WiFi 5GHz CSI-Based Single-AP Localization With Centimeter-Level Median Error

Abstract

Localization schemes based on one-way Channel Frequency Response (CFR) are limited in terms of the available coherent bandwidth due to random timing and phase offsets. Two-way CFR-based localization schemes utilize higher coherent bandwidth but suffer from Angle-of-Arrival (AoA) estimation ambiguity. In this paper, we investigate efficient localization schemes for ubiquitous WiFi networks in the 5GHz unlicensed frequency band by utilizing both the one-way and two-way WiFi Channel State Information (CSI) to overcome the AoA estimation ambiguity arising from the two-way-channel-based localization algorithm. We use the two-way CFR to increase the total available coherent bandwidth through stitching and the one-way CFR to resolve the AoA estimation ambiguity. We also propose complexity reduction methods for our localization scheme. Our proposed reduced-complexity localization scheme achieves 58cm median localization accuracy for our collected WiFi data and simulation environment in multipath scenarios with a 98% reduction in computational complexity compared to the state-of-the-art.

Keywords

• Localization
• angle estimation
• delay estimation
• WiFi 5GHz
• super-resolution
• centimeter-level