Camera-Sensing-Assisted Fast mmWave Beam Tracking for Connected Automated Vehicles

Abstract

The millimeter-wave (mmWave) technology can support the wide-band communication requirements for the environment raw perception information sharing to guarantee the safety of connected automated vehicles (CAVs). But, how to achieve the fast and robust mmWave beam tracking is a challenging problem in a high mobility CAVs scenario. Therefore, this article proposes a camera-sensing-assisted joint offline and online beam tracking (CS-JBT) algorithm in the mmWave frequency band. First, the mobility model is designed to predict the next moment of CAV and reduce the beam searching space overhead. In the offline learning phase of the proposed CS-JBT algorithm, the optimal beam pair under the current mobility state is obtained to guarantee the timeliness of the beam tracking process. Furthermore, to solve the mobility state deviation problem caused by the camera sensing error, the online learning phase is designed to achieve the optimal beam tracking performance efficiently for the practical scenario. Simulation and hardware testbed results verify that the proposed CS-JBT algorithm can minimize the beam tracking latency from 500 to 20 ms in contrast to the existing VBC-PF algorithm, while achieving a stable throughput over 2.5 Gb/s in the 28-GHz mmWave frequency band.