Experimental Evaluation of Enhanced Antenna Switching for CFO Mitigation in DoA Estimation

Abstract

Direction-of-arrival (DoA) estimation in the Internet of Things (IoT) and embedded devices is typically performed using single-RF chain systems and antenna switching using time-division multiplexing. The accuracy of the DoA estimation is reduced due to the additional phase shifts between the samples on the antenna elements caused by the carrier frequency offset (CFO). We propose to use optimized antenna switching patterns (ASPs) to mitigate the effect of CFO on accuracy in DoA estimation using a multiple signal classification (MUSIC) algorithm. We evaluated two switching methods referred to as EvenCFO-SP and Mirror-SP with simulations and confirmed the validity with measurements. Performance is analyzed and compared with a standard sequential sampling (SS) method. Uniform linear and circular array configurations are considered to evaluate the impact of the noise and the CFO value on estimation accuracy. The results show that the optimized ASPs outperform the SS method, with lower performance gain at small signal-to-noise ratios. The ASP sensitivity to the CFO value is studied, and the frequency bounds within which the optimized ASPs maintain high estimation accuracy are identified. A coarse CFO correction, effective for frequency estimation errors within the kHz range, extends the frequency bounds. Compared to a fine calibration, this reduces computational requirements, making it a viable option for embedded and IoT devices with limited hardware resources.