Intelligent Adaptive Streaming for Mobile and Immersive Video
School of Engineering and Technology
Dr Ayub Bokai
Synopsis
Video already dominates internet traffic, and new formats such as 360-degree, virtual and augmented reality demand far more data while networks stay unpredictable, especially on the move. This project develops intelligent methods for delivering multimedia efficiently and at high quality, using machine learning to adapt to changing network conditions, to user behaviour, and to what the viewer actually perceives. The area is broad and can be shaped toward prediction, systems building, or experimental evaluation, and connects naturally to mobile networks, edge computing and quality of experience. The group's earlier work, published in IEEE Transactions on Multimedia and drawing interest from companies such as Netflix, together with real bandwidth traces collected from vehicles around Sydney, provides strong foundations and examples such as adaptive streaming for immersive and mobile video.
multimedia networking, adaptive video streaming, 360-degree video, quality of experience, machine learning, mobile networks, edge computing
Doctorate; Either Masters or Doctorate; Masters
Other Special Notes
Possible directions include: predicting where a viewer will look and using it to control bitrate for 360-degree and VR video; real-time foveated streaming that concentrates quality where the viewer is watching, for mobile and XR devices; reinforcement learning and Markov Decision Process methods for adaptive bitrate selection in fluctuating networks; quality-of-experience modelling and measurement; and broader problems in multimedia delivery, mobile networking and edge computing. Current adaptive-streaming projects and real network datasets provide examples and starting points.
