About this example
This page is an example literature review chapter for a doctoral thesis in electronic engineering or computer science on video compression. The thesis studies how the newest international video coding standard, Versatile Video Coding (VVC), can be deployed efficiently in consumer devices, and the review sets out how the standard developed, what makes it more efficient than its predecessors, what is known about its cost in computation, and how far it has moved from laboratory to real services.
It is an illustrative teaching example, written afresh for this page, not a real student's work. Read each part, then the "What to notice" note beneath it. Do not submit any part of it as your own work; your university's academic integrity rules apply.
The example: scope and approach
2.1 Scope of the review
This chapter reviews the literature on VVC under four headings: the lineage of hybrid video coding standards (2.2), the coding tools that distinguish VVC (2.3), evidence on compression efficiency and complexity (2.4), and implementations and early deployment (2.5). Section 2.6 identifies the gap this thesis addresses. Because standardisation and deployment are documented largely outside academic journals, the review combines peer-reviewed papers with technical reports, preprints and industry accounts of trials, an approach known as a multivocal review (Garousi et al., 2019). Sources were located by database searches and by following references from the standard's overview papers, and each was classified by type so that the weight given to it could be judged (Kitchenham & Charters, 2007).
The example: lineage and coding tools
2.2 From H.264/AVC to VVC
Every major video coding standard since the 1990s has followed the same hybrid design: each picture is divided into blocks, each block is predicted from previously coded data, and only the prediction error is transformed, quantised and entropy coded, with the decoder repeating the prediction to reconstruct the picture. Sullivan and Wiegand (2005) describe how these concepts came together in H.264/AVC, which became the dominant format for broadcast and internet video. Its successor, High Efficiency Video Coding (HEVC), kept the architecture but enlarged and diversified the block structures. VVC was developed jointly by ITU-T and ISO/IEC and approved in July 2020 as ITU-T H.266 and ISO/IEC 23090-3 (Hamidouche et al., 2021).
- 1Partition picture into blocks
- 2Predict each block (intra or inter)
- 3Transform and quantise the residual
- 4Entropy code
- 5Reconstruct and filter (in-loop)
- 6Store as reference for later prediction
2.3 What distinguishes VVC
VVC refines each stage of the hybrid loop rather than replacing it. Its overview paper describes more flexible block partitioning, additional intra- and inter-prediction modes, new transform options and further in-loop filtering, together with features designed for applications its predecessors handled poorly: high-resolution and high-dynamic-range video, adaptive streaming with changes of resolution, and 360-degree immersive video (Bross et al., 2021). The name "versatile" reflects this second aim. For a thesis on consumer deployment, the important point is that efficiency gains come from many tools, each adding a modest saving and some computational cost, rather than from a single new technique.
The example: efficiency, complexity and deployment
2.4 Compression efficiency and complexity
The headline claim for VVC is a bit-rate reduction in the region of 50% compared with HEVC at equal video quality (Bross et al., 2021). Because such figures depend on test material, quality metric and encoder configuration, they should be read with their test conditions. Hamidouche et al. (2021), for example, assessed the contribution of the main tools by switching them off in the reference encoder on 42 ultra-high-definition sequences in the random-access configuration, and reported the loss of efficiency separately for several objective quality metrics. The same review notes that the gains come with greater computational demand, particularly at the encoder (Hamidouche et al., 2021), which is why practical implementations matter.
The choice of metric deserves attention in its own right. Peak signal-to-noise ratio (PSNR) measures pixel-level error and is the traditional basis for comparing codecs, while perceptual metrics such as VMAF and multi-scale structural similarity (MS-SSIM) aim to track what viewers notice. The tool-off experiment reported a larger loss of efficiency under PSNR than under VMAF or MS-SSIM (Hamidouche et al., 2021), which suggests that some tools improve the measured signal more than the perceived picture. For a thesis concerned with consumer viewing, this means that efficiency claims based on PSNR alone may overstate the benefit users experience, and subjective testing or perceptual metrics are needed alongside it.
2.5 Implementations and early deployment
Within months of standardisation, optimised software decoders such as OpenVVC and Fraunhofer's VVdeC were reported to decode in real time, and three trials in 2020 demonstrated end-to-end use: satellite over-the-air broadcasting and over-the-top (internet) delivery in June, and a live 4K over-the-top channel in September (Hamidouche et al., 2021). These accounts show technical feasibility, but most come from the organisations involved and report demonstrations rather than measured performance in use, so they are weighted as practitioner evidence in this review.
| Theme | Main sources | Type of evidence | Weight in this review |
|---|---|---|---|
| Standard design and features | (Bross et al., 2021) | Peer-reviewed overview by the standard's editors | High for what the standard contains |
| Efficiency and tool contributions | (Bross et al., 2021; Hamidouche et al., 2021) | Reference-encoder experiments under common test conditions | High, within the stated conditions |
| Real-time implementation | (Hamidouche et al., 2021) | Developer reports and preprint review | Moderate; independent benchmarks needed |
| Commercial trials | (Hamidouche et al., 2021) | Accounts of demonstrations | Low to moderate; no field measurements |
2.6 Summary and gap
In summary, the sources reviewed agree on three points. VVC keeps the hybrid architecture of its predecessors and gains efficiency through many incremental refinements. Its efficiency advantage over HEVC is substantial under standard test conditions but depends on metric and content. And it moved quickly from standardisation to working decoders and demonstrations, although published evidence on performance in everyday use remains thin.
The literature establishes what VVC can achieve in controlled experiments and that real-time decoding is possible on general-purpose processors. Much less is known about the trade-off between compression gain and energy use on battery-powered consumer devices, measured under realistic viewing conditions. This thesis therefore asks how the choice of VVC decoder configuration affects decoding energy and perceived quality on mobile devices.
How to adapt this to your own study
- Read the key reviews in your area closely, then close them and plan your own structure around your research question.
- Explain shared background once, in your own words, and draw your own diagrams.
- Group the literature by theme and say what each group shows for your study.
- Report every quantitative claim with its source and conditions.
- Decide in advance how you will weigh grey literature, and state it (Garousi et al., 2019).
- Check each reference against the publisher's record before it goes in your list.
- End with a gap that your thesis will address (Boote & Beile, 2005).
Common mistakes in this chapter
- Restructuring a published review and presenting it as your own: this is plagiarism even with citations.
- Reusing published figures without permission or attribution.
- Describing tools one by one without saying why they matter for your question.
- Quoting forecasts or contested estimates as current facts.
- Copying reference lists from another paper, including its errors.
- Treating company announcements as equivalent to peer-reviewed measurement.
| What examiners look for | Where this example does it | Check in your own draft |
|---|---|---|
| Original synthesis | Sections organised around the thesis question | Is the structure yours, not a source's? |
| Clear scope and approach | 2.1 states themes and review type | Have you said how sources were found and weighed? |
| Accurate, attributed claims | Figures given with source and test conditions | Can each number be traced? |
| Critical weighing of evidence | Table 2.1 grades evidence by type | Do you say how strong each source is? |
| Own figures or permitted reuse | Figure 2.1 drawn for this example | Do you have rights to every figure? |
| A gap that motivates the study | 2.6 | Does your gap lead to your research question? |
Key takeaways
- Build the review around your question, not around the structure of one source.
- Draw your own figures and cite the source of every idea (Hamidouche et al., 2021).
- Give each efficiency claim with its test conditions (Bross et al., 2021).
- State how grey literature is found and weighed (Garousi et al., 2019).
- Check every reference before it enters your list.
