一种OpenFlow网络策略冲突解决方案  

Policy Conflicts Resolution in the OpenFlow Network

作　　者：叶星 罗兴国[1] 李丰 YE Xing;LUO Xingguo;LI Feng(Information Engineering University,Zhengzhou 450002,China;Unit 61646,Beijing 100192,China)

机构地区：[1]信息工程大学 [2]61646部队

出　　处：《信息工程大学学报》2018年第2期234-239,共6页Journal of Information Engineering University

摘　　要：针对高层次的决策存在策略冲突,尝试构建自动化推理系统(game-based dedicated expert system,GB-DES)解决当高层策略组合成底层规则时产生的冲突。首先定义了头字段、流类型、流空间、动作、策略以及Open Flow规则等概念;其次构建包含不同推理规则的知识库,用以解决潜在冲突的决策,而针对完全冲突的决策,使用零和博弈从不兼容策略中选择最优策略;最后给出GB-DES的实现原型。Recently,developers tend to develop the modular function of the control plane based on high-level language in the software defined network,and the conflicts among the high-level policies inevitably exist when a plurality of independent functional modules are combined to a complex business processing pipeline. Facing this kind of policy conflicts,this paper attempts to build gamebased dedicated expert system（ GB-DES） to automatically reason and solve the conflict while highlevel policies are combined to the underlying rules. Firstly,we define several concepts such as header field,flow pattern,flow space,action,policy,OF rules and so on; secondly,we build the knowledge base containing different inference rules to solve the potential conflict of policy,while for those completely incompatible conflict policies,we take the most reasonable strategy from incompatible strategy by zero-sum game; finally,we give the prototype of GB-DES.

关 键 词：软件定义网络 策略组合 冲突解决 零和博弈 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]