Our team’s research topic is related to Internet-based video streaming. We now aim at Virtual Reality content and HTTP/2 protocol, which are future technologies for Internet-based video streaming. Other previous researches & projects can be found here.
Multimedia streaming has recently contributed a significant part of the Internet traffic. Sandvine in its 2016 Global Internet Phenomena Report predicted that real time entertainment services (i.e., streaming audios and videos) account 80% of North American downstream in peak hours on fixed access networks in 2020, up from 71% in 2016. To provide users with a high quality of service, a streaming provider should answer two critical questions:
- How to estimate the user experience given a video clip?
- How to deliver the video over the Internet with the highest possible video quality?
1. Quality modelling, evaluation and improvement:
- Quality modelling and evaluation: We conducted multiple subjective tests, where a group of participants is required to evaluate the quality of a video clip. From the experimental results, we proposed a quality model to predict the user experience. Since 2017, we shifted focus from 2D-video content to Virtual Reality (VR) content. Some experiments will be conducted to find the key influence factors for VR content.
- Quality improvement for video streaming: A lot of quality models were proposed to estimate quality of a streaming session, but a few simple model can support real-time streaming services. In our studies, we proposed a lightweight quality model which considers the video-encoding parameters to fast and efficiently predict the user experience. We have a plan to study a lightweight model to support VR content.
- HTTP adaptive streaming (HAS): This streaming has emerged as a de factor standard for delivering videos over the Internet, since it is firewall-friendly and more cost-effective than conventional RTP streaming. The big challenge of any Internet-based video streaming (including HAS) is network fluctuation, leading to the quality variations. In our studies, we aimed to provide seamless streaming with consistent quality.
- Adaptive streaming of Variable Bitrate videos: This study is to address a question on which type of video encoding should be used.
- Adaptive streaming over HTTP/2: This study aims to improve video quality by leveraging new HTTP/2 features.
- HTTP live streaming: Currently, live streaming over the Internet is very common. We develop streaming services towards to high-quality seamless streaming.
- Streaming of Virtual Reality content: We now focus on the VR content. Because the data size of VR content is significantly larger than the typical Internet connections, HAS of VR content is open issue.
- Energy-efficient streaming for mobile clients: Due to battery limitations of mobile devices, energy saving is very important for any mobile streaming service. We proposed streaming solutions to achieve the tradeoff between the high video quality and high energy saving.
- Home surveillance: Using HTTP adaptive streaming, we offered an efficient and significant cost-effective solution to users. For this project, we created a prototype of a system in laboratory environment.