- Enabling Traffic Classification for Mobile Video Applications
- YouTube’s Adaptive Streaming Behavior and Its Implications
- Candid with YouTube: Adaptive Streaming Behavior and Implications on Data Consumption, Abhijit Mondal, Satadal Sengupta, B.R. Reddy, M.J.V. Koundinya, Chander G., Pradipta De, Niloy Ganguly, Sandip Chakraborty, ACM SIGMM Workshop on Network and Operating Systems Support for Digital Audio and Video (NOSSDAV), 2017.
- MoViDiff: Enabling Service Differentiation for Mobile Video Apps, Satadal Sengupta, Vinay Kumar Yadav, Yash Saraf, Harshit Gupta, Niloy Ganguly, Sandip Chakraborty, Pradipta De, 15th IFIP/IEEE International Symposium on Integrated Network Management (IM), 2017.
- MoViDiff: Enabling Service Differentiation for Mobile Video Apps
In this work, we focused on mobile video applications, and showed that characteristics of the flows originating from these apps differ with respect to packet-size, which is a packet data agnostic feature. We used this feature to train classifiers that can identify flows belonging to streaming video and interactive video. The classifier forms the core of a service differentiation platform, called MoViDiff, that runs on a wireless gateway, and apportions bandwidth among different video applications from an end device as per user preference. We show that MoViDiff is able to achieve an average classification accuracy of 96%, with the maximum accuracy reaching as high as 98%.
- Candid with YouTube: Adaptive Streaming Behavior and Implications on Data Consumption
In this work, we studied the internal working of YouTube’s bitrate adaptation algorithm, by identifying important parameters and exploring their roles. We observed that YouTube adapts segment length in addition to quality level, a behavior not been reported earlier. As an implication, we observed that data wastage for a playback session is significantly lower than estimated previously. We further provided an analytical model, augmented with a machine learning based classifier, to predict data consumption in advance for a video playback session.
- Information Technology Research Academy (ITRA) funded project “Post Disaster Situation Analysis and Resource Management using Delay-Tolerant Peer-to-Peer Wireless Networks (DISARM)”
Members & Collaborations
- Satadal Sengupta, Member, MS Student
- Abhijit Mondal, Member, PhD Student
- Sandip Chakraborty, Member, Assistant Professor
- Niloy Ganguly, Member, Professor
- Pradipta De, Collaborator, Georgia Southern University