RF site surveys, measured not assumed

A range figure on a datasheet is measured in open air with clear line of sight. No working site is open air. Between a tag in a plant room and a gateway at height there is reinforced concrete, structural steel, foil-backed insulation, and whatever else on the site already transmits. Fitting the kit and finding out is the costly, time-consuming way to learn that. Attending the site with the right equipment and finding out first avoids it. We do it properly so that on installation day there are no surprises.

Three things decide whether a link holds: how much signal arrives, how much noise it arrives into, and how much headroom sits between them. Test tags go where tags will really sit, inside the plant room, down in the chamber, behind the switchgear, and received signal is measured at each candidate gateway position. That gives measured path loss for that path in that building, which is a different number from the one a model predicts. The noise floor is measured too, and measured across a period that includes normal operation, because interference from process equipment appears when the equipment runs and disappears when it stops. A band swept during a quiet hour describes a site that does not exist during production. Headroom is what the design is built on. A link engineered to the edge works on the day and fails when a door closes or a vehicle parks.

The wavelength at 433 MHz is roughly 69 cm against roughly 35 cm at 868 MHz. The longer wave diffracts more readily around obstructions and passes more effectively through gaps in structure, and free-space path loss rises with the square of frequency, so 868 MHz starts around 6 dB behind before anything else is counted. In buildings with real structure between the tag and the gateway that margin is often the difference between a link that holds and one that does not. Both bands are licence-free in the UK. Our deployments are predominantly 433 MHz because the positions our clients need monitored are difficult ones. Where a site already runs 868 MHz equipment, integrating with it can be assessed during the survey.

Attenuation is where predictions fail. Reinforced concrete, structural steel, lift shafts and below-ground positions all take signal out, and foil-backed insulation is worse than most people expect, because a metallised layer behaves as a reflector across a whole elevation. Mounting matters just as much and gets less attention. A tag fixed to a metal pole because the pole happened to be next to the equipment will perform differently from the same tag fixed a short distance away, and an antenna mounted hard against metal can lose most of what it should radiate. Polarisation counts as well, since a mismatch between transmit and receive orientation gives away signal for nothing. Someone without RF experience will mount kit where it is convenient. Every one of those decisions has a consequence, and the survey is where they get made deliberately.

A survey captures a site on one day, and sites move. Loading bays fill and empty. Stock stacks change height. Racking gets reconfigured, plant is replaced, vehicles park in new places. A large metal box parked in a gap that was clear during the survey will change what arrives at the gateway. So the survey looks at how the site behaves rather than only how it looks, and the design carries enough headroom to absorb the movement that is normal for that operation.

Some of the most useful information on a survey does not come off an instrument. It comes from the people who work there. They know which bay fills up, when the racking gets moved, which plant runs at night, what changed last year and what is changing next. Being on site to see it and to ask is why the design fits the operation rather than the drawing. This is the part that cannot be done remotely.

A gateway position that works on radio is not always one the building allows. It needs a 5 V supply from a mains plug, a DC rack feed or solar, somewhere it can be mounted, access for maintenance, and height usually helps. Every deployment includes its own gateway, so coverage does not depend on a third-party network reaching your site. The survey reconciles where the radio wants the gateway with where the building will accept it.

The report states gateway count and siting, tag count, the tag types matched to what is being monitored, the positions that need attention, and the measurements behind each conclusion. Because the numbers are measured they can be checked and challenged. The report then informs a pilot, and the pilot proves the design in the field before a full rollout.