基于ARMA模型预测的交换机流表更新算法

针对SDN网络中交换机在网络流量高峰期流表匹配率低以及控制器负载过重的问题，提出了一种基于自回归移动平均（ARMA）模型预测的交换机流表更新算法。算法首先收集每个取样周期内的新增流表项数量作为历史数据，然后使用ARMA模型对收集的历史数据进行分析，预测下一个周期内新增加的流表项数量，并结合当前流表空间的使用情况，清除交换机中过去一段时间内使用频率较低的流表项。采用真实数据中心网络数据的模拟实验结果表明，与流表更新的一般方法相比，该算法有效地提高了交换机流表的匹配率，并减少了交换机与控制器之间交互的次数，降低了控制器端的负载。

Flow Table Updating Algorithm Based on ARMA Model Prediction

In order to solve the problem of insufficient resources of flow table during tthe peak period of network traffic and the overload of controller in the Software Defined Network(SDN), a flow table updating algorithm based on ARMA model prediction for SDN switch is proposed. The algorithm firstly collects the number of new flow entries in each unit time, then uses ARMA model to analyze the collected historical data, and predicts the number of additional flow entries in the next unit time. Finally, according to the usage of the flow table space, the low frequency flow entries in the past period of time are eliminated dynamically. The experimental results show that compared with the general method of flow table updating, the proposed algorithm can improve the matching rate of flow table and can decrease the frequency of interaction between the switch and the controller, and the burden of the controller is reduced as well.