What antenna spacing should an in-building DAS use per floor, and is there a rule of thumb?

For UHF two-way radio at 400-470 MHz, a workable starting point is one ceiling-mounted omni antenna per 1,000-1,500 sq m of open floor plate, tightening to one per 400-700 sq m where the floor is partitioned into cellular offices or back-of-house rooms. In practice that lands most antennas 20-35 m apart in open space and 15-20 m apart in partitioned space. But spacing is set by what sits between the antenna and the radio, not by the distance on the plan - count walls before you count metres, and treat any number produced before a site survey as a budgeting figure, not a design.

Is there a usable rule of thumb before a site survey?

Yes, and I use one on every first-pass estimate, because a client pricing a tender needs an antenna count before anyone has walked the building. The rule is per-area, not per-distance: one antenna per 1,000-1,500 sq m open plan, one per 400-700 sq m partitioned, plus dedicated antennas for enclosed vertical spaces. On a typical 1,200 sq m Singapore office floor plate that is one to three antennas per floor depending on fit-out, which is usually within one unit of what the survey ends up confirming.

What the rule of thumb cannot do is tell you where those antennas go. Two floors of identical area can need two antennas and six antennas respectively, because one is open-plan and the other is a hotel corridor with 200 mm reinforced-concrete walls between every room. The area figure sizes the bill of materials. The survey sizes the drawing.

Why does distance matter less than walls at UHF?

Because at 450 MHz the structure is a bigger term in the link budget than the range. Indoors, path loss follows roughly a 3.0-3.5 exponent, so tripling the distance from an antenna - 10 m out to 30 m - costs about 15 dB. A single 200 mm reinforced-concrete wall costs 15-25 dB on its own. One wall therefore takes more link budget than tripling the distance in clear air, which is why an antenna sited in the middle of a corridor with concrete either side covers less floor area than an antenna 30 m away across an open floor plate.

The working numbers I design to at 400-470 MHz:

• Plasterboard partition: 3-5 dB

• Hollow block or brick wall: 6-8 dB

• 200 mm reinforced-concrete wall or floor slab: 15-25 dB

• Reinforced-concrete lift shaft or enclosed stairwell: 25-35 dB

• Body loss, portable worn at the hip: 3 dB

• Design threshold for reliable DMR voice: -95 dBm at the portable, with 10 dB fade margin for commercial systems and 15 dB where the system is life-safety critical

Those figures are why the per-floor answer changes with construction rather than with square metres. The same slab attenuation is what makes floor-to-floor bleed unreliable as a coverage strategy, which I have written about separately in how much signal a radio loses through each concrete floor slab or basement wall.

How many antennas does a typical floor plate need?

For an open-plan commercial floor of 1,000-1,500 sq m, one antenna near the core usually covers it, delivering roughly -60 to -70 dBm at the far corners against a -95 dBm threshold. That is comfortable downlink margin, and it is why open floors are cheap to cover.

Partition that same floor into a hotel guest corridor, a hospital ward or a serviced-office fit-out and the count goes to three to five antennas, sited in the corridor with the assumption that each guest room costs one wall of budget. In a basement carpark, a single antenna covers 25-35 m along a clear drive aisle but far less across rows of parked vehicles and RC downstand beams, so I place antennas along the aisles rather than on a uniform grid.

One design constraint applies whatever the count: the coupler and tap schedule must land every antenna port within about plus or minus 3 dB of every other. A DAS with one antenna at +8 dBm and another at -6 dBm has a coverage hole that no amount of extra antennas will fix, because the problem is distribution, not spacing.

Which areas always need their own antenna, regardless of spacing?

Enclosed reinforced-concrete stairwells, lift shafts and lift cars, sprinkler and pump rooms, basement plant rooms, refuse chambers, and cold rooms or freezer stores with insulated metal panel construction. Each is effectively a shielded box, and at 25-35 dB of shell attenuation no adjacent antenna reaches inside. These are also the areas where a radio is most likely to be needed in an emergency, so they belong in scope explicitly, not as something to be discovered at acceptance testing.

How do you check the spacing is right once it is installed?

By grid survey, not by walking around listening. Signal is logged cell by cell across each floor in both directions, uplink and downlink, and assessed against the agreed threshold and audio-quality criterion. If the spacing was wrong, the survey shows it as a cluster of adjacent failing cells rather than scattered marginal points. The method is set out in how in-building two-way radio coverage is measured and passed.

Related questions

Can I just space antennas evenly on a grid across every floor? Only on genuinely uniform floors such as open warehouses or open-plan offices. On any floor with concrete cores, corridors or shielded rooms, an even grid over-serves the open areas and under-serves the enclosed ones, which is the most common cause of a failed floor at acceptance.

Does a higher-gain antenna let me use fewer of them? Rarely indoors. Going from a 2 dBi omni to a 7 dBi directional buys 5 dB, which is less than one concrete wall, and narrows the pattern so the coverage is more directional rather than larger. Adding an antenna where the wall is usually beats adding gain where it is not.

Do antennas need to line up vertically between floors? No, and deliberately offsetting them is often better. Stacking antennas floor over floor concentrates the strongest signal in a single vertical column and leaves the far corners of each plate weakest; staggering them uses the slab loss as isolation instead of fighting it.

How Suneast approaches this

We size the antenna count from the floor plate area and construction type at tender stage, then confirm it with a physical survey before any cable is pulled - reading floor dimensions, core positions and ELV riser access off the drawings, and verifying wall build-ups on site. The design output is an antenna placement plan with setting-out distances a contractor can build from, plus a coupler and tap schedule that keeps every antenna port within plus or minus 3 dB. Coverage is then proven by a bidirectional grid survey at handover rather than asserted from the design. We work across Singapore and Indonesia on UHF DMR and analogue two-way radio systems, including IMDA licensing for the repeater feeding the DAS.