Transmission strategies for wireless multi-user, multiple-input, multiple-output communication channels
Multiple-Input, Multiple-Output (MIMO) processing techniques for wireless communication are of interest for next-generation systems because of their potential to dramatically improve capacity in some propagation environments. When used in applications such as wireless LAN and cellular telephony, the MIMO processing methods must be adapted for the situation where a base station is communicating with many users simultaneously. This dissertation focuses on the downlink of such a channel, where the base station and all of the users have antenna arrays. If the transmitter has advance knowledge of the users' channel transfer functions, it can use that information to minimize the interuser interference due to the signals that are simultaneously transmitted to other users. If the transmitter assumes that all receivers treat the interference as noise, finding a solution that optimizes the use of resources is very difficult. This work proposes two classes of solutions to this problem. First, by forcing some or all of the interference to zero, it is possible to achieve a sub-optimal solution in closed-form. Second, a class of iterative solutions can be derived by extending optimal algorithms for multi-user downlink beamforming to accommodate receivers with multiple antennas. The closed-form solutions generally require less computation, but the iterative solutions offer improved performance are more robust to channel estimation errors, and thus may be more useful in practical applications. The performance of these algorithms were tested under realistic channel conditions by testing them on channels derived from both measurement data and a statistical model of an indoor propagation environment. These tests demonstrated both the ability of the channel to support multiple users, and the expected amount of channel estimation error due to movement of the users, with promising results. The success of any multi-user MIMO processing algorithm is ultimately dependent on the degree of correlation between the users' channels. If a base station is required to support a large number of users, one way to ensure minimal correlation between users' channels is to select groups of users whose channels are most compatible. The globally optimal solution to this problem is not possible without an exhaustive search, so a channel allocation algorithm is proposed that attempts to intelligently select groups of users at a more reasonable computational cost.