pith. sign in

arxiv: 2412.19446 · v2 · submitted 2024-12-27 · 💻 cs.DC · cs.ET· cs.GR· cs.MM

Stimpack: An Adaptive Rendering Optimization System for Scalable Cloud Gaming

Jin Heo , Vic Wang , Ketan Bhardwaj , Ada Gavrilovska This is my paper

Pith reviewed 2026-05-23 07:35 UTC · model grok-4.3

classification 💻 cs.DC cs.ETcs.GRcs.MM
keywords cloud gamingrendering optimizationadaptive qualityresource efficiencyperceived qualitylossy compressionedge serversmulti-user systems
0
0 comments X

The pith

When network compression degrades rendered game streams, extra rendering quality brings little user benefit and wastes server resources.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that in cloud gaming, where games are rendered on remote servers and streamed to users, the benefits of high-quality rendering are often lost due to lossy compression on the network. This leads to inefficient use of computing resources at edge servers that have limited capacity. Stimpack addresses this by adaptively choosing rendering quality based on a measure of how efficiently those resources translate into perceived quality. By doing so, it improves overall service quality and allows the same hardware to support more simultaneous users. A user study confirms better experiences, making this relevant for scaling interactive cloud services.

Core claim

High-quality rendering becomes ineffective for improving user-perceived quality when the content passes through lossy compression in network delivery. Stimpack therefore adaptively optimizes rendering quality by using a mechanism to quantify the efficiency of resource usage, which maximizes overall system utility in multi-user cloud gaming scenarios. Evaluations of the open-sourced system demonstrate up to 24% higher service quality and the ability to serve twice as many users with the same resources.

What carries the argument

A mechanism that quantifies the efficiency of resource usage to balance server-side rendering costs against user-perceived quality and maximize system utility.

Load-bearing premise

The resource efficiency quantification accurately predicts user-perceived quality improvements across various games, networks, and user loads without adding much overhead.

What would settle it

An experiment where Stimpack's adaptive rendering fails to increase the number of supported users or user satisfaction scores compared to a baseline of always-maximum quality rendering would falsify the claim.

Figures

Figures reproduced from arXiv: 2412.19446 by Ada Gavrilovska, Jin Heo, Ketan Bhardwaj, Vic Wang.

Figure 2
Figure 2. Figure 2: General architecture of cloud gaming We demonstrate the effectiveness of Adrenaline through the evaluations and user study with two off-the-shelf games and a Adrenaline plugin for Unreal Engine [8]. Overall, this paper makes the following contributions: • We identify the opportunity to improve the resource effi￾ciency of edge game servers: the visual quality loss due to compression varies by rendering qual… view at source ↗
Figure 3
Figure 3. Figure 3: The visualization of multi-application rendering on a GPU the number of users served above their FPS threshold, while minimizing the visual quality loss. Observation for Adrenaline. The insight is that the opera￾tion on the server may not be fully effective on the user side. The key observation for Adrenaline is that the quality loss due to compression is more significant for frames of higher RQ than frame… view at source ↗
Figure 4
Figure 4. Figure 4: The FPS and frame quality measurements of the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: The screenshots of 3D scenes used for predicting the frame quality with given RQs and QPs 4.2 User-side Visual Quality Prediction Adrenaline’s design is based on the premise that server-side prediction of user-side visual quality is achievable using given RQs and QPs. To realize effective prediction, we address two key questions: the design of the prediction method and the feasibility of accurate server-si… view at source ↗
Figure 7
Figure 7. Figure 7: The FPS traces with and without the backoff instance, the user’s FPS is above the threshold with the current RQ but can become lower with the one-level-higher RQ. Such frequent and oscillatory RQ changes can hugely de￾grade the user’s gaming experience. Changing RQ presents overhead because the rendering contents on GPU should be reloaded and reconfigured for the new RQ; the FPS drops (downward spikes) in … view at source ↗
Figure 8
Figure 8. Figure 8: The screenshots of sample games Experiment Scenarios. We designed three experimental scenarios for evaluation, each involving a server accommodat￾ing up to 6 users divided into two groups (A and B) of 3 users each. Within each group, users have different network condi￾tions: Good (G), Fair (F), and Poor (P) with corresponding QPs as outlined in [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: The FPS traces of Village Shooter with and without Adrenaline. The dashed line is the FPS threshold (30), and the blue background color indicates all users are served with higher FPS than the threshold, while the red indicates the opposite. 0 20 40 60 80 100 120 140 Time (s) 0 20 40 60 80 100 120 Frame Per Second (FPS) User1A (G) User2A (F) User3A (P) User4B (G) User5B (F) User6B (P) (a) The highest-only c… view at source ↗
Figure 10
Figure 10. Figure 10: The FPS traces of Mountain Hiker with and without Adrenaline. The dashed line is the FPS threshold (30), and the blue background color indicates all users are served with higher FPS than the threshold, while the red indicates the opposite. 0 20 40 60 80 100 120 140 Time (s) 0 20 40 60 80 100 GPU Utilization (%) Adrenaline The highest-only case (a) GPU usage with Village Shooter (matched to [PITH_FULL_IMA… view at source ↗
Figure 11
Figure 11. Figure 11: The server’s GPU usage of Adrenaline and the highest-only case than demoting users with better network conditions. However, User1A and User2A are further demoted to Medium as their FPS remains below the threshold. Then, the under-threshold FPS issue is resolved at 80 seconds. After ensuring all users are with above-threshold FPS, Adrenaline begins promoting users’ RQs in subsequent op￾timization rounds. G… view at source ↗
Figure 12
Figure 12. Figure 12: Service quality score comparison of Adrenaline and the other baselines. The dashed line corresponds to the score of [PITH_FULL_IMAGE:figures/full_fig_p011_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: The winning rate of Adrenaline against the other base￾lines from the user study under Scenario 3 (Mixed-game Case) [PITH_FULL_IMAGE:figures/full_fig_p012_13.png] view at source ↗
read the original abstract

