Over the last ten years, multiple-in-multiple-out (MIMO) systems have gained significant attention due to their promising improvement in system performance and capacity. A lot of work has been done on the transmit preprocessing, and several precoding schemes such as zero-forcing (ZF), minimum mean square error (MMSE) and Tomlison-Harashima precoding (THP) have been proposed for multi-user MIMO downlink. In this paper, we propose a new modulation-dependent precoding scheme. Compared to the conventional precoding scheme which is independent of modulation signal, our new precoding scheme optimally selects modulation dimension to optimize system performance and maximize capacity. It is found that the new scheme can provide much better performance and much larger capacity than the conventional precoding schemes for multi-user MIMO downlink. Because our scheme scarifies modulation dimension and equivalently increases spatial dimension, it can significantly reduce interference and obtain overall performance gain. We also consider its application to distributed MIMO system with multiple transmitters and multiple receivers, and obtain significant performance improvement over the recently proposed interference alignment approach.

Bin Li received Ph.D. degree in 1993. He was a visiting professor with the School of Engineering Science, Simon Fraser University, Canada. From 1997 to 2001, he was a member of technical staff in Advanced Technology Dept. of Wireless Division, Nortel Networks, Ottawa, Canada. From 2001 to 2005, he was a senior staff engineer in Interdigital, NY, USA. Since November 2005, he has been a senior expert in Wireless Research Dept., Huawei Technologies, Shenzhen, China. He published more than 50 technical papers in communications area. He holds more than 30 US patents, two of which are essential patents on 3G system. His research interests are various techniques in wireless communications including modulation/detection, channel coding, turbo codes and MIMO techniques.