Enabling Distributed Intelligence in the Internet of Things with IOTA Tangle and Mobile Agents

Widespread adoption of smart Internet of Things (IoT) devices is accelerating research for new techniques to make IoT applications scalable, energy-efficient, and capable of working in mission-critical use cases. IoT devices are characterised by limited resources, such as power consumption and memory storage, which is a fundamental challenge of IoT applications. Distributed Intelligence (DI) is an area of research within the field of IoT and is seen as a practical route towards the decentralisation of IoT architectures. Enabling DI is a challenging task in IoT because it needs to ensure scalability, and energy-efficiency due to resource constraints. These challenges requires a new solutions to be investigated. There is a wide body of literature about DI in the IoT. These approaches are dealing with a particular challenge and identified as an effective and efficient in achieving that challenge. However, there are few attempts to enable DI in IoT. The aim of this thesis is to develop a scalable, and energy efficient solution for enabling DI in the IoT. This aim is achieved through the development of high-level and low-level intelligence techniques to support DI. This thesis contributes towards the design of a new framework that ensures scalability and energy-efficiency of IoT applications. The developed hybrid Mobile-Agent Distributed Intelligence Tangle-Based framework (MADIT) represents the novel contribution of the work. The aim of which is to offer low-level and high-level intelligence for IoT applications. The low-level intelligence along with IOTA Tangle-based intelligence form the distributed intelligence in the IoT domain. The low-level intelligence is achieved through the use of multi-mobile agents to collect transactions data and high-level intelligence is achieved through the use of Tangle-based architecture. The framework evaluates a Proof-of-Work computation offloading mechanism that performs costly computations on behalf of constrained IoT devices for efficacy with regard to energy efficiency and transaction throughput. The Proof-of-Work offloading computation mechanism improves efficiency and speed of processing, while saving energy consumption. In addition, this thesis proposes a new energy efficient Graph-based Static Mutli-Mobile Agent Itinerary Planning approach (GSMIP). The GSMIP applies Directed Acyclic Graph (DAG) related techniques and divides sensor nodes into different groups based on the routes defined by mobile agents itineraries. Mobile agents follow the predefined routes and only collect data from the groups they are responsible for. The proposed solution can be easily generalised to different application domains, and is less complex than many other existing approaches. The simplicity of the solutions neither demands great computational efforts nor large amounts of energy consumption. The experimental findings demonstrate the effectiveness and superiority of the proposed approach compared to the existing approaches in terms of energy consumption and task delay (time).