Video traffic engineering for IPTV services

点击翻译按钮获取中文摘要

Video traffic is expected to occupy a significant portion of multimedia data in the future. This motivates the research in the delivery of broadcast TV over IP networks (IPTV). Video compression techniques are thus used to reduce the uncompressed raw video streams (~25 Mbps) to compressed streams (~4Mbps for MPEG-2 and less than 2Mbps for MPEG-4 codecs suite) to enable the multiplexing of several streams over current telecommunication infrastructure. Intuitively, this endeavour requires a strong synergy between the two research communities: the video compression and coding community and the networking community. This research aims to increase the dialogue between the two by building on the current literature and proposing (i) new video trace formats and (ii) new rate-control protocols for mass-delivery of video IP packets. The key challenge posed by the transmission of video traffic over IP networks is its high bit-rate variability and sensitivity to jitter. Unlike voice and data VBR models, characterizing the statistical behaviour of video traffic has been mostly based so far on empirical analysis and/or mathematical modeling that assume no packet losses in the transport network. The difficulty lies in the fact that video bit-rate variation depends on both (1) media content (e.g., amount of motion) and (2) compression technique used. Complications also stem from the lack of adequate mapping between transport network Quality of Service (QoS) parameters (packet-losses or bandwidth constraints) and decoded video quality as perceived by the end-user termed "Quality of Experience" (QoE).