Use RF calculator results to reject inconsistent assumptions, compare defined scenarios, and identify what needs further investigation. Do not treat a capacity estimate, throughput figure, or positive link margin as a promise that a radio system will deliver the required service.
The distinction is conditionality: every result applies only to the inputs, reference points, and operating conditions represented by its model. Shannon capacity is a theoretical limit for a specified channel. A throughput estimate depends on resource allocation and implementation assumptions. A free-space link budget calculates received power without establishing how often that power will be available. These outputs answer different questions; they are not interchangeable measures of link performance.
Distinguish a theoretical bound from a planning estimate
“Engineering bound” should not imply that every calculator produces a mathematically guaranteed upper or lower limit.
| Result | What it can establish | What it cannot establish |
|---|---|---|
| Shannon capacity | An information-theory ceiling for the specified bandwidth, signal-to-noise ratio, and additive white Gaussian noise channel | The throughput of a particular radio, scheduled user, or application |
| Cellular throughput estimate | A conditional rate based on the stated resource and efficiency assumptions | A guaranteed busy-hour rate or minimum user experience |
| Free-space received power | A reference received level for the stated frequency, distance, gains, and losses | Site coverage, outage probability, or annual availability |
Shannon’s result is a capacity limit under its channel assumptions. Free-space attenuation, by contrast, is a propagation reference: it is not a universal worst-case or best-case received-power bound for a real site. The underlying definitions come from Shannon’s channel-capacity theorem and ITU-R P.525’s free-space model.
A planning estimate becomes useful when its conditions are explicit enough to support a decision. “The target exceeds this channel’s theoretical capacity” can justify revisiting a proposal. “The target is below capacity” only means that this particular theoretical test has not ruled it out.
Define the requirement before comparing results
Write the performance requirement in a form that can be compared with the calculated quantity. At minimum, specify:
- Direction: uplink, downlink, or a defined bidirectional service.
- Boundary: one radio path, one spatial stream, one carrier, one sector, or one user.
- Rate definition: physical-layer information rate, scheduled payload rate, or application goodput.
- Operating condition: channel bandwidth, radio mode, interference condition, and resource allocation.
- Time basis: a short test interval, busy-hour performance, or a specified availability objective.
A 100 Mbps channel-capacity result and a 100 Mbps application requirement do not have the same boundary. Application delivery also depends on how much of the channel is available to that user and what happens between the radio interface and the measurement endpoint.
Keep an input record alongside each calculation: value, unit, source, reference point, and whether the value is measured, specified, or assumed. A default value is not evidence that the same condition exists at the proposed site.
Use Shannon capacity as a feasibility ceiling
The Shannon Channel Capacity calculator is described as calculating theoretical additive white Gaussian noise, or AWGN, channel capacity from bandwidth and SNR. Its result should be interpreted as an information-theory planning bound, not an equipment throughput rating.
For the band-limited AWGN model:
where:
- (C) is channel capacity in bits per second.
- (B) is channel bandwidth in hertz.
- (\gamma=S/N) is the linear signal-to-noise power ratio within that bandwidth.
Convert SNR expressed in decibels before substitution:
Shannon’s theorem concerns reliable communication with sufficiently capable coding under the specified noise and power assumptions. It does not specify the modulation, coding implementation, latency, or finite-length performance of a commercial radio.
A reproducible capacity comparison
Assume one AWGN channel with 20 MHz bandwidth. The SNR values below are illustrative assumptions, not measurements or equipment specifications.
At 15 dB SNR:
\approx100.6\text{ Mbps}
Changing only SNR produces:
| Assumed SNR | Calculated capacity |
|---|---|
| 10 dB | 69.2 Mbps |
| 15 dB | 100.6 Mbps |
| 20 dB | 133.2 Mbps |
For a 100 Mbps application requirement, the 15 dB case does not provide a credible delivery margin: the requirement nearly consumes the entire idealized channel capacity before accounting for practical resource use. The 20 dB case provides more theoretical headroom, but still does not establish a deliverable application rate.
These numbers are reproducible calculations using Shannon’s formula. They are not observations of a deployed link.
Keep bandwidth and SNR physically consistent
Signal and noise must refer to the same receiver boundary and bandwidth. Do not combine received signal power integrated over one bandwidth with noise measured over another.
