Video Distortion Estimation and Content-Aware QoS Strategies for Video Streaming over Wireless Networks

Abstract

This chapter describes several advanced techniques for estimating the video distortion deriving from multiple video packet losses. It provides different usage scenarios, where the Peak Signal to Noise Ratio (PSNR) video metric may be used for improving the end user quality. The key idea of the presented applications is to effectively use the distortion information associated to each video packet. This allows one to perform optimal decisions in the selection of the more suitable packets to transmit. During the encoding process, the encoder estimates first the loss impact (for instance the amount of error propagation) of each packet. Afterwards, it generates side information as a “hint” for making video content aware transmission decisions. In this way, it is possible to define new scheduling schemes that give more priority to the packets with higher loss impact, and to assign fewer resources to the packets with lower loss impact. To this end, the usage of hint tracks, introduced in the MPEG-4 systems part, provides a syntactic means for storing scheduling information about media packets that significantly simplifies the operations of a streaming server. Moreover, the prioritization scheme may be used to minimize the overall error propagation under the delay constraint imposed by the video presentation deadline. The chapter also reviews recent research advances in the field of QoS mechanisms that adopt video specific metrics to improve the end user perceived quality.