一种基于马尔科夫决策过程的多态路由派生方法  

A Polymorphic Routing Derivation Mechanism Based on Markov Decision Process

作　　者：张岩[1] 兰巨龙[1] 王鹏[1] 胡宇翔[1] 

机构地区：[1]国家数字交换系统工程技术研究中心,郑州450002

出　　处：《电信科学》2015年第6期64-70,共7页Telecommunications Science

基　　金：国家重点基础研究发展计划("973"计划)基金资助项目(No.2012CB315901;No.2013CB329104);国家自然科学基金资助项目(No.61309019;No.61372121);国家高技术研究发展计划("863"计划)基金资助项目(No.2013AA013505)~~

摘　　要：基于路由结构的自组织和自调节思想来为多样化业务需求提供个性化定制路由服务,多态路由机制为满足目前业务和未来新业务的不同路由服务需求提供了一种有效解决途径。基于多态路由机制中派生过程的无记忆序贯决策特性.将其建模为马尔科夫决策过程,并提出一种基于增强型学习的多态派生方法。实验结果表明,该多态派生方法能够灵活有效地为多态路由机制构建满足需求的多样化路由服务。Based on routing structure of self-organization and self-regulation to provide personalized routing service for the diverse business requirements, polymorphic routing mechanism provides an effective solution to meet the current and future business which needs diverse routing services. Considering the memoryless and sequential decision-making features of the polymorphic derivation process, it was modeled as a Markov decision process, and a polymorphic derivation method based on the enhanced learning was put forward. The experimental results show that the polymorphic derivation method can flexibly and effectively construct the diverse routing services for polymorphic routing mechanism.

关 键 词：多态路由 派生 马尔科夫决策过程 增强型学习 

分 类 号：TN912.34[电子电信—通信与信息系统]