NGS:Montage: An Integrated End-to-End Design and Development Framework for Wireless Networks

随着今天的无线网络从电路交换语音向基于互联网协议（IP）的流量过渡，对新兴的端到端用户应用程序的性能进行准确和快速的建模是一个至关重要的需求。令人惊讶的是，今天的无线系统在开发无线标准（例如，WCDMA、EDGE和IEEE 802.11a）时，将网络堆栈的所有高层协议视为不可见和不相关的。未来的无线网络必须放弃这种陈旧的分离观点，必须使用跨层知识来实现下一代无处不在的无线应用程序。事实上，设计和部署新一代无线网络服务的主要障碍之一是缺乏用于端到端系统建模和仿真的复杂计算工具。下一代仿真环境必须同时解决通道和协议层次的物理层所需的建模保真度，同时，交互式地利用在更高层使用的软件协议中的这些功能。当使用这种方法时，就有可能为下一代无线数据网络的开发创建一个功能强大、准确的端到端建模环境。本提案将开发第一个学术的端到端多层无线网络模拟器和仿真器。蒙太奇是一种集成协议开发环境，将支持无线网络协议的进化设计、开发和测试。它将使全国的无线和自组织移动网络研究人员和系统运营商能够在单一集成的公共领域环境中设计、开发和分析无线硬件、网络协议、中间件算法和网络应用。本提案的主要研究问题包括统一的多建模方法（仿真+仿真+直接代码执行）、智能调度算法、构建动态组合环境以及整合各种网络协议层的复杂性能模型。因此，蒙太奇解决了TPES（性能工程系统技术）和cadd（复杂应用设计和支持系统）目标的NSF下一代软件（NGS）计划。蒙太奇的发展推动了复杂系统集成的新方法和新技术的发展。从几个现有的和语义上复杂的系统中创建一个集成的系统是不可能的。本课题的研究任务来源于系统的接口和集成，这些系统具有不同的语义基础、不同的执行模型，并且跨越多个抽象层次。一项任务是设计接口规范语言，使其能够跨多个语义、执行模型和抽象级别表示翻译。将接口语言应用于接口和集成软件系统，共同实现蒙太奇方法，将是本研究的主要组成部分。作为应用案例研究，该项目将解决大规模自组织无线网络以及从2.5G向3G过渡的大型商用蜂窝网络的关键验证和设计问题。今天，这些计算能力并不存在。更广泛的影响：蒙太奇将作为一个公共领域的门户网站，为网络研究人员、教育工作者和商业运营商提供定制的设计场景接口。蒙太奇还应作为衡量网络性能的国家档案和用户社区选择的设计库。蒙太奇将使研究人员能够设置虚拟测试平台，验证他们的建模，并为未来的无线网络设计和测试中间件算法。主要研究人员在为教育和培训目的创造有价值的研究产品和学习材料方面有着良好的记录。此外，调查人员还证明了通过创建高科技衍生企业来促进经济增长的记录。蒙太奇已经得到了主要工业团体的支持（见支持信），并将成为研究人员和教育工作者的宝贵资源。

As today's wireless networks transition from circuit-switched voice to internet protocol (IP) based traffic, there is a critical need to accurately and rapidly model the performance of emerging end-to-end user applications. Surprisingly, today's wireless systems treat all the higher layer protocols of the network stack as invisible and irrelevant in the development of wireless standards (e.g., WCDMA, EDGE, and IEEE 802.11a). Future wireless networks must abandon this archaic viewpoint of separation, and must use cross-layer knowledge to enable the next generations of ubiquitous wireless applications. In fact, one of the major stumbling blocks to designing and deploying the new generation of wireless network services is the lack of sophisticated computational tools for end-to-end systems modeling and simulation. Next generation simulation environments must address both the fidelity of modeling needed for the channel and physical levels of the protocol hierarchy and, at the same time, interactively exploit these featuresin software protocols employed at the higher layers. When such an approach is used, it becomes possible to create a powerful, accurate, end-to-end modeling environment for the development of the next generation of wireless data networks. This proposal will develop the first academic end-to-end multi-layer simulator and emulator for wireless networks. Montage -an integrated protocol development environment -will support the evolutionary design, development, and test of wireless network protocols. It will enable wireless and ad-hoc mobile network researchers and system operators across the country to design, develop, and analyze wireless hardware, network protocols, middleware algorithms, and network applications in a single integrated public-domain environment. Key research issues addressed in this proposal include a unified multi-modeling methodology (simulation + emulation + direct code execution), algorithms for smart scheduling, building a dynamic compositional environment, and incorporating sophisticated performance models of various network protocol layers. Montage thus addresses both the TPES (Technology for Performance Engineered Systems) and the CADDS (Complex Application Design and Support Systems) goals of the NSF Next Generation Software (NGS) program. Development of Montage drives development of new methods and techniques for integration of complex systems. It is simply not possible to create an integrated system from several existing and semantically complex systems on an ad hoc basis. The research tasks of this project derive from interfacing and integration of systems, which have quite different semantic bases, varied execution models, and span multiple levels of abstraction. One task is to design interface specification languages that will enable representation of the translations across the multiple semantics, execution models and levels of abstraction. Applying the interface language to interface and integrate the software systems, which together implement the Montage methodology, will be a major component of this research. As application case studies, this project will address key validation and design problems for large-scale ad-hoc wireless networks, as well as large commercial cellular networks that are transitioning from 2.5G to 3G. Today, these computational capabilities do not exist. Broader Impacts: Montage will be made available as a public domain web portal with customized design scenario interfaces for network researchers, educators, and commercial carriers. Montage shall also serve as a national archive for measured network performance and as a repository of designs selected by the user community. Montage will enable researchers to setup virtual testbeds, validate their modeling, and design and test middleware algorithms for wireless networks of the future. The principal investigators have an excellent track record for creating valuable research products and learning materials for education and training purposes. Furthermore, the investigators have proven records for fostering economic growth by creating high-tech spinout companies. Montage already has the support of key industrial constituencies (see supporting letters), and will emerge as a valuable resource for researchers and educators.