重采样与机器学习结合的防火墙链接动态分配

发布时间：2019-03-11 09:48

【摘要】：通过对网络防火墙连接的动态分配设计,提高网络安全性能和对攻击数据的检测性能,传统方法采用多级处理的有限排队理论的动态分配设计方案,导致防火墙对攻击序列排队解析模型处理混乱,攻击捕获性能不好。提出一种基于重采样与机器学习结合的防火墙链接动态分配设计方法。基于重采样与机器学习结合方法,进行防火墙协议的动态链接分配,数据输入包输入防火墙模型中,排队等待处理,定义数据防火墙链接信息流全极点中心矩阵,基于贝叶斯参数估计进行攻击信息特征挖掘,通过重采样技术提高防火墙的抗干扰能力,求得重采样频谱。提高数据包信息的有效到达率,网络防火墙的链接动态分配转化为对数据集从数据向量点积,优化的动态分配设计。实验得出,采用该算法,能有效提高对攻击数据的检测性能,差错率较低,性能优越。
[Abstract]:Through the dynamic allocation design of network firewall connection, the network security performance and the detection performance of attack data are improved. The traditional method adopts the dynamic allocation scheme of the limited queue theory of multi-level processing. It leads to the confusion of the firewall processing to the attack sequence queue analysis model, and the bad performance of attack capture. A dynamic allocation method for firewall links based on resampling and machine learning is proposed. Based on the method of combination of resampling and machine learning, the dynamic link assignment of firewall protocol is carried out. The data input packet is input into the firewall model, the queue is waiting to be processed, and the all-pole central matrix of data firewall link information flow is defined. Based on Bayesian parameter estimation, attack information feature mining is carried out, and the anti-jamming ability of firewall is improved by resampling technology, and the resampling spectrum is obtained. To improve the effective arrival rate of packet information, the link dynamic allocation of network firewall is transformed into an optimized dynamic allocation design of data set from data vector point product. The experimental results show that this algorithm can effectively improve the detection performance of attack data, and the error rate is low and the performance is superior.
【作者单位】：重庆商务职业学院;
【分类号】：TP393.08

【参考文献】