A Trust-Based Security Model to Detect Misbehaving Nodes in Internet of Things (IoT) Environment using Logistic Regression

Abstract

Ensuring authentication in the Internet of Things (IoT) environment is a crucial task because of its unique characteristics which include sensing, intelligence, large scale, selfconfiguring, connectivity, heterogeneity, open and dynamic environment. Besides, every object in the IoT environment should trust other devices with no recommendation or prior knowledge for any network operations. Hence, those characteristics and blindness in communication make security violations in the form of various attacks. Therefore, a trustbased solution is necessary for ensuring security in the IoT environment. Trust is considered as a computational measure represented through a relationship between trustor and trustee, explained in a particular context valued through trust metrics and evaluated by a trust mechanism. The proposed logistic regression-based trust model provides an efficient way to identify and isolate the misbehaving nodes in the RPL (Routing Protocol for Low Power Lossy Networks) based IoT network. It is one of the popularly used routing protocols in IoT, that builds a path especially for the constrained nodes in IoT environments. However, it is vulnerable to many attacks. The proposed model classifies and predicts the node’s behavior (trusted or malicious). This model uses the logistic regression model to predict the node’s behavior based on the integrated trust value which is computed from the direct trust, reputation score, and experience trust. It is primarily designed to address the black hole attack in the IoT environment. The mathematical analysis shows the possibility of the proposed work and the simulation results show the proposed model is better than the existing similar work.