Implementing Multi Agent Systems (MAS)-based trust and reputation in smart IoT environments : A thesis submitted in partial fulfilment of the requirements for the Degree of Doctor of Philosophy at Lincoln University

Abstract

The Internet of Things (IoT) provides advanced services by interconnecting a huge number of heterogeneous smart things (virtual or physical devices) through existing interoperable information and communication technologies. As IoT devices become more intelligent, they will have the ability to communicate and cooperate with each other. In doing so, enormous amount of sensitive data will flow within the network such as a credit card information, medical data, factory details, pictures and videos. With sensitive data flowing through the network, privacy becomes one of most important issues facing IoT. Studies of data sensitivity and privacy indicate the importance of evaluating the trustworthiness of IoT participants to maximize the satisfaction and the performance of the IoT applications. It is also important to maintain successful collaboration between the devices deployed in the network and ensure all devices operate in a trustworthy manner. This research aims to determine: How to select the best service provider in an IoT environment based on the trustworthiness and the reputation of the service provider? To achieve this, we proposed an IoT agent-based decentralized trust and reputation model IoT-CADM (Comprehensive Agent-based Decision-making Model for IoT) to select the best service providers for a particular service based on multi-context quality of services. IoT-CADM as a novel trust and reputation model, is developed for the smart multi-agent IoT environment to gather information from entities and score them using a new trust and reputation scoring mechanism. IoT-CADM aims to ensure that the service consumers are serviced by the best service providers in the IoT environment which in turn maximizes the service consumers’ satisfaction, which lead the IoT entities to operate and make-decisions on behalf of its owner in a trustworthy manner. To evaluate the performance of the proposed model against some other well-known models like ReGreT, SIoT, and R-D-C, we implemented a scenario based on the SIPOC Supply Chain approach developed using an agent development framework called JADE. This research used the TOPSIS approach to compare and rank the performance of these models based on different parameters that have been chosen carefully for fair comparison. The TOPSIS result confirmed that the proposed IoT-CADM has the highest performance. In addition, the model can be tuned to its parameters weight to adapt to varying scenarios in honest and dishonest agents’ environments.