Compressed sensing (CS)-based beam alignment is a promising solution for rapid link configuration in millimeter wave (mmWave) systems that use large arrays. Translating CS to practical mmWave radios, however, can be challenging due to carrier frequency offset (CFO). Standard sparse recovery techniques that use random sampling strategies to acquire channel measurements can fail even if there is a slight mismatch in carrier frequencies. In this paper, we show that restricting the randomness in compressive sampling to local sets can achieve robustness to structured errors due to CFO. The proposed approach requires fewer channel measurements than comparable algorithms and has the same complexity as standard CS.
Paper
Full text
Localized Random Sampling for Robust Compressive Beam Alignment
Semantic Scholar · Engineering · 2019
Abstract
Compressed sensing (CS)-based beam alignment is a promising solution for rapid link configuration in millimeter wave (mmWave) systems that use large arrays. Translating CS to practical mmWave radios, however, can be challenging due to carrier frequency offset (CFO). Standard sparse recovery techniques that use random sampling strategies to acquire channel measurements can fail even if there is a slight mismatch in carrier frequencies. In this paper, we show that restricting the randomness in compressive sampling to local sets can achieve robustness to structured errors due to CFO. The proposed approach requires fewer channel measurements than comparable algorithms and has the same complexity as standard CS.