In recent years, cyber‐physical systems (CPS) have been widely used in different problem domains. The hardware and software components in a CPS are deeply intertwined at various levels of abstraction under changing contexts to achieve the desired goals. One way to manage this complexity is through the use of modular architecture that enables portions of a CPS to be upgraded, replaced, or fixed in a plug‐and‐play manner. In practice, nominally modular architectures depart from this ideal. Thus, there is a need to identify, measure, and understand the causes of these departures. Techniques have been developed to accomplish this for pure software systems, but can these techniques be adapted to infer hardware‐related modularity violations in a CPS? To investigate this question, we adapted methods from pure‐software systems and analyzed two open‐source CPS: OpenWrt and MD PnP. We found that the software architecture of these systems is well modularized. However, we were able to detect and measure modularity violations associated with hardware and other domain‐related concepts. Furthermore, software components related to hardware were more likely to change frequently compared to general software components. Together, these findings suggest that it is feasible to identify a subset of hardware‐driven modularity violations using techniques adapted from pure software systems.

中文翻译：

---

网络物理系统中违反模块化的案例研究

近年来，网络物理系统（CPS）已广泛用于不同的问题领域。在不断变化的上下文中，CPS中的硬件和软件组件在各种抽象级别上都深深地交织在一起，以实现所需的目标。解决此复杂性的一种方法是使用模块化体系结构，该体系结构使CPS的各个部分可以即插即用的方式进行升级，更换或固定。实际上，名义上的模块化体系结构偏离了这一理想。因此，需要识别，测量和理解这些偏离的原因。已经开发出可以在纯软件系统上实现此目的的技术，但是可以将这些技术改编为推断CPS中与硬件相关的模块化违规吗？为了调查这个问题，我们改编了纯软件系统中的方法，并分析了两个开源CPS：OpenWrt和MD PnP。我们发现这些系统的软件架构已很好地模块化。但是，我们能够检测和测量与硬件和其他域相关概念相关的模块化违规情况。此外，与通用软件组件相比，与硬件相关的软件组件更容易频繁更改。总之，这些发现表明，使用从纯软件系统改编的技术来识别硬件驱动的模块化违规的子集是可行的。与通用软件组件相比，与硬件相关的软件组件更容易频繁更改。总之，这些发现表明，使用从纯软件系统改编的技术来识别硬件驱动的模块化违规的子集是可行的。与通用软件组件相比，与硬件相关的软件组件更容易频繁更改。总之，这些发现表明，使用从纯软件系统改编的技术来识别硬件驱动的模块化违规的子集是可行的。