Mobile Traffic Identification Based On Application'S Network Signature

Network traffic of different mobile applications identification is a fundamental component in many areas, such as network management. Recent research cannot differentiate the common HTTP requests such as GET*.jpg from third party content provider. Hence, they cannot measure how much of the traffic is generated by a certain app. In this paper, we propose a new network signature for mobile apps to solve the common HTTP requests problem. If we discovered one HTTP request, other requests should be expected around it. Based on this observation, we extended the signature extraction from a single HTTP request into the HTTP request set generated by a user's one operation. We correlated the reference information in HTTP traffic and examined the content return by the server to form this set. With this signature, we can not only improve the recognition rate of the app's traffic, but also identify the flow size of the app.