A study on scheduling algorithms for high-speed switching networks based on input-queuing
Weixin Mu · Caai Transactions on Intelligent Systems · 2008
Most high-speed switching networks adopt a switching fabric with fixed-length cells,and their performance depends heavily upon queuing strategies and the cell scheduling algorithm.Only when the input-queuing strategy is combined with a proper switching algorithm can throughput and time delay for the switching fabric be optimized.This paper mainly discusses virtual output queuing(VOQ)-based algorithms,among them the maximum number matching algorithm,the maximum weight matching algorithm,the stable combination matching algorithm,and the Hopfield neural network(HNN) scheduling algorithm.The mechanisms,performance and implementational complexity of these algorithms are compared and analyzed.The working modes of distributed and centralized scheduling algorithms are analyzed.Finally,from the findings in our research,it is concluded that the HNN algorithm can realize other algorithms by defining a priority function.