In distributed multimedia applications, content is often delivered to users in a degraded form due to network-induced lossy compression. Real-time and interactive use cases like cloud gaming, which render content on the fly, require low latency and are hosted at resource-constrained edge servers. We present a new insight: when rendered content is delivered over a network with lossy compression, high-quality rendering can be ineffective in improving user-perceived quality, leading to a poor return on computing resources. Leveraging this observation, we built Stimpack, a novel system that adaptively optimizes game rendering quality by balancing server-side rendering costs against user-perceived quality. The system uses a mechanism that quantifies the efficiency of resource usage to maximize overall system utility in multi-user scenarios. Our open-sourced implementation and extensive evaluations show that Stimpack achieves up to 24% higher service quality and serves twice as many users with the same resources compared to baselines. A user study further validates that Stimpack provides a measurably better user experience.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The paper claims that when rendered content is delivered over a network with lossy compression, high-quality rendering can be ineffective in improving user-perceived quality, leading to a poor return on computing resources. It presents Stimpack, a system that adaptively optimizes game rendering quality by balancing server-side rendering costs against user-perceived quality using a mechanism that quantifies the efficiency of resource usage to maximize overall system utility in multi-user scenarios. Evaluations show up to 24% higher service quality and serving twice as many users with the same resources compared to baselines, supported by open-sourced code and a user study.

Significance. If the efficiency quantification mechanism accurately predicts user-perceived quality across various games, network conditions, and multi-user scenarios without significant overhead, this work could substantially improve the scalability of cloud gaming by optimizing resource allocation at edge servers. The open-sourcing of the implementation and the inclusion of a user study are strengths that support reproducibility and practical applicability.

major comments (3)
  1. [§3] The efficiency of resource usage mechanism is central to the contribution, but the manuscript provides no detail on its derivation, cross-validation, error bounds, or sensitivity to game genre, codec, or loss patterns. This is load-bearing for the claim that it reliably maps rendering cost to QoE.
  2. [§5] The reported gains (24% higher service quality, 2x users) are presented without full details on baselines, statistical analysis, or the specific conditions tested, making it impossible to confirm robustness or rule out post-hoc choices.
  3. [User study] The user study is cited to validate better user experience, but lacks information on participant count, methodology, or statistical significance, which is necessary to support the practical claims.
minor comments (1)
  1. [Abstract] Clarify the definition of 'service quality' and how it is measured, as it is used in the quantitative claims.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We agree that additional details are needed to strengthen the presentation of the efficiency mechanism, evaluation results, and user study. We will revise the manuscript accordingly to address these points.

read point-by-point responses

  1. Referee: [§3] The efficiency of resource usage mechanism is central to the contribution, but the manuscript provides no detail on its derivation, cross-validation, error bounds, or sensitivity to game genre, codec, or loss patterns. This is load-bearing for the claim that it reliably maps rendering cost to QoE.

