CoCPN-ng – Cooperative Cyber-Physical Networking: Next Generation

从抽象的观点来看，网络物理系统可以被建模为一组共存的共享通信资源的应用程序控制环路。CoCPN的核心思想是通过平衡每个环路应达到的控制质量(QoC)来实现应用控制环路之间的通信资源的协作共享。在当前的CoCPN项目中，我们考虑了独立的控制应用程序，每个应用程序都由一个应用程序控制回路组成。CoCPN-NG的目标是进一步改进CoCPN以能够应对由多个可能相互依赖的应用控制回路组成的复杂的、弹性的控制应用，例如在新兴的智能工厂中。与第一个供资期间不同，CoCPN-NG考虑了相互依赖的应用控制回路，即在物理(模拟)世界中相互影响的回路，例如可以相互接触的倒立摆。由于在未来的网络物理系统中设想的动态增加，系统的详细手动预配置不被视为重要选项。因此，在CoCPN-NG中，我们计划为CoCPN-Translator提供自学能力。我们将研究非参数在线回归技术作为一种可能的方法。此外，基于我们的模拟和评估框架CoCPN-Sim的系统模拟实验，将研究应用控制环路之间的相互依赖对新开发的度量QoC和QM的影响。在应用控制方面，将寻求更复杂的控制方法，将完整的控制律发送到执行器，而不是简单的输入序列。此外，将研究如何在通信系统中利用所发送的控制数据分组的控制应用特定的值(例如，关键程度)，以便转发最有价值的信息并丢弃价值较小的信息。这需要将被研究的通信系统中合适的机制(例如，ARQ机制、面向值的排队)。基于这些机制，我们将设计一种适合弹性控制应用需求的控制感知传输协议。我们将进行系统的仿真研究，以考察在通信条件变化的情况下，CoCPN-NG在控制应用数量不同的场景中的适用性。我们将使用在第一个资助期成功开发的CoCPN-SIM。研究适当的权衡，例如关于交换的信息的细节与可实现的QOC，是拟议项目的一个重要目标。此外，还将研究关于通信和控制复杂性以及资源的可用性(例如，资源受限的传感器和执行器的计算、存储和通信能力)的权衡。

From an abstract point of view a cyber-physical system can be modeled as a set of coexisting application control loops that share communication resources. The key idea of CoCPN is to implement cooperative sharing of communication resources among application control loops by balancing the quality of control (QoC) that each loop shall reach. Within the current project CoCPN we considered independent control applications each consisting of a single application control loop. The objective of CoCPN-ng is to further improve CoCPN to be able to cope with complex, elastic control applications that consist of multiple probably interdependent application control loops, e.g., in emerging smart factories. In contrast to the first funding period, CoCPN-ng considers interdependent application control loops, i.e., loops that affect each other in the physical (analog) world, e.g., inverted pendulums that could touch each other. Due to the increased dynamics envisioned in future cyber-physical systems, detailed manual pre-configurations of the system are not considered as a vital option. Therefore, within CoCPN-ng we plan to provide the CoCPN-translator with self-learning capabilities. We will investigate nonparametric online regression techniques as one possible approach. Furthermore, the impact of the interdependencies between application control loops on the newly developed metrics QoC and QM will be investigated based on systematic simulation experiments with our simulation and evaluation framework CoCPN-Sim. Regarding application control, more sophisticated control approaches will be pursued that send complete control laws to the actuators instead of plain input sequences. Furthermore, it will be researched how the control application specific “value” (e.g., criticality) of a transmitted control data packet can be utilized in the communication system in order to forward the most valuable information and discard information of less value. This requires suited mechanisms in the communication system (e.g., ARQ mechanisms, value-oriented queueing) that will be researched. Based on these mechanisms, we will design a control-aware transport protocol that is tailored to the needs of elastic control applications. We will conduct systematic simulation studies to investigate the applicability of CoCPN-ng in scenarios with varying numbers of control applications under changing communication conditions. We will use CoCPN-Sim that was successfully developed in the first funding period. Researching suitable trade-offs, for example regarding the detail of information being exchanged versus the achievable QoC is an important goal of the proposed project. Furthermore, trade-offs regarding communication and control complexity and the availability of resources (e.g., computing, storage, and communication capacity of resource-constrained sensors and actuators) will be investigated.