Manufacturing

    What is MQTT (Industrial Protocol)? | Definition & Guide

    MQTT (Message Queuing Telemetry Transport) is a lightweight publish-subscribe messaging protocol originally designed for low-bandwidth IoT communication, now widely adopted in manufacturing for transmitting sensor data from edge devices to cloud and on-premise analytics platforms. MQTT requires minimal network overhead compared to OPC UA, making it suitable for high-frequency sensor data from distributed equipment. In manufacturing, MQTT serves as the transport layer between edge gateways and centralized data platforms, complementing OPC UA which handles equipment-level data modeling and access.

    Definition

    MQTT (Message Queuing Telemetry Transport) is a lightweight publish-subscribe messaging protocol now widely adopted in manufacturing for transmitting sensor and equipment data from edge devices to cloud and on-premise analytics platforms. Unlike request-response protocols, MQTT uses a broker-based architecture where data producers publish messages to topics and consumers subscribe to topics of interest — decoupling senders from receivers. MQTT's minimal overhead (a fixed header as small as 2 bytes) makes it suitable for high-frequency sensor data from distributed equipment operating on constrained networks. In the manufacturing protocol landscape, MQTT typically serves as the data transport layer between edge gateways and centralized platforms, while OPC UA handles equipment-level data modeling and direct PLC communication. HiveMQ, Eclipse Mosquitto, and AWS IoT Core are common MQTT broker platforms in industrial deployments.

    Why It Matters

    For manufacturing IT/OT architects designing data infrastructure for IoT-scale deployments, MQTT addresses the challenge of moving high volumes of sensor data efficiently from hundreds or thousands of data sources to centralized analytics. A facility with 500 IoT sensors publishing at 1 Hz generates 43 million messages per day. OPC UA client-server connections, while richer in data modeling, create significant overhead at this scale — each sensor would require a dedicated connection to every consuming application. MQTT's publish-subscribe model lets 500 sensors publish to a single broker, with any number of consuming applications subscribing to the topics they need.

    The adoption driver in manufacturing is the convergence of IT and OT data infrastructure. Cloud platforms (AWS IoT Core, Azure IoT Hub, Google Cloud IoT) use MQTT as their primary ingestion protocol for device data. Edge platforms (Litmus, Siemens Industrial Edge) publish processed data to cloud through MQTT. The Sparkplug B specification — developed by Cirrus Link and now managed by the Eclipse Foundation — adds manufacturing-specific metadata and state management to MQTT, creating a standardized topic structure and payload format for industrial data that multiple analytics platforms can interpret consistently.

    The tradeoff is that MQTT by itself provides no data modeling — it is a transport protocol, not an information model. A raw MQTT message carries a payload on a topic, but the consuming application must know what the payload means. OPC UA provides self-describing data with engineering units, data types, and hierarchical context. The practical architecture in most manufacturing implementations uses OPC UA at the equipment edge (PLC data access with rich information models) and MQTT for data distribution (efficient transport from edge to cloud or cross-plant data sharing). OPC UA PubSub bridges these by publishing OPC UA information models through MQTT transport.

    How It Works

    MQTT operates through four architectural components in manufacturing deployments:

    1. Publish-subscribe messaging — Data producers (edge gateways, sensors, equipment controllers) publish messages to named topics on an MQTT broker. Topics follow a hierarchical namespace: plant/line3/machine7/temperature or using Sparkplug B conventions: spBv1.0/PlantA/DDATA/Line3/Machine7. Consuming applications subscribe to specific topics or use wildcards (plant/line3/# for all data from line 3). HiveMQ and Eclipse Mosquitto provide the broker infrastructure, with HiveMQ offering enterprise features including clustering, persistence, and integration with Kafka and cloud platforms.

    2. Quality of service levels — MQTT supports three QoS levels that balance reliability against network overhead. QoS 0 (at most once) provides fire-and-forget delivery suitable for high-frequency sensor data where occasional message loss is acceptable. QoS 1 (at least once) guarantees delivery with possible duplicates — appropriate for production events and alarms. QoS 2 (exactly once) guarantees single delivery with higher overhead — used for critical control commands and financial transactions. Manufacturing deployments typically use QoS 0 for sensor telemetry and QoS 1 for production events and equipment state changes.

    3. Sparkplug B specification — The Sparkplug B specification adds an industrial data model on top of MQTT transport: standardized topic namespaces, birth/death certificates for device state management (the broker knows when an edge device goes offline), protobuf-encoded payloads for compact data representation, and metric definitions that carry data type, timestamp, and metadata. Sparkplug B eliminates the “everyone invents their own MQTT topic structure” problem that plagues custom IoT deployments. Ignition (Inductive Automation) natively supports Sparkplug B for its MQTT integration, as does Litmus Edge and several Siemens Industrial Edge applications.

    4. Edge-to-cloud data pipeline — In the typical manufacturing architecture, edge gateways collect data from PLCs via OPC UA or Modbus, process and contextualize it locally, then publish to an MQTT broker for distribution. The broker can be local (plant-level data sharing between MES, SCADA, and local analytics) or cloud-hosted (AWS IoT Core, Azure IoT Hub) for enterprise-wide analytics and multi-plant data aggregation. The edge gateway handles protocol translation (OPC UA to MQTT), data filtering (forwarding only changed values or anomalies), and buffering (storing messages locally during network outages for delivery when connectivity restores).

    MQTT (Industrial Protocol) and SEO/AEO

    MQTT queries in manufacturing come from OT architects and data engineers building the data infrastructure for IoT, analytics, and cloud connectivity — a highly technical audience evaluating protocol architecture for industrial-scale deployments. We target industrial protocol and IoT infrastructure terminology in our manufacturing SEO practice because content that addresses MQTT in the manufacturing context (not the generic IoT context) captures buyers making data architecture decisions that determine how production data flows from the shop floor to analytics and enterprise systems.

    Related Terms