    Authors: We agree that the manuscript would benefit from expanded details on this central mechanism. In the revised version, we will add a new subsection in §3 that includes: (1) the full mathematical derivation of the efficiency metric, (2) cross-validation results on held-out data, (3) reported error bounds, and (4) sensitivity analysis across game genres, codecs, and loss patterns. This will directly support the QoE mapping claim. revision: yes

  2. Referee: [§5] The reported gains (24% higher service quality, 2x users) are presented without full details on baselines, statistical analysis, or the specific conditions tested, making it impossible to confirm robustness or rule out post-hoc choices.

    Authors: We acknowledge the need for greater transparency. In the revision of §5, we will include: complete specifications of all baselines and their configurations, full statistical analysis (means, variances, confidence intervals, and any significance tests), and exhaustive details on the tested conditions, games, network parameters, and resource settings. We will also explicitly state that the reported gains reflect the primary, pre-specified evaluation scenarios. revision: yes

  3. Referee: [User study] The user study is cited to validate better user experience, but lacks information on participant count, methodology, or statistical significance, which is necessary to support the practical claims.

    Authors: We agree that the user study section requires more detail to support the claims. In the revised manuscript, we will expand the user study description to report the exact participant count, a complete methodology (recruitment, tasks, rating scales, and procedure), and the statistical analysis including significance testing and effect sizes. revision: yes

Circularity Check

0 steps flagged

No circularity: mechanism presented as independent contribution with external validation

full rationale

The provided abstract and context describe Stimpack's efficiency quantification mechanism and multi-user utility maximization as a novel system contribution, evaluated via open-sourced implementation, extensive experiments, and a user study showing 24% quality gain and 2x users served. No equations, fitted parameters renamed as predictions, or self-citation chains are quoted that reduce the central claims to inputs by construction. The derivation chain is self-contained against external benchmarks (user study, baselines), satisfying the default expectation of no significant circularity.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central claim rests on the ability to quantify rendering-to-perceived-quality efficiency and on the assumption that this quantification can be used to optimize multi-user utility; these elements are introduced by the paper rather than derived from prior literature.

free parameters (1)
  • efficiency thresholds or utility weighting parameters
    Parameters that control when to lower rendering quality and how to allocate saved resources across users; likely tuned during system design or evaluation.
axioms (1)
  • domain assumption User-perceived quality after lossy compression can be reliably estimated from rendering settings and network conditions
    Invoked to justify the adaptive decisions and the efficiency metric.
invented entities (1)
  • efficiency of resource usage mechanism no independent evidence
    purpose: Quantifies how effectively server rendering resources translate into delivered user quality under compression
    New construct introduced to drive the optimization decisions.

pith-pipeline@v0.9.0 · 5719 in / 1356 out tokens · 27460 ms · 2026-05-23T07:35:42.645047+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    Delay- aware resource provisioning for cost-efficient cloud gam- ing

    Mohaddeseh Basiri and Abbas Rasoolzadegan. Delay- aware resource provisioning for cost-efficient cloud gam- ing. IEEE Transactions on Circuits and Systems for Video Technology, 28(4):972–983, 2016

    work page 2016

  2. [2]

    T-gaming: A cost- efficient cloud gaming system at scale

    Hao Chen, Xu Zhang, Yiling Xu, Ju Ren, Jingtao Fan, Zhan Ma, and Wenjun Zhang. T-gaming: A cost- efficient cloud gaming system at scale. IEEE Transac- tions on Parallel and Distributed Systems, 30(12):2849– 2865, 2019

    work page 2019

  3. [3]

    The brewing storm in cloud gam- ing: A measurement study on cloud to end-user latency

