针对V2V毫米波通信中传统波束搜索方法开销较大、实时性差的问题,结合视觉感知技术,提出采用一种基于视觉感知的V2V毫米波波束对准方法,以降低波束对准的时延和开销.该方法首先利用车载单目相机采集的实时场景图像,通过FCOS3D目标检测算法,估计场景中车辆目标的三维位置;而后,对车辆目标位置进行坐标变换获得空间角度和距本车体的距离,并由此推导出模拟波束成形矢量,实现波束对准.仿真结果表明,该方法能够通过高精度的目标位置估计,在较低硬件开销的情况下,完成相对精确的波束对准.
Aiming at the problems of high overhead and poor real-time performance of the traditional beam search method in V2V millimeter-wave communication,a V2V millimeter-wave beam alignment method based on the visual perception is proposed to reduce the time delay and overhead of beam alignment in combination with visual perception technology.Specifically,this method first leverages the real-time scene image collected by the vehicle-mounted monocular camera to estimate the 3D position of the vehicle target through the FCOS3D target detection algorithm.Then the coordinate transformation of the vehicle target position is performed to obtain the spatial angle and the distance from the vehicle,and the analog beamforming vectors are derived to realize the beam alignment.Simulation results show that the proposed method can accomplish relatively accurate beam alignment with high accuracy target position estimation and low hardware overhead.