基于BP-DBN的认知网络端到端态势评估算法  

Cognitive Network End to End Situational Evaluation Algorithm Based on BP-DBN

作　　者：蒋云洁 王莉[1] 

机构地区：[1]南京邮电大学,江苏南京210003

出　　处：《计算机技术与发展》2014年第11期148-151,共4页Computer Technology and Development

基　　金：国家"863"高技术发展计划项目(2009AA01Z212)

摘　　要：文中提出了一种基于BP-深度信念网络(BP-DBN)的端到端态势评估算法,实现网络端到端态势等级判定。基于提出的分布式态势评估架构,使用BP-DBN分别构建认知域网元评估值、局部态势评估值和端到端态势评估值三者间的映射关系,最后实现端到端态势等级定性评估。实验结果表明,基于少量标记训练样本,BP-DBN测试错误率低,能够保证评估准确性,同时提出的评估算法能够有效评估端到端网络态势等级。Cognitive network end to end situational evaluation algorithm based on BP-DBN is proposed to judge the situational level. Based on distributed situational evaluation architecture,BP-DBN is used to construct mapping relations among network element evalua-tion value in cognitive domain,local situational evaluation value and global situational evaluation value. End to end situational evaluation can be qualitatively evaluated. Simulation results show that test error rate of BP-DBN is low based on less labeled samples,which can en-sure accuracy of evaluation,and the algorithm proposed can estimate end to end situational level effectively.

关 键 词：认知网络 BP-深度信念网络 端到端态势评估 受限波尔兹曼机 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]