Defending against Link Quality Routing Attacks in Wireless Sensor Networks

Abstract

Wireless sensor networks are gaining popularity due to their ease of deployment, low cost, portability, and scalability. However, the sensor nodes, which are the components of wireless sensor networks, are limited in communication, processing, and storage capabilities. They collect data from their surrounding environments and relay it using wireless communication to a central controller, which is more capable of processing and storing data. The wireless communication exposes the wireless sensor networks to different types of routing attacks. The goals of this dissertation are: study the two types of link quality routing protocols, identify the vulnerabilities of each type, and develop intrusion detection mechanisms to detect any malicious node that uses the identified vulnerabilities. Analysis and simulation are adopted as the research methodology. The first type of protocols requires the sensor nodes to cooperate to compute link qualities. An intrusion detection mechanism is proposed that introduces a gap in the sequence number of packets to detect a malicious node that advertises false link quality values. Hence, sensor nodes can expect the values of the link qualities of their neighbours to detect violators. The second type of protocols does not require the sensor nodes to cooperate. However, a malicious node may use false values for the parameters of the routing protocol, such as advertising false routing cost or sending frequent beacons. An intrusion detection system with two modules is proposed to detect this malicious node. The first module applies the watchdog concept to detect false routing costs and the second module applies a state machine to test the frequency of broadcasting beacons. The two intrusion detection systems and a routing protocol from each type are simulated in ns-2. The simulation results of the first intrusion detection system show that a small gap in the sequence numbers helps the sensor nodes to detect the malicious node effectively. The simulation results of the second intrusion detection system show that the success of the watchdog mechanism is dependent on the density of the sensor nodes in the network. However, the state machine mechanism helps the sensor nodes to detect the malicious node regardless of their density.