Enhanced MIMO wireless video communication using multiple-description coding

Significant enhancements in spectral efficiency and error-resilience must be obtained from emerging wireless video systems if high-quality multimedia transmission is to become cost-effective. MIMO (multiple-input-multiple-output) technology promises to deliver the sustainable high transmission quality required for such applications. MIMO is a generic technology which, depending on the actual architecture deployed and the channel conditions encountered, produces variable error conditions. In particular, the way these errors interact with the transmitted data will depend on both the data source and the coding mechanism employed. In this context, this paper focuses on the interactions when MIMO systems are used to transmit compressed video.In terms of multimedia transmission, spatial multiplexing (SM) has often been proposed as the most suitable MIMO technique. Most SM-based video transport schemes focus on the advantages of multiplexing gain, which are achieved at the expense of relatively high SNR values. This paper proposes the use of multiple-description video coding (MDC) as a means of emulating the spatial diversity lacking in SM systems, to provide a new class of wireless video transmission algorithm. Singular value decomposition (SVD) is employed to create orthogonal sub-channels which provide an efficient means of mapping video content to the wireless channels.Extensive simulations demonstrate the effectiveness of MDC as a video decomposition which couples well with the underpinning SVD architecture. In addition, resource allocation in MIMO-SVD systems is examined in the context of transmitted video quality, and further enhancements have been achieved for low SNR values. Results indicate improvements in average PSNR of the decoded test sequences of around 5–7 dB, compared to standard, single-description video transmission. This is further enhanced by an additional 2–3 dB in the case of channels with low SNR values through the use of unequal power allocation. This approach is particularly well suited to MDC. The results presented are supported by significant enhancements in subjective quality and form the basis of a strong case in favour of MDC-based MIMO systems.