Context-oriented trust computation model for industrial Internet of Things

Abstract

The Industrial Internet of Things (IIoT) has revolutionized the industrial sector by providing advanced and intelligent applications. The objects and nodes communicate with one another to collect, exchange, and analyze a large amount of sensing data using techno-social systems, thereby challenging the security and trustworthiness of the data. To achieve effective communication in IIoT, trustworthy relationships must be established among these objects. This makes trust an important security parameter in an IoT-based environment to achieve secure and reliable service communication at the edge nodes. In this paper, we propose an adaptive Context-Based Trust Evaluation System (CTES), which calculates distributed trust at the node level to achieve edge intelligence. Each edge node takes recommendations from its context-similar nodes to calculate the trust of serving nodes. This collaborative trust calculation mechanism helps in filtering out malicious nodes in the network. The weighing factor “$\mathrm{\mu }$” is dynamically assigned based on the previously calculated trust score experienced by the edge node. This research also focuses on formal verification of the proposed CTES model. We analyze the efficiency of CTES in terms of accuracy, dynamic assignment of $\mathrm{\mu }$, and resiliency against Ballot Stuffing and Bad Mouthing attacks to avoid malicious nodes. The results ensure the significance of the proposed CTES model for dynamic assignment of $\mathrm{\mu }$ and provide satisfactory results against EigenTrust, ServiceTrust, and ServiceTrust$++$ in terms of detecting malicious nodes and isolating them from providing recommendations.

Introduction

Industry 4.0 is the most significant industrial revolution, which is focused on creating smart factories by using smart machines [1]. Generally, Industry 4.0 is used for the automation and exchange of data between smart machines for manufacturing purposes, which comprise the Internet of Things (IoT), Industrial Internet of Things (IIoT), and Cyber–Physical Systems (CPS). The IoT connects physical objects through the Internet using sensors, RFID tags, and various smart devices. Sensing devices are used to get the stimulus from the environment and respond to the system. IoT has become a fundamental part of the smart environments and provides multiple services in safety, transport, healthcare, surveillance systems, education, and more importantly, in the industrial domain. The communication in IIoT is based on IoT-enabled devices that can run numerous applications for collaborative communication in smart manufacturing to generate large amounts of data. This data needs to be trustworthy and secure by isolating false data generated by the malicious nodes. This can be achieved through edge intelligence, which refers to the process of data collection, analysis, and related calculations at the node that captures or generates the data. IIoT is providing solutions to smart manufacturing in combination with security mechanisms to ensure the reliability of data and to improve communication between smart machines. Although this makes IoT helpful in everyday life, it also opens the doors of threats and vulnerabilities [2]. For the service exchange plan, devices need to be in contact with each other to share the data to provide services in industrial applications. There should be mutual trust between the service requester and the service provider. Therefore, it is imperative to measure the reliability of service providers.

Trust formation is important and can have different definitions according to the requirement. Trust is defined as the only requirement that is used to access the resources and information that need to be shared. The best definition of trust found in the IoT literature is, “Trust is the edge that links the intelligent object with the technological ecosystem” [3]. The presence of trust is helpful in making decisions in the network that comprises multiple systems, which makes it a significant part of the system for devices to perform the requested services. In an IoT network, trust is established among different nodes to complete the requested task. The important factor is when they need to associate with the unknown devices. At that time, devices need to have some trust establishment procedure to communicate with unfamiliar ones. IoT is a blend of various types of devices in the same network; thus, trustworthy communication between all associated entities is important [3].

This research proposes an intelligent trust model, which helps the edge nodes to request services from a reliable service provider instead of a malicious one. Since data is the most significant part of an IIoT architecture, we must consider the trust among entities to share the data for any given task. It is important to consider the context of a service provider in a network for trust calculation. The nodes that are used in the trust calculation process belong to various contextual environments; therefore, the Trust Management System (TMS) must include the context-similar nodes only while formulating the trust. A context in our research is to consider multiple factors, as discussed in Section 3, while taking the recommendations from other nodes. Context represents the environment of users and servers under which they interact to establish a service interaction. This represents the current user environment, server location, and Quality of Service (QoS) parameters.

The main contributions of this research are as follows:

• We propose a context-based adaptive IoT trust model for edge intelligence. The novelty is in using collaborative filtering to collect trust feedback from nodes that get services under the same context.

• We develop a direct trust calculation mechanism using actual user satisfaction experience, selected between the range (0–1), based on the percentage of positive observations instead of binary representation, which was previously used in the literature.

• We develop an adaptive calculation mechanism to allocate weights to direct trust and indirect observations based upon the current user experience while considering the context.

• The proposed Context-based Trust Evaluation System (CTES) model is designed and verified using High-Level Petri Nets (HLPN).

• We evaluate CTES to analyze its effectiveness in avoiding malicious nodes against ballot stuffing and bad-mouthing attacks.

The rest of the paper is organized as follows: The introduction section is followed by Section 2, which presents the related work in the domain of IoT. Section 3 describes the proposed CTES model, which gives a clear explanation and working of the mentioned techniques for trust calculation. Section 4 presents the formal modeling and verification of the proposed CTES model. Section 5 provides the performance evaluation of CTES on IoT edge nodes in terms of their accuracy and dynamic assignment of weights to calculate the trust score. It also demonstrates the effectiveness of malicious node avoidance using CTES in the presence of ballot stuffing and bad-mouthing attacks. Finally, Section 6 concludes the paper with our findings and future directions.