This project will investigate the design and development of an end-to-end, distributed, Internet of Things (IoT) infrastructure to support sustainable wildlife conservation. This project would enable the lead-supervisor to start a larger collaboration between academics in the Schools of Computer Science and Informatics (COMSC) and Biosciences (BIOSI). The PhD student will research on using IoT technologies to facilitate and augment wildlife conservation activities in-the-wild. The primary testbed for this project will be Cardiff University’s Danau Girang Field Centre (DGFC) in the Lower Kinabatangan Wildlife Sanctuary (LKWS; Sabah, Malaysia).
One main challenge in performing data driven research in-the-wild is deployment of efficient and effective sensing infrastructure to collect and analyse data in a sustainable manner. Traditional data analytics approaches are designed to send all the data to the cloud for decision making; however, areas such as LKWS and other protected areas in Sabah, network communication is very poor or not existent. Further, continuous data communication consumes a significant amount of energy which most of the edge nodes cannot afford. Therefore, traditional approaches are ineffective and unsustainable.
Traditionally, self-organizing (dynamic orchestration) algorithms are designed to work assisted by a central node (e.g. cloud). In this project, (novelty) our aim is to develop algorithms that can run on resource constrained edge nodes without the support of a central node.
In this project, our focus will be on (exact focus depends on candidate’s skill-set): • Architecture: Design and development of self-organizing and reconfigurable IoT infrastructure that integrates resources from multiple layers (sensing, edge/fog, cloud). Low-cost and off-the-shelf components will be used to develop the prototype. • Algorithms: o Design and development of distributed algorithms that can dynamically orchestrate IoT resources on the edge to satisfy a given sensing requirement without continuous connectivity to the cloud. o These algorithms will also determine how to distribute data analytics workloads (among heterogeneous edge nodes) in an optimal way to satisfy given requirements and real-world constraints.