On the traffic-distribution characteristics of parallel switching architectures

In this paper, we investigate the performance characteristics of parallel s witching architectures constructed by a stack of multistage switching netwo rks. We first find that the performances of the previously proposed paralle l switching architectures are much worse than the expected ones from analyt ic models which are based on the assumption that traffic is uniformly distr ibuted at each stage of a switching network. We show that this phenomenon i s closely related to a traffic-distribution capability of a parallel switch ing system and has a large influence on the performance. From these results , we then propose an architectural solution based on the Generalized Shuffl e Network (GSN) and analyze its performance by proposing a new iterative an alysis method. The proposed architecture uses self-routing and deflection r outing, and inherently has a traffic-distribution capability to improve swi tch performances such as cell loss and delay in a cost-effective manner. Fr om the comparison of simulation and analysis results, it is shown that the developed models are quite accurate in predicting the performance of a new parallel switching system.