CAJ | 학술논문

针对业界常用的定长马尔可夫模型判别伪装入侵检测存在的不足，以及不定阶马尔可夫模型在存储方面存在的限制，提出一种基于PSA的用户伪装攻击方法，阐明训练和预测步骤和方法。并通过实验证明，该方法具有稳定性强，判别精度高的特点，适合应用于在线检测用户伪装攻击。
침대업계상용적정장마이가부모형판별위장입침검측존재적불족，이급불정계마이가부모형재존저방면존재적한제，제출일충기우PSA적용호위장공격방법，천명훈련화예측보취화방법。병통과실험증명，해방법구유은정성강，판별정도고적특점，괄합응용우재선검측용호위장공격。
Aiming at the deficiency of the fixed length Markov model for the common intrusion detection methods and limitation of the adventitious variance length Markov model in terms of storage, a user impersonation attack method based on PSA is presented and the steps and methods of training and prediction is illustrated in this paper. The experimental results show that the method has a feature of strong stability and higher accuracy of discrimination, which is suitable for online de-tection of impersonation attack.