基于FPGA的二次加权NB网络流量分类方法

Method of network traffic classification using doubly weighted Nave bayes based on FPGA

针对传统网络流量分类方法不能很好地满足高速网络流量分类的实时性和准确性要求,提出了一种基于FPGA的二次加权朴素贝叶斯网络流量分类方法。该方法通过对样本属性和样本类别进行加权,并在FPGA上进行建模,从而实现网络流量的实时、准确分类。仿真实验结果表明,该方法与传统网络流量分类方法相比,提高分类精度的同时具有更好的实时性,在100MHz时钟下其分类速度约为纯软件实现分类速度的260倍。

In order to solve the problems in the rapid developing network that the general software methods of traffic classification cannot meet the requirements of real-time, a method of Naive Bayes based on FPGA for network traffic classification is proposed. Sample properties weighted evaluation and sample space weighted evaluation are carried out, which modeled on the FPGA. Hence, this model not only takes into the consideration classification accuracy but also meets the requirement of real-time. Simulation result shows that this method can improve the classification precision and have a better real-time compared with traditional network traffic classification. And in the 100MHz clock time its classification rate is about 260 times than the classification rate imDlementing bv nure software.