Performance Analysis of Multiple Input-Queuing Scheduling Employing Neural Network in ATM Switches.

In this paper, an improved multiple input-queuing (IMIQ) fabric and scheduling algorithm by employing a new energy function based on Hopfield neural network (HNN) in asynchronous transfer mode (ATM) Switches is proposed. The policy of more than one cell transferred in each input port during every time slot is adopted. Performances such as throughput, cell loss and cell delay of this new approach are analyzed and compared with other methods. The study shows that the performances of the new method are better than the others. And the scale of the HNN used in our new approach is much smaller than the one used in Virtual Output-Queuing (VOQ). In addition, due to the HNN model is able to be implemented by circuit or optoelectronic device easily, our approach can be applied to large-scale ATM switches optimization scheduling on line.