多决策属性的网络协议识别算法  

Network Protocol Identification Algorithms Based on Multi-attribute Decision Making Method

作　　者：栾方军[1] 王健[1] 邱海斌[2] 刘天波[1] 

机构地区：[1]沈阳建筑大学信息与控制工程学院,沈阳110168 [2]沈阳广播电视大学人事处,沈阳110003

出　　处：《小型微型计算机系统》2014年第2期365-367,共3页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61070024)资助

摘　　要：物理信息融合系统一经提出就引起了各国研究者的广泛关注,与传统的控制系统不同,它可以看成一个融合了大量异构网络的大型控制网络系统.在这样的系统中对通信网络中的协议实时有效安全地传输提出较高的要求.为提高协议识别的效率,减轻网络负载,在网络流特征提取的基础上,提出对协议进行识别的变精度粗糙集和熵值法相结合的权值算法和神经网络构造优化算法.实验结果表明通过改变α,β值,动态地调整属性权值计算的结果,提高了选择决策的准确度.基于变精度和熵值法的权值计算并结合神经网络的协议识别模型,相较于一般的识别算法,其识别准确度更高、更容易实现.Cyber-physical system is put forward has caused the attention of all countries. Being different of the traditional control sys- tem, cyber-physical system can be thought of as a large control network system that integrated a large number of heterogeneous net- work. In this system put forward higher requirement for network protocol transmission effectively, securely and real-time. In order to improve the efficiency of the recognition and reduce the network load, proposing weight calculation based on variable rough set and the method of entropy ,and then proposing the neural network structure algorithm that all are based on characteristics of network flow in protocol identification.

关 键 词：物理信息融合系统 变精度粗糙集 熵值法 神经网络 协议识别 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]