Technical Paper Abstract
EtherNet/IP devices are becoming more complex, and the associated device configuration in modern automation systems must balance rich, device-specific behavior with cloud-scale deployment, interoperability, and predictable host integration. The paper will describe an approach that combines stateless business logic, schema-versioned APIs to address this challenge. In this model, device vendor-supplied applications remain stateless with interactions where the data models are provided through asynchronous events, while controller configuration applications own project state, sequencing, and consistency boundaries. Operational workflows are exchanged as CloudEvents over MQTT using versioned schemas, with explicit separation between transport responsibilities and profile logic. Enabling moving communications to the Edge, CIP direct-to-device operations use correlated request/response tokens and repeatable transport contracts, enabling deterministic execution for one-shot and polling scenarios without introducing cross-request state leakage in services.
In parallel, the proposed solution preserves compatibility with established artifacts and practices, including static descriptors such as EDS and EDDL-derived structures, while introducing dynamic registration and extensible JSON schema contracts for host/device coordination. The resulting architecture is expected to improve interoperability across hosts and device families, reduce integration risk for device software developers, and provide a scalable baseline for more complex configuration, online diagnostics, and multi-step device workflows to be defined by device vendors. This paper provides an overview of the architecture decisions, implementation implications, and ecosystem benefits for vendors, host developers, and end users.
The approach is deliberately additive rather than disruptive. Static descriptors such as EDS files and EDDL-derived structures continue to serve as authoritative sources; dynamic registration and extensible JSON contracts are layered on top of them for host and device coordination. The intended result is better interoperability across hosts and device families, lower integration risk for device software developers, and a scalable baseline for the more demanding scenarios, complex configuration, online diagnostics, and multistep device workflows that device vendors increasingly need to express.
Purely data-driven descriptors such as EDS and EDDL work well for simple parameter sets. The difficulty is that modeling every device with a single, static schema follows a steep complexity curve. As devices add conditional parameter visibility, cross-parameter dependencies, multi-step calibration workflows, and vendor-specific business rules, the descriptor grammar must grow without bound, and every host must implement an ever-expanding interpreter to keep pace. This is the structural cause behind the fragmentation and cost documented in Section II: each new class of device behavior forces coordinated change across every host that claims to support it. In practice this is felt most acutely with EDDL; for EDS it applies mainly in the rarer cases where the descriptor grammar itself must be extended with new sections or keywords.
Keywords
Common Industrial Protocol (CIP), EtherNet/IP, OpenAPI Specification, REST API, Device Identity, Device Discovery, Interoperability, Industrial Automation, Electronic Data Sheet (EDS), Data Harvesting, Predictive Maintenance, Digital Twin
Paper and presentation from the 2026 ODVA Industry Conference & 24th Annual Meeting: to be published November 2
Michael Miller, Senior System Architect, Rockwell Automation, Inc.
Paul Brooks, Senior Manager, Open Architecture Management, Rockwell Automation, Inc.