Localizing Worst-Parent Rank Attack Using Intelligent Edges of Smart Buildings

Abstract

Routing Protocol for Low power and lossy networks (RPL) is the underlying protocol of 6LoWPAN, a core communication standard for the Internet of Things (IoT). The concept of rank is used in RPL for routing and its optimization. However, the rank provides an avenue for “Rank Attacks (RA)” like worst parent, increased rank, and decreased rank, which are internal to 6LowPAN. In this paper, a smart building scenario with intelligent edges is considered for detecting the worst-parent attacks of RPL. Novelty of our approach is that, in addition to detecting the attack, the impact of the malicious node location on the average statistics of control messages is used to identify its locality, using Machine Learning (ML) classification at the edge of the network. An “impact factor” has been defined and analyzed in terms of the forwarding load of the node. Simulation results obtained for a given static node topology, show a localization accuracy of 0.67 using Support Vector Method (SVM), while the accuracy obtained using eXtreme Gradient Boost (XGBoost) is 0.88. Discussion on how the technique is applicable to any generalized network topology, to reduce the number of packet level inspections for malicious node detection, is also presented.