Trust Management Approach for Detection of Malicious Devices in SIoT

Abstract

Internet of Things (IoT) is an innovative era of interrelated devices to provide services to other devices or users. In Social Internet of Thing (SIoT), social networking aspect is used for building relationships between devices. For providing or utilizing services, devices need to trust each other in complex and heterogeneous environments. Separating benign and malicious devices in SIoT is a prime security objective. In literature, several works proposed trust computation models based on trust features. But these models fail to identify malicious devices. This paper focuses on detection of malicious devices. In this paper, basic fundamentals, properties, models and attacks of trust in SIoT are discussed. Up-to-date research distributions on trust management and trust attacks are reviewed and idea of Trust Management using Machine Learning Algorithm (TM-MLA) is proposed for identification of malicious devices.

A Review of Security Challenges and Solutions in Wireless Sensor Networks

“…In smart cities, low-data-rate WSNs are employed for monitoring and controlling various applications. The sensor nodes serve as the foundational technology infrastructure in the IoT [4,31]. However, there are challenges such as power management, security, and data management that researchers are developing new techniques and technologies to address [4,31].…”

“…Thus, the trust and reputation concepts can be used to determine which sensor nodes to trust and which sensor nodes to avoid in the network. Trust and reputation are mathematical tools that represent a sensor node's opinion of another sensor node [16,19,31,33]. However, trust and reputation management models are divided into three types: distributed, centralized, and hybrid [13,31,34,35].…”

“…Trust and reputation are mathematical tools that represent a sensor node's opinion of another sensor node [16,19,31,33]. However, trust and reputation management models are divided into three types: distributed, centralized, and hybrid [13,31,34,35]. In WSNs, trust and reputation models can be used for various purposes, such as: ➢ Routing: trust and reputation can be used to select the most reliable and trustworthy paths for routing data in the network.…”

A Trust Management Model for IoT Devices and Services Based on the Multi-Criteria Decision-Making Approach and Deep Long Short-Term Memory Technique

“…Similarly, a study in [26] developed an approach for trust management in social IoT devices. The approach contained three main stages: (1) trust composition stage-at this stage, various attributes are chosen as per the attack context.…”

“…A comparison was performed between LSTM and the MLP, and the ANN, which were used in the literature as deep learning architectures (MLP in [26] and ANN in [25]) based on the model setup in Section 5.1. Theoretically, to select a suitable model, it is important to take the dataset patterns into consideration because fitting the data with the models is a condition for ensuring its effective performance.…”