The Short Answer
Fade margin is the number of decibels by which the received signal sits above the weakest level the receiver can still decode. It is the reserve a radio link holds in hand to ride out fading from multipath, rain and atmospheric refraction without dropping the connection.
Fade margin (dB) = Received signal level − Receiver threshold
The received signal level is the power your link budget predicts on a calm, clear day. The receiver threshold, also called receiver sensitivity, is the weakest signal the radio can still demodulate at the chosen modulation and error rate. Subtract one from the other and the difference is how far the signal is allowed to fade before the link fails.
How much you need is set by the availability target. In the multipath regime, every extra 10 dB of fade margin makes the link roughly ten times more available, which is why a high reliability microwave hop is usually engineered for 20 to 40 dB. Below that, the link works on a fine day and falls over in weather.
Fade margin starts from the received signal level, which is the output of a link budget:
Received signal level (dBm) = Ptx + Gtx − Ltx − Path loss + Grx − Lrx
The fade margin is then the gap to the threshold:
Fade margin (dB) = Received signal level − Receiver threshold
- Received signal level is the predicted receive power in dBm, the clear-air number
- Receiver threshold is the sensitivity in dBm for the target bit error rate, usually quoted at 1 in 10⁻⁶, and it depends on the modulation and channel bandwidth
A worked link that receives −51 dBm into a radio with a −67 dBm threshold has a fade margin of 16 dB. That 16 dB is not spare loudness the link enjoys all the time. It is the depth of the deepest fade the link can absorb before the receiver loses lock.
One distinction is worth fixing early. The headroom you measure, received signal level minus threshold, is the link margin. The portion of that headroom you deliberately reserve to survive fading is the fade margin. On a sound design the two are the same number, because the whole of the margin is there to be spent on fades. The link is healthy when the link margin is at least as large as the fade margin the availability target demands.
What Fade Margin Protects Against
The received signal level is a single clear-air figure, but a real path is never static. Five mechanisms move the signal around, and fade margin is the single reserve that has to cover all of them.
| Fading mechanism | What causes it | Where it dominates |
|---|
| Multipath, flat | Reflected and refracted copies arrive out of phase and cancel | Below about 10 GHz, long flat paths, over water |
| Frequency selective | Wideband signal fades unevenly across the channel | Wide channels, long paths, high capacity links |
| Rain attenuation | Raindrops absorb and scatter the signal | Above about 10 GHz, heavy rain climates |
| Atmospheric refraction | Changing air density bends the beam up or down | Calm, humid nights, coastal and tropical paths |
| Gaseous absorption | Oxygen and water vapour absorb energy | Near the 22 GHz and 60 GHz absorption lines |
Multipath, or flat fading, is the usual problem below 10 GHz. The atmosphere splits into layers that send the beam along slightly different routes. When two copies arrive out of phase, they cancel. These fades are deep, fast and short, lasting seconds. They are why a link with thin margin can flicker on a still, clear night.
Frequency selective, or dispersive, fading appears once the channel is wide. Parts of the band fade by different amounts at the same moment, which distorts the signal rather than just weakening it. A separate dispersive fade margin covers it, described later for the links that need it.
Rain attenuation takes over above about 10 GHz. Water is lossy at these frequencies, so a heavy cell on the path can add tens of decibels of loss for as long as it lasts. It does not flicker like multipath. It ramps up, holds, then clears, and on the high bands it sets the budget.
Atmospheric refraction bends the beam up or down. The design assumes an effective earth radius factor, k, of four thirds. On a still, humid night k can drop toward two thirds, lifting the apparent terrain into the path, or rise into ducting that steers the beam off the far antenna. The worst case clearance check uses k of two thirds for this reason.
Gaseous absorption from oxygen and water vapour is minor on most bands. It climbs near the 22 GHz water line and the 60 GHz oxygen line, where it has to be carried in the budget.
The Two Fade Margins That Matter
For most narrowband links, the flat fade margin tells the whole story.
Flat, or thermal, fade margin is the one already defined: received signal level minus the receiver threshold. It governs how the link copes with a signal that drops evenly across the channel. If you are planning a basic or narrowband link, this is the number to work with, and you can skip to the next section.
