Analysis and decentralized synthesis of distributed control systems

在上一次NSERC的资金再分配工作中，关于电气和计算机工程的意见书强调了这样一个事实，在2000年生产的83亿个CPU中，80%进入了嵌入式系统，6%进入了机器人，12%进入了汽车。它询问了无数的智能家电将如何互联，它们将如何通信，以及它们将使用什么样的管理结构，并强调这些设计挑战需要对概念(而不是纯技术)性质的新见解，电话呼叫处理说明了其中的一些问题。电话转接、呼叫等待和数百种其他支持数字网络通信的服务(必然)是由各种不同的供应商作为单独的功能模块独立开发的，并且可能在网络中的不同点在不同的平台上实现。在花费了开发和实现的成本之后，它们经常以不可预见和不受欢迎的方式相互作用。这种所谓的功能交互问题由电信行业引起研究人员的注意，仍然是一个昂贵的实际问题和活跃的研究课题。本文提出的研究将为分析和设计复杂的、分布式的系统提供系统的手段，允许及早发现和解决像功能交互这样的问题。与应用领域或技术无关，研究结果将以高层次的“离散事件”系统模型来表述。参与这项研究及相关项目的名学生和博士后研究员将继续从事工业自动化、软件工程和进一步的学术研究。

In the last NSERC funding reallocations exercise, the submission on electrical and computer engineering highlighted the fact that, of the 8.3 billion CPUs produced in 2000, 80% found their way into embedded systems, 6% into robots and 12% into automobiles.  It asked how myriad intelligent appliances would be interconnected, how they would communicate, and what kind of management structures they would use, and stressed that these design challenges required new insights of a conceptual (as opposed to purely technological) nature, pointing to a coming sea change "where the emphasis shifts from information and communication technologies to information and communication per se.''  Telephone call processing illustrates some of the problems.  Services such as call forwarding, call waiting, and the hundreds of others that support communication over digital networks, are (of necessity) developed independently, as separate functional modules, by a variety of different vendors, and may be implemented on different platforms at various points in a network.  After the expense of their development and implementation, they are often found to interact with one another in unforeseen and undesirable fashion.  This so-called "feature interaction problem'' was brought to the attention of researchers by the telecommunications industry, and remains both a costly practical problem and an active topic of research.The research proposed here will contribute systematic means of analyzing and designing complex, distributed systems, allowing the early detection and resolution of problems like feature interaction.  Independent of application domain or technology, the results will be formulated in terms of high-level ''discrete-event'' system models.   Students and postdoctoral fellows who have been involved with this research and related projects have gone on to work in industrial automation, in software engineering, and in further academic research.