IoT sensors are useful due to their ability to accurately measure environmental parameters on the ground level. However, deploying large-scale IoT networks in forest environments are quite challenging as sensors can only monitor a small geographical area. Therefore, the only way to scale up is to deploy a large number of sensor nodes, which is difficult due to hardware costs, deployment costs, difficulty in providing energy and developing network communication in harsh environments with high humidity and a lot of physical obstacles and so on. Another alternative mechanism is to use drones to observe a forest area. However, drones have limitations where they only see first from the top and cannot directly sense what is happening deep inside the jungles at the soil level. In this project, our objective is to combine both IoT and drone imaging data. Drones bring the scalability aspect, and IoT technologies bring the accuracy aspect. To combine, we will deploy both IoT and drones in a test forest environment to train a model capable of using drone images to predict the outcomes of IoT sensors and, subsequently, the forest health index. 

Our approach is to develop a Forest Observatory and develop data-driven predictive analytics to predict poaching incidents. Forest Observatory is a Linked Datastore that integrates heterogeneous data. We consider Forest Observatory as an extension of Urban Observatories which aim at gathering real-time urban data across cities. Collecting data in forests is much more challenging than in cities due to the lack of infrastructure. However, while we expect to deploy an Internet of Things (IoT) infrastructure to enable poaching monitoring, we should utilise already collected data sets to develop poaching predictive models. For example, DGFC has data sets collected by researchers for wildlife species monitoring over the last decade, such as animal collar data, camera traps, satellite imagery, LiDAR and environmental data, with each data set generated using different time frames, durations, geographic areas. To develop a Forest Observatory, we aim to integrate various data sets collected by the bioscience researchers at DGFC into a unified linked data store. We use semantic data integration techniques while conforming to the data modelling standards (e.g., ontologies) and needs of bioscience research --towards developing a model and novel tools that are exportable to other world areas where poaching is a threat to wildlife conservation.

This primary project objective is to make the Linked Data more accessible and allow the non-technical end-user (here we mean by end-user the Bioscience researchers) to perform their job more efficiently. The end-user should accomplish their regular job faster by using our friendly user interface. Non-technical users will not need to have any experience using SPARQL or any other query language to retrieve the data. Besides, expert users will perform their job easier in less time. This project composes three main objectives: (i) Review of existing techniques on how to make Linked Data accessible (ii) A User-Interface to help the end-users to access the Linked Data easily (iii) An Intelligent User-Interface to access the Linked Data that will involve conversational AI to enhance users' experience.

Sensor networks are becoming more prevalent as IoT devices monitor an increasing amount of information sources.  In most cases, the device data can be processed at source or relayed back to base using high bandwidth, low latency techniques.  This could be via Wi-Fi or 4/5G cellular network communication links.  For instances where these techniques are not available or consume too much power, other technologies, such as LoRaWAN or SigFOX, are commercially available. However, situations still call for low power long-range communications that these techniques cannot fulfil. The hypothesis of this project is to prove that low power, low bandwidth digital signal can achieve a reliable long-distance IoT sensor communication medium in harsh RF environments or over ranges that are not suitable for traditional IoT sensor networks. 

We conducted two full-day workshops to explore and identify research challenges that could potentially be addressed using the Internet of Things (IoT) technologies. We identified two significant areas to focus on during this workshop: (1) Sensing Infrastructure and (2) Data Science. We also discussed citizen engagement research, where we could work with local schools and universities to share our technical expertise with the local community. In the long term, such activities will help local communities to develop technologies to solve their problems.

This project aims to study how BLE beacons based technology can be used to track poachers. Tracking technologies are currently limited to satellites or cellular towers; few viable alternatives exist for environments that do not permit access to these signals. This project implements and extensively tests the use of Bluetooth low energy (BLE) to track vehicles.  The results offer insights into how effective Bluetooth beacons are in a detection situation --for where the beacon and receiver are in range for a short period and how different obstructions will affect the range and strength of the signal.

Poaching has been an ever-present threat to many species worldwide for many decades—traditional anti-poaching initiatives target catching poachers. However, the challenge is far more complicated than catching individual poachers. Poaching is an industry that requires thorough investigation. There are many stakeholders, directly and indirectly, involved in poaching activities (e.g., some local restaurants illegally serving meat to tourists). Therefore, we need a unified understanding of all stakeholders to stop or severely decapitate the poaching industry. Thus, we developed an SMS (short message service) base low-cost tracking system (SMS-TRACCER) to track poachers.

This study presents a novel approach and a proof-of-concept for a hardware-efficient automated license plate recognition system for a constrained environment with limited resources. The proposed solution is purely implemented for low resource edge devices and performed well for extreme illumination changes such as day and nighttime. The generalisability of the proposed models have been achieved by using a novel set of neural networks for different hardware configurations based on the computational capabilities and low cost. The accuracy, energy efficiency, communication and computational latency of the proposed models are validated using different license plate datasets in the daytime and nighttime and real-time. The obtained results have shown competitive performance to the state-of-the-art server-grade hardware solutions.