Indoor Localization via Candidate Fingerprints and Genetic Algorithm

摘要

WiFi-based indoor localization was proposed to be a practical method to locate WiFi-enabled devices due to the popularity of WiFi networks. However, it suffers from large localization errors (6 ～ 10 m). In this paper, we propose a novel localization scheme: indoor localization using candidate fingerprints (CFs) and genetic algorithm (GA). We come up with candidate fingerprints (CFs) selection to increase the probability of obtaining the best location estimations of indoor devices. Furthermore the GA are used to search for the optimal combination of CFs of each device using the relative distance constraint information. In addition, we provide an analytical model for selecting CFs to predict the probability of CFs could cover their true location of target device. The experimental results on realistic data set indicate that our method can reduce the 50% and 80% errors to 1.6 m and 2.4 m respectively. And typical running times for our simulations are only within a few seconds (less than 5 s).