NSF REU site: asset: advanced secured sensor enabling technologies

The National Science Foundation Division of Computing and Communication Foundations awarded Florida International University $464,999 on October 1, 2026, to operate an NSF Research Experiences for Undergraduates (REU) Site focused on trustworthy artificial intelligence for advanced and secure sensing at the edge.

The ASSET-AI REU Site engages ten undergraduate students annually in collaborative research and development with FIU faculty mentors and graduate students on projects integrating sensing networks, AI, edge computing, cybersecurity, and modern language models. The research addresses how AI-enabled sensing systems can make accurate, timely decisions on resource-constrained edge devices while maintaining auditability and how compact models can support on-device reasoning under strict latency, energy, and memory constraints. The program also investigates privacy-aware and federated learning approaches for distributed sensing systems. The award supports workforce development in AI and cybersecurity research, which the program identifies as critical national priority areas.

Performance occurs in Miami, Florida, with a period of performance running through September 30, 2029. Florida International University is a public research university and Hispanic-Serving Institution that operates as a prime recipient across diverse federal portfolios, with substantial grant activity concentrated in the National Science Foundation, National Institutes of Health, and NOAA.

The funding supports the Computer and Information Science and Engineering program (CFDA 47.070), which funds investigator-initiated research and education in computing, communications, and information science and engineering.

Abstract

The research experiences for undergraduates site at florida international university, known as advanced secured sensor enabling technologies ? AI (asset-AI), centers on trustworthy AI for advanced and secure sensing at the edge. Advanced sensing systems now support health care, transport, robotics, infrastructure, disaster response, and environmental monitoring. These systems increasingly rely on artificial intelligence (AI). AI can help turn large, fast, and mixed data streams into earlier warnings, better detection of unusual events, and faster decisions. IT can also help systems adapt as conditions change. But this promise is hard to realize in practice. Many systems must run on small, distributed devices with tight limits on power, memory, computing, and bandwidth. They must also protect privacy and keep working when data are noisy, missing, changing, or under attack. This REU site with its research and education program addresses a timely national need in AI and cybersecurity. IT contributes to building the future workforce for research and development in these critical priority areas. The asset-AI REU site engages a cohort of ten undergraduate students each year with research teams at FIU in research and development of advanced information technologies. The undergraduate students collaborate with faculty mentors and graduate students on projects that integrate sensing networks and AI with edge computing, cybersecurity, and modern language models to enable real-time on-device reasoning under strict resource constraints. The research supported by the asset-AI site asks how AI-enabled sensing systems can make accurate, timely decisions on edge devices with limited resources and still be checked and audited. IT studies three questions. First, how can compact models support on-device reasoning under strict limits on latency, energy, and memory? Second, how can privacy-aware and federated learning remain effective when data are distributed and non-IID? Third, how can sensing pipelines stay robust and traceable under spoofing, poisoning, and distribution drift? The research draws upon advances in AI, adaptive inference, machine learning, networking, and privacy. Research topics include trust-aware validation of streaming IOT data; on-device small language models; explainable RF device fingerprinting; privacy-aware learning for connected transportation; context-aware mobile and wearable sensing; adaptive wireless networking; and resilient coordination in sensor networks and drone swarms. Empirical testing and evaluation methods include simulation; hardware-in-the-loop testing; realistic workloads; stress tests; ablation studies; and repeatable benchmark comparisons. Expected outcomes include new algorithms; system designs; benchmarks; and open research tools for trustworthy AI-enabled sensing. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.