(NECO) Collaborative Research: Reliability Modeling for Large-Scale Networking System (LSNS), and Self-Improvement in LSNS

本项目构建大规模网络系统（LSNS）可靠性和性能的整体模型，开发基于模型的自我改进（MBSI）技术，实现大规模网络系统的高可靠性、高性能和流畅的网络通信。该项目解决了LSNS可靠性和性能整体模型的精确制定、可靠性和性能的分析评估、减少网络流量拥堵的方案等问题，并进一步利用自我改进技术自动收集数据、建立模型、评估设计和优化任务。对LSNS的建模和评价主要采用图论、贝叶斯方法、最大熵原理、通用生成函数、蒙特卡罗模拟等方法，对自我完善函数采用自主计算技术。因此，本研究进一步推进了可靠性方面的理论、算法和技术，填补了LSNS可靠性建模和自我改进的空白。预期结果：(1)将设计新颖的配方，以有效地模拟具有所有基本成分的LSNS；(2)提出了一种新型的混合随机算法（HSA）来评估LSNS的任务可靠性和性能；(3)设计新的LSNS优化方案；(4)开发一种新的MBSI技术，以自我提高可靠性、性能和网络通信；(5)研究成果将应用于多种LSNSs，如NASA的外太空探索、远程医疗、网格计算等；(6)开发一套软件工具。(7)研究成果将通过期刊/会议出版物和pi网站传播。

This project builds a holistic model of reliability and performance for large-scale networking system (LSNS) and develops model-based self-improvement (MBSI) technology for high reliability, high performance, and smooth network communication in LSNS. This project addresses a number of problems, including the precise formulation of a holistic model of the LSNS reliability and performance, the analytical evaluation of the reliability and performance, the scheme to reduce the jams of network traffics, and further uses the self-improvement technology to automatically glean data, build models, evaluate designs, and optimize tasks. The methods, like graph theory, Bayesian approach, maximum entropy principle, universal generating function, and Monte Carlo simulation, are mainly adopted in modeling and evaluating the LSNS, while autonomic computing technologies are implemented for the self-improvement functions. As a result, this research further advances the theory, algorithm and technology in reliability, and fills the gap of reliability modeling and self-improvement in LSNS. Expected results: (1) Novel formulations will be designed to effectively model a LSNS with all essential components; (2) An innovative HSA (Hybrid Stochastic Algorithm) will be developed to evaluate the task reliability and performance for LSNS; (3) New optimization schemes for LSNS will be designed; (4) A novel MBSI technology will be developed for self-improving the reliability, performance, and network communication; (5) Research outcomes will be applied in a variety of LSNSs, such as NASA's outer-space exploration, tele-medicines, grid computing, etc; (6) A set of software tools will be developed. (7) Research results will be disseminated through journal/conference publications and PIs' websites.