TL;DR
- Choose by floor area, peak user density, and whether the deployment must serve more than one operator.
- Above roughly 20,000 m², or where multi-operator neutrality is required, a DAS is usually correct; below that, small cells are cheaper to install and to change.
- Repeaters amplify interference as faithfully as signal, and many operators refuse them as policy rather than as a technical judgement.
What decides between DAS and small cells?
Three inputs decide it: total floor area, peak user density per floor, and whether the deployment must serve more than one operator. Above roughly 20,000 m² of covered area, or where multi-operator neutrality is a requirement, a distributed antenna system is usually the right answer because the head-end can be shared and the radio equipment stays in one place. Below that, small cells are cheaper to install and far cheaper to change.
User density matters independently of area. A 5,000 m² exhibition hall at full occupancy has the same capacity demand as a 40,000 m² office building, but with a much smaller footprint to cable. That combination favours small cells with sectorised coverage over a DAS.
| Option | Typical area | Relative cost | Operator approval | Main risk |
|---|---|---|---|---|
| Passive DAS | 10,000–100,000 m² | High install, low operating | Design review with each operator | Loss budget limits usable size; inflexible once built |
| Active DAS / digital | 20,000–500,000 m² | Highest | Full coordination, capacity planning per operator | Cost overrun on cabling and power |
| Small cells | 500–20,000 m² | Low to moderate | Neighbour and PCI coordination per cell | Interference with the macro layer; many cells to manage |
| Repeater | 200–2,000 m² | Lowest | Often refused outright | Amplifies interference; degrades donor cell |
| Femtocell (operator-issued) | 100–1,000 m² | Very low to customer | Operator-managed | Backhaul dependent; limited capacity |
When is a repeater still defensible?
Almost never on a public network, and the reasons are technical rather than regulatory. A repeater cannot distinguish signal from interference, so it amplifies both. It adds noise to the donor cell's uplink, it extends the donor cell's footprint into places the donor was not planned to serve, and it raises the donor's interference floor for every other user. The result is a small local improvement paid for by a network-wide degradation.
The narrow case where a repeater works is a single-operator, low-traffic, physically isolated space where no backhaul exists and the donor cell is lightly loaded — a basement car park in a rural town, for example. Even then, the operator's approval is the binding constraint, and many refuse repeaters as policy.
How do you size an indoor system before a site survey?
Estimate the link budget from the outside in. Take the macro RSRP at the building facade from a prediction or a walk test, subtract the penetration loss of the facade construction, subtract the internal partition loss per wall the signal must cross to reach the worst-served point, and compare with the target indoor level. The gap tells you whether you need a new signal source or only better distribution.
Then convert to antenna count. Each indoor antenna covers a radius determined by the same link budget, typically 15–25 m in open office space and 8–12 m through dense partitions. Antenna count times per-antenna cost gives the order of magnitude of the distribution cost, which usually dominates the decision long before the radio equipment does.
What does the operator need to approve?
For small cells: physical cell identity allocation, neighbour list entries in both directions, tracking area assignment, and confirmation that the new cells will not create handover ping-pong with the macro layer. For a DAS: the design, the loss budget, the equipment list, and the capacity assumption per operator. For anything that radiates into a licensed band, the operator's radio planning team owns the decision, and their review cycle is usually the longest item on the project schedule.
Start that conversation before the commercial proposal is signed, not after. A design that is technically excellent and unapprovable has zero value, and the approval criteria differ enough between operators that a design accepted by one is routinely rejected by the next.