For guaranteed quality of service (QoS) and sufficient
bandwidth in a communication network which provides an
integrated multimedia service, it is important to obtain an
analytical and tractable model of the compressed MPEG data.
This paper presents a statistical approach to a group of picture
(GOP) MPEG frame size model to increase network traffic
performance in a communication network. We extract MPEG
frame data from commercial DVD movies and make probability
histograms to analyze the statistical characteristics of MPEG
frame data. Six candidates of probability distributions are
considered here and their parameters are obtained from the
empirical data using the maximum likelihood estimation (MLE).
This paper shows that the lognormal distribution is the best
fitting model of MPEG-2 total frame data.