基于CS-CPSO与SVM融合的WSNs入侵检测算法

摘要

In order to improve the detection precision and convergence rate of the intrusion detection algorithm for wireless sensor networks(WSNs) based on fusion of particle swarm optimization (PSO) algorithm and support vector machine(SVM),a WSNs intrusion detection algorithm based on complete sine-mapping chaotic CS-CPSO-SVM based on fusion of CS-CPSO and SVM is proposed.Chaotic PSO algorithm based on CS-CPSO is used to optimize the parameter of SVM,and the sine-mapping chaotic search is applied to not only the generation of initial population and chaotic perturbation of local optimal for PSO algorithm,but also the optimization of the inertia weight and the generation of the random constant and the learning factor,moreover,multiple initial values are used to generate a number of chaotic orbits.With the KDDCUP99 data set as experimental data,the oretical analysis and simulation results show that the proposed method can effectively detect the intrusion behavior,and has a good detection precision and convergence speed.%为了提高基本粒子群优化(PSO)算法与支持向量机(SVM)融合的无线传感网络(WSNs)入侵检测算法的检测精度与收敛速度,提出了一种基于完全正弦映射混沌粒子群优化(cS-cPSO)算法与SVM融合的WSNs入侵检测算法(CS-CPSO-SVM).采用CS-CPSO算法优化SVM参数,不仅将正弦映射混沌搜索应用于粒子群算法中初始种群与局部最优解混沌扰动的产生,且将其用于惯性权重的优化以及随机常数和学习因子的产生,并用多个初始值分别迭代生成多条混沌轨道.以KDDCUP99数据集作为实验数据,经理论分析与仿真实验表明:该方法可以有效地检测入侵行为,并具有良好的检测精度与收敛速度.