Research On The Fusion Of Key Management And Intrusion Detection In Wireless Sensor Networks

With the development of wireless sensor technology, low-power electronics, RFtechnology and embedded technology, wireless sensor networks has become the mostactive emerging interdisciplinary field of research, which has a huge value in themarket and broad development prospects. It is widely used in military defense,medical monitoring, environmental monitoring, traffic management and many otherfields. Sensor nodes are usually deployed in unattended and non-secure area, whichare limited in node energy, computing and communication capabilities. They havepoor safety performance and are faced with a variety of security threats. Therefore,the security issue in wireless sensor networks has become a hot spot of most concern.Designing a reasonable and effective security solutions has become the target of manyscholars eager to study.Key management and intrusion detection are two security defense in wirelesssensor network security. The former withstand attacks outside the network and thelatter against attacks within the network. So far, many distinctive key managementscheme have been raised to against external attacks, but the ability to withstandattacks from inside the network were not taken into account.In this thesis, a lightweight and self-against key management scheme is proposedfor security requirements of the wireless sensor networks and the above problems.The method contains a two-stage authentication method based on random numbersand a abnormal behavior detection module based on analysis of packets. In theauthentication mechanism, the base station is a trusted third party and randomnumbers are links. It completes key pre-distribution while providing effectiveauthentication for nodes in wireless sensor network. Member nodes record behavioralcharacteristics of the cluster heads by analyzing the packets and provide the basis forkey update. The key management method has some advantages on networkconnectivity, survivability and energy overhead compared to the probability-basedrandom key pre-distribution scheme. The two-stage authentication method does notincrease the communication overhead of the key management scheme but savingenergy. The abnormal behavior detection module with no separate monitoring node,smaller energy cost, can withstand simple attacks to cluster head node such as replayattack, select forward attacks, and so on. The program is suitable for small andmedium clustering wireless sensor networks and it is scalable to larger clusteringwireless sensor networks.