Calculation-based Beam Training Scheme for Massive MIMO Systems

In future communication systems, the exploration of high-frequency communication, such as mmWave and THz communication, has garnered significant attention due to its potential to enhance channel capacity. However, the utilization of high-frequency signals brings about the challenge of severe path loss. Fortunately, the introduction of massive MIMO technology offers a promising solution by enabling the generation of directional beams with a high array gain, effectively compensating for the substantial path loss. For accurate beamforming in massive MIMO systems, the acquisition of reliable Channel State Information (CSI) plays a crucial role. Beam training, a conventional method of obtaining CSI, involves an exhaustive search of the entire spatial domain using predesigned codewords from a codebook. The codeword with the highest received power is subsequently chosen as the direction for the intended users. However, this conventional beam training approach is limited by the angular resolution of the available codewords, which restricts the precision of beam alignment. Furthermore, as the number of antennas increases, the beam training overhead escalates proportionally. To address these limitations, this paper proposes a novel beam training scheme that achieves efficient beam training in only four time slots, regardless of the angular resolution of the codewords. The proposed scheme offers improved accuracy, circumventing the constraints imposed by the codeword resolution. Simulation results illustrate that the proposed method attains near-optimal beam training performance while requiring significantly reduced beam training overhead. By introducing this innovative beam training scheme, we aim to contribute to the advancement of massive MIMO systems, ensuring enhanced channel capacity and more precise beamforming capabilities, even in scenarios with high-frequency communications and severe path loss.