Projects and publications.
Can a low-power tag detect a nearby FMCW radar and work out its configuration with no prior knowledge of it?
A passive tag uses delay lines to turn GHz radar chirps into kHz beat tones, detecting radars with 99% accuracy up to 15 m and estimating their chirp slope and direction.
Can a single mmWave radar find a child left in a car, including in a rear-facing seat it cannot see directly?
Passive reflectors cover the radar’s blind spots, and a multi-task model detects child presence and estimates breathing rate, with over 97% detection accuracy and under 6 breaths per minute of error.
How can low-power backscatter tags be located accurately in multipath-rich industrial environments?
A machine learning framework performs time-difference-of-arrival ranging from pairs of channel impulse responses, halving positioning error compared with conventional methods in real industrial tests, and a calibration step adapts it to new sites with little data.
Full list on Google Scholar. * equal contribution.
Physics Unrolled Neural Operator for Wireless Field Modeling
NeurIPS 2026
Generalizable Machine Learning-based Bistatic Ranging for Backscatter Tag Localization
SPIE Defense + Security 2026 Paper
“My Credentials, Please”: Detecting Filesystem Capability Laundering in MCP Servers at Scale
arXiv preprint arXiv:2610.22301
ChirpEye: Passive Sensing and Profiling of FMCW Radars with a Resource-constrained Tag
Atlas: Ensuring Accuracy for Privacy-Preserving Federated IoT Applications
BSENSE: In-vehicle Child Detection and Vital Sign Monitoring with a Single mmWave Radar and Synthetic Reflectors
Extended-Range Two-way Radar Backscatter Communication with Low-Power IoT Tags
Personalized State Anxiety Detection: An Empirical Study with Linguistic Biomarkers and Machine Learning Pipeline
SRDA: Mobile Sensing based Fluid Overload Detection for End Stage Kidney Disease Patients using Sensor Relation Dual Autoencoder
GNNs in IoT: A Survey
PFed-LDP: A Personalized Federated Differential Privacy Framework for IoT Sensing
Mobile Sensing in the COVID-19 Era
Dynamic Network Anomaly Modeling of Cell-Phone Call Detail Records for Infectious Disease Surveillance
Graph Auto-Encoder via Neighborhood Wasserstein Reconstruction
Using Ubiquitous Mobile Sensing and Temporal Sensor-Relation GNN to Predict Fluid Intake of End Stage Kidney Patients