Proactive eviction of flow entry for SDN based on hidden Markov model

With the fast development of software defined network(SDN),numerous researches have been conducted for maximizing the performance of SDN.Currently,flow tables are utilized in OpenFlows witch for routing.Due to the space limitation of flow table and switch capacity,various issues exist in dealing with the flows.The existing schemes typically employ reactive approach such that the selection of evicted entries occurs when timeout or table miss occurs.In this paper a proactive approach is proposed based on the prediction of the probability of matching of the entries.Here eviction occurs proactively when the utilization of flow table exceeds a threshold,and the flow entry of the lowest matching probability is evicted.The matching probability is estimated using hidden Markov model(HMM).Computersimulation reveals that it significantly enhances the prediction accuracy and decreases the number of table misses compared to the standard Hard timeout scheme and Flow master scheme.

PopFlow:a novel flow management scheme for SDN switch of multiple flow tables based on flow popularity

Pipeline processing is applied to mutiple flow tables(MFT)in the switch of software-defined network(SDN)to increase the throughput of the flows.However,the processing time of each flow increases as the size or number of flow tables gets larger.In this paper we propose a novel approach called PopFlow where a table keeping popular flow entries is located up front in the pipeline,and an express path is provided for the flow matching the table.A Markov model is employed for the selection of popular entries considering the match latency and match frequency,and Queuing theory is used to model the flow processing time of the existing MFT-based schemes and the proposed scheme.Computer simulation reveals that the proposed scheme substantially reduces the flow processing time compared to the existing schemes,and the diference gets more significant as the flow arrival rate increases.

Determining node duty cycle using Q-learning and linear regression for WSN

Wireless sensor network(WSN)is effective for monitoring the target environment,which consists of a large number of sensor nodes of limited energy.An efficient medium access control(MAC)protocol is thus imperative to maximize the energy efficiency and performance of WSN.The most existing MAC protocols are based on the scheduling of sleep and active period of the nodes,and do not consider the relationship between the load condition and performance.In this paper a novel scheme is proposed to properly determine the duty cycle of the WSN nodes according to the load,which employs the Q-leaming technique and function approximation with linear regression.This allows low-latency energy-efficient scheduling for a wide range of traffic conditions,and effectively overcomes the limitation of Q-learning with the problem of continuous state-action space.NS3 simulation reveals that the proposed scheme significantly improves the throughput,latency,and energy efficiency compared to the existing fully active scheme and S-MAC.