For wideband, high capacity links
The rest of this section deals with dispersion, which mainly matters on wide channel, high throughput microwave systems. If that is not your link, you can move on without missing anything you need.
Dispersive fade margin describes how much frequency selective distortion the radio’s equaliser can correct before the error rate runs away. It is a property of the radio and the channel width, listed on the datasheet, and it does not change with signal strength.
The two margins combine in power rather than in decibels, because the link fails when either one wins:
10^(−Net fade margin / 10) = 10^(−Flat fade margin / 10) + 10^(−Dispersive fade margin / 10)
Putting numbers in shows why this matters. With a flat fade margin of 35 dB and a dispersive fade margin of 25 dB:
10^(−35/10) + 10^(−25/10) = 0.000316 + 0.00316 = 0.00348, so the net fade margin = −10 · log₁₀(0.00348) = 24.6 dB
The net result sits just below the smaller of the two. The link is held back by dispersion at 25 dB, and adding transmit power or bigger antennas barely moves it. The fix for a dispersion limited link is a better equaliser, a narrower channel, or diversity. This net figure, not the flat margin alone, is what sets the multipath outage on a wideband link.
How Much Fade Margin Do You Need
Fade margin is not a fixed target. It is whatever the availability requirement works out to for that path, frequency and climate. Availability is usually quoted in nines.
| Availability | Unavailable per year | Typical application |
|---|
| 99.9% | 8 hours 46 minutes | Best effort access, non critical links |
| 99.99% | 52.6 minutes | Standard carrier backhaul, enterprise |
| 99.999% | 5.3 minutes | Core transport, critical infrastructure |
| 99.9999% | 31.5 seconds | Protected core, safety of life networks |
The bridge from a fade margin in decibels to an availability in nines is the multipath outage relationship in ITU-R Recommendation P.530. In the deep fade region, the fraction of time the signal is faded below a depth A is:
Pw = P₀ · 10^(−A / 10)
A is the flat fade margin in decibels, and P₀ is the path’s multipath occurrence factor, which grows with path length, frequency and how flat or reflective the route is. Putting numbers in: a path with P₀ of 0.01, which is one percent, and a 20 dB fade margin gives Pw = 0.01 · 10^(−2) = 0.0001. That is an outage of 0.01% of the time, which is 99.99% availability.
The shape of that equation gives the most quoted fact about fade margin:
The ten decibel rule
In the multipath regime, every extra 10 dB of fade margin divides the outage time by ten. One more nine of availability costs about 10 dB of margin.
Warning: this rule is for multipath only
The ten decibels per nine rule holds when multipath is the limiting fade, which is the case below about 10 GHz. It does not apply to rain limited links on the higher bands, where the margin is set by rain rate statistics and the curve is far steeper. It is also a deep fade approximation, so treat it as a quick planning guide rather than a replacement for a full P.530 calculation.
Turn the equation around and it sizes the margin directly:
A = 10 · log₁₀(P₀ / Pw)
For the same path with P₀ of 0.01, a 99.99% target means an outage Pw of 0.0001, so the required margin is:
A = 10 · log₁₀(0.01 / 0.0001) = 10 · log₁₀(100) = 20 dB
Ask for one more nine, 99.999%, and Pw drops to 0.00001, so A becomes 10 · log₁₀(1000), which is 30 dB. The same 10 dB step buys one more nine each time.
These worked numbers are why the rough targets engineers carry tend to look the way they do.
| Scenario | Typical required fade margin |
|---|
| Short, low frequency, non critical link | 10 to 15 dB |
| Standard microwave backhaul at 99.99% | 20 to 30 dB |
| Long or high capacity hop at 99.999% | 30 to 40 dB |
| Rain limited high band link, tropical climate | 35 dB and beyond, set by rain not multipath |
Worked Example: From Margin to Availability
Take the 5.8 GHz point to point backhaul from the link budget worked example: it receives −51.2 dBm against a −67.0 dBm threshold, so the flat fade margin is 15.8 dB.
