Machine Learning Based Intelligent RPL Attack Detection System for IoT Networks

Routing Protocol for Low-Power and Lossy Networks (RPL) has been used in the Internet of Things (IoT) with Wireless Sensor Networks (WSNs), but it is does not give much focus to protect against routing attacks. Therefore, it is possible for an attacker to utilize the RPL routing system as an initial platform for devastating and crippling attacks on an IoT network. In IoT based networks, RPL provides a minimal level of protection against a wide variety of attacks, which are unique to RPL and are launched in WSNs. Moreover, the traditional Internet and routing security solutions also have memory, processing, and resource limitations that make them ineffective for IoT devices. Several mitigation schemes, such as those based on rule-based learning algorithms, Intrusion Detection Systems (IDSs), and trust computation and management techniques were proposed in the past to improve the safety of IoT networks and routing, but they do not provide the required security. To overcome these security issues, we propose an intelligent IDS in this article to detect and isolate the RPL attacks. For this purpose, we propose a new classification algorithm based on neural networks and genetic algorithms. Moreover, the proposed Neuro Genetic Classification Algorithm (NGCA) detects the nodes which launch RPL and other attacks more accurately and hence it increases the network security in IoT. From the experiments done with NSL-KDD data-set, it is proved that the proposed NGCA is detecting the attacks with higher accuracy than the other classifiers namely Decision Trees, Logistic Regression and Support Vector Machines. It also helps to reduce the RPL attacks in the routing process with enhanced detection rate with reduced false positives.