With the growing demand for Internet video applications, last-mile Internet Service Providers (ISPs) are under increasing pressure to efficiently provision and manage their networks. For an efficient network optimization, it is crucial for network operators to have a deep understanding of end-user video Quality of Experience (QoE). However, assessing video QoE is challenging for ISPs, as they typically lack access to either endpoints. This talk will present methods for using network measurement data to estimate objective video QoE metrics. First, I will introduce eMIMIC, a novel approach to inferring QoE for encrypted HTTP-based Adaptive video Streaming (HAS). eMIMIC leverages knowledge of the HAS streaming protocol and exploits the traffic patterns it generates in the network. Next, I will discuss a machine learning-based technique to infer key QoE metrics for video conferencing applications. This method is effective even without access to application-layer headers, which are both cumbersome to collect and are also becoming less accessible due to the use of proprietary protocols. The talk will conclude with an overview of other ongoing projects in my group.
Prof. Tarun Mangla, Assistant Professor, IIT Delhi
Tarun Mangla is an Assistant Professor in the Department of Computer Science at IIT Delhi. His research focuses on using data-driven methods to understand and improve Internet performance and accessibility. He completed his Ph.D. at Georgia Tech in 2020 and then spent three years as a Postdoctoral Scholar at the University of Chicago. He earned his bachelor's degree from IIT Delhi in 2014. Tarun is a recipient of the Best Paper Award at IFIP TMA 2018.