DiffServ on-board satellite switching based on cellular neural networks

In modern satellite communication systems the Quality of Service (QoS) management has became a crucial topic due to the increasing interest in multimedia traffic. The actual trends is to consider the satellite networks as an integrated part of the terrestrial data networks. In IP networks, the Differentiated Service (DiffServ) approach seems to be the best to satisfy the QoS constraints, due to its end-to-end philosophy. Actual trend is to consider satellite on-board switching capabilities for managing multibeam inputs and outputs. In particular this paper deals with a the proposal of a new Cellular Neural Network (CNN) for the on-board switching problem in order to reduce the computational complexity; several traffic classes, according to the DiffServ approach, have been considered and the switch takes into account their priority, queue length and time spent inside queues. Numerical results show performance similar to the optimal switching solution, but with a cellular neural network that is a more flexible structure. Simulation results have been driven with memoryless distribution and heavy-tailed distribution for several input buffer size and switch dimension.