Delay jitter correlation analysis for traffic transmission on high speed networks

Studies the time-dynamic behavior of delay jitter as captured by the autocorrelation function; the second-order statistics provide information relating to consecutive cell loss in real-time services. The authors derive an expression for delay jitter correlation of a stationary traffic stream in an MMPP/M/1/K system with FIFO service discipline; use of a type of deterministic rate server yields the same correlation function but with reduced variance. The jitter second-order statistics are shown to depend strongly on those of the queue, and the queue second-order statistics depend strongly on those of the input traffic. The authors show that the shape of the jitter correlation is determined by the weighted difference in queue correlations, and they derive a method for obtaining the weights for an MMPP traffic stream. In the numerical study emphasis is placed on a periodic traffic stream (such as CBR video) multiplexed on a major communication node; the authors investigate the impact of input traffic correlations and queueing system parameters on the queue length and delay jitter second-order statistics. They also briefly consider the delay jitter correlation from a binary Markov source (voice model)