Dynamic indoor localization techniques based on Rssi in WLAN environment

Location-based services are coming closer and closer to our daily life. And one of many key techniques concerning that is indoor localization. But there are still many problems not yet solved on the accuracy of indoor localization. This paper proposes several localization techniques using received signal strength in indoor wireless local networks, based on Bayesian model. In training phase, information of neighboring grid locations' samples are considered in calculating the Gauss distribution of signals received from some access point for each grid. In positioning phase, anchor points selection and access point selection are done before positioning. After finishing the prior probability and likelihood calculation, a candidate set of grids is dynamically generated based on the prior estimated location. And a global positioning on all trained grids will be performed if the positioning result gets trapped in some confined areas, in these areas the posterior probability would be less than some threshold value concluded from the statistical results. Our experimental results, conducted on a real office environment, indicate that these dynamic positioning techniques would evidently improve the accuracy of localization.