Cross-layer scheduling with content and packet priorities for optimal video streaming over 1xEV-DO

Abstract

Maximization of received video quality and application-level service fairness are the two principal objectives of multi-user wireless video streaming. The user and packet scheduling mechanisms employed are the determining factors on the communication system performance and must utilize multiple layers of the OSI protocol stack. The semantic and decodability (concealment related) importance and hence priorities of video packets can be considered at the application layer. In this paper, the use of video content and packet priorities for multi-objective optimized (MOO) scheduling in 1xEV-DO system is introduced. Rate adaptive AVC/H.264 encoding is used for content adaptation and a user with the least buffer fullness, best channel throughput and highest video packet importance is targeted for scheduling. Hence, losses are forced to occur at packets with low semantic/decodability importance. Results show that the proposed framework achieves 1-to-2 dB's better PSNR in high importance temporal regions compared to the state-of-the-art CBR encoding case.