基于点火耦合神经网络的多约束QoS路由选择算法

A multiple constrained QoS routing based on firing coupled neural networks

本文针对多约束QoS路由选择问题,将其转化为一个多约束的赋权图最短路问题,并建立点火耦合神经网络,通过在其上所具有的自动波生成和传播特性,并在自动波的传播过程中随时监督约束的满足情况,及时取消不满足约束的自动波,从而最先到达目的节点的自动波所走过的路径即为多约束QoS的最优路径。该算法具有高度的并行性,并总是获得全局最优解,所需的迭代次数相对其他算法而言也是最少的。最后本文给出了实验结果及与其他算法的比较。

This paper aims at solving QoS(quality of service) routing with multiple constraints. Firstly, the solution of the QoS routing is converted to a shortest path problem for a weighted graph with multiple constraints. Then firing coupled neural network(FCNN) is applied, where autowave can be generated by the firing of the neuron corresponding to the start vertex in the graph, and such autowave can travel through the firing coupling between neurons. The autowaves are checked whether all of the constrains are satisfied at any time, and if at least one of them is not for some autowave, that autowave is omitted. The route of the autowave that firstly arrives at the neuron corresponding to the end vertex in the graph is the solution of the multiple constrained QoS routing problem. The algorithm presented here is of high parallelism and the route obtained from the running of FCNN is always globally optimal one for an appropriate choice of time step. The number of iterations necessary is much less than those compared with other existing methods. Finally, experimental result is presented showing that the algorithm is effective and efficient in finding the optimal route for multiple constrained QoS problems.