Optimum training for CSI acquisition in cognitive radio channels

One of the main issues in the Overlay Cognitive Radio Paradigm (a framework where the secondary transmitter partially cooperates with the primary user) is the acquisition of channel state information at the secondary transmitter. In [1] a simple interaction framework that could allow the estimation of the channel values at the secondary transmitter was presented, although its performance was not characterized. In this paper, we extend the framework in [1] to the MIMO and SISO Time-Varying channels, present closed form expressions for the mean square error of these channel estimates, and derive optimum training sequences to minimize the estimation variance.