Real Time Prediction of the Mobile MIMO channel
Contact: A/Prof Paul Teal
The development of multiple input, multiple output (MIMO, or multi-antenna) systems promises to revolutionise radio communications, allowing far greater bandwidths than was considered possible before the pioneering work of Foschini and Gans, and Telatar: the channel capacity scales linearly with the minimum of the number of antennas at the two ends of a link. There are several different approaches to MIMO systems - the two main contenders being space time coding, and the use of singular value decomposition (SVD). The latter approach is closer to the means used for deriving the theoretical channel capacity, and promises greater performance, but suffers from the requirement of channel knowledge at the transmit end of a link. This is very difficult to achieve in a mobile environment. The mobile wireless channel is constantly changing, so any channel knowledge is out of date before it can be used. A potential method of overcoming this problem is to predict a short time into the future what the channel state till be.
Quite a lot of work has already been done on the prediction problem. However, the work to date has been based on single antennas. Most of the promising results so far have been based on simulation rather than channel measurement. Although more information is required for a multi-antenna system, there is the potential for a greater ``awareness'' by the system of the scattering environment in which it is operating. Also, existing approaches have not used Bayesian estimation techniques. This project is extending channel prediction using Bayesian array signal processing techniques and establishing theoretical performance bounds.