| With the rapid development of network technology and computer network business, more and more OoS (Quality of Service) algorithms are proposed to meet the network transmission service quality requirements. In various QoS techniques, QoS routing technology is one of the highly concerned technology. A basic QoS problem is the shortest path (SP) problem. Pulsed Couple Neural Network (PCNN) which is parallel coupling between neurons, and has auto-wave characteristic, can realize parallel process of the network. This is PCNN’s unique advantage in the application of searching for the shortest path. The existing PCNN finding shortest path algorithm only takes into account the path length, without considering other parameters. In this paper, we use a simplified Unit-Linking Pulse Coupled Neural Network(Unit-Linking PCNN), and take into account the bandwidth remaining rate together with the path length, add these two parameters to control neurons’threshold to finding the shortest path.Main work of this paper is described as follow:(1) Based on the Unit-Linking pulse coupled neural network, add the network parameter (bandwidth remaining rate) to the PCNN neuron model to control the threshold, then control the sequence of firing time. Improve PCNN model to make it more suitable for the application of the network, thus can effectively realize the purpose of finding the shortest path.(2) Respectively use linear threshold decline and exponential threshold decline. The experiments show that in the loose environment of the network, exponential decline can more quickly find the shortest path. But in the crowed situation, linear decline has certain advantages. Generally, use exponential threshold decline is more suitable.(3) Aiming at the dynamic changes in the network, according to the parallel processing advantage of PCNN, proposed the corresponding solution. The method can find the new shortest path accurately and quickly after the network parameters changed. |
PDF Full Text Request