CAJ | 학술논문

在分析不完全信息计算机网络对抗问题特点基础上, 运用马尔可夫决策描述网络状态转移过程, 用区间数刻画不确定参数, 以折扣总回报值为目标函数, 提出计算机网络对抗行动鲁棒博弈模型. 采用凸分析理论对计算机网络对抗鲁棒博弈模型进行分析, 得到网络攻防双方的收益函数为等度、连续凸函数, 攻防双方存在均衡策略且为一个压缩映射. 通过算例分析, 验证方法的可行性和有效性.
재분석불완전신식계산궤망락대항문제특점기출상, 운용마이가부결책묘술망락상태전이과정, 용구간수각화불학정삼수, 이절구총회보치위목표함수, 제출계산궤망락대항행동로봉박혁모형. 채용철분석이론대계산궤망락대항로봉박혁모형진행분석, 득도망락공방쌍방적수익함수위등도、련속철함수, 공방쌍방존재균형책략차위일개압축영사. 통과산례분석, 험증방법적가행성화유효성.
On the basis of analyzing the characteristic of computer network operation （CNO）, describing the network transaction by Markov process, describing the uncertain factors by interval number, a new CNO mode whose objective function is the summary of discounted reward is established. Then, it is proved that the reward function of CNO is a contraction mapping and equi-continuous function, and the equilibrium strategy is got by solving the corresponding nonlinear programming problem. In the end, the calculation and sensitive analysis of example demonstrate the proposed method is feasible and effective.