Bandwidth sensitivity also depends on what remains fixed. With fixed SNR, the formula makes capacity proportional to bandwidth. With fixed received signal power and constant white-noise spectral density, doubling bandwidth doubles noise power and reduces SNR.
For example, starting with 20 MHz and 20 dB SNR:
- Holding SNR at 20 dB while doubling bandwidth gives approximately 266.3 Mbps.
- Holding signal power fixed while doubling noise bandwidth gives a linear SNR of 50 and capacity of approximately 226.9 Mbps.
Those are different scenarios, not competing answers. Shannon explicitly discusses the increase in integrated noise power with bandwidth.
Likewise, substituting SINR for SNR requires a justified treatment of interference. A calculation that represents interference as equivalent Gaussian noise remains conditional on that approximation; it does not reproduce arbitrary interference behavior merely because the input is labeled in decibels.
Interpret cellular throughput through its resource assumptions
For a result from the Cellular Link Throughput Estimate, first establish what the displayed rate includes. The tool’s title alone does not establish whether its output represents a carrier total, a single user, a physical-layer rate, or application goodput. Its current input fields and implementation were not available for verification here, so the relationship below is an explicit planning example, not a claim about that calculator’s implemented formula.
A transparent resource-based estimate can be written as:
where:
- (R) is estimated scheduled payload rate in bits per second.
- (B) is the bandwidth associated with the efficiency definition, in hertz.
- (\eta_{\mathrm{agg}}) is aggregate payload spectral efficiency in bits per second per hertz, including any spatial layers already represented.
- (f_{\mathrm{dir}}) is the usable time fraction for the direction being estimated.
- (f_{\mathrm{user}}) is the user’s share of those resources.
This is a bookkeeping model. It requires evidence for the efficiency and allocation assumptions; it is not a substitute for a radio-specific performance model.
For an illustrative calculation, assume:
\eta_{\mathrm{agg}}=3\text{ bit/s/Hz},\quad
f_{\mathrm{dir}}=0.75,\quad
f_{\mathrm{user}}=0.50
Then:
=22.5\text{ Mbps}
Here, 3 bit/s/Hz is assumed to represent payload efficiency after the coding and radio overhead included in its definition, but before directional and user allocation. Applying another blanket “radio efficiency” factor for those same losses would double-count them.
Before using any throughput result, resolve these boundary questions:
- Does spectral efficiency include coding, reference signals, control resources, and guard resources?
- Is efficiency per spatial layer or already aggregated across layers?
- Does the directional allocation include switching and guard time?
- Is the rate for all scheduled users or one user’s resource share?
- Are retransmissions and higher-layer overhead already reflected in the selected efficiency?
Do not multiply by antenna count when the efficiency already includes spatial layers. Do not multiply by a downlink fraction when the supplied rate is already downlink-only. Where a definition is unavailable, retain the result as a scenario rather than converting it into a service commitment.
Use the RF link budget to calculate received power—not availability
The RF Link Budget calculator exposes these inputs: transmit power in dBm, transmit and receive antenna gains in dBi, frequency in MHz, path length in kilometers, and other losses in dB. Its published calculation is:
32.44+20\log_{10}(f_{\mathrm{MHz}})
+20\log_{10}(d_{\mathrm{km}})
Here, (P_r) and (P_t) are powers in dBm; antenna gains are in dBi; and loss terms are in dB. The frequency and distance units are part of the equation. Entering GHz into the MHz expression changes the answer by 60 dB.
ITU-R P.525-5, Annex Section 2.3, gives the underlying free-space relationship:
with distance (d) and wavelength (\lambda) in the same units. Its practical-unit expression uses a rounded constant of 32.4 for MHz and kilometers; the calculator uses 32.44. That small rounding difference should not be interpreted as meaningful site-performance precision.
Reproduce a defined link-budget scenario
Assume:
| Input | Illustrative value and boundary |
|---|---|
| Transmit power | 23 dBm at the transmit antenna input |
| Transmit antenna gain | 23 dBi toward the receiver |
| Receive antenna gain | 23 dBi toward the transmitter |
| Frequency | 5,800 MHz |
| Distance | 10 km |
| Other losses | 2 dB between the stated antenna reference points, excluding transmit feeder loss already removed from the power input |
Using the calculator’s published expression:
=32.44+20\log_{10}(5800)+20\log_{10}(10)
\approx127.71\text{ dB}
\approx-60.71\text{ dBm}
As a numerical cross-check, calculating wavelength from frequency and using (20\log_{10}(4\pi d/\lambda)) gives approximately 127.72 dB. The agreement supports consistency of the arithmetic and unit conversion; it does not verify the path or establish agreement with an interactively tested calculator output.
