Enhancing WiFi-fingerprinting accuracy using RSS calibration in dual-band environments

Wi-Fi radio signals are commonly used to localize mobile users indoors. The use of a dual-band channel (2.4 GHz and 5 GHz) in mobile devices, however, significantly affects the accuracy of localization, because scanning all the channels requires a relatively long delay of approximately 5 seconds. During this scan time, a user may move tens of meters, depending on his or her walking speed. In this paper, we propose the self-calibration of Wi-Fi signals in a dual-band channel to solve the received signal strength (RSS) delay problem in Wi-Fi-based indoor localization. We first investigate the causes of RSS delay by analyzing Wi-Fi driver implementation and observing users' walking behaviors. The proposed system comprises four components: a delay detector, a speed calculator, a search window manager, and a window selector. The delay detector estimates the delay in Wi-Fi scanning at each access point. The speed calculator estimates a user's walking speed using an accelerometer. The search window manager quantifies the size of the reference fingerprints from the radio map based on the update delay and movement speed. The window selector revises the signals using the RSS compensation model. The experiment results in two buildings show that the proposed system greatly improves the accuracy of indoor localization.