1.     Introduction

All popular entertainment services today involve video streaming over HTTP. The tools you have experimented with in the IMS laboratory exercises allow you to transcode video with multiple codecs (via FFmpeg), create DASH manifest files for delivery (via MP4BOX), experience multiple delivery conditions (via a DASH player and the browser’s developer ànetwork settings), and measure video quality (via the Netflix VMAF library).

The goal of this assignment is to get you to explore in more detail:

• video encoding options and challenges,
• HTTP content delivery options and challenges, and
• quality-vs-bitrate-vs-complexity trade-offs

for state-of-the-art internet multimedia systems and compile results and an assignment report based on your assignment proposal and related experimentation.

2.     Project Specification

All IMS assignments are individual (one per student). Within the assignment, you are first expected to familiarise yourself with the tools and methods described above and demonstrated during the laboratory sessions. You are then expected to read about the latest developments on media streaming over IP for entertainment, surveillance, gaming or other media services by surveying related top-tier technical journals, conferences and reputable online resources (e.g., blogs of technology companies like Netflix, Facebook, Google, etc.).

Assignment summary:

• Based on:
  • the IMS lectures;
  • the IMS lab session;
  • reading of research literature and understanding of media delivery services, select a topic from the list of topics of the “Requirements” section;

• Analyse and design your own solution on this topic with emphasis on actual experimentation with FFmpeg, MP4BOX, VMAF, DASH players (or similar tools of your choice), and prepare a short pitch of your idea to be presented during the tutorial session; this is similar to presenting your new idea to technology venture capitalists: it must explain why the idea is new (within the context of this module) and important for media streaming, and how it is going to be delivered by you during the timeframe of the IMS assignment;
• Deploy this concept and measure performance or feature improvement (or do a parametric study) by devising appropriate quantitative metrics;
• Compile the final report of your project and any supplementary evidence (source code, experiments, quantitative user survey results, ).

Requirements:

• You must test your idea quantitatively, by devising appropriate distortion, delay, bandwidth and/or quality measurement tests (or other metrics that fit your domain) and quantify the performance in your final report in comparison to the performance obtained with some baseline solution, or other solutions in the research literature, or in proprietary or open-source applications. Specific evaluation criteria should be presented and discussed by each student in their tutorial presentation, where the module organiser will provide feedback on their suitability for the specific task addressed by the student’s

(ii)    Your idea should be within the scope of one the following topics:

• Study of bitrate-quality-complexity of several video encoders supported by FFmpeg (or libav or similar), such as H.264/AVC, HEVC, VP9 or even AV1 for specialized types of media (e.g., surveillance content, other types of content such as medical or gaming content).

3.     Goals of the project

This project has several goals:

• To further develop your skills in programming and the Linux operating system
• To further develop your skills on network (client-server) programming and real- time constraints
• To read state-of-the-art research papers on multimedia streaming, content encoding, quality assessment, and understand the challenges of this research area
• To demonstrate your R&D skills by showing the deployment of novel concepts from the literature within a practical framework
• To develop your skills in presenting technical ideas within the context of a short technical pitch to potential industrial funders or investors

4.     Evaluation

The final mark breakdown is:

10% Ten-minute presentation on the planned R&D work during the IMS  tutorial.  The envisaged idea to be deployed is validated there by the Module Organiser and comments are provided for the remaining part of the work.

60% Final report including abstract, introduction, algorithmic and implementation description, experimental validation, conclusion and references. Your final report must be in the format of an IEEE-style paper. It must not exceed 14 double-column single-spaced pages including figures, tables and references: https://journals.ieeeauthorcenter.ieee.org/create-your-ieee-article/authoring-tools- and-templates/ieee-article-templates/templates-for-transactions/

Your code and instructions on how to set up the environment, tools and code must be made available in a private GitHub repository (or similar) for inspection during the assignment marking. In addition, as a minimum, your experimental validation must include experimentation and graphs/tables with your own measurements and associated analysis (e.g., averages or other statistics appropriate to quantify quality, latency, bitrate, etc., as appropriate to your assignment goals) and cannot be a survey paper of external measurements.

30% Practical deployment in an automated manner. Full marks for this part will be given to assignments that port and validate the idea within a practical test-bed that can be used by third parties with setup and usage instructions. This is similar to “early prototype” stage in industry. This can take the form of a github repository for the setup and installation of tools and execution of the code, with provided README and examples of usage, along with graphs or other related material.

5.     Further Reading

Beyond the module material available on Moodle, consultation on the latest developments on IP-based media streaming can be made via the following journals and conferences (available from IEEExplore and Elsevier Science via UCL subscription):

• IEEE Transactions on Multimedia, IEEE Trans. on Mobile Computing, IEEE/ACM Trans. on Networking, IEEE Transactions on Circuits and Systems for Video Technology, IEEE Journal on Selected Areas in Communications, IEEE Transactions on Image Processing, Elsevier Signal Processing: Image Communications,
• IEEE INFOCOM and IEEE SIGCOMM conferences: http://www.ieee-infocom.org/, http://conferences.sigcomm.org/sigcomm/
• The ffmpeg library: http://www.ffmpeg.org/
• Visual Quality Standardization meetings from VQEG: https://www.its.bldrdoc.gov/vqeg/meetings-home.aspx
• Technology blogs from video teams of Google, Netflix, Facebook, and others

6.     Other Actions

As a side target, which is not part of the project mark, the best IMS assignments may be invited to go through an integration phase (which will be guided by the IMS module manager) in order to be incorporated in a future open-source testbed (see here for such a past example). It is emphasised however that this will be done on a voluntary basis and has no impact on the IMS module mark.