Computer vision plays a crucial role in enabling connected autonomous vehicles (CAVs) to observe and comprehend their surroundings. The computer vision tasks are typically based on convolutional neural networks (CNNs)...
详细信息
Computer vision plays a crucial role in enabling connected autonomous vehicles (CAVs) to observe and comprehend their surroundings. The computer vision tasks are typically based on convolutional neural networks (CNNs). However, CNNs often require significant processing power. Techniques like early exiting and split computing enhance CNN taskexecution latency and adaptability to varying environmental conditions. Since the split computing introduces additional overhead for offloading of the task from the CAV to an edge servers, we incorporate multiple autoencoders within each split point to enhance the adaptability of splitting under varying environmental conditions. However, the autoencoders introduce an additional layer of complexity related to the selection of the optimal compression strategy alongside the splitting and exiting decisions. To tackle this challenge, we introduce a novel approach based on the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm. This algorithm dynamically and jointly determines the most suitable exit point, split point, and autoencoder. Furthermore, the MADDPG-based approach considers other CAVs when selecting action, promoting cooperation among CAVs. Our results demonstrate that the proposed approach reduces latency up to 44.4% while maintaining at least comparable or even higher accuracy of the computed vision outcome compared to the state-of-the-art solutions.
The development of information technology and intelligence in urban rail transit has attracted great attention. The requirement for compute-intensive and delay-sensitive services is increasing rapidly. As a new networ...
详细信息
The development of information technology and intelligence in urban rail transit has attracted great attention. The requirement for compute-intensive and delay-sensitive services is increasing rapidly. As a new network paradigm, Multi-access Edge Computing (MEC) can offer an environment for information technology service and extend cloud computing capability to the edge of mobile communication network. So communication data can be timely processed near the source to effectively solve the limitation of on-board resources. In this paper, we consider the heterogeneous characteristics of services and different requirements of tasks in urban rail transit system, and a MEC system architecture based on task classification is designed. To effectively utilize communication and computing resources, a resource allocation strategy based on task Classification Twin Delayed Deep Deterministic policy gradient (TC-TD3) algorithm is proposed, which includes communication and computing resources allocation. In addition, to address the offloading selection issue of multi-MEC servers, a task offloading algorithm based on balancing tasks priority of MEC is designed. In the system model, considering the random arrival time of tasks and the dynamic resource allocation at each time slot, we propose a dynamic task execution model which is more suitable for the urban rail transit system. The findings of the simulation demonstrate that the proposed scheme can significantly improve the task completion rate and reduce the task processing delay of different tasks.
暂无评论