Device-Free Simultaneous Wireless Localization and Activity Recognition With Wavelet Feature

Device-Free simultaneous wireless Localization and Activity Recognition (DFLAR) is a promising novel technique that empowers wireless networks with the ability to perceive the location and activity of a target within its deployment area while not equipping the target with a device. This technique turns traditional wireless networks into smart context-aware networks and will play an important role in many smart applications, e.g., smart city, smart space, and smart house. Essentially, DFLAR utilizes the shadowing effect incurred by the target on wireless links to realize localization and activity recognition. The feature utilized to characterize the shadowing effect is crucial for DFLAR. Traditional methods use time-domain features to characterize the shadowing effect. In this paper, we explore the method of realizing DFLAR with a wavelet feature. Compared with the time-domain feature, the wavelet feature could characterize link measurement in both the time and frequency domains, which could provide in-depth robust discriminative information and, therefore, improve the performance of the DFLAR system. Meanwhile, we also design a two-stage strategy to realize multitarget DFLAR with the feature map built by one target only, which reduces the training complexity remarkably. The experimental results in a clutter indoor scenario show that it could achieve location estimation and activity recognition accuracy of higher than 90%.