An Effective Adaptive Autocorrelation-Based Neighboring Matching Location-Aware Computing in WLANs

Abstract

The recent advances of ubiquitous wireless infrastructures and requirements for high speed context-aware computing have created the opportunities to supply the high efficient location based service (LBS) in indoor wireless local area network (WLAN) environment. Because of the serious multi-path effect, unpredictable co-channel interference and inherent equipment noise, the measured signal strengths vary a lot in the real-world indoor environment. And this strength variation will also result in the performance deterioration of radio map-based neighboring matching algorithm. In response to this compelling problem, we propose the adoption of adaptive autocorrelation-based signal preprocessing method as a specific solution by effectively eliminating the singular strength from the original fingerprint set. Finally, the feasibility and effectiveness of autocorrelation-based preprocessing are also verified by decreasing about 33.4% and 32.9% of errors in k nearest neighbor (KNN) and weighted KNN (WKNN).