RI: Tera-Pixels: Using High-Resolution Pervasive Displays to Transform Collaboration and Teaching

该项目旨在为北卡罗来纳大学计算机科学系的办公室、教室和公共空间配备多投影仪显示器，这些显示器作为单个广域逻辑计算机显示器运行，取代这些空间中现有的传统基于CRT的显示器。这一基础设施将使计算机图形学、网络、多媒体网络、协作系统和人机交互等领域的基础研究成为可能。该研究将包括一个集成的计算机系统研究工作到未来的显示环境和大规模的使用实验，以评估我们将构建的环境的影响。作为一个计算机系统的研究项目，我们寻求探索新的技术和系统架构，以支持高分辨率的分布式计算机显示器。这个问题是计算机工业和计算机科学与工程领域未来增长和发展的基础。现代计算机显示架构是20世纪40年代老式电视标准的直接结果，这些标准在今天的数字世界中已经过时。光栅显示器以恒定速率从共享处理器内存块中定期刷新的概念是使用阴极射线管（CRT）作为早期计算机显示器的直接结果。此外，这种“光栅扫描”方法假定具有最小状态和低级接口的哑设备。我们相信在显示器中提供本地智能将使显示设备的概念与驱动它的特定计算机分离。我们设想将显示器视为独立于向其呈现数据的特定计算平台而运行的共享资源。我们的方法提供了无缝和对称的访问显示器从任何主机，并支持多个主机同时连接到一个显示器。我们建议通过显示信息的有效传输，并通过使用有效的传输和网络级机制的服务质量来实现这一目标。为显示接口开发更高级别的语义将提高带宽利用率。高级接口和本地智能的结合还将改善交互，使内容流的呈现适应特定显示平台的能力。换句话说，智能显示器将根据显示器是在低分辨率移动终端上还是在墙壁大小的会议室显示器上来定制其呈现。同样，它也会根据可用的通信带宽来调整显示。在这项计算机系统研究中，我们提出了四项新的倡议。我们将研究和开发新的智能显示架构，能够克服传统的帧缓冲器架构的局限性。我们将开发更高级别和更高带宽效率的协议，用于将数据传输到可视显示器。我们将扩展数字显示互连，使其更类似于实时网络，从而提高其实用性和灵活性。最后，我们将开发虚拟化显示所需的同步和视觉上下文切换功能，以便它们可以用作分布式资源。 第二，作为一个大规模的实验，在办公室和教室使用广域显示器，我们试图研究我们提出的显示架构对个人工作习惯和生产力，教学，学生辅导，以及小组和同行合作的影响。我们将进行传统的用户研究，以评估我们建议构建的环境的影响和效率。无处不在的显示器，无论大小，都与特定的计算机分离，将改变我们生活，工作和学习的方式。在工作场所，这些显示器将增强协作，并减轻计算机之间同步和传输文件的需求。在家里，智能显示器将提供前所未有的信息访问。例如，通常显示家庭照片的显示器可以查询天气预报。在学校，大型投影显示器有可能通过从任意来源呈现真正的多媒体（多个独立的媒体流和窗口）来彻底改变教育和协作。

This project seeks to outfit offices, classrooms, and public spaces in the Computer Science Department at UNC with multi-projector displays that operate as a single wide-area logical computer display replacing the traditional CRT-based displays now found in these spaces. This infrastructure will enable fundamental basic research in computer graphics, networking, multimedia networking, collaborative systems, and human-computer interaction. The research will consist of an integrated computer systems research effort into the display environment of the future and a large-scale usage experiment to assess the impact of the environments we will construct.As a computer systems research project we seek to explore new technologies and system architectures for supporting high-resolution distributed computer displays. This problem is fundamental to the future growth and development of the computer industry and the fields of computer science and engineering. Modern computer display architecture is a direct result of 1940's vintage television standards, which are outdated in today's digital world. The notion of a raster display that is periodically refreshed from a block of shared processor memory at a constant rate is a direct result of using Cathode Ray Tubes (CRTs) as early computer monitors. Moreover, this "raster-scan" approach assumes a dumb device with minimal state and a low-level interface. We believe providing local intelligence within displays will decouple the notion of a display device from the specific computer driving it.We envision treating displays as shared resources that operate independently of the specific computing platforms that present data to them. Our approach provides seamless and symmetric access to the display from any host, and supports multiple simultaneous host connections to a single display. We propose to achieve this via the efficient transport of displayed information, and through the use of effective transport and network-level mechanisms for quality-of-service. Developing higher-level semantics for display interfaces will improve bandwidth utilization. The combination of higher-level interfaces and local intelligence will also improve interaction with the ability to adapt the presentation of a content stream to the specific display platform. In other words, an intelligent display will tailor its presentation according to whether the display is on a low-resolution mobile device or a wall-sized boardroom display. It will, likewise, adapt its display according to available communication bandwidth.In this computer systems research we are proposing four new initiatives. We will investigate and develop new intelligent display architectures capable of overcoming the limitations of traditional frame-buffer architectures. We will develop higher-level and more bandwidth-efficient protocols for communicating data to visual displays. We will extend digital display interconnections so that they are more akin to real-time networks, thus increasing their utility and flexibility. Lastly, we will develop the synchronization and visual context-switching capabilities necessary to virtualize displays so that they can be used as distributed resources. Second, as a large-scale experiment into the use of wide-area displays in offices and classrooms we seek to study the effect of our proposed display architecture on personal work habits and productivity, teaching, student mentoring, and group and peer collaboration. We will undertake traditional user studies to assess the impact and efficacy of the environments we propose to construct.Broader Impact. Ubiquitous displays, large and small, and decoupled from specific computers, will transform the way we live, work, and learn. In the workplace, these displays will enhance collaboration, and alleviate the need to synchronize and transfer files between computers. At home, intelligent displays will provide unprecedented access to information. For instance, displays, which normally show family photographs, could be queried for weather forecasts. At school large projection displays have the potential to revolutionize education and collaboration by enabling the presentation of true multimedia (multiple, independent media streams and windows) from arbitrary sources.