    Sharon Choy, Bernard Wong, Gwendal Simon, and Catherine Rosenberg. The brewing storm in cloud gam- ing: A measurement study on cloud to end-user latency. In 2012 11th Annual Workshop on Network and Systems Support for Games (NetGames), pages 1–6. IEEE, 2012

    work page 2012

  4. [4]

    The server allocation problem for session- based multiplayer cloud gaming

    Yunhua Deng, Yusen Li, Ronald Seet, Xueyan Tang, and Wentong Cai. The server allocation problem for session- based multiplayer cloud gaming. IEEE Transactions on Multimedia, 20(5):1233–1245, 2017

    work page 2017

  5. [5]

    Server allocation for multiplayer cloud gaming

    Yunhua Deng, Yusen Li, Xueyan Tang, and Wentong Cai. Server allocation for multiplayer cloud gaming. In Proceedings of the 24th ACM international conference on Multimedia, pages 918–927, 2016

    work page 2016

  6. [6]

    Godot engine - free and open source 2d and 3d game engine

    Godot Foundation. Godot engine - free and open source 2d and 3d game engine. https://godotengine.org/, 2024

    work page 2024

  7. [7]

    Medieval houses modular vol 2

    Game Asset Factory. Medieval houses modular vol 2. https://www.unrealengine.com/marketplace/en- US/item/96a5ab5b99ef4c1ab36a1bbc5a36074a, 2024

    work page 2024

  8. [8]

    Unreal engine: The most powerful real-time 3d creation platform

    Epic Games. Unreal engine: The most powerful real-time 3d creation platform. https://www.unrealengine.com, 2024

    work page 2024

  9. [9]

    Cost- efficient and quality of experience-aware provisioning of virtual machines for multiplayer cloud gaming in geographically distributed data centers

    Yongqiang Gao, Lin Wang, and Jiantao Zhou. Cost- efficient and quality of experience-aware provisioning of virtual machines for multiplayer cloud gaming in geographically distributed data centers. IEEE Access, 7:142574–142585, 2019

    work page 2019

  10. [10]

    Webrtc, real-time communication for the web

    Google. Webrtc, real-time communication for the web. https://webrtc.org/, 2024

    work page 2024

  11. [11]

    An analysis of cloud gaming platforms behavior under different network constraints

    Philippe Graff, Xavier Marchal, Thibault Cholez, Stéphane Tuffin, Bertrand Mathieu, and Olivier Festor. An analysis of cloud gaming platforms behavior under different network constraints. In 2021 17th Interna- tional Conference on Network and Service Management (CNSM), pages 551–557. IEEE, 2021

    work page 2021

  12. [12]

    djay: Enabling high-density multi-tenancy for cloud gaming servers with dynamic cost-benefit gpu load balancing

    Sergey Grizan, David Chu, Alec Wolman, and Roger Wattenhofer. djay: Enabling high-density multi-tenancy for cloud gaming servers with dynamic cost-benefit gpu load balancing. In Proceedings of the sixth ACM sym- posium on cloud computing, pages 58–70, 2015

    work page 2015

  13. [13]

    Energy-efficient sla guarantees for virtualized gpu in cloud gaming

    Haibing Guan, Jianguo Yao, Zhengwei Qi, and Runze Wang. Energy-efficient sla guarantees for virtualized gpu in cloud gaming. IEEE Transactions on Parallel and Distributed Systems, 26(9):2434–2443, 2014

    work page 2014

  14. [14]

    Nvidia grid: Graphics accelerated vdi with the visual performance of a workstation

    Alex Herrera. Nvidia grid: Graphics accelerated vdi with the visual performance of a workstation. Nvidia Corp, pages 1–18, 2014. 13

    work page 2014

  15. [15]

    A google congestion control algorithm for real-time communication on the world wide web

    Stefan Holmer, H Lundin, G Carlucci, L De Cicco, and S Mascolo. A google congestion control algorithm for real-time communication on the world wide web. IETF Draft, June, 2015

    work page 2015

  16. [16]

