Task Offloading Decision Making for IoV Based on Deep Reinforcement Learning
With the rise of in-vehicle applications, vehicles face increasing limitations in computing power, storage, and energy. To address the growing needs of compute-intensive tasks, cloud-edge collaborative computing has become an essential approach. However, current task offloading methods often assume full task offloading, which does not suit practical cases like segmented data processing in autonomous driving, making it challenging to determine the best offloading rate.
Moreover, many existing solutions lack a priority model driven by task resource demands, which limits their ability to balance efficient offloading with sensible resource allocation.
Proposed Models and Solution
- Communication model
- Energy consumption model
- Cost model
- Priority model
- Task offloading model
The paper introduces a task offloading decision scheme based on deep reinforcement learning algorithms, allowing optimal offloading strategies to be chosen dynamically.
Experimental results demonstrate that, compared to existing methods, the proposed approach significantly improves performance.
By integrating diverse models and leveraging deep reinforcement learning, the scheme better manages resource allocation and offloading rates in practical Internet of Vehicles (IoV) scenarios.
Author's summary: This study presents a deep reinforcement learning-based framework that dynamically optimizes task offloading in vehicles, significantly enhancing resource management and performance in IoV environments.