Research And Data Analysis Of Network Quality Measurement Method Based On Active And Passive Combination

Recently,with the rapid development of network technology,the network users have an explosive growth from fixed network to mobile network.In particular,the emergence of network applications profoundly changes people's status.People can shop,watch videos,browse news,and socialize on the Internet and so on.The convenience of the network provides a powerful guarantee for people to pursue a more efficient life.Network traffic has a doubled growth with the Internet penetration.Analyzing user traffic and grasping the user demand have become the top priority of network operators.The wide range of network services makes the network structure more complex.Controlling the network quality and managing the network system play significant role in improving competitiveness among operators.Therefore,it is helpful to improve the user experience and promote the further development of the network society by monitoring the service quality of users accessing the network,mining the factors that affect the quality of the network,and detecting the network anomalies.Firstly,this thesis summarizes the background and current studies on the network quality measurement method.Secondly,we focus on the merits and demerits of active measurement method and passive measurement method.Based on this,we design a network quality measurement system combining active and passive method.The system includes four major modules:data acquisition module,data management module,data analysis module and data visualization module.Thirdly,we have an overall analysis of end-to-end network quality from three factors:the busy period,with or without cache,different business categories.By defining the first packet delay to specify the network quality and analyzing the influential factors of the delay.Then we find that the network delay has a linear growth with the server distance.Finally,it is effective to find out the cause of the network from our data by the unsupervised anomaly detection model based on clustering.In summary,this thesis presents a method of the network quality measurement,and we can obtain the delay data of each phase when users accessing the network.So it is easy to find out the main factors that affect the network quality by analyzing above data.We find that correlation exists between the network quality and other indicators from multi-perspective,and detect the abnormal data in network to find the causes of exceptional network.Above findings are of great significance for mining the data value in networks.