Behavior signature for big data traffic identification
With the rapid development of the Internet and the popularization of multimedia services, Internet traffic has become a big data traffic that its volume, variety and velocity has dramatically increased. This phenomenon causes several limitations in traffic classification such as increased computational complexity and difficult real-time control. In this paper, we propose a behavior signature for application-level traffic identification to overcome these limitations. The proposed behavior signature is the identity pattern of traffic behavior appearing in the first few request packets of plural traffic flows when a specific function is conducted by an application. This is in contrast to the previous signature techniques that usually use a singular packet or flow for feature extraction and traffic identification. In order to prove the feasibility of the proposed behavior signature, we present the experimental results based on five popular applications.