Digital Twin-Enhanced Framework for TCP Throughput Map Construction in Dynamic IoV

With the integration of 6G networks and Internet of Vehicles (IoV), constructing reliable Transmission Control Proto-col (TCP) throughput maps is crucial for intelligent transportation systems. These maps provide critical insights into network performance, enabling efficient data transmission planning and informed decision-making. However, the dynamic nature of IoV systems, along with resource constraints poses challenges for map construction. To address these challenges, we develop a novel framework for constructing TCP throughput maps. The framework uses base station sensing to collect vehicle positions and communication resources to gather TCP throughput data. A Digital Twin integrates those data over the map in real-time and predicts TCP throughput at every location with prediction uncertainty. A key aspect of this framework is the selection of vehicles to measure TCP throughput, minimizing prediction uncertainty and measurement overhead. We present a method that models the TCP throughput map using Gaussian Process Regression, which accounts for uncertainty. And to take into account the dynamic nature of IoV environments, we also propose the Fixed-Observation Rolling Optimization Algorithm (FOROA) to dynamically select vehicles that measure communication per-formance. Simulation experiments show that the FOROA reduces map uncertainty 41% faster than the previous method when reducing map uncertainty by 40%.