Management of Flow Table of SDN for Proactive Eviction Using Fuzzy Logic

摘要

With the development of Software Defined Network (SDN), numerous researches have been conducted for improving the performance of SDN. In SDN flow table is used in OpenFlow switch for the routing of the packets. Due to the space limitation of flow table and switch capacity, various issues need to be resolved for effectively dealing with a large number of flows. The existing schemes typically employ a reactive approach such that evicted entries are decided only when timeout or table miss occurs. In this paper a novel proactive eviction scheme is proposed which employs hidden Markov model (HMM) to predict the probability of table miss of the entries. If the probability exceeds the preset threshold, fuzzy logic is used to select the entries for eviction considering match priority, idle time, and the number of unmatched flows via a new notion called eviction index. The proposed scheme is for efficient flow entry eviction before table miss actually occurs, which eventually increases the speed of flow management in SDN switch. Computer simulation reveals that the proposed scheme increases the match probability and prediction accuracy, and reduces the number of misses at least 10% compared to three existing entry eviction schemes.