Traffic Convergence Detection in IoT LLNs: A Multilayer Perceptron based Mechanism

Abstract

The Low power and Lossy Networks (LLNs) form an important segment of the Internet of Things (IoT). LLNs comprise of sensors and RFIDs which are constrained and not IP-enabled. IPv6 over Low power Personal Area Networks (6LoWPAN) enables connectivity of constrained non IP-enabled devices to the Internet. The routing protocol used in 6LoWPAN is IPv6 Routing Protocol over Low power and lossy networks (RPL). Though RPL meets all the routing requirements of LLNs, it is prone to several attacks. Among the several RPL attacks, misappropriation attacks are those which disrupt the legitimate path of traffic flow in the lossy network and causes convergence of a large section of traffic towards a particular malicious node. Also, misappropriation attacks can make the LLNs vulnerable to several other security attacks. Hence, it is important to timely detect misappropriation attacks. In this paper, we propose a mechanism to detect misappropriation attacks in IoT LLNs. Our approach makes use of Multilayer Perceptron (MLP) neural network as a classification tool. The MLP classifies the network data as normal or as under attack. Our proposed mechanism also identifies the nodes affected by the attack and identifies the attacker node.