Caching-Assisted Collaborative Task Offloading for Vehicular Edge Computing: A Deep Reinforcement Learning-Based Approach

doi:10.1109/TMC.2025.3650617

Caching-Assisted Collaborative Task Offloading for Vehicular Edge Computing: A Deep Reinforcement Learning-Based Approach

2026-01-05

0

PRE

AI

DOI：10.1109/TMC.2025.3650617

Original

Original request for help

Save

Abstract

En 中文

Collaborative task offloading in vehicular edge computing (VEC) primarily emphasizes the diversity of offloading destinations, such as cloud centers, roadside units (RSUs), and other entities with underutilized resources. However, it often neglects the collaborative potential among vehicles whose tasks are associated with the same service. In this paper, we propose a collaborative task offloading strategy from the perspective of vehicles with offloading requests. Vehicles collectively accomplish task offloading by dividing responsibilities for specific service component offloading. To enhance the performance of the VEC system, we introduce a caching-assisted collaborative task offloading strategy. An optimization problem is formulated to minimize the response latency of tasks in VEC. Due to the complexity of solving this Mixed Integer Nonlinear Programming (MINLP) problem, we decompose it into three subproblems: the Task Offloading and Service Caching (TOSA) problem, the Computing Resource Allocation (RA) problem, and the Service Component Assignment (CA) problem. We address the RA problem using a Lagrangian duality-based approach, solve the CA problem with a heuristic algorithm, and tackle the TOSA problem using a Proximal Policy Optimization (PPO)-based deep reinforcement learning (DRL) algorithm. Extensive simulations are conducted to evaluate the performance of the proposed strategy. The simulation results demonstrate that our solution outperforms existing methods in multiple dimensions, including convergence rate, average response latency, and task success rate.

Keywords:

Vehicular edge computing

collaborative offloading

service caching

resource allocation

deep reinforcement learning

Caching-Assisted Collaborative Task Offloading for Vehicular Edge Computing: A Deep Reinforcement Learning-Based Approach

Abstract

Journal

Organization