Analysis of an Isolated Stream with Correlation at a Finite Capacity Queue with Multiple Input Streams

This paper analyzes a discrete-time, finite capacity, deterministic service time queueing system (MMBP + Ml\!fBP(XljD(N)jljK) with superposing non-renewal arrival processes. This queueing system models a switch in high speed networks, where multiple applications which generate correlated and bursty traffic are supported simultaneously using fast packet switching. A designated arrival stream (VC-stream) which is rnodeleled as a Markov Modulated Bernoulli Process (MMBP) is considered for analyzing the effect of the network traffic characteristics on its performance measures. The aggregate of the remaining arrival streams (EXT-stream) is modeled as an MMBP with batch arrivals. Using exact analysis we derive the queue length distribution at the VC-stream arrival instants, and also obtain the interdeparture time distribution of the VC-stream.