A hidden Markov model based intrusion detection system for wireless sensor networks

Wireless sensor network (WSN) technology is being increasingly used for data collection in critical infrastructures (CIs). This paper presents an intrusion detection system (IDS), which is able to protect a CI from attacks directed to its WSN-based parts. By providing accurate and timely detection of malicious activities, the proposed IDS solution ultimately results in a dramatic improvement in terms of protection, since opportunities are given for performing proper remediation/reconfiguration actions, which counter the attack and/or allow the system to tolerate it. The proposed solution has the important advantage of exploiting the high accuracy of hidden Markov models as an effective means of detecting malicious activities. We present the basic ideas, discuss the main implementation issues, and perform a preliminary experimental campaign, with respect to sinkhole attacks, one of the most serious attacks to WSNs.