    En- abling adaptive cloud gaming in an open-source cloud gaming platform

    Hua-Jun Hong, Chih-Fan Hsu, Tsung-Han Tsai, Chun- Ying Huang, Kuan-Ta Chen, and Cheng-Hsin Hsu. En- abling adaptive cloud gaming in an open-source cloud gaming platform. IEEE Transactions on Circuits and Systems for Video Technology, 25(12):2078–2091, 2015

    work page 2078

  17. [17]

    Zeromq, an open-source universal messaging library

    iMatix. Zeromq, an open-source universal messaging library. https://zeromq.org/, 2024

    work page 2024

  18. [18]

    Towards minimizing resource usage with qos guarantee in cloud gaming

    Yusen Li, Changjian Zhao, Xueyan Tang, Wentong Cai, Xiaoguang Liu, Gang Wang, and Xiaoli Gong. Towards minimizing resource usage with qos guarantee in cloud gaming. IEEE Transactions on Parallel and Distributed Systems, 32(2):426–440, 2020

    work page 2020

  19. [19]

    What is frame rate? https://www.adobe.com/creativecloud/video/discover/frame- rate.html, 2024

    Margaret Kurniawan, Hiroshi Hara. What is frame rate? https://www.adobe.com/creativecloud/video/discover/frame- rate.html, 2024

    work page 2024

  20. [20]

    Xbox cloud gaming

    Microsoft. Xbox cloud gaming. https://www.xbox.com/en-us/play, 2024

    work page 2024

  21. [21]

    Geforce now - the next generation in cloud gaming

    Nvidia Corporation. Geforce now - the next generation in cloud gaming. https://www.nvidia.com/en-us/geforce- now/, 2024

    work page 2024

  22. [22]

    Percep- tual quality assessment of video considering both frame rate and quantization artifacts

    Yen-Fu Ou, Zhan Ma, Tao Liu, and Yao Wang. Percep- tual quality assessment of video considering both frame rate and quantization artifacts. IEEE Transactions on Circuits and Systems for Video Technology, 21(3):286– 298, 2010

    work page 2010

  23. [23]

    Scikit-learn: Machine learning in python

    Fabian Pedregosa, Gaël Varoquaux, Alexandre Gram- fort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vin- cent Dubourg, et al. Scikit-learn: Machine learning in python. the Journal of machine Learning research, 12:2825–2830, 2011

    work page 2011

  24. [24]

    Vgris: Virtualized gpu resource isolation and scheduling in cloud gaming

    Zhengwei Qi, Jianguo Yao, Chao Zhang, Miao Yu, Zhizhou Yang, and Haibing Guan. Vgris: Virtualized gpu resource isolation and scheduling in cloud gaming. ACM Transactions on Architecture and Code Optimiza- tion (TACO), 11(2):1–25, 2014

    work page 2014

  25. [25]

    The logarithmic nature of qoe and the role of the weber-fechner law in qoe assessment

    Peter Reichl, Sebastian Egger, Raimund Schatz, and Alessandro D’Alconzo. The logarithmic nature of qoe and the role of the weber-fechner law in qoe assessment. In 2010 IEEE International Conference on Communica- tions, pages 1–5. IEEE, 2010

    work page 2010

  26. [26]

    A latency compensation technique based on game characteristics to mitigate the influence of delay on cloud gaming quality of experience

    Saeed Shafiee Sabet, Steven Schmidt, Saman Zad- tootaghaj, Babak Naderi, Carsten Griwodz, and Sebas- tian Möller. A latency compensation technique based on game characteristics to mitigate the influence of delay on cloud gaming quality of experience. In Proceedings of the 11th ACM Multimedia Systems Conference, pages 15–25, 2020

    work page 2020

  27. [27]

    Cloud gaming: architecture and performance.IEEE network, 27(4):16–21, 2013

    Ryan Shea, Jiangchuan Liu, Edith C-H Ngai, and Yong Cui. Cloud gaming: architecture and performance.IEEE network, 27(4):16–21, 2013

    work page 2013

  28. [28]

    Qoe-aware resource allocation for multiple cloud gam- ing users sharing a bottleneck link

    Ivan Slivar, Lea Skorin-Kapov, and Mirko Suznjevic. Qoe-aware resource allocation for multiple cloud gam- ing users sharing a bottleneck link. In 2019 22nd con- ference on innovation in clouds, internet and networks and workshops (ICIN), pages 118–123. IEEE, 2019

    work page 2019

  29. [29]

