Simulation and evaluation of security and intrusion detection in IEEE 802.15.4 network.

摘要

With the fast growth of wireless sensor technology, the idea of using Wireless Sensor Networks (WSNs) in more critical applications of real life is becoming popular. WSNs are restricted in energy, memory and bandwidth which makes them particularly vulnerable against intrusions. Security is important and needs to be considered for any WSN. The standard specification of IEEE 802.15.4; as one technology of such networks, introduces many constraints which makes the application of security even harder. In addition, WSNs are vulnerable to an intruder's malicious attacks because of their characteristics and the nature of applications they are used for. In this thesis, I describe a secure data exchange protocol including a key exchange mechanism based on the ZigBee specification and built on top of IEEE 802.15.4 link layer. All nodes will apply power management technique based on the constant event sensing reliability required by the coordinator. Power management generates random sleep times by every node which on average fairly distributes the sensing load among the nodes. Key exchange is initiated by cluster coordinator after some given number of sensing packets have been received by the coordinator. On the other hand cheap devices and accessibility of these sensors, encourage adversaries to launch physical attacks. This can be done by compromising a node inside the network or introducing an extra malicious node to the network. I have simulated key exchange and power management technique into an IEEE 802.15.4 cluster's reliable sensing function. I evaluate the impact of security function and its periodicity on cluster performance. I have also simulated attacks in which sensor nodes are compromised and could be accessed by intruders to maliciously fool the network. I have proposed and simulated intrusion detection mechanisms utilized to detect such attacks. The results show high completeness of detection and suggest parameter selection values to keep high level of performance.