Performance evaluation of algorithms for input-queued switch scheduling

The paper presents the most up-to-date randomized algorithm for evaluating the performance of the efficient randomized algorithms for input-queued switch scheduling. Six techniques of randomized algorithm as random-I,II,III,IV,Laura,Serena have been implemented and their performance have been evaluated. The first four techniques gives throughput up to 81%, but they experience long delays when packets are matched. So to overcome the above problem, the technique named ´Laura´ has been implemented and proven that the technique achieves a throughput up to 90%. Simulation shows that it provides delays close to that of MWM and out performs all other known low complexity scheduling algorithms. ´Laura´ needs an external source of randomness to obtain random matchings each time, which can cause some implementation difficulties. To avoid this problem, an enhanced version of Laura called ´Serena´ has been proposed which exploits the randomness present in the arrivals process to determine good random matchings and achieves a throughput of 100%. Fortuitously, this improves performance, because the arrivals are precisely what increase the weight of edges, using them leads to better schedule.