WiFi-Based Channel Impulse Response Estimation and Localization via\n Multi-Band Splicing

Abstract

Using commodity WiFi data for applications such as indoor localization,\nobject identification and tracking and channel sounding has recently gained\nconsiderable attention. We study the problem of channel impulse response (CIR)\nestimation from commodity WiFi channel state information (CSI). The accuracy of\na CIR estimation method in this setup is limited by both the available channel\nbandwidth as well as various CSI distortions induced by the underlying\nhardware. We propose a multi-band splicing method that increases channel\nbandwidth by combining CSI data across multiple frequency bands. In order to\ncompensate for the CSI distortions, we develop a per-band processing algorithm\nthat is able to estimate the distortion parameters and remove them to yield the\n"clean" CSI. This algorithm incorporates the atomic norm denoising sparse\nrecovery method to exploit channel sparsity. Splicing clean CSI over M\nfrequency bands, we use orthogonal matching pursuit (OMP) as an estimation\nmethod to recover the sparse CIR with high (M-fold) resolution. Unlike previous\nworks in the literature, our method does not appeal to any limiting assumption\non the CIR (other than the widely accepted sparsity assumption) or any ad hoc\nprocessing for distortion removal. We show, empirically, that the proposed\nmethod outperforms the state of the art in terms of localization accuracy.\n