A Study On The Effect Of Chinese Holidays On IP Backbone Traffic

Temporal characteristics of network traffic analysis is a traditional problem in networkmeasurement. In general, changes in behavior of a large number of users can lead to networktraffic changes. Flow cycle characteristics in some mobile communication networks have beenobserved, the user daily or weekly behaviors. In this paper, we use a IP backbone networktraffic as the research object, research method through the timing analysisproblem Chinese holidays on the impact of traffic.This paper gets a523day a group of IP backbone network traffic from a telecomoperators, including the daily peak value of12backbone node. We developed a data analysisalgorithm, trying to analyze the change characteristics of node flow backbone in Chinese SpringFestival and National Day. We first on the network traffic data are processed by wavelettransform, and the wavelet transform modulus maxima method for singular point of data flow, andthen were treated with repair of missing or anomalous data. Next, we performed flowdecompositionof network traffic, to obtain the overall growth trend and cyclical components. Theperiodic component of our overlay network obtained the periodic flow template each backbonenode3and5weeks. Finally, data flow and node template we within the data set during the SpringFestival and National Day comparison, fluctuations in network traffic analysis during the holidays.The results showed that, seven days of the holiday will be Chinese regular periodiccomponents on each node of the affected. Enter after the Spring Festival, the flow rate of eachnode will be different degrees of decline; population migration and the reason forthis China Spring Festival, the festival began, most people may be reunited with their families inthe busy, Internet traffic corresponding less than usual. After entering the National Day, each nodeof the flow rate will be increased to varying degrees, but periodic flow basically did not change theflow of the weekly; the reason, people have more time during the holidays can get to theInternet, so the flow results are presented in the uniform improvement, peak andvalley period flow a week of value characteristics affected by the National Day holiday.