NSF NAIRR Pilot Award

FedAlign-HAR: Federated Security Evaluation of Sensor-Text Alignment in Wearable Human Activity Recognition

CARBONDALE, Ill. — June 14, 2026

Dr. Abdur Rahman Bin Shahid, Assistant Professor in the School of Computing and director of the SHIELD Lab, has received a computing allocation through the U.S. National Science Foundation (NSF)’s National Artificial Intelligence Research Resource (NAIRR) Pilot program.

Artificial intelligence is quickly moving from screens and servers into the devices people carry and wear every day. Smartphones, smartwatches, and wearable sensors are beginning to do more than record steps, heart rate, or motion. When combined with language-based AI, these devices could one day help explain what a person is doing, recognize changes in health or stress, support safer workplaces, assist older adults, and provide more personalized help in daily life.

Through the awarded program, Dr. Shahid and his PhD students will study how to make this emerging generation of wearable AI more secure, private, and trustworthy. The project will examine security risks in federated wearable AI systems, where models can learn from many users and devices without collecting raw personal sensor data in one central location. While this approach can help protect privacy, it also creates new risks if malicious devices or users attempt to corrupt the learning process.

The project, titled “FedAlign-HAR: Federated Security Evaluation of Sensor-Text Alignment in Wearable Human Activity Recognition,” will investigate whether malicious participants can disrupt the connection between sensor signals and natural-language activity descriptions. Such attacks could undermine the reliability of wearable AI systems used in healthcare, workplace safety, elder care, and other real-world applications. If left unaddressed, these vulnerabilities could lead to incorrect decisions, reduced trust in AI-assisted technologies, and broader risks when wearable systems are used to support health, safety, and human well-being.

The NAIRR Pilot allocation will support the development of an initial federated wearable AI research pipeline, early attack and defense studies, and future research on secure multimodal wearable agentic AI. The project has been awarded 2,000 GPU-hours on Purdue Anvil AI supercomputer through the NAIRR Start-Up program.

Media Contact: School of Computing Communications • socinfo@siu.edu