基于RBF神经网络的攻防博弈模型  

ATTACK-DEFENSE GAME MODEL BASED ON RBF NEURAL NETWORK

作　　者：娄燕强[1] 宋如顺[1] 马永彩[1] 

机构地区：[1]南京师范大学数学与计算机学院,江苏南京210097

出　　处：《计算机应用与软件》2011年第1期99-101,共3页Computer Applications and Software

基　　金：国家"211工程"建设项目(181070H901)

摘　　要：为了阐明网络攻防过程中博弈双方如何确定对方的类型,从而选择行动策略,提出了基于RBF神经网络的攻防博弈模型。首先使用两人随机博弈模型来分析网络攻防双方的特点,揭示制约双方选择策略的因素;通过精炼贝叶斯纳什均衡求得博弈双方选择的最优策略;最后,根据可疑者的行动策略和系统的状况,使用RBF神经网络对其类型进行推理。In order to elucidate how to determine the types between game sides in the process of network attack and defense and then to choose the action strategy,the attack-defense game model based on RBF neural network is put forward.Firstly,the two-player stochastic game model is used to analyze the characteristics of offensive and defensive sides,to reveal the restrict factors of strategies selection.Then,the optimal strategies chosen by both sides can be got through perfect Bayesian Nash equilibrium.At last,the RBF neural network is adopted to reason out the types of the suspects according to their action strategies and system status.

关 键 词：博弈模型 网络攻防 RBF神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]