13 LoRaWAN Link Budget and ADR
Signal Evidence, Link Margin, Data-Rate Choice, ADR Commands, and Release Records
13.1 Start Simple
Start with two linked questions. Can the selected radio setting be heard with enough margin at the gateway? If the device later changes data rate or power, is ADR using evidence from a stable situation? Link budget and ADR belong together because one estimates the path and the other changes behavior based on observed path history. Neither should be approved as a slogan.
Phoebe’s Why
This page’s own link-budget derivation lives elsewhere on the site; what it does not derive is why ADR needs the 10 dB safety buffer it already uses, rather than moving a device the instant its margin merely clears the receiver’s hard floor. A LoRa chirp sweeps the whole channel bandwidth once per symbol, and a higher spreading factor is literally a slower chirp – each SF step doubles how long that sweep takes, and airtime held at the same transmit power is energy spent. A device sitting exactly on a margin boundary with no buffer does not settle on the cheaper setting; it can bounce between two spreading factors uplink to uplink, and every bounce that fails costs a wasted transmission on top of the successful one that follows.
The Derivation
A LoRa symbol sweeps the full channel bandwidth exactly once, so its duration is set by spreading factor and bandwidth alone:
\[T_{sym}(SF,BW) = \frac{2^{SF}}{BW}\]
Transmit energy for a burst of \(N\) symbols at conducted power \(P_{tx}\):
\[E_{tx} \approx P_{tx}\,N\,T_{sym}(SF)\]
Stepping one SF doubles the symbol time, so it doubles the energy of an otherwise identical packet:
\[\frac{E_{tx}(SF+1)}{E_{tx}(SF)} = \frac{T_{sym}(SF+1)}{T_{sym}(SF)} = 2\]
Worked Numbers: SF7 vs. SF8 On This Page’s Own 14 dBm Setting
- Symbol time, EU868’s 125 kHz channel: \(T_{sym}(7) = 2^7/125{,}000 = 1.024\) ms; \(T_{sym}(8) = 2^8/125{,}000 = 2.048\) ms – exactly double, the ladder this page’s ADR logic silently depends on.
- Energy per packet, this page’s own 14.0 dBm EIRP (\(=25.1\) mW conducted) and an illustrative 15-symbol short uplink: \(E_7 = 25.1\times15\times1.024\times10^{-3} = 0.386\) mJ; \(E_8 = 25.1\times15\times2.048\times10^{-3} = 0.772\) mJ – double, as the derivation predicts.
- The cost of one failed guess: if a device at the margin boundary tries the cheaper SF7 setting, is missed by the gateway, and has to retry at the safer SF8, the round trip costs \(E_7+E_8 = 0.386+0.772 = 1.158\) mJ – exactly \(50.0\%\) more than simply sending the one SF8 packet (\(0.772\) mJ) in the first place, with nothing gained for the extra half.
- Why the buffer, not the boundary, is the setting: that \(50.0\%\) tax is paid every time a boundary-hugging guess fails, which is precisely why this page’s own margin target sits 10 dB above the hard receiver requirement rather than at it – the buffer is not caution for its own sake, it is the difference between one settled transmission and a string of wasted ones.
Overview: Link Budget Is Evidence, Not a Guarantee
A LoRaWAN link-budget review asks whether the received signal has enough margin above the receiver requirement for the selected radio setting. It combines configured transmit behavior, antenna path, path-loss assumptions, gateway evidence, and receiver threshold.
ADR, or Adaptive Data Rate, then uses recent uplink evidence to request better data-rate, transmit-power, channel-mask, or repetition behavior. ADR should not be approved as a generic battery or capacity feature; it is only as good as the evidence window it uses.
A simple review example makes the boundary visible. Suppose a device transmits at 14 dBm EIRP, the installation record allows 1 dB for enclosure and connector loss, the path survey estimates 126 dB of path loss, and the gateway antenna contributes 2 dB. The expected received level is about -111 dBm. If the selected data rate needs roughly -123 dBm and the project requires 10 dB of margin, the paper margin is only about 2 dB above target. That does not reject the design by itself, but it tells the reviewer to demand representative gateway samples before approving a less robust setting. It also names exactly which assumption to revisit when enclosure, antenna height, gateway diversity, or regional data-rate policy changes.
Configured Input
Transmit power, data rate or spreading factor, antenna path, channel plan, message profile, and confirmed-message policy.
