TL;DR
- Availability is set by fade margin, not by received level: a link at −45 dBm with 12 dB of margin is worse than one at −60 dBm with 30 dB.
- Above 15 GHz rain attenuation dominates and must come from the local R0.01 rainfall rate, not from a default template.
- Path clearance matters more than distance: a hop over growing vegetation closes today and fails in two years.
What does a microwave path budget actually contain?
A path budget starts with transmit power, subtracts feeder and connector loss, adds both antenna gains, and subtracts free-space path loss at the operating frequency and hop length. What remains is the received signal level, and the difference between that level and the receiver threshold for the chosen modulation is the flat fade margin.
Two numbers decide whether the link is real. The first is the fade margin against the required availability. The second is the rain attenuation for the local rainfall region at that frequency. A hop can show a comfortable −45 dBm receive level and still fail, because at 23 GHz in a high-rainfall region the rain event eats 25 dB of it for a few hours a year.
| Band | Typical hop length | Rain sensitivity | Practical note |
|---|---|---|---|
| 6–8 GHz | 30–50 km | Low | Long haul and difficult terrain; scarce and expensive licences |
| 11–15 GHz | 15–30 km | Moderate | The workhorse band for regional backhaul |
| 18–23 GHz | 5–15 km | High | Urban access; rain margin dominates the design |
| 26–42 GHz | 1–5 km | Very high | Dense urban small-cell backhaul; short hops only |
| 70–80 GHz | 0.3–2 km | Very high | Fibre-equivalent capacity, line-of-sight mandatory |
How much fade margin do you need?
Fade margin follows from the availability target. A link specified at 99.9% availability needs far less margin than one at 99.999%, and the difference in required margin is not linear — it grows quickly as the target tightens. The calculation itself is standardised in ITU-R P.530, which combines multipath fading for narrowband and wideband cases with rain attenuation from the rainfall intensity exceeded for 0.01% of the year.
In practice, the availability target should come from the traffic carried, not from a default. A microwave hop carrying a rural 3-sector LTE site and a hop aggregating thirty sites into a core site do not deserve the same target, and specifying the higher target everywhere adds antenna size and cost across the whole network for no measurable service benefit.
Why does rain rate matter more than distance above 15 GHz?
Free-space loss grows with the logarithm of distance, so doubling a hop adds about 6 dB. Rain attenuation grows roughly linearly with path length in the rain cell but is driven primarily by rainfall intensity, and intensity varies by more than an order of magnitude between climatic zones. Two 8 km hops at 23 GHz — one in a temperate coastal city and one in a tropical delta — can differ by 20 dB of required margin.
What kills links that pass the calculation?
Clearance is the first. A path needs 60% of the first Fresnel zone clear of terrain, buildings and vegetation, and vegetation grows. A link planned over young trees closes for two years and then stops closing. Where clearance cannot be achieved, the design must reduce hop length or lower the frequency rather than accept a partial obstruction.
Alignment and mounting are the second. A 0.6 m dish at 23 GHz has a beamwidth of about 1.5°, and a two-degree pointing error costs several dB. On shared towers with wind loading, the mount stiffness matters as much as the initial alignment. Interference is the third: licence-free bands and lightly coordinated bands accumulate interferers that no single-link calculation sees, so the noise floor must be measured, not assumed.
How do you plan a microwave layer for a new rollout?
Start from the aggregation topology, not from individual links. A star from each site to the nearest hub is cheap but concentrates risk; a ring around each hub survives a single link failure. For networks where a hub failure removes a whole town, the ring is not optional.
Then group sites by required capacity and distance band, and assign frequency bands to those groups before designing any single hop. Designing hop by hop and discovering afterwards that half the network sits in one congested band is the standard failure of ad-hoc microwave planning. Finally, reserve capacity headroom on aggregation hops for the LTE-to-5G transition: the access layer will add carriers while the backhaul topology stays fixed.