Indoor Positioning Algorithm Study Based On Wi-Fi

In recent years, location-based services used for equipment positioning and resource management has become an indispensable basic services with the rapid development of IOT(internet of things), intelligent transportation and medical wisdom system. Although the GPS global positioning technology has been developed, there are still some apparent defects;For example, this technology cannot work in indoors, under bridges and buildings intensive areas due to the blocking for satellite signals. The Wi-Fi-based positioning technology has advantages of widely deployed, wide covering range, long transmission distance, fast information transmission speed and low cost, and attracted lots of attention.This paper makes a comprehensive research on Wi-Fi indoor positioning technologies and related algorithms and draws a conclusion that Wi-Fi indoor positioning is available and has a normal distribution. Indoor positioning signal method uses extended Kalman filter, Gaussian smoothing filter and mean filter to smooth the signal. The new idea that using corrective devices to update attenuation coefficients in real time has been proposed in ranging location. The results of this method show the improvement on positioning accuracy. According to the cumulative distribution function(CDF, Cumulative Distribution Function) of the distance errors, we draw a conclusion that the combination of Kalman filter and Gaussian smoothing filter is optimal. Meanwhile, we study the algorithms of buildings floors location and analyze the principle of measuring height with the pressure sensors and acceleration sensors to achieve indoor positioning in the vertical direction through the conversion between altitude and storey interval. A high probability interval screening method is adopted in fingerprint according to the law that the normal distribution of indoor signal, redrawing the room collection area and proposing the weight block method.Through the practical test of the above improved methods, the cumulative probability of ranging positioning error within 1.5 meters can reach more than 95% when selecting the combination of Kalman filter and Gaussian smoothing filter.The average accuracy rate of fingerprint algorithm is close to 80%. The test results show that many measures we take in this paper could improve the indoor positioning accuracy and precision effectively.Android platform may extend the pressure sensor for buildings floors positioning and add wall attenuation factor in the attenuation model In future studies. Optimization fingerprint database, shorten fingerprint algorithm positioning cycle. Initial coordinates of anchor nodes can refer GPS positioning results, or to determine its coordinates according to building size.