|
Since the late 20th century , the rapid development of information technology , embedded systems , wireless communications , distributed information processing technology and microelectronic mechanical systems and other technologies have developed rapidly , with perception, computing and wireless network communications capabilities for wireless sensor networks (Wireless Sensor Network, referred WSN) has aroused great concern. WSN through various integrated micro-sensor collaboration , real-time monitoring, sensing and collecting a variety of environmental or monitoring object. WSN logical and objective information on the physical world, the world together, changed the way people interact with the natural world , the expansion of the human ability to understand the world . With the rapid development of wireless sensor networks and applications, including network topology , node localization , communication protocols, network security has become a research hotspot. Especially in areas such as military applications , sensor nodes deployed in enemy controlled most of the region , node data acquisition and transmission is not to be found , and the enemy's grasp. Otherwise, once the sensor network is under attack and destruction by enemy mastered the information , the consequences could be disastrous. Since the sensor nodes and WSN networks specificity, the decision can not be copied using traditional network security measures . In combination with existing wireless sensor network intrusion detection technology , we propose a cluster-based wireless sensor networks hybrid intrusion detection system. The system model includes three modules , including misuse and anomaly detection module is used to discover intrusion detection module , decision module for feedback to help network managers to keep abreast of the situation , to strengthen the security of wireless sensor network review . Misuse detection module uses BP neural network as the core, the experiment using KDD CUP'99 data sets, using MATLAB software and additional momentum BP neural network BP neural network was trained and tested , the results showed that the detection accuracy of the two methods are is relatively high , and the additional momentum method convergence speed as more suitable for practical application .
|