Walking direction detection using received signal strengths in correlated RF links

Walking direction detection is valuable in determining the number of subjects and pedestrian tracking by using radio frequency (RF) tomographic networks. In this paper, we use two correlated links based on the most existing network to sense the walking movement and extract the short-time variances of received signal strength (RSS) on both links as the motion feature. Several groups of correlated links deployed in different heights and with different geometric structures are experimentally compared for walking direction acquisition. Two detection methods involving the cross-correlation based time delay estimation and dynamic time warping (DTW) distance based matching are built for identifying the direction of walking motion. The experimental results in indoor line-of-sight (LOS) and non-line-of-sight (NLOS) scenarios confirm the effectiveness of the proposed sensing method.