Asynchronous Video Coding for Wireless Transport

Wireless access to continuous-medis services such as video, voice, and audio is becoming increasingly prevalent. Interactive video services such as video conferencing and multimedia editing is one such service, but existing compression standards (designed for wired, circuit-switclrcd services) are unsatisfactory for wireless packet-video services. We propose a novel strategy for video transport using a layered source coder in conjunction with a variableQOS, multiple-substream abstraction for the transport. This abstraction addresses specifically the need to obtain simultaneously high spectral efficiency, good subjective quality, and low perceptual delay on awireless channel. It also addresses the heterogenous transport resulting from the concatenation of a wireless access link with a broadband backbone network. We use asynchronous video (ASV) reconstruction, running counter to current techniqucs, which use strictly synchronous (frame-by-frame ) video processing. By doing so we hope to achieve a perceptual delay that is much lower than the worst-cese tronsport delay. (By "perceptual delay" , we refer to the effective end-to-end latency observed by the user,for example es represented by the audio delay required to maintain lip synchronization.) By identifying packets to the transport with relaxed reliability andlor delay requirements (through the subsream identifier), the transport (particularly wireless) can achieve high traffic capacity. Reasonable and promising simulation results are achieved, although much work remains on achieving significant video compression in this environmcnt.