Design Of Scheduling Scheme In Hybrid Beamforming System Based On Large Scale Antenna Array

The fifth generation (5G) mobile communication system demands higher performance, which has aroused the industry's interest in millimeter-wave communication and large-scale MIMO. Hybrid beamforming technology could take advantages of both technologies and thus became a research hotspot. In hybrid beamforming technology, user scheduling is a key problem and a practical scheme is needed.In this paper, the modeling of 60GHz millimeter-wave channel is discussed first, and the influence of the fixed subarray formation on the correlation of channel is studied. For hybrid beamforming system with fixed subarray formation, this paper found that user scheduling is a combinatorial problem. The exhaustive method is needed to find the optimal solution, and the scheduling scheme based on exhaustive method is given. Aiming at the shortcomings of the exponential complexity of the exhaustive scheme, this paper proposes a greedy scheduling scheme using channel correlation based on fixed subarray formation.The results of computer simulation show that the greedy scheduling scheme can complete the scheduling task in linear complexity, but the performance is lost.At high signal-to-noise ratio, the system spectral efficiency can reach 60% of the optimal solution.Based on the greedy algorithm, this paper further studies the performance bottleneck of the scheduling scheme. Through theoretical analysis, I have found more essential factors of the performance in system, namely the average singular value and condition number of the equivalent channel. A better scheduling criterion is proposed. According to this criterion, this paper introduces the clustering algorithm into the greedy scheduling scheme, and an enhanced scheduling scheme based on clustering analysis is proposed. Computer simulation results show that the scheduling scheme based on clustering analysis is better than the greedy scheduling scheme when the complexities are both linear. At high signal-to-no-ise ratio, the system spectral efficiency of clustering analysis based scheduling scheme can reach more than 70% of the exhaustive method based scheme, which is nearly 10% higher than the greedy scheduling scheme. Therefore, this proposed scheduling scheme based on clustering analysis is practical.