Performance of traffic management strategies for interconnected high-speed local and metropolitan area networks

Abstract

Interconnected high-speed local and metropolitan area networks are considered in which call-oriented resource allocation mechanisms are used for traffic management. Management domains are defined for the network, and call requests must be admitted by the appropriate domain manager. Assuming Poisson arrival rates and exponential service times, the state process can be modeled as a Markov process with the state defined as the number of active calls of each call type at time t. An equilibrium distribution can be found for this process, but the state space quickly becomes intractable with increases in link capacities and in the number of subnetworks. To reduce the state space, the network is partitioned into subnetwork groups, assuming independence. The probability of blocking can be calculated iteratively and leads to an approximate equilibrium distribution of the system. From these values the desired performance measures are computed: network utilization, delay, throughput, and packet loss probabilities. An example is presented which illustrates the effects of various surveillance strategies and decision policies on network performance. Implementing the iterative procedure using parallel processing is considered. >