通信学报
通信學報
통신학보
JOURNAL OF CHINA INSTITUTE OF COMMUNICATIONS
2010年
1期
65-72
,共8页
聂仁灿%周冬明%赵东风%谭颖芳
聶仁燦%週鼕明%趙東風%譚穎芳
섭인찬%주동명%조동풍%담영방
竞争型脉冲耦合神经网络%多约束QoS路由%脉冲波任务%最短路径
競爭型脈遲耦閤神經網絡%多約束QoS路由%脈遲波任務%最短路徑
경쟁형맥충우합신경망락%다약속QoS로유%맥충파임무%최단로경
competitive PCNN%multiple constrained QoS route%pulse-wave task%shortest path
在脉冲耦合神经网络的基础上提出了竞争型脉冲耦合神经网络模型,分析了该模型用于求解网络最短路由时的脉冲波传播特性,并提出了脉冲波任务的产生、分解和状态转换理论,在模型中实现了脉冲波的多约束传播,成功地应用于网络多约束QoS路由问题的求解,并可得到全局最优解.仿真实验表明,与其他算法相比,该方法的计算迭代次数最少,且减少较多;而且迭代次数只与网络路由图中源点与目的点之间的最优QoS路由长度有关,而与节点数、链路数和网络的分布构成复杂性无关,体现出较好的计算性能和优势.
在脈遲耦閤神經網絡的基礎上提齣瞭競爭型脈遲耦閤神經網絡模型,分析瞭該模型用于求解網絡最短路由時的脈遲波傳播特性,併提齣瞭脈遲波任務的產生、分解和狀態轉換理論,在模型中實現瞭脈遲波的多約束傳播,成功地應用于網絡多約束QoS路由問題的求解,併可得到全跼最優解.倣真實驗錶明,與其他算法相比,該方法的計算迭代次數最少,且減少較多;而且迭代次數隻與網絡路由圖中源點與目的點之間的最優QoS路由長度有關,而與節點數、鏈路數和網絡的分佈構成複雜性無關,體現齣較好的計算性能和優勢.
재맥충우합신경망락적기출상제출료경쟁형맥충우합신경망락모형,분석료해모형용우구해망락최단로유시적맥충파전파특성,병제출료맥충파임무적산생、분해화상태전환이론,재모형중실현료맥충파적다약속전파,성공지응용우망락다약속QoS로유문제적구해,병가득도전국최우해.방진실험표명,여기타산법상비,해방법적계산질대차수최소,차감소교다;이차질대차수지여망락로유도중원점여목적점지간적최우QoS로유장도유관,이여절점수、련로수화망락적분포구성복잡성무관,체현출교호적계산성능화우세.
Based on the PCNN (pulse-coupled neural network), the CPCNN (competitive PCNN) model was proposed. First, the properties of pulse-wave propagation in the CPCNN were analyzed for the solution of network shortest route. Then the theories of generation, decomposing and states conversion of pulse-wave task were suggested to reafize the pulse-wave propagation constrained. Finally, the CPCNN was successfully applied to solve multiple constrained QoS route with globally optimal result. The computer simulation shows that the number of iterations for proposed algorithm is the least, and the difference is fairly obvious compared with former methods, it only depends on the length of the globally optimal QoS route between source node and destination node, not the number of nodes and edges, and the complexity of network distribution structure. Thus, the method has better computational performance and dominance.