A specular approach to MIMO frequency-selective channel tracking and prediction

We propose a specular approach to MIMO channel tracking that yields a parsimonious channel representation where the time and space components of the channel variations are separated by singular value decomposition (SVD). We show that this separation enables simple short-term channel prediction. This is especially useful for systems using channel state information at the transmitter, since the latency necessary to the feedback process can be mitigated. We validate our approach through the use of computer simulations.