Microwave Backhaul Rain Fade Availability Case Study

Microwave backhaul links can look healthy during commissioning and still miss their availability target during heavy rain. The problem is not that the link never closes. The problem is that weather consumes fade margin, the modem falls to a lower modulation and coding mode, capacity drops below the service requirement, and higher-layer queues begin to fail before the radio fully loses carrier.

This case study follows an 18\ \text{GHz} point-to-point backhaul link serving a remote access site. The link passed clear-sky commissioning with good received signal level, but operations later report repeated evening service degradation during thunderstorms.

The purpose is to connect RF link budget, rain attenuation, adaptive modulation, network capacity, outage accounting, and validation evidence into one availability decision.

Case Context

The link carries access traffic from a remote site to an aggregation site. The service has an availability target of 99.99\% for the committed backhaul capacity. The radio supports adaptive modulation and remains synchronized during many rain events, but the service is considered unavailable when usable capacity falls below the committed level.

Item	Value
Frequency	18\ \text{GHz}
Path length	12.0\ \text{km}
Channel bandwidth	56\ \text{MHz}
Transmitter power	20\ \text{dBm}
Antenna gain at each end	39\ \text{dBi}
Feeder and connector loss at each end	2\ \text{dB}
Miscellaneous clear-sky allowance	3\ \text{dB}
Receiver noise figure	5\ \text{dB}
Committed service capacity	120\ \text{Mbit/s}
High mode capacity	300\ \text{Mbit/s}
Medium mode capacity	180\ \text{Mbit/s}
Robust mode capacity	50\ \text{Mbit/s}
Required SNR for medium mode	20\ \text{dB}
Implementation and aging reserve	4\ \text{dB}
Observed degraded minutes in one wet month	138\ \text{min}

The important distinction is between link-up availability and service availability. A radio can remain up in robust mode while the committed service is unavailable because the offered load cannot be carried without queueing and packet loss.

Clear-Sky Link Budget

Free-space path loss using distance in kilometers and frequency in MHz is:

For:

the loss is:

Clear-sky received power is:

Substitute the values:

This level is consistent with a healthy clear-sky microwave link, assuming the antenna gains, feeder losses, and alignment are real.

Noise Floor and Clear-Sky SNR

Receiver noise floor is:

For:

With NF=5\ \text{dB}:

Clear-sky SNR is:

The medium mode needed to carry the committed service requires:

With implementation and aging reserve:

The link has about 19.4\ \text{dB} of clear-sky service margin above the medium-mode requirement. That sounds comfortable until rain attenuation is checked.

Rain Attenuation Screen

During the outage review, local weather records and radio logs show that the worst events correspond to intense convective rain along the path. Use a simplified rain-fade screen:

where \gamma_R is specific attenuation and d_{eff} is the effective rain path length. For this case, the path review uses:

and:

Therefore:

The screened rain attenuation is larger than the clear-sky service margin:

This predicts that the link will fall below the medium mode during severe rain. The radio may remain connected, but the committed capacity cannot be guaranteed.

Rain-Event Received Level

The received power during the screened rain condition is:

Rain-event SNR is:

After the 4\ \text{dB} reserve:

The medium mode requires 20\ \text{dB} usable SNR, so it is not available. The robust mode remains available, but robust mode carries only:

That is below the committed:

The physical link is degraded, not fully down. The service is still unavailable for the committed-capacity requirement.

Capacity and Queueing Consequence

During evening storms, offered traffic often reaches:

In robust mode, link capacity is:

The excess offered traffic is:

If this condition lasts 10\ \text{min}, the backlog generated is:

Convert to gigabytes:

A buffer cannot absorb this without large latency and eventual loss. Real-time traffic fails before the radio declares link down. This explains why user complaints, packet loss, and high jitter appear while the radio management system still reports carrier lock.

Availability Accounting

For a 30-day month:

The observed degraded time below committed capacity is:

Service availability for the month is:

So:

The monthly downtime allowed by a 99.99\% target is:

The observed degraded time is:

about 32 times the monthly outage budget. The failure is not a minor nuisance; the design basis does not meet the availability requirement for the installed path and weather exposure.

