Semantic reasoning on the edge of internet of things

Abstract

The Internet of Things (IoT) is a paradigm where physical objects are connected with each other with identifying, sensing, networking and processing capabilities over the Internet. Millions of new devices will be added into IoT network thus generating huge amount of data. How to represent, store, interconnect, search, and organize information generated by IoT devices become a challenge. Semantic technologies could play an important role by encoding meaning into data to enable a computer system to possess knowledge and reasoning. The vast amount of devices and data are also challenges. Edge Computing reduces both network latency and resource consumptions by deploying services and distributing computing tasks from the core network to the edge.

We recognize four challenges from IoT systems. First the centralized server may generate long latency because of physical distances. Second concern is that the resource-constrained IoT devices have limited computing ability in processing heavy tasks. Third, the data generated by heterogeneous devices can hardly be understood and utilized by other devices or systems. Our research focuses on these challenges and provide a solution based on Edge computing and semantic technologies.

We utilize Edge computing and semantic reasoning into IoT. Edge computing distributes tasks to the reasoning devices, which we call the Edge nodes. They are close to the terminal devices and provide services. The newly added resources could balance the workload of the systems and improve the computing capability. We annotate meaning into the data with Resource Description Framework thus providing an approach for heterogeneous machines to understand and utilize the data. We use semantic reasoning as a general purpose intelligent processing method.

The thesis work focuses on studying semantic reasoning performance in IoT system with Edge computing paradigm. We develop an Edge based IoT system with semantic technologies. The system deploys semantic reasoning services on Edge nodes. Based on IoT system, we design five experiments to evaluate the performance of the integrated IoT system. We demonstrate how could the Edge computing paradigm facilitate IoT in terms of data transforming, semantic reasoning and service experience. We analyze how to improve the performance by properly distributing the task for Cloud and Edge nodes. The thesis work result shows that the Edge computing could improve the performance of the semantic reasoning in IoT.