Industrial IoT for Smart Factories: Connecting 20-Year-Old Machines Without Replacing Them

The dirty secret of every "factory of the future" rendering is that real factories are twenty years old on average and profitable precisely because those machines are paid off. The question that matters is never "when do we replace the plant" — it is "how do we get live data out of what we already own". That is what industrial IoT actually does for a living.

The workhorse is the retrofit I/O module. A compact box wires into the signals even the oldest machine already has — run contacts, counters, alarms — and publishes them as clean digital events: running, stopped, cycle completed, fault. No PLC replacement, no controls project, no production stop; installation is measured in hours per machine. From the MES's point of view, a 1998 press and a 2024 machining center now speak the same language.

Energy is usually the second retrofit, because it carries its own business case. Class 0.2S energy analyzers per machine — or LoRaWAN meters where pulling cable is impractical — turn electricity from one monthly invoice into a per-machine, per-product measurement. Idle-load waste and cosφ penalty exposure appear on the first dashboards; deployments typically recover 10-15% of energy cost, and ISO 50001 evidence becomes a by-product instead of a project.

For anything that moves rather than runs, identification and location close the loop. RFID readers at stations confirm which pallet, which tool, which mold actually arrived; UWB-based RTLS tracks forklifts and pallets to within 10-30 centimeters, and smart gates post warehouse movements to the ERP automatically. The manual booking — the scan someone forgot, the transfer typed at shift end — quietly disappears.

Protocol religion is a trap worth naming. The practical stack is boring on purpose: Modbus and OPC UA to talk to machines and meters, MQTT to move events, LoRaWAN where wires cannot go. Any vendor who leads with a proprietary protocol is selling a future hostage situation; insist on standards and the hardware market stays competitive for you.

What separates IIoT that transforms from IIoT that decorates is where the data lands. Sensors feeding a standalone dashboard produce prettier ignorance. The same sensors feeding the MES change decisions: machine states become real OEE, energy joins the product cost, locations drive dispatching, and the AI layer finally has ground truth to learn from. Connectivity is the means; the shared live data model is the point.

A realistic rollout, from deployments we have run: week one, retrofit I/O and meters on one critical line; week two, states and energy live in the MES, first real OEE baseline; weeks three to four, extend to the bottleneck line and add RFID at the handover points. No shutdown, no controls rewrite — and the payback discussion changes from slideware to a measured number on the wall.

Meta Smart Factory ships this as one kit: META Smart I/O modules, iMeter-class energy analyzers, LoRaWAN devices, RFID and UWB RTLS — pre-validated against the platform, with five-year warranty, feeding the same MES, energy and AI modules everything else uses. Legacy machines included, by design.