Diagnosis

The evidence supports rain fade as the dominant cause:

Evidence	Interpretation
clear-sky received level matches commissioning	no persistent alignment or feeder-loss fault
degraded events correlate with intense rain cells	weather is the trigger
SNR falls before packet loss and MCS fallback	RF margin is being consumed
carrier remains locked in robust mode	physical link is not always hard down
packet loss and jitter rise when capacity drops	service layer is overloaded during fallback
spectrum scans show no matching interferer during events	interference is not the primary cause

The root problem is not that adaptive modulation failed. Adaptive modulation preserved a lower-rate link. The root problem is that the path was accepted without enough fade margin, diversity, or traffic policy to meet the committed service during severe rain.

Corrective Options

Possible corrective actions should be evaluated as engineering alternatives, not as cosmetic tuning:

1. Increase antenna gain. Larger antennas at both ends can add several dB of margin. A 3\ \text{dB} gain increase at each end adds about 6\ \text{dB} total link margin, but tower loading, beamwidth, alignment tolerance, licensing, and wind survival must be checked.
2. Shorten the hop. Adding an intermediate site can reduce path loss and rain path length per hop. This is often more robust than forcing one long high-frequency hop to meet a high availability target.
3. Use a lower frequency band. Lower microwave bands are usually less sensitive to rain fade, but spectrum availability, antenna size, licensing, and interference must be reviewed.
4. Add a diverse fiber or radio route. Availability can be improved by route diversity if failure modes are independent. A backup path with the same rain exposure may not solve the problem.
5. Shape traffic during fallback. Priority traffic can be protected when capacity drops, but traffic shaping does not make the committed high-capacity service available.
6. Set earlier alarms. Alarm thresholds should warn when fade margin is being consumed, not only when the link drops.

For this case, the preferred technical path is larger antennas plus fallback traffic policy as an immediate mitigation, followed by a route-diversity review for the committed service.

Mitigation Screen

If both antennas are upgraded by 3\ \text{dB} gain, clear-sky received level improves by:

Clear-sky service margin becomes:

Against the screened rain attenuation:

This provides positive but thin margin for the screened event. It may be acceptable as an interim mitigation if monitoring confirms the rain model is conservative enough. It is not a strong long-term answer if the service contract needs high confidence under rarer storms, because only 2.3\ \text{dB} remains.

If a midpoint relay splits the path into two 6\ \text{km} hops, free-space loss per hop is reduced by about:

and each hop also has a shorter effective rain length. The trade-off is added equipment, site power, operations burden, and another failure point. The better design depends on total service availability, not one RF number.

Validation After Mitigation

The link should not be released from corrective action based only on a stronger clear-sky received level. Validation should prove that the service behaves acceptably during margin loss.

Acceptance evidence should include:

1. updated link budget with antenna, feeder, alignment, rain, implementation, and aging assumptions separated;
2. measured received signal level after final alignment at both ends;
3. alarm thresholds tied to received level, SNR, MCS state, error seconds, and capacity state;
4. controlled attenuation or vendor-supported fade test showing MCS transitions and traffic shaping behavior;
5. service test in high, medium, and robust modes with throughput, latency, jitter, and packet-loss records;
6. monitoring report over a wet-weather observation period, including rain-event received level and degraded minutes;
7. availability calculation based on committed capacity, not only radio carrier uptime;
8. route-diversity decision if the residual outage budget is still too small.

The acceptance criterion should be explicit: the link is accepted only if degraded minutes below committed capacity remain within the outage budget or if a documented backup route carries the committed service during fade events.

Engineering Lessons

The first lesson is that fade margin must be tied to an availability requirement. A link that has enough SNR in clear weather may still be a poor service link in a high-rain path.

The second lesson is that adaptive modulation changes the failure mode. It can prevent a hard outage while creating a lower-capacity state that still violates the service requirement.

The third lesson is that availability should be counted at the service boundary. If the committed traffic cannot be carried, the service is unavailable even if the RF carrier is still locked.

Good microwave backhaul engineering therefore separates clear-sky margin, rain fade, implementation reserve, MCS capacity, traffic policy, and route diversity. The design is complete only when the physical link and the network service meet the same availability claim.