Research On Traffic Modeling And Network Performance Analysis Of Modern Internet

With the rapid development of network communication technology,especially mobile Internet technology,the number of network application services and users has grown rapidly,modern Internet applications have undergone tremendous changes.Compared with the traditional Internet traffic,the modern Internet contains more traffic such as instant message and video and audio.Changes in network traffic and user participation behaviors have caused the characteristics of modern Internet traffic to be different from traditional traffic characteristics,which has an important impact on network traffic supervision and network management.This thesis takes modern Internet traffic as the research object,carries out traffic characteristic analysis and traffic modeling,and analyzes the network performance indicators of modern Internet traffic.The development of modern Internet applications has been analyzed and finds that instant messaging and online video applications have become the main services of the modern Internet.Therefore,the We Chat text traffic,Facebook video traffic,Skype audio traffic in instant messaging,and Xunlei video traffic and Vimeo video traffic in online video applications are used as the research data.Mathematical statistical processing is performed on the foreign public data sets and the traffic data captured in the laboratorye,and the time series of the number of data packets and packet lengths arriving at different time scales are obtained.Firstly,using MATLAB software to draw an intuitive diagram of the number of data packets arriving at different time scales.Through intuitive observation,it is found that modern Internet traffic has simultaneously characteristics of self-similarity and burstiness.Then,the R/S analysis method is used to calculate the hurst parameter values of modern Internet traffic under different time scales to determine whether the modern Internet traffic has self-similarity.The research results show that modern Internet traffic has a self-similar long correlation.The cause why modern Internet traffic has a self-similar long correlation is analized.Then the alpha stable distribution is used to verify the burstiness of modern Internet traffic.The quantile method is used to estimate the burst parameter values of modern Internet traffic,and the probability density function is used to fit the actual traffic sequence and the QQ diagram is used to prove alpha the stable distribution can be used to describe the burst size of modern Internet traffic.Aiming at the characteristics of modern Internet traffic with both self-similarity and burstiness,skewed LFSN traffic model is used to model modern Internet traffic.For comparison,the FBM self-similar traffic model and heavy-tailed distribution model are also used to model modern Internet traffic.Through model comparison analysis and goodness of fit of !,it is proved that the skewed LFSN traffic model can simultaneously characterize the self-similarity and burstiness of modern Internet traffic,and has a higher degree of fit than FBM self-similar traffic model and heavy-tailed distribution model.Based on the approximate expressions of network performance indicators of the skewed LFSN model,the average delay and packet loss rate performance indicators of modern Internet traffic are analyzed.Finally,OPNET simulation software is used to establish a G/D/1 queue model for modern Internet traffic,and the average delay and packet loss rate are simulated,and compared with the traffic analysis results based on the skew LFSN model.The simulation results prove the feasibility and effect of modern Internet traffic model based on skewed LFSN traffic model.