基于PCA的无线传感器网络入侵检测系统

摘要

Wireless Sensor Networks(WSN)has disadvantages of energy limited, computing power limited and vulnera-ble to attacks, so it proposes an intrusion detection system based on PCA and fuzzy cluster analysis. It extracts the charac-teristic values from PCA from the original attacking data and combines with the fuzzy clustering algorithm to analyze and classify the test data. It chooses VC++6.0, MATLAB R2010b and NS2 simulation platforms, assesses the detection rate, error rate performance of the program. The simulation results show that the program reduces the data dimension effectively, optimizes the characteristics of space by exacting the characteristic values from the known attacking types of training sets, and combines with the fuzzy clustering analysis method to improve the accuracy of the classification algorithm.%针对无线传感器网络（Wireless Sensor Network，WSN）能量有限、计算能力有限和极易受到攻击等缺陷，提出一种结合主成分分析法（Principal Component Analysis，PCA）和模糊聚类分析法的入侵检测系统。通过对原始攻击数据进行主成分分析提取其特征值，再结合模糊聚类算法对测试数据进行分析并分类。选用VC++6.0、MATLAB R2010b和NS2仿真平台，评估方案的检测率，误差率等性能和能量消耗。仿真结果表明，该方案将已知攻击类型的训练集进行特征提取有效地降低了数据维数，优化了特征空间，并结合模糊聚类分析法提高了分类算法的准确率。