NeTS: Small: Meta-Modelling for Complex Engineered Networks

复杂的工程网络在日常生活中无处不在。一些例子包括互联网、电网和交通网络。复杂性不仅来自网络的规模，还来自网络的结构、运作、随时间的演变，以及人们参与其设计和运作的事实。理解复杂的工程网络具有挑战性;它们的规模和涌现特性意味着传统的设计，分析和建模技术不适合充分表征其行为和基本属性。该项目开发了一种新的工具，用于在无线网络的背景下进行严格的设计，分析和建模。目标是提高对影响绩效的因素和相互作用的理解。使用NSF全球网络创新环境（GENI）网络的模拟和测试平台的实验正在进行中。定位阵列（LA）的制定重点是识别因素的相互作用，而不是测量。因此，使用LA的设计在实验因子的数量上呈几何增长。这使得实验中的因素实际上多了一个数量级。因此，LAs有可能在巨大的因子空间中改变实验，例如在复杂的工程网络中发现的那些。一个迭代的方法，类似于压缩感知中使用的方法，被应用在模型开发中;这也解决了无线网络中新的数据驱动的数学模型的巨大挑战。评估开发的模型的有效性和鲁棒性对于了解其质量和用于网络优化、管理和控制的有用性至关重要。总体而言，该项目有助于理解塑造现代社会的复杂工程网络。

Complex engineered networks are pervasive in everyday life. A few examples include the internet, power grid, and transportation networks. The complexity arises not just from the size of the network, but also its structure, operation, evolution over time, and the fact that people are involved in its design and operation. Understanding complex engineered networks is challenging; their size and emergent properties means that traditional techniques of design, analysis, and modelling are poorly suited to adequately characterizing their behaviour and fundamental properties. This project develops a new tool for rigorous design, analysis, and modelling in the context of wireless networks. The goal is to improve understanding of the factors and interactions that impact performance. Experimentation in simulation and testbeds, using the NSF Global Environment for Network Innovations (GENI) network, is ongoing.Locating arrays (LAs) are formulated to focus on identification of factor interactions rather than measurement. Consequently, designs using LAs grow logarithmically in the number of experimental factors. This makes practical an order of magnitude more factors in experimentation. Hence, LAs have the potential to transform experimentation in huge factor spaces such as those found in complex engineered networks. An iterative approach, similar to that used in compressive sensing, is applied in model development; this also addresses a grand challenge in wireless networks for new data-driven mathematical models. Assessing the validity and robustness of the models developed is essential to understanding their quality and usefulness for optimization, management, and control of the network. Overall, this project contributes to understanding of complex engineered networks that shape modern society.