Path Evidence
Gateway reports, survey records, map assumptions, enclosure notes, antenna placement, and any known obstruction pattern.
Margin Decision
Received evidence compared with the receiver requirement and the safety target selected for the deployment.
ADR Boundary
Network-service commands are reviewed against representative uplink history and verified with follow-up traffic.
Use conservative wording. "This packet was received" is not the same as "this setting has representative margin." The release record should separate measured evidence from assumptions.
Practitioner: Review Margin Before Changing Settings
Margin is the gap between what the receiver needs and what the device path appears to deliver. A strong review does not stop at packet delivery; it asks how much room is left after normal placement, enclosure, antenna, gateway, and seasonal conditions are considered.
In practice, the ADR decision should name both the proposed change and the evidence window. If the network server wants to reduce transmit power by 6 dB after 20 strong courtyard uplinks, the reviewer should check whether those uplinks cover the expected gateway, weather, enclosure, and payload conditions. The approval should then require follow-up uplinks after the LinkADR response. If the same device later moves behind a metal cabinet or switches to a larger confirmed payload, the old evidence window no longer proves the new margin. The record should say whether the action is automatic, manually approved, or disabled for that route.
For fixed courtyard sensors, repeated gateway records from stable locations can justify ADR when margin remains acceptable and follow-up uplinks verify the new setting. For a moving asset, recent uplinks from one stop may not represent the next stop, so ADR should be disabled, bounded, or manually reviewed until route evidence is representative.
Under the Hood: Airtime, Exceptions, and Release Records
More robust settings usually consume more shared channel time. Less robust settings may reduce airtime but need enough margin. The reviewer should connect link-budget decisions to airtime-budget decisions instead of approving them separately.
Every number this chapter uses — path loss, sensitivity, margin — comes from three short equations. New to dB entirely? Start with the Decibels and the Link Budget foundations interactive, which teaches the idea from zero. Then, if you want the mathematics behind the calculator rather than just its output, the optional panel below derives the equations and works one real link end to end. You can skip both without losing the chapter's thread.
The release record should make tradeoffs auditable. A change from a slower spreading-factor setting to a faster one may reduce airtime and gateway load, but it can also shrink margin enough that retransmissions erase the benefit. A useful record therefore includes the before-and-after setting, the measured RSSI/SNR range, the receiver requirement used, the margin target, the expected airtime impact, and the rollback trigger. For example, "accept faster data rate while median margin stays above target and confirmed retries stay below the release limit" is reviewable. "ADR saves battery" is not. The same record should name who watches the exception signal: operations may own gateway outages, firmware may own ADR response handling, and the application team may own retry or command-expiry policy. That ownership keeps a radio optimization from becoming an untracked production behavior change.
Measured Inputs
Gateway evidence, device setting, antenna path, channel-plan context, and decision window.
Assumptions
Path-loss estimate, enclosure placement, obstruction pattern, and receiver requirement source.
Verification
Follow-up uplinks after a setting or ADR change, plus rollback behavior when evidence degrades.
Ownership
Named owner for route, gateway, enclosure, firmware, network-service, or application changes.
Retest when environment, route, antenna, enclosure, gateway path, channel plan, payload pattern, firmware, network-service ADR policy, or ownership changes. The point is not to freeze one setting forever; it is to keep the approval tied to current evidence.
A moving asset, a basement meter, and a fixed courtyard sensor should not share one ADR conclusion unless their evidence windows represent the same future path. Exception handling is part of the release, not an afterthought.
13.2 Summary
LoRaWAN link-budget review turns radio settings, path assumptions, gateway evidence, and receiver requirements into a margin decision. ADR can improve settings when the evidence is representative, but it needs suitability review, command acknowledgement, follow-up uplinks, and rollback behavior.
Good release records separate measured evidence from assumptions, connect margin to airtime impact, document exceptions, name an owner, and state retest triggers.
13.3 Key Takeaway
Review link budget and ADR together. Coverage margin, data-rate choice, gateway evidence, airtime impact, exceptions, and follow-up verification decide whether a LoRaWAN setting is release-ready.
13.4 See Also
Keep downlink timing separate from radio setting selection.
Connect gateway placement and star-of-stars topology to evidence quality.
Review activation and session behavior alongside radio evidence.
Test whether the release record handles common deployment failure cases.