Two Layers Trust-Based Intrusion Prevention System for Wireless Sensor Networks

Abstract

Security of a wireless sensor network is aimed at ensuring information confidentiality, authentication, integrity, availability and freshness is an important factor considering the criticality of the information being relayed. Hence, the need for an intrusion detection/prevention system. Conventional intrusion avoidance measures, such as encryption and authentication are not sufficient because they become useless in the event of a sensor node being compromised, hence, can only be seen as a first line of defence in the network after which intrusion detection schemes follow. In this paper, two layers trust-based intrusion detection system was developed for wireless sensor networks. A trust-based model is presented to detect intrusions to the network. Scenarios were created by using different set of weights. By injecting 2%, 5% and 10% malicious nodes from the 100 nodes considered, the results obtained were carefully observed. For scenario 2 (S2) with 2% and 5% malicious nodes injected, the model achieved the best result in all cases with an average detection accuracy of 97.8% while scenario 3 (S3) with 10% of malicious nodes introduced recorded the best performance with an average accuracy of 96%. Hence, the model will be suitable with combination of weights in S2 with small networks but when the scale of the network increases, the set of weights in S3 are best with the model.