Modeling of flows based on behavior of applications

The best way to make a modeling of traffic to obtain characteristics of traffic and manage the network in the Internet. These days cloud services, which provide many services, are widely used. And many applications create some connections in a short term to speed up user´s response time. For this reason, there are many bursty flows, which are generated in a short term. So far, there are many studies about modeling of flows by taking bursty packets in to account. However it is difficult to make a model of arrival processes of flows by previously methods, because bursty flows are generated automatically by many applications. Then we focus on the inter-arrival time of flows. We analyze factors affecting flow generation. Then we show the two kinds of flows based on the factors. Next we analyze arrival patterns of flows based on user operation. Finally we show that dependencies are between arrival patterns of flows and the behavior of applications by user operations. And we approximate distribution of inter-arrival time of flows.