CAJ | 학술논문

针对无线传感器网络节点能量少、存储空间小、计算能力有限的特性，本文提出了基于遗传算法(GA)和LMBP神经网络融合算法的入侵检测模型。与传统方法相比，该模型利用神经网络离线学习建立检测模型，无需储存大量的入侵行为特征，节省了存储资源。同时，采用多层合作检测机制，与采用混杂模式获取数据的方法相比，减少了能源消耗。仿真结果表明，GA-LMBP入侵检测模型在性能、能耗、存储开销、检测率和误检率都优于传统方法。
침대무선전감기망락절점능량소、존저공간소、계산능력유한적특성，본문제출료기우유전산법(GA)화LMBP신경망락융합산법적입침검측모형。여전통방법상비，해모형이용신경망락리선학습건립검측모형，무수저존대량적입침행위특정，절성료존저자원。동시，채용다층합작검측궤제，여채용혼잡모식획취수거적방법상비，감소료능원소모。방진결과표명，GA-LMBP입침검측모형재성능、능모、존저개소、검측솔화오검솔도우우전통방법。
According to the energy constrained, low-storage space and limited computing ability of wireless sen-sor network nodes, an intrusion detection model based on genetic algorithm (GA) and LMBP neural network was proposed. Compared with traditional methods, the program takes advantage of offline learning neural network algo-rithm to build detection model without storing large amounts of intrusion features, saving storage resources. Com-pared with the use of promiscuous mode capturing data, multi-detection cooperative mechanism reduces energy con-sumption. Simulation results show that GA-LMBP intrusion detection model in terms of performance, energy con-sumption, storage costs, the detection rate and false detection rate is better than those of traditional methods.