Modeling of correlated arrival processes in the Internet

The correlations in the arrival processes in the Internet are very significant since voice, video and data are transported simultaneously over high speed links. Detailed knowledge of the behavior of the arrival processes is of paramount importance in traffic engineering and network optimization. In this paper, we propose to use Markov-modulated arrival processes to approximate the correlated arrival processes in the Internet. Two approximate procedures are developed to evaluate the parameters of the underlying Markov chain and the conditional density functions given a few moments and the lag-1 autocorrelation of the arrival process. The resulting queueing system can be analyzed using the existing techniques for the G/G/1 queue with Markov-modulated arrival processes. We also show how our model can be used in connection admission control