WI-Fi based localization using RSSI fingerprinting

Abstract

Indoor localization has become inevitable in the technological sector especially when a majority of stand-alone as well as commercial applications require accurate location estimate for the users. Since GPS signals are insufficient for indoor localization of object Therefore, Wi-Fi based fingerprinting have received great attention for such localization, among the available Wi-Fi fingerprint-based localization techniques, K-Nearest Neighbor (KNN) is most popular. However, existing techniques failed to consider the fact that equal RSS spaces at different RSS level may not mean identical geometric distances in complex indoor environment in calculating the signal distance among the different RSS vectors. To address this issue, Feature-Scaling-based K-Nearest Neighbor (FS-KNN) algorithm is proposed for achieving improved localization accuracy. In this research, we have proposed the Genetic algorithm (GA) based localization to minimize the error distance between the estimated position and actual position. Genetic algorithm solves problems through natural selection and evolutionary processes based on genes. It works with a population of solutions and attempts to guide the research toward improvement, using the survival of the fitness, which can improve the accuracy, stability and robustness of the localization gadget receiving RF signal from access point. A series of cumulative distribution plots rectify the performance measurement of this research and show that our proposed technique give better results in comparison to the previous techniques. The simulations were performed in MATLAB