IoT integration

A sensor reporting every ten seconds produces just over 8,600 readings a day. With a hundred sensors, that is over 860,000 a day. Receiving them is a known problem with known solutions. The questions that take time are what to keep, for how long, and who should act on them.

Expect the connection to drop

Equipment sits in basements, on roofs and in vehicles. Collection needs to buffer locally when the link goes and send on when it comes back. How long it must cope without contact, whether the order of readings matters and how the receiver recognises a reading sent twice are requirements we set together with you before anything is built.

Alarm locally when it is urgent

A threshold evaluated only in the cloud will not alarm if it is the connection that has failed. Anything that must alarm even then should alarm locally, on the equipment or a gateway next to it. Trends, history and anything that is not time-critical belong centrally. If the alarm concerns safety, it has its own requirements, and they come before everything else here.

From reading to action

The data is collected and a dashboard gets built. If no process receives the alarms, the project stops there: a coloured box someone has to happen to look at. So we connect readings to case management or process automation, so that an exceeded limit becomes a work order with an owner.

Once there is enough history, the same data can flag equipment that starts behaving differently before it breaks. It does not replace the maintenance you are obliged to do on schedule, but it can guide what you look at in between. The order matters: collection and action first, analysis after.