Improvement of Network Application Type Identification Based on Measure of Medium Truth Degree
Weiqing Cheng · Journal of Nanjing University of Posts and Telecommunications · 2012
Internet application identification is important in network management.This paper aims to improve application type identification of Internet flows related to medium mathematics.It proposes a novel application type identification method based on MMTD(Measure of Medium Truth Degree).The efficiency of the original method shows the potential and advantages.But the parameters used in the method were probably based on network expertise and the experience of the artificial statistics and analysis of features of flows.The lack of systemic method of calculations of parameters affected the use and promotion of the identification method.This paper improves the original method in followings.First,it proposes mappings between truth scales and metrics values that reflect the distribution characteristics of metrics of objects and the corresponding method of calculation of truth scales.Second,it proposes a method of metrics selection and a method of weights calculation based on intensity degrees of distribution characteristics of metrics.Third,it proposes a method that can calculate feature parameters of internet flow attributes by rigorous statistics and data processing methods.The experimental result demonstrates efficiency of the proposed method.