Wireless sensor networks

The usual argument for a wireless system is that nobody has to walk round with a cable. That is true and it is the least important benefit. The real case is alarms: knowing that a store went above 70 percent relative humidity last night, rather than discovering it during the next quarterly download after three weeks of mould growth.

What you actually gain

Time to react. A failed dehumidifier in a damp store is a serious problem after a fortnight and a manageable one after a day. Only continuous monitoring with an alert closes that gap.

Evidence during an incident. When a heating system fails over a holiday weekend, a continuous record with timestamps is what allows an insurer, a contractor or a funder to be shown exactly what happened and when.

Data from places you cannot reach. A sensor high in a roof void or inside a sealed case can report without anyone climbing or opening anything.

Removal of the human failure mode. Monthly downloads get missed. They get missed most reliably during the busy periods, which are frequently also the risky ones.

What it costs beyond the hardware

Subscription. Most systems charge annually per sensor or per site for the platform. Over a ten-year life this frequently exceeds the hardware cost, and it should be in the business case from the start.

Infrastructure. Gateways need power and a network connection, and both need to exist where the sensors are, which in a stone-walled store is not always straightforward.

Radio reality. Thick masonry, metal racking and lead-lined rooms all attenuate radio badly. Ask for a trial rather than accepting a coverage claim, and test the worst location rather than the easiest one.

Battery management. Continuous transmission drains batteries faster than a logger writing to memory. Multiply the number of sensors by the replacement interval and see whether you have swapped a monthly download round for a monthly battery round.

Lock-in. Proprietary sensors that only talk to one platform mean the platform’s commercial decisions become your problem.

The questions to ask before committing

• What happens to readings when the gateway or the internet link is down? Buffered and backfilled, or lost?
• Can we export the full raw record, in an open format, without the vendor’s involvement?
• What is the stated sensor accuracy and drift, and is there a calibration path? The wireless part changes nothing about drift.
• Who receives alarms, at what threshold, and what happens when that person leaves?
• What is the total cost over ten years including subscription and batteries?
• Can we trial two sensors in the worst radio location in the building before ordering the rest?

The hybrid that usually wins

Very few collections need everything on a network. The arrangement that tends to give the best value is mixed: wireless sensors with alarms in the few spaces where active equipment is holding a condition or where a failure would be expensive, and standalone loggers everywhere else.

That keeps the subscription small, puts the alarm where it earns its keep, and leaves the rest of the building on instruments that keep recording through a power cut and owe nothing to a supplier’s continued existence.