Future broad-band communications networks are expected to be dominated by video and image traffic. Variable bit-rate (VBR) video compression is generally preferred to constant bit-rate (CBR) compression because constant image quality can be provided. In contrast, CBR transport is preferred to VBR transport from the networking standpoint because of its simplicity. This paper studies the important issue of adapting VBR compressed video for transport over a CBR channel. We focus on temporal traffic smoothing using an elastic buffer. The target image quality and the output rate of the video encoder is controlled by feedback based on the buffer-occupancy level. Previous adaptation schemes are not readily analyzable. An analyzable control-theoretic adaptation framework is proposed. It allows systematic and quantitative investigation of issues such as stability, robustness against scene changes, robustness against image-quality oscillations due to coding-mode switching, and tradeoffs between image-quality and buffer-occupancy (delay) fluctuations. Perhaps more importantly, the framework opens up many new possibilities for further research.

A new transform, called the rounding transform (RT), is introduced which employs a pair of rounding operations namely, rounding up and rounding down operations. The RT is of interest in pyramid data structured coding for progressive and lossless image transmission, since the total number of data to be transmitted is the same as that of the original without overhead bits and since it has the advantages of both implementation with various elementary block sizes and filters. Additionally, the previous methods, reduced difference pyramid (RDP) and hierarchy embedded differential image (HEDI) are proved to be only simple cases of the rounding transform. Computer simulations prove that several kinds of rounding transform can be found, which perform better than the previous methods.

Future broadband integrated services networks based on the asynchronous transfer mode (ATM) technology are expected to carry information from a large variety of different services and applications. This paper investigates video aggregation, a concept that integrates compression and statistical multiplexing of video information for transport over a communication network. We focus on the transmission of a group of video sessions as a bundle, the practical examples of which include entertainment-video broadcast and video-on-demand (VoD). In this situation, the advantage of constant bit-rate (CBR) transport (which facilitates simple network management and operation) and the advantage of variable bit-rate (VBR) video compression (which yields smoother image quality) can be achieved simultaneously. We show that it is better to integrate compression and statistical multiplexing before the bundle of video traffic enters the network than performing them as independent processes. We present experimental results which indicate the advantages of video aggregation in terms of superior image quality and efficient bandwidth usage.