Below 10 GHz this link is multipath limited, so rain is not the worry, the still-night fade is. Suppose the path analysis gives a multipath occurrence factor P₀ of one percent. The outage fraction at the 15.8 dB margin is:
Pw = 0.01 · 10^(−15.8 / 10) = 0.01 · 0.0263 = 0.000263
That is an availability of 99.974%, or about two and a half hours of multipath outage in a worst month. The link clears 99.97% comfortably, but it falls short of a 99.99% contract, which on this path needs 20 dB. The gap is only about 4 dB, so the design is close. A slightly larger dish at one end, a lower loss feeder, or a step down to a more robust modulation that lowers the threshold all close it without rebuilding the link.
The equation has done something useful here. It turned an availability target into a specific decibel figure, and it showed that the shortfall is a few decibels rather than a different link design. Working the number is what tells you the difference between the two.
Rain Changes The Rules Above 10 GHz
Everything above assumes multipath is the limiting fade, which holds below about 10 GHz. On the higher bands, such as 11, 18, 23 and 38 GHz, rain takes over and the ten decibel rule no longer applies.
Rain attenuation does not follow the neat exponential of multipath. The loss per kilometre of path, called the specific attenuation, follows the ITU-R P.838 power law:
γ = k · R^α
R is the rain rate in millimetres per hour, and k and α are coefficients set by frequency and polarisation. As a rough feel for the numbers, an 18 GHz path in heavy rain of about 50 mm per hour sheds in the region of 5 dB for every kilometre, so a 5 km hop can lose more than 20 dB while the cell sits on it.
The margin a rain limited link needs is whatever loss the rain produces at the rain rate exceeded for the target slice of time. For 99.99% availability, that is the rain rate exceeded 0.01% of the year, which comes from the ITU-R P.837 rain maps.
So a rain limited link is sized by climate and path length, not by the multipath occurrence factor. Two identical radios on identical 5 km hops can need very different margins if one is in temperate Tasmania and the other in the tropical Top End. The rain fade calculator works the P.838 and P.837 chain through for a given frequency, polarisation, path length and rain rate, and returns the rain margin the availability target demands.
Fade Margin In Australian Conditions
Australia spans two very different propagation worlds, and fade margin has to be set for the one the link lives in.
The temperate south, covering most of the populated southeast, the southwest and Tasmania, has modest rain rates. The rain rate exceeded 0.01% of the year sits in the rough region of 30 to 50 mm per hour, so even links at 18 and 23 GHz can be closed with manageable margins over sensible hop lengths.
The tropical north, across the Top End, the Kimberley and far north Queensland, is a different proposition. Monsoonal downpours push the 0.01% rain rate well above 100 mm per hour in places, and the rain margin needed at 18 GHz and above climbs accordingly. The usual answers are shorter hops, lower frequency bands where the licensing allows, larger antennas, and where the geometry suits it, rain aware adaptive modulation that trades throughput for threshold during the heaviest cells.
Multipath risk is its own map. Long paths over water, across flat inland plains, and along humid coastal corridors on calm nights are where multipath fading bites, and those paths want generous flat fade margin and often space diversity regardless of the band. The link planner carries the budget onto the real terrain profile, flags Fresnel clearance, and computes the multipath and rain outage for the path so the fade margin is checked against the actual route rather than a rule of thumb.
How To Increase Fade Margin
When a design comes up short, there is a clear order to reach for, cheapest and most effective first.
- Larger antennas. Antenna gain rises with the square of the dish diameter, so doubling the diameter adds about 6 dB, and it helps at both ends. This is usually the most cost effective decibel on a microwave hop.
- Lower feeder loss. Mounting the radio at the antenna, or shortening and upgrading the feeder, recovers loss that was taxing both the transmit and receive sides.
- Higher transmit power, or remove a backoff. More power lifts the received level directly, and automatic transmit power control lets a link sit at a lower everyday power yet ramp up during a fade.
- A more robust modulation. Dropping from a dense modulation to a simpler one lowers the receiver threshold, which adds fade margin at the cost of throughput. Adaptive modulation does this automatically, holding the link up at reduced capacity through a fade instead of dropping it.
- A shorter hop. Halving the path length cuts the free space loss by 6 dB and lowers the multipath occurrence factor sharply, which is why a relay site can rescue a path that will not close in one leap.
- A lower frequency band. Lower frequencies suffer far less rain attenuation, so moving down a band is a direct way to win rain margin where the spectrum is available.
