Abu Dhabi (UrduPoint / Pakistan Point News / WAM – 06 April 2026) Khalifa University of Science and Technology technologyThe company’s Digital Future Institute announced the launch of RF-GPT, a first-of-its-kind RF AI language model capable of interpreting radio signals, overcoming a major limitation in communications AI, where language models typically only work on text and structured network data.
RF-GPT demonstrated consistent performance improvements on RF spectrogram tasks, outperforming existing baseline models by up to 75.4 percent, reflecting a strong understanding of RF. The model also correctly counted the number of signals in the spectrogram approximately 98 percent of the time, an ability that is rarely achieved by general-purpose AI models.
RF-GPT works by converting radio signals into visual patterns that AI systems can interpret. Once converted, these systems analyze patterns and respond to queries about activity within the wireless spectrum using plain language. The base model supports directly United Arab Emirates The National Artificial Intelligence Strategy lays the foundation for more autonomous and intelligent wireless networks.
The project was developed by researchers from Khalifa University, led by Professor Marwan Dabbah, Senior Director of the Digital Future Institute, with contributions from Postdoctoral fellows Hang Zhou and Yu Tian, research scientists Dr. Lina Pariah from Khalifa University, Dr. Samson Lassolcy from the University of Lorraine, and Dr. Zhongwen Huang, along with Ph.D. student Buhao Wang from Zhejiang University.
Professor Ahmed Al-Durra, Associate Dean for Research at Khalifa University, said: “The launch of RF-GPT reflects Khalifa University’s long-term focus on innovation in digital infrastructure to enhance the integration of artificial intelligence across strategic sectors and next-generation communication research, in line with national priorities. Initiatives such as this model contribute to enhancing United Arab EmiratesThe rapidly growing human capital and research capabilities needed to support the country’s evolving digital ecosystem.
Professor Marwan Dabbah said: “RF-GPT represents a turning point in spectral intelligence from Isolated and task-specific RF pipelines toward a unified RF language interface. We gave a language model its first glimpse of the electromagnetic spectrum, and the sight was truly remarkable. By making the physical layer queryable in natural language, we open the door to native AI radio systems, where RF awareness can directly support network optimization and policy decisions, a critical step toward native AI 6G networks.
RF-GPT is trained using approximately 625,000 computer-generated radio signal models and is designed for telecom operators, network engineering teams and spectrum authorities, to support increasingly complex wireless environments. The model demonstrated strong performance across tasks such as identifying signal types, detecting interfering transmissions, recognizing wireless standards, estimating device usage in Wi-Fi networks, and data mining. from 5G signals.