桂林电子科技大学学报2026，Vol.46Issue(4)：338-347,10.DOI:10.16725/j.1673-808X.202362
一种主动感知的SDN交换机流表溢出缓解方法
An active sensing method for mitigating flow table overflow in SDN switches
摘要
Abstract
In OpenFlow-based Software-Defined Networking(SDN),flow tables are usually implemented using ternary content-ad-dressable memory(TCAM)to enable fast rule lookup.However,TCAM capacity is limited and can accommodate only thousands to tens of thousands of flow entries.When a large number of new flows arrive within a short period,flow table overflow can easily oc-cur,generating a large number of Packet-In messages and overloading the SDN controller,thereby degrading overall network perfor-mance.To address this issue,an active-sensing-based adaptive flow-table-overflow mitigation method(ASAOM)is proposed.Based on the historical installation statistics of flow entries,an adaptive triple exponential smoothing algorithm is employed to predict the number of newly installed flow entries in the next sampling interval.By combining the prediction results with the current flow-table capacity,potential overflow events can be detected in advance.Once an impending overflow is identified,flow entries with a low predicted matching probability are proactively replaced using a scoring-based replacement strategy,thereby improving flow-table utilization.Experimental results show that,compared with the conventional P-LRU algorithm,the proposed method reduces approxi-mately 27%of flow-table overflow events and decreases the number of Packet-In messages by about 20%.关键词
软件定义网络/流表溢出/OpenFlow交换机/流表项替换/指数平滑法Key words
software-defined networking/flow table overflow/OpenFlow switch/flow entry replacement/exponential smoothing分类
信息技术与安全科学引用本文复制引用
张诗杰,王勇,刘玉明..一种主动感知的SDN交换机流表溢出缓解方法[J].桂林电子科技大学学报,2026,46(4):338-347,10.基金项目
国家自然科学基金(61831013) （61831013）
广西创新驱动发展专项基金(科技重大专项桂科AA18118031) （科技重大专项桂科AA18118031）
广西研究生创新计划(YCSW2021179) （YCSW2021179）
