A CNN model for ATM cells scheduling in MIN switches

Abstract

The multistage interconnection network (MIN) with self-routing is one of the best structures for implementation of IP/ATM switches, however, its internal cell blocking introduces more complications in the design of optimal scheduling algorithms for IP/ATM switches with quality of service (QoS) guarantees. A new model for cellular neural networks (CNN) is introduced to control traffic cell scheduling in MIN switches. the CNN scheduling algorithm will not only avoid cell blocking in MIN switches and therefore provide the global optimal solutions for traffic cell scheduling with QoS guarantees, but can also be programmed for different QoS reassignment. Our model is also resilient in the presence of switch element failure. The flexibility of our CNN model makes it very useful for applications in unreliable environments where the opportunity for link failure and node relocation are high.