Collaborative research: nets: physics-enhanced AI for spectrum awareness and radiomap-assisted network management

The National Science Foundation Division of Information and Intelligent Systems awarded Georgia State University Research Foundation Inc. $350,000 on October 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) for collaborative research in physics-enhanced artificial intelligence for spectrum awareness and radiomap-assisted network management.

The project develops machine learning methods that integrate wireless propagation principles with advanced machine learning techniques to construct accurate radio maps from sparse measurements. Research focuses on three areas: physics-enhanced learning methods for radio map estimation incorporating domain knowledge into feature extraction and neural network architectures; integration of radio maps into wireless network management for planning, optimization, and diagnosis; and advanced learning methods for joint radio map estimation and network management. The work produces new algorithms, models, evaluation methods, datasets, and software tools for managing future wireless communication systems.

Performance occurs in Atlanta, Georgia with a period of performance from October 1, 2026, through September 30, 2029. The funding supports more effective use of spectrum resources, improves wireless connectivity for emerging applications, and includes student training, curriculum development, and dissemination of research outcomes.

Abstract

Next-generation wireless networks will support rapidly growing numbers of connected devices and users that share wireless spectrum in complex and dynamic environments. However, accurately characterizing radio environments across various spatial locations, terrain conditions, and network activities remains challenging. This project develops new machine learning methods for estimating and utilizing radio maps to support network planning, infrastructure deployment, system optimization, and network fault diagnosis. The project promotes more effective use of scarce spectrum resources, improves wireless connectivity for emerging applications, and prepares the future workforce through student training, curriculum development, and dissemination of research outcomes. This project develops a physics-enhanced artificial intelligence (AI) framework for radio map estimation and radio map-assisted wireless network management. The research integrates wireless propagation principles with advanced machine learning techniques to construct accurate radio maps from sparse measurements. The project investigates three research directions. First, physics-enhanced learning methods are developed for radio map estimation by incorporating domain knowledge into feature extraction, learning objectives, and neural network architectures. Second, radio maps are incorporated into wireless network management as inputs, performance metrics, and environmental descriptors to support network planning, optimization and diagnosis. Third, advanced learning methods are designed for joint radio map estimation and network management. The project advances both wireless networking and AI technologies through the development of new algorithms, models, evaluation methods, datasets, and software tools for effective and reliable management of future wireless communication systems. 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.