Intrusion Detection and Prevention in CoAP Wireless Sensor Networks Using Anomaly Detection

It is well recognized that security will play a major role in enabling most of the applications envisioned for the Internet of Things (IoT). We must also note that most of such applications will employ sensing and actuating devices integrated with the Internet communications infrastructure and, from the minute such devices start to support end-to-end communications with external (Internet) hosts, they will be exposed to all kinds of threats and attacks. With this in mind, we propose an IDS framework for the detection and prevention of attacks in the context of Internet-integrated CoAP communication environments and, in the context of this framework, we implement and experimentally evaluate the effectiveness of anomaly-based intrusion detection, with the goal of detecting Denial of Service (DoS) attacks and attacks against the 6LoWPAN and CoAP communication protocols. From the results obtained in our experimental evaluation we observe that the proposed approach may viably protect devices against the considered attacks. We are able to achieve an accuracy of 93% considering the multi-class problem, thus when the pattern of specific intrusions is known. Considering the binary class problem, which allows us to recognize compromised devices, and though a lower accuracy of 92% is observed, a recall and an F_Measure of 98% were achieved. As far as our knowledge goes, ours is the first proposal targeting the usage of anomaly detection and prevention approaches to deal with application-layer and DoS attacks in 6LoWPAN and CoAP communication environments.