MRI: Acquisition of a High Performance Computer Cluster for Multidisciplinary Computational Research at Prairie View A&M University

这项工作在Prairie View A&amp；M大学提供了一个高性能计算(HPC)集群，以支持广泛的多学科计算密集型研究和研究培训。建议的集群具有节点之间的以太网端口连接、共享网络存储和图形处理器内核上的大规模并行计算。此外，带交换机连接的车载千兆以太网端口可以支持大数据集传输，从而支持实时多媒体网络仿真研究。实时多视点视频应用中的压缩感知(CS)演示需要相当大的计算能力，但对于监控摄像头和传感器网络至关重要。这些影响在国土安全应用中不是微不足道的，特别是在传感器和手持无线电都是移动的战场上。视频传感器检测物理属性，并将数据快速传输到命令控制中心，以便远程执行操作。高性能计算集群使探索CS采样和渲染的并行处理成为可能，而Prairie View集群非常适合在这些应用程序的研究中进行高速并行视频处理。实时仿真不仅有助于了解实时处理对认知无线电应用的影响，而且有助于为现实环境中的应用设计算法。本研究包括研究与教育的融合。最直接和最直接的影响是，从该HPC组所支持的研究项目中获得的研究数据、源代码和课程材料可以在研究社区内免费获取。加强对多媒体信号的实时重构、分析和建模研究，影响到国土安全需求和网络资源的高效利用。此外，开发新课程，扩大研究生和本科生的研究经验，并扩大到高中和中学生，是计划使用Prairie View HPC集群不可或缺的一部分。

This work provides a High Performance Computing (HPC) cluster at Prairie View A&M University to support a broad array of multidisciplinary computing-intensive research and research training.The proposed cluster features Ethernet ports connection between nodes, shared network storage and massive parallel computation on graphics processor cores. Moreover, the on board Gigabit Ethernet ports with switch connection can support large data sets transmission to enable the research in real time multimedia network simulation.Demonstrating compressive sensing (CS) on real time multi-view video application requires considerable computational capability but is critical for surveillance camera and sensor networks. The impacts are nontrivial in homeland security applications especially on the battlefield where sensors and hand-held radios are mobile. Video sensors detect physical properties and rapidly transmit the data to a command control center for remote executable actions. A High Performance Computing cluster makes it possible to explore parallel processing of CS sampling and rendering and the Prairie View cluster is ideally suited for high-speed, parallel video processing in the research of these applications. This cluster enables new research in simulating cognitive radio networks with more complicated situations The real time simulation will help not only understanding how real time processing affects the cognitive radio applications, but also designing algorithms for applications in real world environment. This research includes the integration of research and education. The most direct and immediate impact is that research data, source code, and curriculum materials obtained from the research projects supported by this HPC cluster are freely accessible across the research communities. Strengthening the real time multimedia signal reconstruction, analysis and modeling research impacts homeland security needs and efficient usage of network resources. Additionally, the development of new courses, expanded research experiences for Graduate and undergraduate students, and outreach to high school and middle school students are an integral part of planned use of the Prairie View HPC Cluster.