Skip to content

Product Overview

EMQX Neuron (formerly NeuronEX) is an industrial edge gateway that runs on the plant floor — in manufacturing, energy, and building automation. Device-side protocols are fragmented and data formats are inconsistent, while MES, SCADA, and cloud platforms need standardized real-time data. EMQX Neuron collects from PLCs, CNC machines, robots, and meters, cleans and computes on that data at the edge, and delivers it in the shape the systems above expect.

Field devicesPLCCNC machinesRobotsSmart meters · DCS · SCADAOther data sourcesHTTP APIs · SQL databasesFiles and logs · Video streamsKafka · Redis · CANCollectIngestEMQX NeuronData collection100+ protocols · 50 southbound driversData processing160+ functions · streaming · AI/MLData forwardingpublish out · serve on sitePublishWrite backStoreRead / writeIoT platforms & enterprise systemsMQTT · AWS IoT Core · Azure IoT HubSparkplug B (UNS) · Kafka · HTTPDatabases & storageMySQL · PostgreSQL · SQL Server · OracleInfluxDB · Redis · Files · AWS S3On-site systemsconnect in via OPC UA ServerSCADA · HMI · MESHistorians · HMI software

Core capabilities

Capability
Description
Learn more
Data collectionSouthbound drivers connect 100+ industrial protocols, covering Modbus, OPC UA, EtherNet/IP, IEC 60870, BACnet, Siemens and Mitsubishi PLCs, and a range of CNC controllersSouthbound Drivers
Data processingA built-in stream processing engine with 160+ functions for filtering, transformation, aggregation, and windowing; extensible with Python/C++ functions and AI/ML modelsData Processing
Data forwardingPublish to IoT platforms and enterprise systems, write into databases, or serve data on the plant floor through an OPC UA ServerNorthbound Applications
OperationsA web console for configuration, users and permissions, log download, runtime monitoring and alerts, with master-backup deploymentOperations

Getting started

  • Get one pipeline working firstQuick Start: run an instance in Docker, collect from a simulated device, and forward to MQTT.
  • Deploy to productionInstallation and Deployment: tar.gz, rpm, deb, and Docker, plus master-backup deployment.
  • Check whether your device protocol is supportedSouthbound Drivers: look up your protocol or CNC model in the compatibility tables.
  • Understand how it is put togetherArchitecture.

Data destinations

As the diagram shows, collected and processed data leaves in four directions:

Destination
HowNotes
IoT platformsMQTT, AWS IoT Core, Azure IoT Hub, Sparkplug BBidirectional: publish tag values, and accept write commands from the platform to control devices. Sparkplug B builds a unified namespace (UNS); the docs include verified walkthroughs for Ignition and Cogent DataHub
Enterprise systemsKafka, HTTP, WebSocketFeed the enterprise message bus for MES and data platforms
Databases and storageMySQL, PostgreSQL, SQL Server, Oracle, InfluxDB, Redis, AWS S3Written by a data processing sink, optionally aggregated or downsampled first
Served on the plant floorOPC UA ServerOn-site SCADA, HMI, MES, and historians connect in as clients to read tags and write control commands

Serving OPC UA on the plant floor

The first three push data outward. The OPC UA Server reverses that: EMQX Neuron exposes an OPC UA service, and the SCADA, HMI, MES, and historian systems already installed in the plant connect to it as clients — subscribing to tag changes, reading live values, and writing control commands back down.

On an existing production line, the value is that nothing upstream has to change. SCADA already speaks OPC UA, so once it connects, the hundred-plus device protocols collected below become a single OPC UA data source — Modbus, Siemens S7, Mitsubishi, and CNC devices no longer have to be integrated one by one. Security covers policies such as Basic256Sha256, username and password authentication, and mutual validation with a server certificate and trusted client certificates.

Deployment and performance

  • Low latency: collection intervals down to 100 ms, with processing at the edge instead of round-tripping to the cloud.
  • Lightweight: low memory footprint, x86 and ARM support, runs on industrial PCs and gateway hardware, and deploys via Docker and Kubernetes.