Indoor Localization for Smart-handhelds with Stable Set of Wireless Access Points

Wi-Fi-based indoor localization plays a pivotal role in estimating the location of an indoor environment. This approach depends upon fingerprinting effort. However, fingerprint data are susceptible to factors like time and device heterogeneity. Thus, identifying stable Access Points (APs) is important such that, the signal for those APs do not vary much even when training and test conditions change. Localization with stable APs not only provides sustainable localization accuracy but also saves location prediction time. Consequently, in this paper, we propose an algorithm for identifying stable APs in an real indoor environment of our university campus subject to time and device heterogeneity. Our proposed algorithm is found to select 20 stable APs irrespective of time and device variation out of all APs (46) which give an average error of 1.3 meter and 90.13% localization accuracy using KNN whereas, considering all APs KNN can achieve an accuracy of 84.20% with an average error of 2.10 meter.