基于深度Q网络的云演艺延迟敏感业务QoE优化

QoE optimization of delay-sensitive cloud performance services based on Deep Q-Network

网络中的资源分配问题一直备受关注,特别是在超高清视频流的传输中,对资源的有效管理至关重要。然而,随着网络服务的多样化和不断增加的业务类型,传统的资源分配策略往往显得不够灵活和智能。深度Q网络(Deep Q-Network,DQN)是一种能够自适应地学习和调整资源分配策略的神经网络模型。它基于神经网络与Q-Learning算法,通过不断尝试和学习来决策最佳的资源分配方案。本文旨在研究一种在云演艺场景下基于深度Q网络的延迟敏感业务资源调度算法,以满足当今网络中多样化的业务需求。仿真结果表明,基于深度Q网络的延迟敏感业务资源调度算法使得用户体验质量(Quality of Experience)指标显著提升,表明所提算法能够更好地满足延迟敏感业务的需求。

The problem of resource allocation in the network has been paid much attention,especially in the transmission of ultra-high-definition video streams,so the effective management of resources is very important.However,with the diversification of network services and the increasing types of business,the traditional resource allocation strategy often appears to be not flexible and intelligent enough.Deep QNetwork(DQN)is a kind of neural network model which can learn and adjust resource allocation strategy adaptively.It is based on the neural network and Q-Learning algorithm,through continuous trial and learning to decide the best resource allocation scheme.This paper aimed to study a delay-sensitive service resource scheduling algorithm based on DQN in the cloud performing arts scene,so as to meet the diversified service requirements in today's networks.Simulation results show that the delay-sensitive service resource scheduling algorithm based on DQN can significantly improve the Quality of Experience(QoE),indicating that the proposed algorithm can better meet the needs of delay-sensitive services.