Internet of Things data contextualisation for scalable information processing, security, and privacy

The Internet of Things (IoT) interconnects billions of sensors and other devices (i.e., things) via the internet, enabling novel services and products that are becoming increasingly important for industry, government, education and society in general. It is estimated that by 2025, the number of IoT devices will exceed 50 billion, which is seven times the estimated human population at that time. With such a tremendous increase in the number of IoT devices, the data they generate is also increasing exponentially and needs to be analysed and secured more efficiently. This gives rise to what is appearing to be the most significant challenge for the IoT: Novel, scalable solutions are required to analyse and secure the extraordinary amount of data generated by tens of billions of IoT devices. Currently, no solutions exist in the literature that provide scalable and secure IoT scale data processing.

In this thesis, a novel scalable approach is proposed for processing and securing IoT scale data, which we refer to as contextualisation. The contextualisation solution aims to exclude irrelevant IoT data from processing and address data analysis and security considerations via the use of contextual information. More specifically, contextualisation can effectively reduce the volume, velocity and variety of data that needs to be processed and secured in IoT applications. This contextualisation-based data reduction can subsequently provide IoT applications with the scalability needed for IoT scale knowledge extraction and information security. IoT scale applications, such as smart parking or smart healthcare systems, can benefit from the proposed method, which  improves the scalability of data processing as well as the security and privacy of data.  

The main contributions of this thesis are: 1) An introduction to context and contextualisation for IoT applications; 2) a contextualisation methodology for IoT-based applications that is modelled around observation, orientation, decision and action loops; 3) a collection of contextualisation techniques and a corresponding software platform for IoT data processing (referred to as contextualisation-as-a-service or ConTaaS) that enables highly scalable data analysis, security and privacy solutions; and 4) an evaluation of ConTaaS in several IoT applications to demonstrate that our contextualisation techniques permit data analysis, security and privacy solutions to remain linear, even in situations where the number of IoT data points increases exponentially.