This well-written and very well-organized paper models, analyzes, and evaluates “the performance of a 2-class priority architecture for finite-buffered multistage interconnection networks (MINs).” The approach of the authors was to take into account the previous and current states of each switching element on the network. Thus, they had to process the previous and current states of all packets in the queues and also deal with both high- and low-priority queues where high-priority packets are processed before low-priority ones. They based their operation modeling “on a state diagram [that included] the possible MIN states, transitions, and conditions under which each transition occurs.” They provide a state transition diagram from which they derived equations that “represent the state transition probabilities as clock cycles advance,” and they use these equations to prove the correctness of their model. This analysis is performed for both high-priority and low-priority packet classes using such factors as packet delay and packet throughput.
The simulation performed at the packet level is extensive and “has several parameters such as the buffer length, the number of input and output ports, the number of stages, the offered load, and the ratio of high-priority packets.” The two state diagrams, the 18 simulation figures, the state notations, and the mathematical derivations are clearly explained and correctly referenced. As indicated by the authors, this paper may be useful to network designers to fine-tune parameters for their installations to achieve optimal performance.