Inferensys

Glossary

OPC UA PubSub

A communication model within the OPC Unified Architecture that enables publisher-subscriber messaging patterns, often used to integrate substation data with higher-level enterprise and cloud systems.
Data scientist building training data pipeline on laptop, data preprocessing visible, technical workspace.
Publisher-Subscriber Communication Model

What is OPC UA PubSub?

OPC UA PubSub is a communication model within the OPC Unified Architecture that decouples data producers from consumers using a publisher-subscriber pattern, enabling scalable, one-to-many distribution of industrial data to cloud and enterprise systems.

OPC UA PubSub extends the traditional client-server OPC Unified Architecture by introducing a publisher-subscriber messaging pattern where data sources broadcast information without requiring direct, persistent connections to every recipient. This decoupling allows a single publisher to transmit datasets to multiple subscribers simultaneously, significantly reducing network overhead and configuration complexity in large-scale industrial environments. The model supports two transport protocols: UADP (UA Datagram Protocol) for low-latency, multicast communication on local networks, and AMQP/MQTT for secure, brokered messaging to cloud platforms.

Within substation automation, OPC UA PubSub serves as a critical bridge between operational technology and enterprise systems by mapping IEC 61850 data models onto cloud-native messaging infrastructure. Unlike the real-time, lossless GOOSE and Sampled Values protocols used on the process bus, PubSub is designed for non-deterministic, fire-and-forget telemetry streams where occasional data loss is acceptable. This enables secure, firewall-friendly integration of IED status, measurement data, and event logs with higher-level applications such as SCADA, digital twins, and predictive maintenance analytics without exposing the time-critical substation network.

DECOUPLED DATA DISTRIBUTION

Key Features of OPC UA PubSub

OPC UA PubSub extends the client-server model with a publisher-subscriber paradigm, enabling scalable, one-to-many or many-to-one data distribution critical for integrating substation automation systems with cloud analytics and enterprise platforms.

01

Brokerless UDP Multicast

Leverages User Datagram Protocol (UDP) multicasting on the local substation LAN to distribute data without a central message broker. This eliminates a single point of failure and minimizes latency for high-speed, time-critical data like GOOSE or Sampled Values mapped to PubSub.

  • Zero-broker architecture: Publishers send directly to multicast groups.
  • Sub-microsecond synchronization: Integrates with Precision Time Protocol (PTP) for deterministic delivery.
  • Use case: Replacing traditional IEC 61850 publisher-subscriber mechanisms on the process bus with a unified, OPC UA-encoded data stream.
02

Brokered AMQP/MQTT Transport

Maps OPC UA PubSub datasets onto standard messaging protocols like Advanced Message Queuing Protocol (AMQP) and MQ Telemetry Transport (MQTT) for cloud integration. A central broker handles store-and-forward, topic routing, and subscriber management.

  • Protocol bridging: Translates IEC 61850 Logical Node data into JSON or binary payloads for enterprise brokers.
  • Decoupled consumers: Enables multiple cloud applications to subscribe to a single Merging Unit data stream without direct IED connections.
  • Use case: Streaming substation Disturbance Recorder files in real-time to a central analytics platform via an MQTT broker.
03

DataSet & DataSetMetaData

Defines a structured collection of OPC UA Variable values published as a single atomic message. The DataSetMetaData describes the structure, data types, and semantics of the payload, enabling subscribers to interpret the data without prior out-of-band configuration.

  • Self-describing payloads: Each message carries its own schema, ensuring interoperability.
  • Field mapping: Directly maps attributes from Logical Nodes like XCBR (circuit breaker) position or MMXU (measurement) current.
  • Version control: DataSetMetaData includes a configuration version to manage schema evolution across publishers and subscribers.
04

Publisher & Subscriber Entities

The core architectural components that replace the rigid client-server relationship. A Publisher defines and emits DataSets, while Subscribers dynamically discover and consume them. This decoupling allows for flexible, many-to-many communication patterns.

