<XML><RECORDS><RECORD><REFERENCE_TYPE>10</REFERENCE_TYPE><REFNUM>8089</REFNUM><AUTHORS><AUTHOR>Alzeidi,N.</AUTHOR><AUTHOR>Khonsari,A.</AUTHOR><AUTHOR>Ould-Khaoua,M.</AUTHOR><AUTHOR>Mackenzie,L.M.</AUTHOR></AUTHORS><YEAR>2005</YEAR><TITLE>A New Queueing Model for the Analysis of Virtual Channels Occupancy in Wormhole-Switched Networks</TITLE><PLACE_PUBLISHED>DCS Technical Report Series</PLACE_PUBLISHED><PUBLISHER>Dept of Computing Science, University of Glasgow</PUBLISHER><ISBN>TR-2005-206</ISBN><LABEL>Alzeidi:2005:8089</LABEL><KEYWORDS><KEYWORD>Analytical modelling</KEYWORD></KEYWORDS<ABSTRACT>Arranging physical channels into several virtual channels have been introduced to design deadlockfree routing algorithms and also to overcome the chain blocking problem encountered in wormholeswitched networks. It has been shown in many studies that adding virtual channels significantly improves the performance of the network. Almost all previous analytical studies relayed on a method proposed by Dally to capture the effect of arranging the physical channel into several virtual channels. Dally's method is based on a Markov process and losses its accuracy as the traffic increases. In this study, we propose an alternative approach to deal with virtual channels in analytical performance modelling. Our new method is based on an M/G/1 queuing system. Beside the accuracy that it achieves under low, moderate as well as high traffic, one of the a main advantages of our proposed method is its ability to be customized and adapted to different traffic and network conditions. This is could be achieved by simply changing the service time distribution and inverting a z-transform. Examples of closed-form solutions are outlined to solve the model and also several approximations and algorithmic approaches are devised when close-form solution is not feasible or when the service time distribution is not available. The new method is validated via extensive simulation experiments and compared to Dally's method.</ABSTRACT></RECORD></RECORDS></XML>