A Self-adaptive AP Selection Method Based on Information Theory in the Fingerprint Location Scene

摘要

With the improvement of location technology, the demand for location-based application services grows. The commonly used global navigation satellite system has a poor performance in the center of the city with dense buildings, the interior of buildings and other similar places. WLAN positioning has become a very convenient solution for indoor positioning since it does not need additional positioning equipment. The key to build and optimize the location fingerprint database is to select the appropriate APs. Current AP selection algorithms usually ignore the fact that the wireless transmission environment is very complex leading the signal strength of AP time-varying characteristics; and because of the low-delay requirements of indoor positioning, the traditional algorithms for network access with high complexity are not suitable for AP selection scene. To solve those problems, firstly we analyzed the upper bound of AP selection performance from the perspective of information theory, its mathematical expression is also been given; Then we proposed a novel AP selection model for fingerprint location, which considering both the time-varying signal and complexity cost. From the experimental results, positioning accuracy using the AP selection algorithm model proposed in this paper has been improved by 47% compared with the traditional AHP algorithm.