Software Defined Radio Demonstration of MIMO-OFDM Rate Adaptation

In this paper, we present a framework for rate and power allocation in MIMO-OFDM systems using the V-BLAST and SVD techniques. The performance of these techniques are compared with one another and with waterfilling using channel measurements on our MIMO-OFDM software defined radio testbed. Specifically, we show with measured data that sub-carrier rate adaptation allows for more efficient spectral use. Furthermore, we show that power allocation over streams provides significant gains as well. We analyze the impact of stale channel feedback information on this rate and power allocation framework and demonstrate how matrix channel prediction using a Kalman-filter based, vector autoregressive model can alleviate this problem.