  • Dynamic discovery: Subscribers can locate publishers using standard OPC UA discovery services.
  • Loose coupling: Publishers have no knowledge of subscribers, enabling independent scaling.
  • Use case: A single Phasor Measurement Unit (PMU) publisher streams synchrophasor data to multiple subscribers: a local Wide-Area Monitoring System, a cloud historian, and a real-time oscillation detection application.
05

Security Key Management

Implements message-level security for PubSub data flows, independent of the transport protocol. Publishers sign and encrypt individual messages using symmetric keys distributed via a Security Key Service (SKS).

  • Group security model: A shared symmetric key secures a multicast group, ensuring only authorized subscribers can decrypt data.
  • Compliance: Aligns with IEC 62351 security requirements for power system communications.
  • Use case: Encrypting Sampled Values traffic on a process bus to prevent unauthorized injection of malicious current measurements into protection relays.
06

JSON & UADP Message Mapping

Supports two primary encoding formats for PubSub messages. JSON mapping provides human-readable, web-friendly payloads for enterprise integration, while UADP (UA Binary) offers compact, high-performance encoding for resource-constrained IEDs and low-bandwidth links.

  • JSON: Ideal for cloud ingestion and integration with SCADA analytics platforms.
  • UADP: Optimized for Process Bus applications requiring minimal overhead and deterministic parsing.
  • Use case: An Intelligent Electronic Device (IED) publishes UADP-encoded GOOSE tripping signals on the local LAN while simultaneously publishing JSON-encoded status updates to a cloud broker.
OPC UA PubSub

Frequently Asked Questions

Clear answers to common questions about the publisher-subscriber communication model in OPC Unified Architecture and its role in industrial and utility data integration.

OPC UA PubSub is a communication model within the OPC Unified Architecture that enables publisher-subscriber messaging patterns, decoupling data producers from consumers. Unlike the client-server model, PubSub allows a publisher to send messages to a message-oriented middleware—such as a broker like MQTT or AMQP, or directly via UDP multicast—without knowing the subscribers. Subscribers receive only the data they have expressed interest in through topic-based or content-based filtering. This model is defined in Part 14 of the OPC UA specification and supports both JSON and binary UADP message encoding, making it ideal for large-scale, one-to-many data distribution scenarios like streaming substation telemetry to cloud analytics platforms.

COMMUNICATION PATTERN COMPARISON

OPC UA Client-Server vs. PubSub

Structural comparison of the two OPC UA communication models for substation-to-enterprise data integration

FeatureClient-ServerPubSub

Communication Pattern

Point-to-point request/reply

One-to-many publish/subscribe

Connection Model

Persistent session with stateful context

Connectionless or broker-mediated messaging

Data Delivery

Pull-based (client polls or subscribes)

Push-based (publisher sends to all subscribers)

Network Efficiency

N+1 connections for N clients

Single publish to multiple subscribers

Decoupling

Tight coupling (client knows server address)

Loose coupling (publisher unaware of subscribers)

Use Case in Substations

SCADA polling, engineering access, configuration

Streaming Sampled Values, GOOSE integration, cloud telemetry

Transport Protocols

UA TCP binary protocol, HTTPS

UADP (UDP/IP), MQTT, AMQP

Time Synchronization

Not inherent to transport

Built-in timestamps for isochronous data

Security Model

Per-session authentication and encryption

Per-message security with group key management

IEC 61850 Alignment

MMS-like supervisory control

GOOSE/SV-like multicast data distribution

Scalability

Limited by server connection capacity

Horizontally scalable via broker clustering

Firewall Traversal

Requires port opening per connection

Broker-mediated outbound connections only

Message Ordering

Guaranteed in-order delivery

Configurable (best-effort or guaranteed via broker)

Bandwidth Utilization

Higher overhead per client

Lower overhead for many consumers

Fault Tolerance

Single server point of failure

Redundant publishers and broker clustering

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.