Nwamaka Okafor

Bio: Nwamaka Okafor is a final year PhD student at University College Dublin. She works with the Smartbog Observation Group on Irish peatlands and advised by Dr. Declan Delaney. Her work is supported by Schlumberger Faculty for the Future Fellowship (FFTF) Programme, Environmental Protection Agency (EPA) and TETFUND and focuses on the applications of IoT and machine learning in environmental monitoring. Key to her research is a means to process and efficiently harness inherently noisy data collected from IoT deployments, web/crowdsourced data and remote sensing of ecologically significant sites. Nwamaka holds an MSc in Computer Forensics and Cyber Security (Distinction) from the University of Greenwich, London. During the Covid-19 pandemic, Nwamaka collaborated with other researchers from the Universities of Oxford and Cambridge on a British Telecom’s sponsored project to explore the impact of containment measures on the rate of spread of Covid-19 in selected countries. The project won the overall best project with Real World Impact Award among 44 other projects from across the world. Nwamaka is a recipient of the 2020 Google Women Techmakers award (now Generation Google Scholarship), ACM SRC award and recently listed as one of the eighty women advancing AI in Africa and around the world by African Shapers. Her work has been published in leading journals and conferences including ICML, IEEE-Sensors. Upon completion of her PhD, Nwamaka will join Argonne National Laboratory, IL, USA as a Postdoctoral appointee and would focus on developing novel methods to analyse supercomputer logs, including Aurora and Polaris.

Talk Title: Efficient data processing pipeline for IoT sensor devices in environmental monitoring networks.

Talk Abstract: Low-cost IoT sensors (LCS) have the capacity to provide high resolution spatio-temporal dataset of key variables in environmental monitoring networks. The use of LCS in environmental monitoring, however, continues to raise important questions, especially pertaining to the reliability, field performance and data quality of the sensors. IoT technologies are at the early stage of development in many application areas and as such they are still challenged by a number of issues including data handling and framework. There is currently no universal framework and/or formalized architecture in existence for the application of IoT sensor devices in environmental monitoring and no complete tool has been developed for data handling and analysis.

Developing effective data processing pipeline for IoT sensor data is essential for the adoption and application of LCS in environmental monitoring. In this talk, I will discuss our research efforts aimed at improving field performance and data quality of LCS devices in environmental monitoring networks, through the development of reliable data processing and error correction techniques such as automatic sensor calibration, data fusion, data imputation, outlier detection and effective feature selection.

I will discuss how the capabilities of AI/ML can be leveraged to support the effective application of IoT in ecological sensing. In particular, how these technologies can: 1.) support the operations of LCS in data acquisition models to capture large scale and longitudinal observations, 2.) support the development of specific data handling and Data Quality Improvement (DQI) techniques for LCS, and 3.) serve as effective strategies to support on-site sensor calibration.