CRI: Reconfigurable High Performance Cluster Computing and Medical Engineering Applications

0708232CRI：可重构高性能集群计算与医学工程应用徐成中本项目的目标是为各种互联网服务和实时医学工程应用开发一种可重构的高性能、高能效的网络计算解决方案。多核服务器集群被构建为可重新配置的分布式虚拟机(DVM)以提供共享计算资源，以支持韦恩州立大学的多学科研究，尤其包括以下活动：(1)开发用于高性能和节能计算的资源管理技术；(2)开发用于服务器集群中数据密集型应用的混合存储系统；(3)开发用于可扩展流服务的带宽高效资源管理技术；(4)开发用于机器人手术中快速和准确的端口放置和手臂定位的算法；以及(5)实时麻醉规划、诊断和决策支持。该项目充分利用了有针对性的研究活动的成功。系统研究在处理能力、内存/存储、网络I/O带宽和能效方面开发了新的资源管理技术。医学工程应用需要大量的计算，并且需要检索和处理大量数据。研究活动共享实时和高保证性处理的要求。DVM促进了对新资源管理技术的实验评估，并使其能够在应用程序中试验各种设计选择，并在合理的时间内研究许多重要因素的影响和相互作用。DVM解决方案还将医疗信息处理和患者治疗提升到一个新的及时性和准确性水平。

0708232CRI: Reconfigurable High Performance Cluster Computing and Medical Engineering ApplicationsChengzhong XuThe goal of this project is to develop a reconfigurable high performance and energy efficient networked computing solution for a variety of Internet services and real-time medical engineering applications. A cluster of multi-core servers is constructed as a reconfigurable distributed virtual machine (DVM) to provide a shared computing resource in support of multi-disciplinary research at Wayne State University, including in particular, the following activities: (1) development of resource management technologies for high performance and energy efficient computing; (2) development of hybrid storage systems for data-intensive applications in server clusters; (3) development of bandwidth efficient resource management techniques for scalable streaming services; (4) development of algorithms for fast and accurate port placement and arm positioning in robotic surgery; and (5) real-time anesthesia planning, diagnosis, and decision support. This project leverages the success of the targeted research activities. The systems research develops novel resource management technologies with respect to processing power, memory/storage, network-I/O bandwidth, and energy efficiency. The medical engineering applications require intense computations and entail retrieving and processing of large amounts of data. The research activities share the requirements of real-time and high assurance processing. DVM facilitates experimental evaluation of new resource management technologies and make it possible to experiment with various design choices in applications and to study the impacts of and the interactions among many important factors in reasonable times. The DVM solution also advances medical information processing and patient treatments to a new level of timeliness and accuracy.