The directionality of millimeter-wave (mmWave) communications introduces a significant challenge in serving fast-rotating/moving terminals, e.g., mobile AR/VR, high-speed vehicles, trains, UAVs.This challenge is exacerbated in mmWave systems using analog beamforming, because of the inherent non-convexity in the analog beam tracking problem. In this paper, we obtain the Cramer-Rao lower bound (CRLB) of beam tracking and optimize the analog beamforming vectors to get the minimum CRLB. Then, we develop a low complexity analog beam tracking algorithm that simultaneously optimizes the analog beamforming vector and the estimate of beam direction. Finally, by establishing a new basic theory, we provide the theoretical convergence analysis of the proposed analog beam tracking algorithm, which proves that the minimum CRLB of the MSE is achievable with high probability. Our simulations show that this algorithm can achieve faster tracking speed, higher tracking accuracy and higher data rate than several state-of-the-art algorithms. The key analytical tools used in our algorithm design are stochastic approximation and recursive estimation with a control parameter.
Download