RUI: NeTS: Data-Centric AI for Proactive Telecom Network Operations

Telecommunication networks have become an essential part of everyday life, supporting everything from commerce and education to emergency services and daily connectivity. When network problems occur, users may experience slow connections, interruptions, or complete service outages that affect both personal and professional activities. Many of these problems are difficult to predict because the data collected from telecommunication networks are often incomplete, imbalanced, or unavailable due to privacy concerns. This project develops new methods for improving the quality of network data so that artificial intelligence systems can recognize warning signs of service degradation before customers are affected. The long-term goal is to support more reliable telecommunication networks through better data rather than simply building larger prediction models. The intellectual merit of this research lies in advancing data-centric artificial intelligence for proactive telecommunication network operations. The research is organized into three coordinated thrusts. The first thrust develops methods for generating realistic synthetic network data that preserve the statistical, temporal, and operational characteristics of real network measurements. The second thrust investigates complementary data-centric artificial intelligence techniques such as data augmentation and domain adaptation to improve model robustness across heterogeneous network environments. The third thrust integrates data quality monitoring, adaptive retraining, and closed-loop feedback into network management workflows to create self-improving predictive systems. The broader impacts of this project include improving the reliability and resilience of telecommunication infrastructure while expanding educational opportunities in artificial intelligence and networking, especially at a primarily undergraduate institution. Undergraduate students will participate in research experiences involving data-centric artificial intelligence, network analytics, and data engineering, gaining hands on experience with emerging technologies. Educational modules on data-centric artificial intelligence and network analytics will be incorporated into undergraduate computer science courses and research activities. Code, data, and research results will be shared through open-source repositories and scholarly publications to encourage reuse by researchers, educators, and students. By improving methods for learning from limited and imperfect network data, the project will contribute to more dependable telecommunication services that benefit communities, businesses, and public institutions. 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.