网络对抗攻击中外部注入病毒检测模型仿真

Detection Model Simulation of Network Against Attacks From External Injection Virus

研究网络对抗攻击中外部注入病毒检测问题。在网络对抗中,往往存在从外部人为注入的入侵病毒,病毒特征具有高伪装性,无法形成常规的识别特征,传统的病毒检测方法以固定病毒特征作为依据,一旦外部病毒特征不在其数据库内,将造成检测准确性下降的问题。提出基于RBF前馈式神经网络算法的网络对抗攻击中外部注入病毒检测方法。根据极值距离方法相关理论,能够计算外部注入病毒初始聚类中心,得到病毒检测的初始种群,通过迭代处理的方法获取种群中的差异个体。带入RBF神经网络结构,通过计算隐含层和输出层的输出结果,表示网络对抗攻击中外部注入病毒检测结果。实验结果表明,利用改进算法进行网络对抗攻击中外部注入病毒检测,可以提高检测的准确性。

This paper studied the method of detecting external injected virus in network confrontation attacks. There are intrusive viruses that juiced artificially from outside during network confrontation. One of the characteristics of these kinds of viruses is their high camouflage, for which conventional identification features cannot be formed. While in traditional virus detection methods, fixed virus signatures were regarded as the detective foundation. Once the feature of external injected virus goes out of scope of virus database, the detection accuracy will decrease. In this study we propose a detection method for external injective virus in network confrontation attacks based on the RBF feed-forward neural network algorithm. On the basis of related theories about maximum distance method, the initial cluster centers of external invading virus can be calculated, with which initial population of virus will be obtained, followed by catching the different individuals via iterative process. After putting them into the RBF neural network and then getting the outputs calculated from the hidden layer and output layer, virus detection results will he exhibited. Experimental results show that the detection accuracy is improved by using this improved algorithm in the virus-attacks process of network confrontation.