All work
Industrial operations · Telemetry & monitoring
Industrial Monitoring & Telemetry
Equipment and process telemetry from PLCs and sensors into live dashboards and alarms.
Operational· pending verificationPublic summary — client anonymised
Publication draft: status, contribution and technical details require contributor approval. The architecture and implementation notes below are illustrative, not verified accounts of a client deployment.
Context & challenge
Plant condition is only known when someone walks the site. Faults are found late, and history is incomplete.
Objectives
- Continuous visibility of key process values
- Alarms on threshold and communication failure
- Historical trends for diagnosis
Solution approach
- Modbus polling and digital/analog acquisition at the edge.
- MQTT topic hierarchy with retained last-known state.
- Streaming dashboards with explicit stale-data indication.
Architecture
- Sensors
- PLC / Modbus
- Edge gateway
- MQTT
- Ingest service
- Time-series DB
- Dashboard
- Alarms
Simplified and anonymised. No private infrastructure details are published.
Important technical decisions
- Treat 'no data' as an alarm state, not a blank chart
- Retained messages for instant dashboard state on reconnect
Operational considerations
- Sensor failure vs. process fault
- Clock drift on edge devices
- Network outages
Results
Results pending verification. We only publish outcomes that have been confirmed.
Lessons learned
- Operators trust a dashboard that shows its own data age.
Visual demonstration
An interactive, synthetic demonstration related to this work is available in the laboratory.
Open demonstration