Resumen
Simply speaking, automatic driving requires the calculation of a large amount of traffic data and, finally, the obtainment of the optimal driving route and speed. However, the key technical difficulty is the obtainment of data; thus, radar has become an indispensable hardware for automatic driving. Compared to the optical and infrared radar, millimeter-wave radar is not affected by the shape and color of the target, and it is not affected by the atmospheric turbulence, compared to ultrasonic, and so it has a stable detection performance and good environmental adaptability. It is little affected by changes in the weather, and the external environment, rain, snow, dust, and sunshine have no interference in it. The Doppler frequency shift is large, and the accuracy of the relative velocity measurement is improved. However, one challenge for vehicles in fast environments is millimeter-wave-based communication. Because of the short wavelength of the millimeter wave and the high path and penetration losses, the beamforming technology of a large-scale antenna array plays a key role in the construction and maintenance of millimeter-wave communication links. Millimeter waves have wide channel bandwidths, unique channel characteristics, and hardware limitations, and so there are many challenges in the direct use of beamforming technology in millimeter-wave communication. Traditional beam training cannot meet the requirements of low overhead and low delay. This paper, in order to obtain beam information, introduces the context-awareness module to the deep-learning net, which is derived from past observation data. This paper establishes a model that contains the receiver and the surrounding vehicles to perceive the environment. Then, a long short-term memory (LSTM) neural network is used to foresee the acquired power, which is quantized by several beam powers. According to the conclusion, the prediction accuracy is greatly increased, and the model could yield throughput with almost zero overhead and little performance loss.