Intelligent UAV Trajectory Design and Task Offloading for UAV-Assisted Edge Computing in Urban Road Scenarios
With the rapid proliferation of internet of vehicles applications, vehicle users in urban road scenarios face ever-increasing demands for low-latency and energy-efficient processing of computation-intensive tasks. The traditional fixed terrestrial infrastructure offers limited coverage in complex urban environments, making it difficult to satisfy the differentiated quality of service requirements of large-scale vehicle populations. To fully exploit the advantages of unmanned aerial vehicles (UAVs) in terms of flexible deployment and on-demand service provisioning, we propose a UAV-assisted mobile edge computing architecture tailored for urban road scenarios. By modeling realistic urban road terrain with varying elevations, we construct a two-tier cooperative network consisting of multiple rotary wing UAVs and ground vehicles. Aiming at maximizing the total system energy consumption, we formulate a mixed integer nonlinear programming problem that minimizes total system energy consumption through joint optimization of UAV flight trajectories and vehicle task offloading decisions while comprehensively accounting for task latency constraints, UAV flight velocity constraints, and the impact of three-dimensional terrain on air-to-ground channels. Considering the high-dimensional, non-convex, mixed integer, and strongly coupled nature of the problem, we design a genetic algorithm (GA)-based UAV trajectory design and offload allocation algorithm. The proposed approach encodes UAV trajectories as real-valued vectors and offloading decisions as binary vectors, employs a penalty function method to handle constraints, and achieves efficient global search through tournament selection, single-point crossover, and Gaussian mutation operators. Simulation results verify that the proposed algorithm converges reliably to feasible solutions under varying task data sizes and vehicle densities and achieves up to 20.6% energy savings compared to the benchmark schemes. The experimental results validate the necessity and effectiveness of jointly optimizing UAV trajectory and task offloading in urban road scenarios.