Collaborative Research: VINES: Track 1: NSF-RCF: LLM-guided Adaptive Networking with Edge Intelligence for Emergency Response

Firefighters and other first responders rely on reliable communication to save lives in emergencies, yet the radios and cellular networks they use today are vulnerable to disruption when disasters strike. This project aims to build an intelligent communication network for dynamic, high-stakes environments in emergencies. Using artificial intelligence (AI) to understand and predict changes in fire scenes and first responders’ operations, the network adaptively and proactively reconfigures itself to prevent disrupted communication. This project builds and evaluates a holistic large language model (LLM)-network co-design solution for the fire emergency response vertical to enable ultra-reliable and low-latency communication among humans, robots, and sensors. Three interdependent thrusts structure the work. Thrust 1 provides the semantic data foundation for LLM-enabled proactive networking by developing semantic-aware multimodal sensing and adaptive data transmission mechanisms. Thrust 2 supports proactive network configuration by designing an LLM-enabled network adaptation framework that models responder intent and environmental conditions. Thrust 3 provides the infrastructure needed for LLM-driven proactive networking through adaptive resource allocation and memory management, enabling collaborative LLM inference across edge devices. Validation spans simulations and real-world testbeds, including building an integrated network and fire-emergency simulator, as well as testing on the Aerial Experimentation and Research Platform for Advanced Wireless (AERPAW) and the 6G Test Network Finland (6GTNF) testbeds. This project advances both the science of AI-native networking and its societal impact on supporting connected responders in public safety operations. These innovations can also generalize beyond emergency response to other latency-sensitive verticals, including telesurgery, disaster recovery, industrial automation, and intelligent transportation systems. This international collaborative project brings together researchers from North Carolina State University, Pennsylvania State University with researchers from Aalto University and University of Helsinki in Finland to make emergency communication faster, smarter, and more dependable for the people who protect communities. This project will generate societal and educational benefits through data dissemination, student mentoring, curriculum development, U.S.-Finland collaboration, and partnership with local fire departments. 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.