This work introduces a modular self-supervised learning (SSL) architecture that utilises masked signal modeling (MSM) and RF domain adaptation to enhance classification performance for radar signal recognition in environments with limited RF samples and annotations. This paper (arXiv ver.) was published at the 2025 IEEE International Workshop on Machine Learning for Signal Processing (MLSP) in Istanbul, Turkey.
This paper introduces a challenging dataset (RadDet) for radar spectrum detection. RadDet comprises a large corpus of radar signals occupying a wide frequency band (500 MHz) and is simulated in different radar density environments and signal-to-noise ratio settings. This work (arXiv ver.) was published at the 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Hyderabad, India.
Precise localisation of radar signal activities is a crucial function in defence applications. This paper (arXiv ver.) introduces an end-to-end method to detect, localise and refine pulse activities of interleaved radar signals across an extended time horizon. This work was published at the 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Seoul, South Korea.
Radio signal recognition is an essential part of spectrum management and spectrum access. This paper presents a novel approach to solving radio signal classification and characterisation as a joint problem by introducing the multi-task I/Q Signal Transformer (IQST) deep neural network. This work (arXiv ver.) was published at the 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Rhodes, Greece.
Cellular network technologies like LTE and 5G present an opportunity to explore BVLOS flight operations in dense metropolitan areas and enable the large-scale deployment of aerial delivery platforms such as drones and Urban Air Mobility (UAM) vehicles. This work provides an overview of my team's efforts to evaluate the suitability of LTE as a potential technology to enable effective and safe aerial operations in Singapore.
The Joint Authority for Rulemaking on UAS (JARUS) released a comprehensive process for managing air and ground risk for UAS operations. This paper explores the air risk element, where we introduce a Bayesian framework that explicitly links encounter rate exposure, detection performance, cost, and safety. This work was published at the 2018 IEEE/AIAA Digital Avionics Systems Conference (DASC) in London, UK.
Interviews
A fun interview I did with the Royal Melbourne Institute of Technology (RMIT) University back in 2020, where I shared with university students my graudate engineering journey at Nova Systems together with a bunch of useful tips and tricks on how to make the most out of university. In this interview, you can also learn about my origin story, my work in Singapore, and how I became an engineer.
Talks
This paper introduces a challenging dataset (RadDet) for radar spectrum detection. RadDet comprises a large corpus of radar signals occupying a wide frequency band (500 MHz) and is simulated in different radar density environments and signal-to-noise ratio settings. This work (arXiv ver.) was published at the 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Hyderabad, India.
Contemporary machine learning methods have demonstrated remarkable potential in addressing domain-specific tasks that demand a level of solution accuracy and speed beyond human performance. In this talk, I provide an overview of recent innovations in advancing radio emitter detection, signal characterisation, and source localisation methods. This work was presented at the 2023 Association of Old Crows (AOC) Convention in Adelaide, Australia.
This work proposes a novel approach to solving radio signal classification and characterisation as a joint problem by introducing the multi-task I/Q Signal Transformer (IQST) deep neural network. I gave this talk at the IEEE Integrated Sensing and Communications (ISAC) workshop and our paper (arXiv ver.) was published at the 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Rhodes, Greece.
Building a robust and scalable UTM system is hard. This presentation explores the regulatory and technical challenges surrounding the development of UTM in Australia, Singapore and beyond. I presented this work with my colleague from Revolution Aerospace at the 2022 Australian Association for Uncrewed Systems (AAUS) RPAS in Australian Skies Conference in Canberra, Australia.
Tutorials
Managing large collections of images with conflicting names can become messy over time. In this step-by-step tutorial, we'll develop a simple Python program that scans a file directory and automatically renames image files using their embedded metadata. This method helps streamline file organisation and makes it easier to maintain a clean and consistent photo repository.