Scholarlink

The operation of connected vehicles demands substantial computing power, which cannot be fully supported by on-board units alone. Offloading tasks to network-based computing resources, spanning from the edge to the cloud, presents a promising solution. However, the challenge lies in assigning each request to an appropriate processing unit within this edge-cloud continuum while satisfying strict latency constraints. In this paper, we propose a simplified MILP framework for task offloading across a four-tier hierarchy: vehicle, edge, regional data centers, and cloud. While the MILP approach is impractical for real-time decision-making due to its high computational complexity and the dynamic nature of the problem, it provides an optimal solution that is valuable for benchmarking. We compared the proposed MILP model with four heuristic methods, revealing significant differences in terms of performance. The MILP model achieved a higher success ratio and reduced the average time required to process requests, thereby improving overall system performance.

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

A Simplified MILP Framework for Task Offloading in Vehicular Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ramón J. Durán Barroso,
  • Ignacio de Miguel,
  • Camilo Anzola,
  • Soheil Hosseini,
  • Maryam Masoumi,
  • Hafiza Kanwal Janjua,
  • Miguel Silva,
  • Noemí Merayo,
  • Juan Carlos Aguado,
  • Alfonso Bahillo,
  • Patricia Fernández,
  • Rubén Lorenzo,
  • Evaristo J. Abril

摘要

The operation of connected vehicles demands substantial computing power, which cannot be fully supported by on-board units alone. Offloading tasks to network-based computing resources, spanning from the edge to the cloud, presents a promising solution. However, the challenge lies in assigning each request to an appropriate processing unit within this edge-cloud continuum while satisfying strict latency constraints. In this paper, we propose a simplified MILP framework for task offloading across a four-tier hierarchy: vehicle, edge, regional data centers, and cloud. While the MILP approach is impractical for real-time decision-making due to its high computational complexity and the dynamic nature of the problem, it provides an optimal solution that is valuable for benchmarking. We compared the proposed MILP model with four heuristic methods, revealing significant differences in terms of performance. The MILP model achieved a higher success ratio and reduced the average time required to process requests, thereby improving overall system performance.