Hardware engineers move between physical devices and register-transfer level models all the time, but the conceptual basis of that movement is rarely examined. A paper from Infineon Technologies and the Technical University of Munich argues that abstraction is the defining characteristic of hardware engineering models, not merely a convenience. The authors—Wolfgang Ecker, Natalie Simson, Johannes Ecker, and Endri Kaja—released it on arXiv in October 2026.
The paper surveys abstraction methods that appear in digital design, including lumped models, value-discrete models, and time-discrete models. When those methods are combined with validity constraints and design guidelines, they form what the authors call design disciplines. The work also connects these abstractions to pre-clustered elements such as transistors, gates, registers, and transfer functions, which define abstraction levels such as the gate level.
Abstraction has trade-offs. It necessarily omits details, so models are incomplete in specific ways; the paper invokes statistician George Box's observation that all models are wrong but some are useful. Yet omission is also what makes design tractable: simpler models require fewer details, which makes simulation faster and raises design productivity.
Because this article draws on a single source, there are no competing claims to compare. The paper's contribution is a framework that treats abstraction levels as deliberate engineering choices rather than natural layers of the design stack.