Digital twin simulation models are not static artifacts; they change, get refined, and are redeployed as the systems they mirror evolve. The paper argues that this lifecycle resembles software development, yet the tools used for DEVS-based simulation have not kept pace. While the Discrete Event System Specification (DEVS) provides a rigorous formal foundation, the surrounding infrastructure offers little help for managing model versions or automatically checking that updates behave correctly.
The authors propose a model-centric DevOps architecture designed specifically for DEVS-based digital twin services. By borrowing ideas from continuous integration and deployment, the architecture treats the model as the central artifact that flows through validation and release stages. This is intended to make simulation services more maintainable and reliable as they are updated over time.
The abstract highlights two concrete gaps in current DEVS engines: versioning and automated validation. The proposed architecture directly targets these shortcomings, potentially helping to move digital twin simulation from research prototypes toward production-grade services that can be trusted to evolve safely.