MMWiLoc: A Multi-Sensor Dataset and Robust Device-Free Localization Method Using Commercial Off-The-Shelf Millimeter Wave Wi-Fi Devices

Device-free Wi-Fi sensing has numerous benefits in practical settings, as it eliminates the requirement for dedicated sensing devices and can be accomplished using current low-cost Wi-Fi devices. With the development of Wi-Fi standards, millimeter wave Wi-Fi devices with 60GHz operating frequency and up to 4GHz bandwidth have become commercially available. Although millimeter wave Wi-Fi presents great promise for Device-Free Wi-Fi sensing with increased bandwidth and beam-forming ability, there still lacks a method for localization using millimeter wave Wi-Fi. Here, we present two major contributions: First, we provide a comprehensive multi-sensor dataset that synchronously captures human movement data from millimeter wave Wi-Fi, 2.4GHz Wi-Fi, and millimeter wave radar sensors. This dataset enables direct performance comparisons across different sensing modalities and facilitates reproducible researches in indoor localization. Second, we introduce MMWiLoc, a novel localization method that achieves centimeter-level precision with low computational cost. MMWiLoc incorporates two components: beam pattern calibration using Expectation Maximization and target localization through Multi-Scale Compression Sensing. The system processes beam Signal-to-Noise Ratio (beamSNR) information from the beam-forming process to determine target Angle of Arrival (AoA), which is then fused across devices for localization. Our extensive evaluation demonstrates that MMWiLoc achieves centimeter-level precision, outperforming 2.4GHz Wi-Fi systems while maintaining competitive performance with high-precision radar systems. The dataset and examples processing code will be released after this paper is accepted at https://github.com/wowoyoho/MMWiLoc.