{
"title": "Burstiness Reduction of a Doubly Stochastic AR-Modeled Uniform Activity VBR Video",
"authors": "J. P. Dubois",
"country": null,
"institution": null,
"volume": "10",
"journal": "International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering",
"pagesStart": 1478,
"pagesEnd": 1483,
"ISSN": "1307-6892",
"URL": "http:\/\/waset.org\/publications\/8377",
"abstract": "Stochastic modeling of network traffic is an area of\r\nsignificant research activity for current and future broadband\r\ncommunication networks. Multimedia traffic is statistically\r\ncharacterized by a bursty variable bit rate (VBR) profile. In this\r\npaper, we develop an improved model for uniform activity level\r\nvideo sources in ATM using a doubly stochastic autoregressive\r\nmodel driven by an underlying spatial point process. We then\r\nexamine a number of burstiness metrics such as the peak-to-average\r\nratio (PAR), the temporal autocovariance function (ACF) and the\r\ntraffic measurements histogram. We found that the former measure is\r\nmost suitable for capturing the burstiness of single scene video\r\ntraffic. In the last phase of this work, we analyse statistical\r\nmultiplexing of several constant scene video sources. This proved,\r\nexpectedly, to be advantageous with respect to reducing the\r\nburstiness of the traffic, as long as the sources are statistically\r\nindependent. We observed that the burstiness was rapidly\r\ndiminishing, with the largest gain occuring when only around 5\r\nsources are multiplexed. The novel model used in this paper for\r\ncharacterizing uniform activity video was thus found to be an\r\naccurate model.",
"references": null,
"publisher": "World Academy of Science, Engineering and Technology",
"index": "International Science Index 10, 2007"
}