Performance analysis of Shufflenet with deflection routing

Multi-hop communication networks are drawing increasing interest. Shufflenet is a multi-hop communication network which achieves high performance while overcoming the current shortcomings in device technology. The authors present a new and comprehensive analysis for Shufflenet. By making use of the connectivity properties of any (p, k) Shufflenet, they characterize the probabilistic behavior of a typical packet in the network in terms of a simple discrete-time Markov chain with only 2k states, one of which is an absorbing state. They then derive analytic expressions for several important network performance parameters such as hops distribution, average delay for a packet to travel to its destination and the probability of don't care in each hop that the packet takes. The expressions derived can be applied to any (p, k) Shufflenet. The calculation of throughput of the Shufflenet can then be obtained by finding the relationship between network packet generation probability.<
>