of RSSI Based Indoor Positioning System

The rapid development of Internet Industry has propelled enormous technical innovations and revolutions.To cater the increasing requirements of precise geo-location information in Remote monitor,smart home and intelligent transport system,various applications of indoor location technology,especially those embedded on mobile platforms,has emerged and sparked the research enthusiasms of the whole world.Though nowadays the outdoor geo-location technology is adequate for daily uses with the help of GPS,the indoor geo-location system still failed to meet the basic requirements of precision,simultaneity and robustness for real implementations.Due to the remarkable costs-effectiveness of RSSI based indoor geo-location system,increasing number of researchers are focusing on this area.In this paper we first introduce the widely used fingerprint location algorithm,To promote the precision of location,various machine learning algorithms such as KNN,SVM,decision trees,etc are combined with classic fingerprint location system.Then we implement our algorithm into a WiFi indoor location system and a Zigbee indoor location system respectively.The test results show that our new method increases the accuracy rate effectively.Affected by the indoor environment and conditions,performance test machine learning algorithms employed fluctuations,cannot achieve a stable and reliable positioning accuracy,so we modified the location fingerprint approach through involving compressed sensing technology,propose a de-noise algorithm and an orthogonal pretreatment algorithm.By this mean the problem of applying congressed sensing on location system will be solved.Finally,the results of different algorithms for positioning are compared and analyzed in this paper.