    Mobile cloud gaming: Issues and challenges

    Omar Soliman, Abdelmounaam Rezgui, Hamdy Soli- man, and Najib Manea. Mobile cloud gaming: Issues and challenges. In International Conference on Mobile Web and Information Systems, pages 121–128. Springer, 2013

    work page 2013

  30. [30]

    Stylized fantasy provencal

    StylArts. Stylized fantasy provencal. https://www.unrealengine.com/marketplace/en- US/item/f0f978cdc78348b5b5d0b1c5a11e6d53, 2024

    work page 2024

  31. [31]

    Overview of the high efficiency video coding (hevc) standard

    Gary J Sullivan, Jens-Rainer Ohm, Woo-Jin Han, and Thomas Wiegand. Overview of the high efficiency video coding (hevc) standard. IEEE Transactions on circuits and systems for video technology , 22(12):1649–1668, 2012

    work page 2012

  32. [32]

    Unity real-time development plat- form

    Unity Technologies. Unity real-time development plat- form. https://unity.com, 2024

    work page 2024

  33. [33]

    Adaptive mobile cloud computing to enable rich mobile multimedia applica- tions

    Shaoxuan Wang and Sujit Dey. Adaptive mobile cloud computing to enable rich mobile multimedia applica- tions. IEEE Transactions on Multimedia , 15(4):870– 883, 2013

    work page 2013

  34. [34]

    Overview of the h

    Thomas Wiegand, Gary J Sullivan, Gisle Bjontegaard, and Ajay Luthra. Overview of the h. 264/avc video coding standard. IEEE Transactions on circuits and systems for video technology, 13(7):560–576, 2003

    work page 2003

  35. [35]

    Sara-sdn: state aware re- source allocation in sdn to improve qoe in cloud gaming

    Mehrdad Hosseinnejad Yami, Farhad Pakdaman, and Mahmoud Reza Hashemi. Sara-sdn: state aware re- source allocation in sdn to improve qoe in cloud gaming. In Proceedings of the 25th ACM Workshop on Packet Video, pages 8–14, 2020

    work page 2020

  36. [36]

    vgasa: Adaptive scheduling algorithm of virtualized gpu resource in cloud gaming.IEEE Transac- tions on Parallel and Distributed Systems, 25(11):3036– 3045, 2013

    Chao Zhang, Jianguo Yao, Zhengwei Qi, Miao Yu, and Haibing Guan. vgasa: Adaptive scheduling algorithm of virtualized gpu resource in cloud gaming.IEEE Transac- tions on Parallel and Distributed Systems, 25(11):3036– 3045, 2013. 14

    work page 2013

  37. [37]

    Under the hood: Meta’s cloud gaming infrastructure

    Qunshu Zhang and Xiaoxing Zhu. Under the hood: Meta’s cloud gaming infrastructure. https://engineering.fb.com/2022/06/09/web/cloud- gaming-infrastructure/, 2022

    work page 2022

  38. [38]

    Fine-grained scheduling in cloud gam- ing on heterogeneous cpu-gpu clusters

    Wei Zhang, Xiaofei Liao, Peng Li, Hai Jin, Li Lin, and Bing Bing Zhou. Fine-grained scheduling in cloud gam- ing on heterogeneous cpu-gpu clusters. IEEE Network, 32(1):172–178, 2017

    work page 2017

  39. [39]

    A cloud gaming system based on user-level virtu- alization and its resource scheduling.IEEE Transactions on Parallel and Distributed Systems, 27(5):1239–1252, 2015

    Youhui Zhang, Peng Qu, Jiang Cihang, and Weimin Zheng. A cloud gaming system based on user-level virtu- alization and its resource scheduling.IEEE Transactions on Parallel and Distributed Systems, 27(5):1239–1252, 2015

    work page 2015

  40. [40]

    Toward a practical perceptual video quality metric

    Ioannis Katsavounidis Anush Moorthy Zhi Li, Anne Aaron and Megha Manohara. Toward a practical perceptual video quality metric. The Netflix Tech Blog, 6, 2016. 15

    work page 2016