Abstract
Recent years, the number of mobile uses is increased, then, the number of data packets flowing in the computer networks is also increased. Because the packets flowing in the computer networks become huge number, the packets are congested in the computer networks. If the packets congestion occurs in the computer networks, some packets are trapped in the congested nodes, then, these packets are delayed to be transmitted to the destinations. Further, the packets are removed from the computer networks in the worst case. To overcome these undesirable problems, an efficient routing method which uses the mechanism of the mutually connected neural net-works has been proposed by Horiguchi and Ishioka. This method searches the optimal path for each packet by using the principal of the energy minimization of the mutually connected neural networks. Then, this method has been evaluated by the network models which have fixed number of flowing packets in the computer networks. Recent works for the evaluation of the routing methods, many authors use the computer networks to which the data packet are gradually added. Then, in this paper, we evaluated the routing method for both network models. We show that the performance of the routing method becomes high for regular topologies of the computer networks.
| Translated title of the contribution | A Study of The Packet Routing Method Using Mutually Connected Neural Networks |
|---|---|
| Original language | Japanese |
| Pages (from-to) | 51 - 56 |
| Journal | IEICE technical report. Nonlinear problems |
| Volume | 113 |
| Issue number | 225 |
| State | Published - 26 Sep 2013 |