Statistical division based modeling for multimedia network traffic
B. Li, Bart De Moor · 2003
For Markov modulated Poisson process (MMPP) models, a statistical division (SD) based modeling approach is proposed for multimedia network traffic. Due to its exploiting information from observed data, the SD can make significant reductions of problem-scales and then lead to a fast algorithm in comparison with the previously proposed non-informed division (NID) approach (reduced to 5.84% of the CPU time in the practical case given in this paper). Thus, modeling of MMPP can be largely speeded up to trace multimedia network traffic.