International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 11 | Nov 2024
p-ISSN: 2395-0072
www.irjet.net
Securing the Social Internet of Things: Trust Detection through Machine Learning Techniques Suvarna Rajappa1, Lohith S Y2 Swetha J3 1Lecturer, Department of Computer Science & Engineering, Government Polytechnic Sorab, Karnataka, India 2Lecturer, Department of Computer Science & Engineering, Government Polytechnic Sorab, Karnataka, India 3Lecturer, Department of Electronics & Communication, S J (Government ) Evening Polytechnic Bangalore,
Karnataka, India --------------------------------------------------------------------***----------------------------------------------------------------------Abstract - Social Internet of Things (SIoT) encompasses things in social network-like structures, thus incurring different
security and trust concerns. This research proposes the machine learning-based trust model for detecting trusted and untrusted devices in SIoT based on their evaluations of past interactions and reputation scores. Supplementary to manually trained classifiers and ML algorithms, our suggested model efficiently detects possible threats while generating a minimum of false alarms, thus improving SIoT network availability. Evaluation of the model shows positive results and we explain directions for enhancing trust in highly dynamic IoT settings.
Keywords: Social Internet of Things (SIoT), Trust detection, Machine learning, Device reputation, Anomaly detection, SIoT security, Network security
I. INTRODUCTION The Social Internet of Things (SIoT) is not a new concept in the overall Internet of Things (IoT) perspective. Thus, devices in SIoT create connections and interact similarly to people interacting in social networks with other people. These connections can be based on various factors, including shared functionalities, common ownership, collaborative tasks, or physical proximity. Trust detection in SIoT is critical because it involves determining the trustworthiness of these interconnected devices and their interactions. Trust is central to ensuring security, reliability, and privacy within SIoT networks, where devices are expected to work together and share sensitive information. Machine Learning (ML) is increasingly utilized in trust detection because of its ability to analyze complex patterns, identify anomalies, and make predictions based on vast datasets. This study explores how ML can be applied to trust detection in SIoT to enhance the security and reliability of these networks. It aims to develop and validate a trust detection model that can evaluate the trustiness of devices in a dynamic and social-structured IoT environment. Key challenges and their real-world implications 1. Dynamic and Autonomous Interactions: In contrast to conventional IoT, SIoT nodes actively connect and disconnect with receiving nodes based on social like relationships (e.g., tasks that are being worked on together, shared ownership). This gives variability to the device interactions, and therefore it becomes challenging to reliably determine the trustworthiness of devices. Real-world implication: The increase in possibilities for untrusted devices to enter critical system or sensitive data without outside monitoring. 2. Complexity of Relationship-Based Trust: Traditional social networks estimate the trust using other people’s interactions and this is relatively static. The SIoT trust approach must include device behavior patterns as well as contextbased interactions among these devices, which constantly change as some devices join or leave the network or transform their roles in the network. Real-world implication: Trust assessments could become stale in a short order and therefore create security breaches in critical areas that include healthcare smart cities among others. 3. High Scalability and Real-Time Requirements: Hence, trust models require scalability to accommodate a large number of interactions generated from increasing densities of devices forming SIoT networks at interactive rates. It is far from straightforward, especially in IoT where devices are mostly known to only exchange data with each other. Real-world
© 2024, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 650