Is DAQ 3.0 or a -95 dBm Signal Level the Right Acceptance Criterion for In-Building Radio Coverage?

Both, and they are not alternatives. A -95 dBm signal level is an input measurement: how much RF power arrives at the receiver. DAQ 3.0 is an output measurement: whether the person on the other end can actually understand the speech. A system can deliver -95 dBm across every test cell and still fail on audio quality if the noise floor in that building is elevated, so a specification that names only one of the two is incomplete. Write both into the acceptance criteria, and measure both.

What does -95 dBm actually guarantee?

It guarantees headroom, nothing more. A typical UHF portable has a receiver sensitivity around -119 dBm at 12 dB SINAD, so -95 dBm sits roughly 24 dB above the point where the radio stops working at all. That gap is the fade margin, and it is there to absorb the things a survey cannot capture: a guard turning his body between the radio and the antenna, a fire door closing, a trolley parked in a corridor, a new partition wall next year.

The catch is that -95 dBm is only meaningful against a quiet noise floor, and in a real building the noise floor is rarely quiet. LED drivers, variable-frequency drives on lift and chiller plant, switch-mode power supplies and poorly filtered signage all raise it, and a floor sitting 15 dB above thermal noise turns a comfortable -95 dBm into a marginal signal-to-noise ratio. The power meter still reads -95 dBm. The audio still sounds bad. That is why we treat a level-only specification as a design input rather than an acceptance test.

What does DAQ 3.0 mean, and is 3.0 the right target?

DAQ, or Delivered Audio Quality, comes from TIA TSB-88 and the underlying NTIA work on Project 25 systems. It is a subjective scale from 1 to 5 mapped onto measurable SINAD for analogue and bit error rate for digital. DAQ 1 is unusable: speech is present but not understandable. DAQ 2 is understandable with considerable effort and needs frequent repetition. DAQ 3 is understandable with slight effort, requiring occasional repetition. DAQ 3.4 is understandable without repetition, with some noise or distortion present. DAQ 4 is easily understandable with little noise, and DAQ 5 has no discernible noise or distortion at all.

DAQ 3.0 explicitly tolerates occasional repetition. For a housekeeping channel that is fine. For a fire command post talking to a search team in a smoke-logged stairwell it is not, and that is why the more demanding US jurisdictions specify DAQ 3.4 rather than 3.0 for emergency responder coverage. If you are writing a specification for critical areas, meaning the fire command centre, stairwells, lift lobbies and basement plant rooms, specify 3.4 there and 3.0 for general areas. The cost difference at design stage is small. The cost difference after handover is a re-test.

So what should the specification actually say?

A defensible in-building radio coverage acceptance criterion has four parts, not one. It states a minimum received level in both directions, an audio quality or equivalent SINR target, a coverage percentage, and the grid the test is run on. The figures we work to on Singapore projects:

• Minimum -95 dBm received signal, measured in both directions: downlink at the portable and uplink at the repeater

• DAQ 3.0 in general areas and DAQ 3.4 in critical areas, or the equivalent SINR or BER for the technology in use

• 95% of test cells passing on each floor, and 99% in designated critical areas

• Test cells of roughly 6 m by 6 m (a 20 ft grid), with the cell centre as the measurement point

• Design fade margin of 10 dB for commercial systems and 15 dB for mission-critical systems

Two of those deserve emphasis. Uplink is the one most often omitted and most often the limiting path, because a 4 W portable at hip level with 3 dB of body loss is fighting a 40 W repeater, and a downlink-only survey will happily pass a system that cannot hear its own users. And the percentage rule matters more than it looks: adjacent failed cells are worse than scattered ones, because two neighbouring failures make a contiguous dead area a person can stand in rather than a spot they walk through.

Related questions

Does Singapore mandate -95 dBm or DAQ 3.0? No. Singapore has not adopted an IFC 510-style emergency responder radio coverage code. Obligations here come from IMDA's COPIF and the SCDF Fire Code, and neither prescribes the US numbers. In practice the -95 dBm and DAQ criteria are used contractually, because they are the only well-established figures available. See our note on how Singapore's requirements compare to IFC 510.

Can a system pass on level and fail on audio? Yes, and it is common in car parks and plant rooms. Elevated noise from variable-frequency drives and LED lighting degrades the signal-to-noise ratio while leaving the measured power level untouched. Always measure both, and record the noise floor per floor as a separate line in the test report.

Who decides the criteria if no code applies? The building owner or the specifying consultant, in the tender. If the tender is silent, the contractor will test to whatever is easiest to pass. Our approach to the grid survey and pass/fail method sets these out before installation, not after.

How Suneast handles this

We write the acceptance criteria into the design at survey stage, before any cable is pulled, and we test to them at handover in both directions with the noise floor recorded per floor. Every project is handed over with a per-floor grid result, the measured uplink and downlink levels, and the fade margin against the design threshold, so the figures are auditable rather than asserted. If a floor comes in marginal, we say marginal, and we fix it before the certificate rather than after the first incident.