Research On Indoor Location Based On Amplitude Distribution Of Wi-Fi Signal

Location-based services,with great market demand and application value,showed a rapid growth,receiving widespread concern in various fields.Now,people are increasingly enjoying the convenience and comfort of positioning service when traveling.With the wide promotion of wireless network technology and mobile terminal,reliable and efficient indoor positioning technology plays a key role in the whole process.Most of other types of wireless positioning technologies need to purchase specific equipment,which will increase the amount of investment and is difficult to promote.While the indoor positioning technology,based on wireless Wi-Fi and signal strength,realizes intelligent terminal positioning by using existing router equipment in public places and specific positioning algorithm.However,in practical application,different Wi-Fi signal transmission power from different equipment,as well as variable environments,directly complicate the Wi-Fi signal,affecting positioning accuracy and increasing positioning difficulty.This paper develops an in-depth study and experiment on indoor positioning application and precision issue.Based on basic theory research,it studies the Wi-Fi amplitude characteristics and formulates a specific method of combining spectral clustering and SVR algorithm,which can improve the reliability and effectiveness of indoor positioning.Basic problems of Wi-Fi fingerprint positioning include how to guarantee the accuracy of fingerprint database in the early stage,and how to achieve the matching effect in the later stage.After pre-processing the original data,spectral clustering uses eigengap to optimize NJW algorithm to enhance the reliability of fingerprint database and ensure real-time data.After construction of the database,SVR regression learning will be used to establish the regression equation,waiting for positioning.In online positioning,wireless signal will be detected by handheld mobile terminal,then clustered and extracted for its feature,before output to the SVR positioning function to estimate the user's location.Experiment results in various experiment sites show that indoor positioning accuracy within 2 meters is 78.4% and 93.6% within 3 meters.Compared with previous performance,positioning accuracy has been significantly improved and positioning time shortened.