自体集网络入侵检测中的高效寻优算法仿真

Optimizing Algorithm Simulation of Network Intrusion Detection of Autologous Set

延迟容忍网络环境复杂,自体集数据规模较大造成的时空上的巨大消耗而难以处理,导致传统网格入侵匹配的数据分类过程收敛速度慢、局部最优等缺陷。为了提高网格入侵特征匹配效率,提出一种基于重复博弈的自体集网络入侵检测中的高效寻优算法。算法利用了入侵数据特征在空间上的相对集中性,在分类概率计算中,加入一种约束计算过程的博弈因子,在概率匹配过程中,通过数据博弈消除多次概率对比,并加入博弈约束计算最优反应函数,将最优反应函数求解过程看作是一个博弈的过程,计算分类优化达到的极大值。仿真结果表明,改进算法的分类匹配时间复杂度有了一定程度的降低,优化效果较为明显。

In order to improve the efficiency of the grid intrusion feature matching,we proposed an efficient optimization algorithm based on repeated game network intrusion detection of autologous set.We utilized the characteristics of the data in space invasion relative concentration,and in the classification probability calculation,we added a constraint factor calculation process of the game.In the matching process of the probability,we eliminated the multiple probability comparison through data game,joined the game constrained computing best-response functions,and calculated the maximum value of classification optimization.Simulation results show that the complexity of the classification matches time has reduced to a certain degree in this improved algorithm,and the optimization results are obvious.