Events Calendar

PhD Final Exam – Kevin Gatimu

qMDP: DASH Adaptation using Queueing Theory within a Markov Decision Process

Today's Internet traffic is significantly dominated by video. As the Internet's underlying protocol, HTTP has become the de facto video delivery mechanism. In addition, Dynamic Adaptive Streaming over HTTP (DASH) has emerged as the standard for state-of-the-art video streaming. In DASH, a client makes HTTP requests to a server for video chunks from a selection of different quality versions. Through a dynamic process called bitrate adaptation, the client selects a video quality based on the network bandwidth. Most bitrate adaptation algorithms tend to have significant ad hoc components, which has paved the way for more deterministic solutions based reinforcement learning (RL). However, RL algorithms for DASH tend to trade off between accuracy and efficiency. This work proposes a queueing-theory-based Markov Decision Process (qMDP), which features a simplified state characterization that includes an M/D/1/K queueing model for the client buffer. qMDP also featu! res a training algorithm based on the asynchronous advantage actor-critic algorithm. qMDP converges towards an optimal solution faster than pre-existing RL methods, making it well-suited for online deployment in most DASH streaming systems.

Major Advisor: Ben Lee
Committee: Thinh Nguyen
Committee: Huaping Liu
Committee: Bella Bose
GCR: Janet Nishihara

Friday, December 13, 2019 at 2:00pm to 4:00pm

Kelley Engineering Center, 1007
110 SW Park Terrace, Corvallis, OR 97331

Event Type

Lecture or Presentation

Event Topic

Research

Organization
Electrical Engineering and Computer Science
Contact Name

Calvin Hughes

Contact Email

calvin.hughes@oregonstate.edu

Subscribe
Google Calendar iCal Outlook

Recent Activity