Performance analysis of DQDB metropolitan subnetwork

An extensive performance analysis of the DQDB (distributed queue dual bus) metropolitan area network is presented. The performance is presented in terms of transfer delay, bandwidth allocation, per node and per network total delay, transient time, network size, and maximum local queue length. The effects of varying traffic intensity on the performance are examined. Simulation models for all MAC protocols were developed to investigate the performance of the network. The results demonstrate that the original protocol exhibits the worst performance in all aspects. The bandwidth balancing mechanism protocol proved to be better in bandwidth allocation but at the cost of increasing the average delays. Even though the amount of sacrificed bandwidth is small, it still results in less than satisfactory performance. Only the preservation procedure protocol proved to be superior in its ability to allocate bandwidth fairly among the nodes without incurring increased delays, which was evidenced in the bandwidth balancing mechanism protocol. >