An Extensible Framework for Intelligent Ubiquitous Sensing within Precision Agriculture

Humanity resides in a world that has never been as interconnected, with billions of sensing devices presently deployed in various domains. Governments, industries and researchers alike are leveraging this data to solve real-world problems and advance human knowledge. Decades of research founded on long-term observations from such sensors have foretold the consequences of climate change, with current data highlighting trends in climate instability of notable concern. These are now present on the global stage, endangering food security at an unprecedented scale. As the human population is projected to increase by 25% over the next 30 years to 10 billion (Hickey et al., 2019), additional food demand will strain an already tense, fragile food supply chain to a point where the large-scale loss of human life and environmental damage will become even more common than it is today. These challenges have motivated the conception of Sustainable Agriculture, which seeks to innovate the agricultural industry, allowing more food to be created with fewer resources. Within sustainable agriculture, the term 'precision agriculture' emerged as an approach to use advances in Information Technology to achieve this goal. This thesis considered existing best-of-breed middlewares, exposing some of their frailties when considering their application to the precision agriculture sector. These limitations motivated the conception of Indra, a novel middleware framework that advances state-of-the-art by offering a highly extensible, usable, modular microservice-oriented framework capable of interfacing seamlessly with a diverse range of ubiquitous data. Indra was extensively designed, iteratively implemented and evolved into an application deployed in the wild within CONSUS over a sustained and prolonged period. Indra offered a microservices suite that enabled and accelerated research within CONSUS, enhancing the typical functional offerings of leading-edge middlewares and evidencing appropriateness for the particular operational context that is CONSUS. Indra was extensively evaluated and is proven to be highly usable, extensible and high performing and represents a mature, well-established middleware within CONSUS.