Abstract
Millimeter wave (mmWave) multiple-input and multiple-output (MIMO) system provides high throughput for next generation system. To avoid severe signal attenuation and path loss caused by short wavelength, mmWave MIMO system utilizes large antenna arrays in transmitter and receiver to reduce fading. However, the high cost of radio frequency (RF) chain and data converter limits the number of antennas. Thus, hybrid analog/digital precoding was proposed to reduce the required number of RF chains and data converters under the condition of large antenna arrays. Hybrid precoding MIMO system divides precoding process into analog domain and digital domain in order to reduce the sampled data from digital precoding process. In this thesis, we propose a scalable low complexity hybrid precoding algorithm based on parallel data-stream hardware architecture which enables independently computing for each data stream. Moreover, the proposed scalable precoding algorithm is implemented using TSMC 40 nm CMOS process technology. The proposed precoding processor supports the transmissions of 1 to 4 data streams for 16x16 mmWave MIMO systems. The operating frequency of this chip is 120 MHz and power consumption is 139 mW. The normalized throughput of this chip achieves 6.67M/6.67M/6.67M/6.67M channel-matrices per second in 1/2/3/4 data streams.