Abstract
Millimeter wave (mmWave) multiple-input and multiple-output (MIMO) system with large antenna arrays is a key research topic in the next generation communication systems. Since the high cost and power consumption of mmWave radio frequency chain, the hybrid analog/digital precoding is recently proposed as a proper precoding architecture for mmWave MIMO systems. In this thesis, we propose a modified low complexity hybrid precoding algorithm based on orthogonal matching and local search. Simulations show that the proposed algorithm can reduce the computation complexity by 40% with negligible BER performance loss compared with state-of-the-art algorithm. We also design and implement a hybrid precoding processor based on the proposed low complexity algorithm. The chip is implemented using TSMC 90nm CMOS process technology. The throughput of implementation processor can achieve 11.1M/10.4M/6.9M/4.3M channel-matrices per second in 1/2/3/4 data streams, respectively. Compared with state-of-the-art work, we increase the throughput by 65%/55%/40%/7.5% in 1/2/3/4 data streams with only 28% core area.