Scholarlink

<p>In frequent traffic participation, For-Hire Vehicles (FHVs) can be effectively integrated into Mobile Edge Computing (MEC) systems, thus playing a critical role in urban emergencies when additional computing resources are required. However, there are considerable challenges for FHVs to participate and complete auxiliary tasks. On one hand, the operational status of FHVs dictates their availability for such tasks. On the other hand, task completion latency and FHVs’ revenue must be ensured. To address the aforementioned issues, this paper studies a ride-hailing-enabled MEC system, and deeply couples the operational state of ride-hailing vehicles, cruising routes, task processing locations, and incentive mechanisms. Different from the limitation of existing studies that mostly optimize routing or offloading decisions independently while ignoring the revenue demands of ride-hailing vehicles, this paper realizes the two-way adaptation between vehicle operation strategies and system-level optimization through incentive design. For empty ride-hailing vehicles, an expected revenue-cruise cost compensation mechanism is introduced to enhance their willingness to participate in tasks. For occupied ride-hailing vehicles, task admission rules under passenger service quality constraints are established to balance the conflict between the core passenger-carrying business and auxiliary tasks. On this basis, a multi-objective optimization problem is formulated to minimize task completion latency and maximize FHVs’ revenue. To solve this problem, we design the Dynamic Optimization Scheme Driven by Incentives (DOSDI). First, we design a vehicle path planning algorithm based on Branch and Cut (B&amp;C) to find paths that minimize cruising time and maximize revenue. Then, we develop a task offloading algorithm based on Deep Q-Network (DQN) to optimize the problems of task offloading decisions and processing resources, thereby reducing task completion latency and increasing vehicle revenue. Extensive experimental results demonstrate that our scheme performs better than other algorithms.</p>

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

Dynamic optimization scheme driven by incentives in FHVs-enabled MEC systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月7日

查看全文
  • Jinlei Liu,
  • Linbo Zhai,
  • Ping Zhao,
  • Dongsheng Wu,
  • Jing Yan

摘要

In frequent traffic participation, For-Hire Vehicles (FHVs) can be effectively integrated into Mobile Edge Computing (MEC) systems, thus playing a critical role in urban emergencies when additional computing resources are required. However, there are considerable challenges for FHVs to participate and complete auxiliary tasks. On one hand, the operational status of FHVs dictates their availability for such tasks. On the other hand, task completion latency and FHVs’ revenue must be ensured. To address the aforementioned issues, this paper studies a ride-hailing-enabled MEC system, and deeply couples the operational state of ride-hailing vehicles, cruising routes, task processing locations, and incentive mechanisms. Different from the limitation of existing studies that mostly optimize routing or offloading decisions independently while ignoring the revenue demands of ride-hailing vehicles, this paper realizes the two-way adaptation between vehicle operation strategies and system-level optimization through incentive design. For empty ride-hailing vehicles, an expected revenue-cruise cost compensation mechanism is introduced to enhance their willingness to participate in tasks. For occupied ride-hailing vehicles, task admission rules under passenger service quality constraints are established to balance the conflict between the core passenger-carrying business and auxiliary tasks. On this basis, a multi-objective optimization problem is formulated to minimize task completion latency and maximize FHVs’ revenue. To solve this problem, we design the Dynamic Optimization Scheme Driven by Incentives (DOSDI). First, we design a vehicle path planning algorithm based on Branch and Cut (B&C) to find paths that minimize cruising time and maximize revenue. Then, we develop a task offloading algorithm based on Deep Q-Network (DQN) to optimize the problems of task offloading decisions and processing resources, thereby reducing task completion latency and increasing vehicle revenue. Extensive experimental results demonstrate that our scheme performs better than other algorithms.