A markov - based algorithm for the generation of MPEG sequences matching intra - and inter - GoP correlation

The development of sophisticated multimedia applications requires accurate performance evaluation tools to efficiently design and manage future high bandwidth networks. These tools have to be accurate, in order to closely match the statistical characteristics of network traffic, easy to implement, and efficient. VBR compressed video traffic is expected to represent most of the multimedia traffic loading networks in the immediate future, and the MPEG coding scheme will be the most widely used standard for many types of applications. Although many works in literature present models of video traffic, little effort has been devoted up to now to capturing one of the most important characteristics of MPEG video traffic: intra-GoP correlation. In this paper, after demonstrating that this correlation strongly affects the performance of a network loaded by MPEG video traffic, an accurate, fast and easy generation algorithm is proposed. The algorithm allows a sequence to be generated, imposing the Short Range Dependence (SRD), the Long Range Dependence (LRD) behavior and the probability density function of any given MPEG video sequence. Correspondence address: Prof. Sergio Palazzo Istituto di Informatica e Telecomunicazioni, Università di Catania V.le A. Doria, 6 95125 Catania (Italy) phone: +39 095 7382370 fax: +39 095 338280 e-mail: [email protected]