Can we make FCC Experts out of LLMs?
Atul Bansal, Veronica Muriga, Jason Li, Lucy Duan, Swarun Kumar · 2025
This paper investigates whether Large Language Models (LLMs) can provide wireless expertise, particularly in the context of spectrum regulations. For any wireless system designer, defining an effective set of wireless parameters, such as the frequency operating band and total radiated power, is one of the very first steps in designing any wireless system. The current approach to do this relies on human FCC experts, who have to consult large volumes of techno-legal documentation on government regulations, such as the FCC rulebook and other industry standards. To ensure the exhaustive coverage of various use cases of wireless systems, these documentations tend to be quite long and very hard to parse through, even for an FCC expert. Given the success of LLMs in providing expert and accurate responses to varied other domains, it is natural to wonder if they could also assist wireless system designers in ensuring their compliance to these techno-legal rules and regulations. In this paper, we present WiLL, a system built on a state-of-the-art LLM, designed to be an assistive tool for wireless system designers to ensure their wireless system complies with regulations. Specifically, we focus on ensuring compliance with FCC regulations for various wireless systems that are designed for technologies such as WiFi, Bluetooth, LoRaWAN and Ultra Wide Band (UWB). We observe that WiLL achieves an accuracy of 78.57%, as compared to an average 51.77% accuracy of off-the-shelf state-of-the-art Large Language Models.