Prioritized multimode precoding for joint minimization of source-channel video distortions

Real-time video streaming over wireless channels suffers from channel-induced distortions due to losses and errors as well as source-induced distortions due to lossy compression. This paper proposes a cross-layer design for quality-driven packet prioritization at the physical layer combined with source and link rate adaptation to jointly minimize the source and channel distortions. With a MIMO physical layer, packet prioritization is achieved by mapping video packets to ordered spatial streams to maximize the throughput weighted by the per-packet loss visibility. The MIMO mode controls the number of priority levels and is dynamically optimized to provide the best tradeoff between high source rate and low channel distortion. Jointly with mode adaptation, source rate adaptation is driven by the client by estimating the end-to-end throughput based on the transmission rate. Results show that the proposed design enables composite video quality gains due to prioritized delivery and rate adaptation.