Quality of Sendee (QoS) has become an important topic in modern telecommunication network in order to guarantee multimedia traffic. In IP networks, DiffServ seems to be the best approach to satisfy QoS constraints, due to its end-to-end philosophy. Actual trend is to consider satellite on-board switching capabilities for managing multibeam input and output. In this paper, for reducing computational complexity, a Cellular Neural Network (CNN) has been proposed for the on-board switching problem; several traffic classes have been considered and switching algorithm has been implemented within a CNN taking into account their priority, queue length and time spent inside queues. Numerical results shows performance similar to optimal switching solution, but with a higher flexibility due to neural techniques. Simulation results have been driven with memoryless distribution and heavy-tailed distribution for several input buffer size and switch dimension.
Fantacci R., Gubellini R., Tarchi D., Chiti F., Pecorella T. (2003). An efficient DiffServ switch for satellite communication systems based on cellular neural networks. 345 E 47TH ST, NEW YORK, NY 10017 USA : Institute of Electrical and Electronics Engineers Inc. [10.1109/vetecf.2003.1286063].
An efficient DiffServ switch for satellite communication systems based on cellular neural networks
Tarchi D.;
2003
Abstract
Quality of Sendee (QoS) has become an important topic in modern telecommunication network in order to guarantee multimedia traffic. In IP networks, DiffServ seems to be the best approach to satisfy QoS constraints, due to its end-to-end philosophy. Actual trend is to consider satellite on-board switching capabilities for managing multibeam input and output. In this paper, for reducing computational complexity, a Cellular Neural Network (CNN) has been proposed for the on-board switching problem; several traffic classes have been considered and switching algorithm has been implemented within a CNN taking into account their priority, queue length and time spent inside queues. Numerical results shows performance similar to optimal switching solution, but with a higher flexibility due to neural techniques. Simulation results have been driven with memoryless distribution and heavy-tailed distribution for several input buffer size and switch dimension.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
