Video coding mechanism to predict video traffic in ATM networks

Asynchronous transfer mode (ATM) was proposed as a transport technique for B-ISDN. The main goal of this proposal was to provide flexibility to B-ISDN supporting a high variety of traffic. To avoid congestion in the ATM network users´ traffic has to be predicted. Video is a special variable bit rate (VBR) traffic which is not easily predicted. To meet the requirements of video traffic we suggest its virtual separation from the other traffic types and it will be treated as a video virtual network (VVN) in ATM. Our goal was to find a way to predict VBR-video traffic to achieve the best VVN support. We propose a video shaped buffer (VSB) coding mechanism to be implemented in the video coders maintaining an acceptable delay and a better image quality. With the traffic parameters provided by the VSB mechanism we present the video Markov prediction model (VMPM). With this prediction, and the VVN approach, one can allocate and control network resources for video traffic according to its own characteristics. Besides, network operation, maintenance and control are improved