基于CS-BPNN的网络安全态势评估方法研究

发布时间：2018-07-24 11:23

【摘要】：近年来,各种网络安全事件层出不穷,网络安全问题成为社会关注的热点,如何感知、防御网络威胁成为网络安全研究的重点。网络安全态势评估作为一门综合评估网络安全的技术,是现阶段网络安全领域研究的重点。神经网络作为人工智能的重要组成部分,在网络安全态势评估中扮演着重要角色,其中反向传播(back propagation,BP)神经网络应用最为广泛。但是BP神经网络评估模型的收敛速度较慢,局部搜索和全局搜索能力不足等因素影响评估性能。本文提出基于布谷鸟搜索算法优化BP神经网络(cuckoo search optimize back propagation neural network,CS-BPNN)的网络安全态势评估方法,并对该方法进行实验验证。首先,从理论上介绍网络安全态势评估知识,对网络安全态势评估的定义、目的、作用和常用评估方法作详细地研究分析,对比各方法的优势与不足,为后续提出的网络安全态势评估方法提供理论依据。其次,提出基于CS-BPNN的网络安全态势评估方法,设计评估模型,将评估模型抽象为输入部分、映射部分和输出部分。将映射部分的权值和阈值编码成布谷鸟,使用布谷鸟搜索算法寻找最优的权值和阈值,再经BP算法对误差进行调整。最后,对提出的基于CS-BPNN的网络安全态势评估方法进行实验验证。搭建实验网络,模拟正常用户和攻击者对网络进行访问和攻击,从网络中获取态势数据,与基于粒子群算法优化BP神经网络和基于BP神经网络的网络安全态势评估方法进行对比,使用Matlab软件完成三种方法的设计,实验结果表明本文所提出的方法具有较快的评估速度和较高精度。
[Abstract]:In recent years, a variety of network security incidents emerge in endlessly, network security issues have become the focus of social concern, how to perceive and defend against network threats has become the focus of network security research. Network security situation assessment as a comprehensive evaluation of network security technology is the focus of network security research. As an important part of artificial intelligence, neural network plays an important role in network security situation assessment, in which back propagation BP neural network is the most widely used. However, the convergence speed of BP neural network evaluation model is slow, and the local search and global search ability are not enough to influence the evaluation performance. This paper presents a network security situation assessment method based on Cuckoo search algorithm and optimized BP neural network (cuckoo search optimize back propagation neural network CS-BPNN). The method is verified by experiments. First of all, it introduces the knowledge of network security situation assessment in theory, makes a detailed study and analysis of the definition, purpose, function and common evaluation methods of network security situation assessment, and compares the advantages and disadvantages of each method. It provides the theoretical basis for the following network security situation assessment method. Secondly, the network security situation assessment method based on CS-BPNN is proposed, and the evaluation model is designed. The evaluation model is abstracted into input part, mapping part and output part. The weights and thresholds of the mapping part are coded into cuckoo, and the optimal weights and thresholds are found by using the Cuckoo search algorithm, and then the error is adjusted by BP algorithm. Finally, the proposed network security situation assessment method based on CS-BPNN is verified experimentally. Build the experimental network, simulate the normal users and attackers to access and attack the network, and obtain the situation data from the network. Compared with the methods of optimizing BP neural network based on particle swarm optimization and BP neural network, using Matlab software to complete the design of three methods. The experimental results show that the proposed method has higher evaluation speed and higher accuracy.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.08

【相似文献】