基于伪似然估计算法的网络延迟计算仿真  

Simulation of Network Delay Calculation Based on Pseudo Likelihood Estimation Algorithm

作　　者：金意[1] 

机构地区：[1]华东交通大学现代教育技术中心,江西南昌330013

出　　处：《计算机仿真》2013年第10期333-336,共4页Computer Simulation

摘　　要：在网络实际应用过程中,由于网络规模不断扩大,网络复杂性不断提高,在网络数据传输过程中,需要获取网络中的节点性能参数。但是,网络中的节点存在较大的差异性,将造成节点的性能参数过于复杂,在通信过程中协调性较差,存在冲突现象,导致对这种延迟的估计出现偏差。提出了一种伪似然的网络延迟估计算法。利用小波域数据线性变换模型,计算网络状态参数,为网络延迟计算提供准确的数据基础。统计网络延迟的概率分布情况,将网络延迟计算问题,转化为伪似然估计问题,并对该问题进行求解处理,消除干扰的影响。实验结果表明,利用该算法进行网络延迟计算,能够在网络规模比较大、网络复杂性比较高的情况下有效地获取网络延迟程度,获取的结果与实际的网络延迟分布情况非常接近,真实的反映了网络延迟情况。In the process of network applications, due to the enlarged network scale and increased network com- plexity, in network data transmission process, performance parameters of nodes in the network need to be obtained. However, there is a big difference existing in the network nodes, which will cause the performance parameter of the node too complex, and poor coordination in the process of communication is occurred. Also, there is a conflict phe- nomenon which leads to errors occuring in estimation of the delay. A network delay estimation algorithm for pseudo likelihood was put forward. A wavelet domain data linear transformation model was used to calculate network status parameters, thus providing accurate data basis for the network delay calculation. Probability distribution situation of network delay was counted to transform network delay calculation problem into pseudo likelihood estimation problem, and solving the problem to eliminate interference effects. Experimental results show that using this algorithm to calcu- late the network delay, can access to network latency degree effectively under the condition of the relatively large size of the network and the high network complexity, the obtained results arc closed to the actual distribution network de- lay, which reflects the real network latency.

关 键 词：伪似然估计 网络延迟 计算复杂度 网络规模 

分 类 号：TP353[自动化与计算机技术—计算机系统结构]