Cruising over an area known for interference, two receivers announce the loss of GPS within seconds of each other; the aircraft ahead of you reports a terrain alert over the sea to control. Two distinct phenomena hide behind these symptoms: one deprives you of the signal, the other manufactures a false one for you. Telling them apart on the flight deck decides whether the inertial position will remain usable for the following hour.
Why the satellite signal gives way to a low-power source
GNSS is a measurement system with a very thin power margin. The satellites transmit from roughly twenty thousand kilometres away, and the signal reaching the roof antenna sits, in order of magnitude, around minus one hundred thirty decibels relative to a milliwatt — well below the thermal noise of the receiver chain. The receiver only extracts it because it already knows each satellite’s spreading code in advance and derives a correlation gain from it. The whole robustness of the system rests on that gain, which is finite.
The consequence is mechanical: a low-power transmitter placed in the L band raises the noise floor beyond what correlation can compensate for, and the receiver stops tracking. There is nothing exotic about it, no special equipment involved: simply a link budget that is unfavourable by construction.
What remains is geometry, and it explains what is experienced on the flight deck. A ground transmitter whose effect would be limited to a few miles for a terrestrial user covers, for an aircraft in cruise, the whole cone of direct visibility — on the order of two hundred nautical miles of horizon from a high level. You enter and leave the affected zone along the line of sight, not by ground distance: the signal can return in descent, disappear in a turn when the fuselage masks the antenna, and reappear on a heading change. This intermittency is the signature of an external source, not of failing equipment.
GNSS signals are very weak, so even a small jammer can block them. From the air, a jammer’s signal covers a huge area because it’s in direct line of sight. If your GPS cuts out, then comes back when you turn, it’s probably a jammer, not a broken system.
A Boeing 737 flying at FL370 over a conflict zone loses GPS for 15 minutes. The signal returns when the aircraft descends below 20,000 feet, then drops again when turning east. The crew confirms it’s a ground-based jammer, not a system failure.
Jamming and spoofing: two failures that do not look alike
Jamming is a denial of service: it stops the receiver from working. Spoofing is a deception: it feeds it signals conforming to the format, more powerful than the real ones, and it tracks them in good faith. The first produces an absence, the second an error. Everything else follows from that.
| Criterion | Jamming (denial) | Spoofing (deception) |
|---|---|---|
| What the antenna receives | Noise or a carrier within the service band | Structured signals, valid in format, dominant in power |
| Receiver behaviour | Loss of tracking, no more position solution | Normal tracking, solution computed and declared sound |
| Displayed position | Absent, then an inertial position that drifts | Present, plausible, false |
| Onboard integrity monitoring | Effective: the system knows it no longer knows | Outside its failure model: nothing triggers |
| Effect on the hybrid position | Progressive degradation, of a known rate | Contamination of the inertial solution |
| What control sees | Degraded or lost position report | Report of a false and credible position |
| Crew priority | Monitor the growth of the error | Isolate the source before it contaminates the inertia |
What the crew sees, in the order it appears
Under jamming
The first notable thing is simultaneity. Two or three independent receivers declaring the loss within the same few-second window are not failing together: they are experiencing the same environment. Next come the growing precision announced by the flight management system, a message of inability to meet the required performance, the loss of satellite approach capabilities, a negative integrity prediction, inhibition of the terrain function, and degradation of the position report.

The behaviour remains legible: you know that you no longer know, and the drift of the inertial position follows a known law, growing with the time elapsed since the last satellite update — the order of magnitude retained in operations is a few nautical miles per hour, noticeably more if the alignment was poor or the platform old. So you reason in terms of accumulated error budget, not a single position: what matters is not the displayed figure but the tolerable margin for the segment underway. The return of the signal, for its part, deserves watching: an abrupt reacquisition can produce a position jump and an autopilot reaction.
Under spoofing
At first there is no message at all. The first clue is an inconsistency, and the earliest inconsistencies are not the position itself. The wind computed by the flight management system is the vector difference between airspeed and ground speed: a falsified ground speed produces an aberrant wind, sometimes several hundred knots, while the displayed position remains credible. Estimated times jump. The broadcast time can drift, since the receiver also distributes a reference time. Then come the spectacular manifestations: terrain alert with no relief, proximity alert with no traffic, erratic sequencing of waypoints, disagreement between inertial positions.
