Power Control and Channel Training for MIMO Channels: A DMT Perspective

The achievable diversity and multiplexing tradeoff (DMT) in MIMO fading channels with channel training is analyzed in this paper. We first consider a typical training scenario, where the transmitter transmits training symbols followed by data symbols, and the receiver performs channel estimation using the training symbols and then uses the imperfect channel estimates to decode the data symbols. From the DMT perspective, our results show that as long as the training power is equal to the data power, the obtained DMT result based on imperfect channel state information at receiver (CSIR) is the same as the original DMT result with perfect CSIR given in Zheng and Tse's seminal work . We extend the analysis to two-way training scenarios, with single and multiple training rounds. Our results show that two-way training together with power control can substantially improve the achievable diversity gain. Specifically, the achievable DMT with multiple training rounds can be described as a single straight line; while in the case of single training round, the achievable DMT is much lower and can be described as a single straight line or a collection of line segments, depending on the underlying training strategy and channel qualities.