摘要
合理分配资源,可有效控制光网络的拥塞现象,确保服务质量机制正常运转,提升用户体验。为此,对基于机器学习技术的光网络资源动态分配方法进行研究。通过融合改进的半监督机器学习方法Tri-Training算法与深度包检测技术构建光网络资源分类器,获取光网络流量所属的服务质量要求类别,实现资源类型分类,提升资源分配的合理性;依据资源分类结果,利用基于循环神经网络的光网络资源动态分配算法中的执行者部分优化分配方案,通过输入、动作与效用函数三部分获取混合策略纳什均衡,依据纳什均衡动态分配分类后的光网络资源。实验证明:该方法能够有效进行光网络资源的动态分配,降低业务阻塞率与延时情况的发生率。
Reasonable allocation of resources can effectively control the congestion of optical network,ensure normal operation of the quality of service mechanism,and improve the user experience.Therefore,the dynamic allocation method of optical network resources based on machine learning technology is studied.By combining the improved semi supervised machine learning method,Tri-Training algorithm and deep packet detection technology,the optical network resource classifier is constructed to obtain the QoS requirement category of optical network traffic,so as to realize the resource type classification,and improve the rationality of resource allocation.According to the results of resource classification,the executor part of the dynamic resource allocation algorithm based on recurrent neural network is used to optimize the allocation scheme.The Nash equilibrium of mixed strategy is obtained through three parts of input,action and utility function,and the classified optical network resources are dynamically allocated according to the Nash equilibrium.Experimental results show that this method can effectively allocate the optical network resources dynamically,and reduce the incidence of traffic blocking and delay.
作者
彭雪梅
黄建军
PENG Xuemei;HUANG Jianjun(Nanchang Institute of Technology,Nanchang 330044,China)
出处
《激光杂志》
CAS
北大核心
2022年第7期144-148,共5页
Laser Journal
基金
江西省教育厅科学技术研究项目(No.GJJ191012)。
