LLM-Driven Traffic–Communication–Energy Coordination to Enhance Demand Response of IoV Communication Infrastructures

Demand response (DR) can improve power system flexibility and support low-carbon operation, yet the potential of Internet of Vehicles (IoV) communication infrastructures remains limited by the coupling between traffic distribution and the communication network load. Existing traffic-only or communication-only approaches do not exploit this cross-domain flexibility. This study proposes a sustainable traffic–communication–energy coordinated DR framework that uses dynamic route compensation to reshape communication loads. A Stackelberg mechanism coordinates the IoV operator and vehicle users while a large language model (LLM) generates and corrects adaptive compensation candidates under heterogeneous operating conditions. Compared with the no-guidance and static-compensation baselines, the proposed dynamic strategy redistributes traffic and communication loads, releases communication-side DR capacity, and improves both operator and user performance. Compared with the non-LLM decision baseline, LLM assistance increases the load regulation rate from 21.5% to 36.2%, raises cost savings from 16.8% to 28.5%, and increases the user response rate from 72.4% to 88.7%. The framework provides a sustainable pathway for using existing IoV communication infrastructure as a flexible demand-side resource.