A generalized seasonal persistent model for video traffic
Document typeConference lecture
PublisherSpringer Berlin / Heidelberg
Rights accessRestricted access - confidentiality agreement
Video tra±c generated with modern codecs is known to possess a complex statistical structure that includes short memory, long memory, and seasonal components. We propose a k-factor Generalized Autoregressive Moving Average (k-GARMA) process as a versatile model for video tra±c. We provide a wavelet-based approximate Maximum Likelihood Estimator for such a process, together with some results obtained with H.263 and MPEG-2 video traces.