Application-aware MIMO Video Rate Adaptation

High data rates and multiple channel profiles of a MIMO system are naturally well-suited to carry video content. However, a video communication scheme to exploit these desirable properties of MIMO systems is widely unexplored. Even the most sophisticated MIMO rate adaptation methods rely on channel BER which shows a monotonic decrease as the channel quality improves. However, video quality, for which PSNR is a better measure than BER, does not show a proportional increase with improved channel quality because of error concealment. In this paper, we present a novel application-aware rate adaptation method which can detect variations in a MIMO channel at a video receiver, quantify the impact of these variations on the received video quality, and adaptively select a transmission profile (consisting of modulation, coding and MIMO mode) to provide unprecedented improvements in video quality. The proposed application-aware MIMO Video Rate Adaptation (MVRA) method relies on a comprehensive model of source- and channel-induced distortions in video quality. Using this model, a MIMO receiver can select an appropriate transmission profile on a per-GOP basis. Trace-driven evaluations over an 802.11n channel show that the proposed MVRA method's PSNR is very close to the PSNR of an optimal RA scheme. Comparison with state-of-the-art MIMO RA methods shows that the proposed MVRA method provides consistently better PSNR, with an average improvement of 19.29% (5.5486 dBs) over the best existing MIMO RA scheme.