A WEB OF THINGS ARCHITECTURE FOCUSING ON TRUST

Abstract

Trust has a significant impact on our personal and business relations. It is noteworthy that in the era of things based services, people trust in various companies/organizations for personal and business purposes. They receive their services and consequently, ‘data including personal and business’ is accessible to these service providers which sometimes bring undesirable results. An example of things-based services is remote health monitoring system. In this case also, the sensitive data is accessible to known and occasionally unknown entities. Since health-related services are life-critical, therefore, the quality of such services in term of trust should not be compromised, and the context of the services should be the same as required. Hence, it is fairly said that maintaining trust in the environment of this dominating technology is a tedious task. Web of Things (WoT) simplifies the development of things based services by providing an application layer. It is identified that the existing work focuses on building blocks of WoT architecture. However, the aspect of trust has got less attention in the area of WoT. Consequently, this research proposes a WoT architecture focusing on Trust. It has five layers which are not like those layers defined in the OSI (Open System Interconnection) model. Instead these layers facilitate i) in making things part of the cyber world with the help of the Web technologies, ii) monitoring and evaluating trust at a central trust management layer, iii) receiving and providing recommendations through social relations, iv) and help in discovering and selecting the required services by managing their trustworthiness at user level. For trust monitoring and management, a methodology is introduced in which entities are evaluated on the basis of multi-criteria such as context and Quality of Service (QoS). For this purpose, several attributes related to context and QoS are identified which can affect the i trustworthiness of the entities. Direct and indirect experiences are the sources of evidence of trustworthiness. Moreover, third-party recommendations are also considered. To complete these tasks, algorithms are also designed. The research also proposes a methodology for the identification of service-related attacks which are carried out to malign the trust level of the entities present in the system. To validate the proposed architecture, a prototype is implemented which is for the real-time monitoring of a poultry farm in Pakistan. It demonstrates an instance of the architecture presenting the five layers of the architecture. The test results show the effectiveness of the architecture in maintaining the healthy environment of the poultry farm. In addition, to demonstrate the effectiveness of proposed algorithms, two specially designed environments are implemented where two real datasets are used for experimental purpose. In the first group of experiments, the algorithms for service discovery, selection, and ranking are tested along with query augmentation. In the second group of experiments, a classification approach along with the proposed algorithms is used to identify various types of attacks on the basis of feedbacks. During two kinds of experiments, the proposed algorithms demonstrate their effectiveness in discovering and ranking the services. In addition, identification of the attacks is also demonstrated. The results show that the system outperforms the existing related approaches in terms of the comparative analysis on the basis of performance metrics (such as precision, sensitivity, and accuracy) and parametric metrics (such as availability, interoperability, context-awareness, scalability, and QoS).