The calculator describes its transmit-power input at the transmit antenna reference plane. If the available power specification is instead at the radio output, account for the intervening feeder loss once. Do not reduce transmit power for a feeder and then subtract that same feeder loss again under “other losses.”
Compare received power with a matching receiver threshold
A separate margin calculation is:
Suppose, solely for illustration, that a receiver threshold is −70 dBm for the exact bandwidth, operating mode, and error criterion of interest. The example would give:
That is approximately 9.3 dB above the assumed threshold. It is not an availability percentage, and the threshold is not a verified specification for any device.
For an equipment decision, obtain the threshold for the actual model, firmware or configuration where relevant, bandwidth, modulation and coding mode, and stated error criterion. Also check the applicable maximum input or overload limit. A threshold for a more robust, lower-rate mode does not demonstrate support for a higher-rate requirement.
The RF calculator explicitly excludes terrain, atmospheric loss, fading, polarization, and antenna-pattern effects unless represented through appropriate inputs. Free-space propagation also presumes conditions consistent with its propagation model; a valid numerical input range does not prove that a short path is in the antenna far field or that a long path has adequate clearance.
The calculator cites P.525-4. ITU lists P.525-5 as the in-force edition and identifies P.525-4 as superseded but incorporated by reference in the Radio Regulations. Identify the edition governing a particular assessment rather than inferring project compliance from the calculator’s reference.
Test whether uncertainty changes the decision
Use scenarios to expose assumptions that could reverse the conclusion. Do not label them confidence intervals unless a defensible probability model supports that interpretation.
For the illustrative link budget above, vary the entered additional losses while keeping the other inputs and assumed −70 dBm threshold fixed:
| Additional losses | Calculated received power | Margin above assumed threshold |
|---|---|---|
| 2 dB | −60.7 dBm | 9.3 dB |
| 5 dB | −63.7 dBm | 6.3 dB |
| 8 dB | −66.7 dBm | 3.3 dB |
Each added decibel of loss reduces received power and threshold margin by one decibel under the published link-budget equation. These scenarios show sensitivity to a loss assumption; they do not assign a probability to any row or describe a complete fading process.
Apply the same discipline to capacity and throughput:
- Vary SNR across a justified operating range rather than relying on one favorable sample.
- Change bandwidth together with the associated noise and allocation assumptions.
- Reduce user resource share to represent a defined loaded condition.
- Recalculate both link directions when their powers, gains, receiver characteristics, or interference differ.
If plausible input changes move a proposal from adequate to inadequate, the decision depends on resolving those inputs. More displayed decimal places will not resolve that uncertainty.
Match the evidence to the claim
Use precise language when recording a result:
| Evidence available | Appropriate statement |
|---|---|
| A primary source supports the equation | “The method follows the cited channel or propagation model.” |
| An independent calculation reproduces a number | “The stated inputs reproduce this result.” |
| An actual calculator run agrees with a reference calculation | “The calculator output agrees for this tested case.” |
| Measurements match the modeled configuration and conditions | “The prediction was compared with these measurements under these conditions.” |
| The applicable review or approval process is complete | State the actual approval, its scope, and the responsible authority or party. |
These are different levels of evidence. Numerical agreement cannot establish that an antenna is correctly installed, interference is acceptable, a receiver supports the intended mode, or a service meets its availability target.
A useful decision record therefore connects the requirement to the model and the next necessary evidence:
Under the stated bandwidth, SNR, resource allocation, and loss assumptions, the calculation supports continued evaluation of this option. Equipment selection depends on configuration-specific performance data. A delivery commitment depends on defined loading, path conditions, and acceptance criteria supported by appropriate testing.
For a near-threshold result, preserve the unrounded calculation and compare the remaining margin with unresolved losses and measurement uncertainty. If those quantities have not been established, report “insufficient evidence to confirm margin,” not “pass.”