An autonomic traffic analysis proposal using Machine Learning techniques

Network analysis has recently become in one of the most challenging tasks to handle due to the rapid growth of communication technologies. For network management, accurate identification and classification of network traffic is a key task. For example, identifying traffic from different applications is critical to manage bandwidth resources and to ensure Quality of Service objectives. Machine learning emerges as a suitable tool for traffic classification; however, it requires several steps that must be followed adequately in order to achieve the goals. In this paper, we proposed an architecture to perform traffic analysis based on Machine Learning techniques and autonomic computing. We analyze the procedures to perform Machine Learning over traffic network classification, and at the same time we give guidelines to introduce all these procedures into the architecture proposed. The main contribution of our proposal is the reconfiguration of the traffic classifier that will change according to the knowledge acquired from the traffic analysis process.