Dynamic Network Traffic Identification with Scalable and Resilient Machine Learning

点击翻译按钮获取中文摘要

With the emergence of data-hungry applications, such as big-data AI, augmented reality, wireless multimedia, and Internet-of-Things (IoT), the current generation of network infrastructure will face severe challenges in its struggle to satisfy the exploding service demands. In order to efficiently allocate the available networking resources among these diverse set of applications, traffic classification has become an essential technology for the optimal operation of a network. Moreover, the ever-changing characteristics of new applications demand a flexible and automatic approach. In response to these challenges, and in collaboration with TELUS, we propose a new research project to develop dynamic machine learning (ML) techniques to classify the network traffic of an Internet service provider (ISP). Our long-term objective is to understand the intersection between ML and networked systems and discover new theories and practical ML-based designed guidelines for ISPs. In the shorter term of this project, we focus on (1) developing scalable ML-based designs for dynamic network traffic classification, and (2) designing resilient ML algorithms for robust network traffic classification. Our scientific discoveries will contribute to the fundamental understanding and insights into future network services, and they will provide guidelines for future regulatory policies. Furthermore, through research collaboration and technology transfer, our technical outcomes benefit TELUS and its technology and business partners in Canada, to improve their innovation portfolio, system design, and profitability. This will help maintain their competitive advantage in the global market for the benefit of all Canadians.