Scholarlink

To address the selfishness problem in autonomous vehicles, this paper proposes a credit-value-based vehicular–aerial–edge cooperative task offloading optimization scheme. First, unmanned aerial vehicles(UAVs) are introduced as aerial computing nodes to alleviate the performance bottlenecks caused by limited resources of road side units(RSUs) in complex road environments. Second, a credit-value-based incentive mechanism is designed to encourage vehicles with idle computing resources to actively contribute surplus resources in exchange for credits, which can subsequently be used to purchase computational resources from other vehicles, thereby effectively suppressing selfish behavior. Finally, during the task offloading and execution process, a weighted-sum optimization model is formulated, and an Adaptive Genetic Algorithm-assisted Particle Swarm Optimization(AGA-PSO) algorithm is proposed. Built upon the Particle Swarm Optimization(PSO) framework, AGA-PSO integrates crossover and mutation operations from Genetic Algorithms(GA), while adaptively adjusting inertia weight as well as crossover and mutation probabilities. This design enhances population diversity and prevents premature convergence, thereby improving global search capability while maintaining convergence efficiency. Simulation results demonstrate that the proposed UAV-assisted framework effectively alleviates the bottlenecks associated with relying solely on RSUs as edge computing nodes, significantly improving the quality of vehicular edge cooperative task offloading. For instance, when the number of task vehicles increases to 15, UAV assistance reduces latency by approximately 88.7%, when the task size expands to 9Mbits per vehicle, the latency reduction reaches up to 93.6%, showcasing excellent scalability and stability. Moreover, in comparative evaluations, AGA-PSO consistently outperforms PSO, GA, Simulated Annealing(SA), Differential Evolution(DE), and Ant Colony Algorithm(ACA) in terms of convergence speed, solution accuracy, and stability, thereby validating the effectiveness and efficiency of the proposed scheme.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Credit-Value-Based UAV-Assisted Task Offloading Optimization for Autonomous Vehicles

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yongyi Yang,
  • Tanshen Li,
  • Deming Yin

摘要

To address the selfishness problem in autonomous vehicles, this paper proposes a credit-value-based vehicular–aerial–edge cooperative task offloading optimization scheme. First, unmanned aerial vehicles(UAVs) are introduced as aerial computing nodes to alleviate the performance bottlenecks caused by limited resources of road side units(RSUs) in complex road environments. Second, a credit-value-based incentive mechanism is designed to encourage vehicles with idle computing resources to actively contribute surplus resources in exchange for credits, which can subsequently be used to purchase computational resources from other vehicles, thereby effectively suppressing selfish behavior. Finally, during the task offloading and execution process, a weighted-sum optimization model is formulated, and an Adaptive Genetic Algorithm-assisted Particle Swarm Optimization(AGA-PSO) algorithm is proposed. Built upon the Particle Swarm Optimization(PSO) framework, AGA-PSO integrates crossover and mutation operations from Genetic Algorithms(GA), while adaptively adjusting inertia weight as well as crossover and mutation probabilities. This design enhances population diversity and prevents premature convergence, thereby improving global search capability while maintaining convergence efficiency. Simulation results demonstrate that the proposed UAV-assisted framework effectively alleviates the bottlenecks associated with relying solely on RSUs as edge computing nodes, significantly improving the quality of vehicular edge cooperative task offloading. For instance, when the number of task vehicles increases to 15, UAV assistance reduces latency by approximately 88.7%, when the task size expands to 9Mbits per vehicle, the latency reduction reaches up to 93.6%, showcasing excellent scalability and stability. Moreover, in comparative evaluations, AGA-PSO consistently outperforms PSO, GA, Simulated Annealing(SA), Differential Evolution(DE), and Ant Colony Algorithm(ACA) in terms of convergence speed, solution accuracy, and stability, thereby validating the effectiveness and efficiency of the proposed scheme.