- Space or frequency diversity. Two receive antennas separated vertically, or two channels on different frequencies, give the link independent looks at the path. The fades rarely coincide, so the combiner sees a much shallower effective fade. Diversity is the standard tool against multipath on long and critical paths, and its improvement factor can be equivalent to tens of decibels of extra margin.
One caution ties the list together. Diversity, more power and bigger antennas all fight multipath effectively, but rain attenuates every path and every antenna at once, so diversity does little for it. A rain limited link is improved by frequency, hop length and modulation, not by a second antenna.
Common Fade Margin Mistakes
- Treating a positive link margin as a pass. A link that closes with a few decibels to spare works on the bench and fails in weather. The margin has to be checked against the fade margin the availability target requires, not merely be greater than zero.
- Using the flat fade margin on a wideband link. Once the channel is wide, dispersion can be the limit. Sizing a high capacity link on the flat number alone overstates how robust it is.
- Forgetting that threshold scales with bandwidth. Widening a channel for throughput raises the noise floor and the threshold, which eats fade margin on the same path. The noise floor and receiver sensitivity relationship is where that comes from.
- Applying the ten decibel rule to a rain limited link. The clean ten decibels per nine only holds in the multipath regime. Above 10 GHz the margin is set by the rain rate statistics, and the curve is far steeper.
- Ignoring climate. A margin that is generous in Hobart can be marginal in Darwin. Fade margin is a function of where the link is, not just how it is built.
Frequently Asked Questions
What is fade margin in RF? Fade margin is the number of decibels by which the received signal level exceeds the receiver’s minimum usable threshold. It is the reserve a radio link holds to absorb fading from multipath, rain and atmospheric refraction without the link dropping. It is calculated as the received signal level in dBm minus the receiver sensitivity in dBm.
What is a good fade margin? A good fade margin is one that meets the availability target for that path. For a standard microwave backhaul aiming at 99.99% availability, that is commonly 20 to 30 dB, and a long or critical link at 99.999% can need 30 to 40 dB. A short, low frequency or non critical link may be sound with 10 to 15 dB. The number is set by the requirement, not chosen for comfort.
How much fade margin do I need for 99.99% availability? In the multipath regime it depends on the path’s multipath occurrence factor, but for a typical path it works out near 20 dB, and each additional nine of availability adds about 10 dB. Above 10 GHz the requirement is set instead by the rain rate exceeded 0.01% of the year for that climate, which can push the margin well beyond 30 dB in tropical regions.
What is the difference between fade margin and link margin? Link margin is the measured headroom between the received power and the receiver threshold. Fade margin is the portion of that headroom reserved to ride out fading. On a well designed link the two are the same value, because all of the headroom exists to be spent on fades. The link is sound when the link margin is at least as large as the required fade margin.
Does fade margin protect against rain? Yes, rain attenuation is one of the fades the margin has to cover, and above about 10 GHz it is the dominant one. Below 10 GHz the margin is mostly spent on multipath, and rain is minor. The two are sized differently: multipath margin follows the ten decibels per nine rule, while rain margin is set by the rain rate statistics for the climate and the path length.
How do I increase fade margin? Larger antennas, lower feeder loss, higher transmit power, a more robust or adaptive modulation, a shorter hop, a lower frequency band, and space or frequency diversity all add fade margin. Bigger antennas and diversity are the usual first choices for multipath, while a lower band, shorter hops and adaptive modulation are the levers for a rain limited link.
Build it in noIM₃
The link budget calculator returns the fade margin alongside EIRP, received level and carrier to noise ratio for a path, and the rain fade calculator sizes the rain margin from the ITU-R P.838 and P.837 chain. When the path itself is the question, the link planner carries the whole budget onto real terrain, checks Fresnel clearance, and computes the multipath and rain outage so the fade margin is tested against the actual route and availability target.
Key Takeaway
Fade margin is the reserve that decides whether a link survives the day the weather turns. It is the received signal level minus the receiver threshold, and the right size for it is whatever the availability target works out to for that path, frequency and climate. Below 10 GHz it roughly follows the ten decibels per nine guide, and above 10 GHz it is rain that sets the number. Size it from the requirement, check it against the real path, and the link should ride through the weather it was built for.
