Research Experience

My research combines signal processing, sensing systems, and machine learning to make sensor data more robust, interpretable, and useful in real-world settings.

Wireless sensing and foundation models

I study representation learning for wireless sensing, with a focus on WiFi CSI-based human activity recognition. My final-year project develops a two-stage WiFi-LLM framework that combines masked self-supervised pretraining with alignment to a frozen language model for signal-to-text understanding. The work explores frequency-aware temporal modelling, few-shot transfer, and efficient sensing representations.

Wearable and physiological sensing

I work on contactless and wearable measurement of human physiological signals. Projects in this direction include 60GHz mmWave radar for vital-sign monitoring, ECG-based identity and activity recognition, and production-oriented wearable algorithms for VO2 max estimation and running performance prediction.

Industrial sensing and non-destructive testing

My work on magnetic flux leakage sensing develops robust inspection methods for steel wire ropes. I have explored adaptive multi-scale defect detection, physics-guided unsupervised anomaly detection, and methods for defect localization and quantification under varying sampling and operating conditions.