SHF: Small: Turning Visual Noise into Hardware Efficiency: Viewer-Aware Energy-Quality Adaptive Mobile Video Storage

SHF：小：将视觉噪声转化为硬件效率：观看者感知的能源质量自适应移动视频存储

• 批准号: 1855706
• 负责人: Na Gong
• 金额: $ 30万
• 依托单位: University of South Alabama
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-11-01 至 2022-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1855706&HistoricalAwards=false
• 关键词: SHF; Small; Turning; Visual; Noise

Mobile devices, such as smart phones, are being increasingly utilized for watching videos, since they can be conveniently used for this purpose anywhere anytime, such as commuting on a subway or train, sitting in a waiting room, or lounging at home. Due to the large data size and intensive computation, video processing requires frequent memory access that consumes a large amount of power, limiting battery life and frustrating mobile users. On one hand, memory designers are focusing on hardware-level power-optimization techniques without considering how hardware performance influences viewers' actual experience. On the other hand, the human visual system is limited in its ability to detect subtle degradations in image quality; for example, under conditions of high ambient illumination, such as outdoors in direct sunlight, the veiling luminance (i.e., glare) on the screen of a mobile device can effectively mask imperfections in the image, so that under these circumstances a video can be rendered in lower than full quality without the viewer being able to detect any difference. This isolation between hardware design and viewer experience significantly increases hardware implementation overhead due to overly pessimistic design margins. This project integrates viewer-awareness and hardware adaptation to achieve power optimization without degrading video quality, as perceived by users. The results of this project will impact both basic research on hardware design and human vision, and provide critical viewer awareness data from human subjects, which can be used to engineer better video rendering for increased battery life on mobile devices. The project will directly involve undergraduate and graduate students, including females and Native Americans, in interdisciplinary research. Developing a viewer-aware mobile video-memory solution has proven to be a very challenging problem due to (i) complex existing viewer-experience models; (ii) memory modules without runtime adaptation; and (iii) the difficulty of viewer-experience analysis for hardware designers. This project addresses the problem by (i) focusing on the most influential viewing-context factor impacting viewer experience - ambient luminance; (ii) proposing novel methodologies for adaptive hardware design; and (iii) integrating a unique combination of expertise from the investigators, ranging from psychology to Integrated Circuit design and embedded systems. Specifically, this project will (i) experimentally and mathematically connect viewer experience, ambient illuminance, and memory performance; (ii) develop energy-quality adaptive hardware that can adjust memory usage based on ambient luminance so as to reduce power usage without impacting viewer experience; and (iii) design a mobile video system to fully evaluate the effectiveness of the developed methodologies.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

移动设备，如智能手机，越来越多地被用于观看视频，因为它们可以方便地随时随地用于观看视频，例如在地铁或火车上上下班，坐在候车室，或者在家闲逛。由于庞大的数据量和密集的计算，视频处理需要频繁的内存访问，这消耗了大量的电力，限制了电池寿命，让移动用户感到沮丧。一方面，内存设计人员专注于硬件级的功率优化技术，而不考虑硬件性能如何影响观看者的实际体验。另一方面，人类视觉系统在检测图像质量细微退化的能力上是有限的；例如，在高环境照度的条件下，例如在阳光直射的户外，移动设备屏幕上的遮蔽亮度（即眩光）可以有效地掩盖图像中的缺陷，因此在这些情况下，视频可以以低于全质量的方式呈现，而观看者无法检测到任何差异。由于过于悲观的设计余量，硬件设计和查看器体验之间的这种隔离显著增加了硬件实现开销。该项目集成了观众感知和硬件适应，在不降低用户感知的视频质量的情况下实现功耗优化。该项目的结果将影响硬件设计和人类视觉的基础研究，并提供来自人类受试者的关键观众感知数据，这些数据可用于设计更好的视频渲染，以延长移动设备的电池寿命。该项目将直接涉及本科生和研究生，包括女性和印第安人，进行跨学科研究。开发一个观众感知的移动视频存储器解决方案已经被证明是一个非常具有挑战性的问题，因为(i)复杂的现有观众体验模型；（ii）没有运行时适应性的内存模块；（三）硬件设计者对观众体验分析的难度。本项目通过以下方式解决了这个问题：(i)关注影响观众体验的最具影响力的观看环境因素——环境亮度；（ii）提出自适应硬件设计的新方法；（iii）整合调查者独特的专业知识组合，从心理学到集成电路设计和嵌入式系统。具体来说，该项目将(i)通过实验和数学方法将观众体验、环境照度和记忆性能联系起来；（ii）开发能够根据环境亮度调整内存使用的节能自适应硬件，从而在不影响观看者体验的情况下减少功耗；（iii）设计一个移动视频系统，以充分评估所开发方法的有效性。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Flexible Low-Cost Power-Efficient Video Memory With ECC-Adaptation

具有 ECC 适配功能的灵活低成本高能效视频内存

• DOI: 10.1109/tvlsi.2021.3098533
• 发表时间: 2021
• 期刊: IEEE Transactions on Very Large Scale Integration (VLSI
• 作者: Das, Hritom;Haidous, Ali Ahmad;Smith, Scott C.;Gong, Na
• 通讯作者: Gong, Na

Content-Adaptable ROI-Aware Video Storage for Power-Quality Scalable Mobile Streaming

• DOI: 10.1109/access.2022.3156274
• 发表时间: 2022
• 期刊: IEEE Access
• 影响因子: 3.9
• 作者: Ali Ahmad Haidous;W. Oswald;Hritom Das;Na Gong

Application-Aware Quality-Energy Optimization: Mathematical Models Enabled Simultaneous Quality and Energy-Sensitive Optimal Memory Design

应用感知的质量-能量优化：数学模型支持同时进行质量和能量敏感的最佳存储器设计

• DOI: 10.1109/tsusc.2020.2999882
• 发表时间: 2021
• 期刊: IEEE Transactions on Sustainable Computing
• 影响因子: 3.9
• 作者: Xu, Yiwen;Das, Hritom;Gong, Na
• 通讯作者: Gong, Na

Content-Adaptive Memory for Viewer-Aware Energy-Quality Scalable Mobile Video Systems

• DOI: 10.1109/access.2019.2908997
• 发表时间: 2019
• 期刊: IEEE Access
• 影响因子: 3.9
• 作者: J. Edstrom;Y. Gong;Ali Ahmad Haidous;Brittney Humphrey;M. McCourt;Yiwen Xu;Jinhui Wang;Na Gong

Memory Optimization for Energy-Efficient Differentially Private Deep Learning

• DOI: 10.1109/tvlsi.2019.2946128
• 发表时间: 2020-02
• 期刊: IEEE Transactions on Very Large Scale Integration (VLSI) Systems
• 影响因子: 2.8
• 作者: J. Edstrom;Hritom Das;Yiwen Xu;Na Gong