What a link budget is for
A link budget answers one question: how much path loss can this link tolerate before the service fails? That number — the maximum allowable path loss — is what a propagation model converts into a distance, and therefore into a site count.
It is built by accounting for every gain and every loss between the transmitter output and the receiver input. Get the accounting right and the site count is defensible. Omit a margin and the network is short of coverage everywhere, uniformly, in a way that is expensive to discover after build.
Transmit side: EIRP
EIRP (dBm) = P_tx (dBm) + G_ant (dBi) − L_cable (dB)
Transmit power at the port, plus the antenna gain that concentrates it, less the feeder and connector loss on the way. Everything downstream is measured against this single figure.
Receive side: sensitivity
Receiver sensitivity is the weakest signal that still supports the target modulation. It is the thermal noise floor, raised by the receiver noise figure, plus the minimum SINR the modulation scheme needs.
Sensitivity = (−174 + 10·log₁₀(BW_Hz) + NF) + SINR_min
10 MHz LTE, 7 dB noise figure, QPSK at −5 dB minimum SINR gives −174 + 70 + 7 − 5 = −102 dBm. Raise the target modulation and sensitivity worsens — higher order modulation buys throughput with coverage.
Maximum allowable path loss
MAPL = EIRP − Receiver sensitivity − Margins
The margins term is the part most often underestimated. On an indoor-served design it can exceed 30 dB, which is more than the entire antenna gain.
The margins, and why they dominate
Margins are not padding. Each one accounts for a specific physical effect that the median path loss model does not capture.
| Component | Typical value | What it accounts for |
|---|---|---|
| Transmit power, per port | 20 W (43 dBm) | Sets EIRP together with antenna gain. |
| Antenna gain | 16–21 dBi | Roughly 1 dBi buys about 11 per cent more range, all else equal. |
| Shadow fade margin | 8–10 dB | Log-normal variation around the median path loss. Sets the location probability — typically 90 to 95 per cent. |
| Body loss | 3 dB | Absorption by the user holding the handset. |
| Penetration loss, outdoor | 0–5 dB | Vehicles, foliage and street clutter. |
| Penetration loss, indoor concrete | 15–25 dB | Building fabric. The largest single term in most urban budgets, and the one that decides whether indoor coverage is achievable from outside at all. |
| Interference margin | 2–5 dB | Rise over thermal from co-channel neighbours. Grows as the network loads. |
State the assumption with the answer
A cell radius is meaningless without the margin set that produced it. Always publish the target — for example 95 per cent location probability, indoor, 20 dB penetration — alongside the number. Two budgets that differ only in assumed penetration loss can differ by a factor of two in site count.
From path loss to cell radius
R = 10^((MAPL − A) / B) km
A and B are the constants of the chosen propagation model — for Okumura-Hata they come from the frequency, base station height and environment terms. The inversion is only as good as the model behind it, so the model must match the clutter.
A worked example
An outdoor-to-indoor design on a 10 MHz LTE carrier, targeting QPSK at the edge:
- EIRP — 43 dBm transmit + 18 dBi antenna − 2 dB feeder = 59 dBm
- Sensitivity — −174 + 70 + 7 − 5 = −102 dBm
- Margins — 9 dB shadow fade + 3 dB body + 20 dB penetration + 3 dB interference = 35 dB
- MAPL — 59 − (−102) − 35 = 126 dB
126 dB is a modest budget, and that is the point of doing the arithmetic: the 20 dB of building penetration has consumed more than the antenna contributed. The honest conclusion from a budget like this is often that indoor coverage needs an indoor solution rather than a taller mast.