遗传投影寻踪确定权重的网络仿真评价方法  

Evaluation of Network Performance Using Genetic Projection Pursuit Model Weighted Method

作　　者：曹庆璞[1] 董淑福[1] 罗赟骞[1] 苏广军 

机构地区：[1]空军工程大学电讯工程学院,陕西西安710077 [2]山东省济宁市人民防空办公室,山东济宁272137

出　　处：《计算机仿真》2010年第10期103-106,262,共5页Computer Simulation

基　　金：国家综合业务网理论及关键技术重点实验室开放基金(ISN-9-08)

摘　　要：针对网络运行质量评价中多元指标权重难以确定的问题,为了提高网络传输速度等性能,采用投影寻踪方法对各指标分类特征进行发现以确定指标客观权重,并利用混合遗传算法解决难以合理求解的问题。采用NS2网络仿真获得确定指标客观权重所需要的网络运行样本数据。仿真实验表明,与主观赋权法相比方法确定的权重客观地反映了各指标之间的相对重要程度,能够为网络运行质量的计算提供重要依据,证明了方法的有效性和可行性。Aimed at solving the problem of determining the weight of multivariate indexes,the projection pursuit method was introduced to find the index classificatory character for determining the index objective weight,then the improved genetic algorithm was adopted to solve the problem rationally.It used the network simulation to derive the samples which can validate the feasibility of the method.The case study demonstrates that the genetic projection pursuit method can objectively reflect the related importance of each index compared with subjective weighting methods,which could provide important proof for evaluation of network performance,which proved the effectiveness and feasibility of the method.

关 键 词：网络管理 运行质量评价 遗传投影寻踪 权重 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]