The trap of the alert dismissed too quickly. Rejecting a terrain alert because you believe yourself spoofed, or durably inhibiting the terrain function to silence the warnings: two symmetric mistakes. A terrain alert is treated as genuine until the position has been confirmed by a means independent of the satellite: barometric altitude, known relief, bearing and distance from an identified ground station. Inhibition, where approved documentation provides for it, is decided with full awareness of the actual vertical margin and is restored as soon as the cause disappears.
Why onboard integrity does not protect against spoofing
Autonomous integrity monitoring works by consistency: it compares redundant measurements and detects the one that deviates from the others. Its failure model is the failure of one satellite, possibly two, within an otherwise healthy constellation. A well-built spoofing attack does not present an outlying measurement — it presents an entirely consistent constellation pointing to a false position. The residuals stay small, integrity is declared available, and the green flag you are looking at means « these measurements are mutually consistent », never « these signals are authentic ».
Satellite augmentation does not change the nature of the problem: it broadcasts corrections and health information about the constellation, not proof of the origin of the signal received on board. Authentication mechanisms for the navigation message exist at the level of certain constellations, but their use by certified avionics presupposes an evolution of the equipment and its approval: it cannot be assumed to be available on an aircraft in line service. Conversely, a multi-frequency, multi-constellation receiver is far harder to deny service to with a narrow-band jammer — the robustness gained bears on denial, much less on deception.
The second mechanism costs even more. On modern architectures, the inertial position is continuously corrected by GNSS: the tighter the coupling, the faster a false position propagates into the estimated state. A platform that has integrated several minutes of erroneous correction does not become clean again once the input is removed — the error is already in the state. Hence the importance of early isolation: it is not a procedural refinement, it is the only action that preserves a usable reference for the rest of the flight.
What is lost, what remains
GNSS disturbance is not a global failure of the aircraft. Reasoning by source avoids panic and shortens the decision.
| Function | Primary source | Behaviour under disturbance |
|---|---|---|
| Altitude and flight level | Static pressure, barometric setting | Unchanged: altimetry does not come from the satellite |
| Airspeed, Mach number | Total and static pressures | Unchanged |
| Attitudes, heading, accelerations | Inertia | Unchanged |
| Navigation position | Hybridisation of inertia, satellite, radio | Degraded under jamming, false under spoofing |
| Ground speed, computed wind, estimated times | Mostly satellite | First revealers of the anomaly |
| Terrain function | Position, database, geometric altitude | Inhibition or unjustified alerts |
| Satellite-guided approaches | Satellite and augmentation | Unavailable |
| Conventional radio guidance | Ground transmitters | Available if in service |
| Position report under surveillance | Satellite position | False or absent: control’s picture is affected too |
| Timestamping of onboard systems | Satellite time | Jump or drift possible |
The right reflex: cross-check against a physically independent source. Comparing the navigation position to that of a second system fed by the same signal proves nothing and gives false reassurance. A valid cross-check is made against a bearing and distance from an identified ground station, a radio position computed by two stations, or what control sees on radar — bearing in mind that surveillance based on the onboard report is itself contaminated too.
The order of actions, and the reason for that order
- Fly first on references that do not depend on the satellite: attitudes, heading, barometric altitude. The trajectory is not negotiable during the diagnosis.
- Note the simultaneity. Several receivers affected at the same instant point to the environment, a single one affected points to the equipment.
- Cross-check before concluding, against an independent source, and record the time and position thus obtained: it will be the only sure reference for the report.
- Qualify the regime. No position solution: jamming. Solution present but inconsistent with the cross-check: spoofing.
- Under spoofing, isolate the source according to the type’s approved procedure, without waiting for perfect confirmation: every minute of coupling degrades the inertial reference. Under jamming, let the reversion take place and monitor the drift.
- Explicitly give up on what relies on the satellite: satellite-guided approach, tightened navigation performance, a departure or arrival dependent on GNSS. Tell control in terms of inability to meet the required performance, and request a conventional routing.
- Resume the decision on destination and alternate. A conventional approach with no vertical guidance raises the minima and lengthens the trajectory: the margin is calculated before the descent, not during it.
- Log during the event, not after the flight: time, independent position, level, heading, exact messages, systems affected, duration, and the moment and manner of the return to normal.
On the ground: return to service and the value of the report
A receiver that has tracked false signals can retain an erroneous time or orbit data, which lengthens reacquisition on the next flight, or even prevents it until a reset. A contaminated inertial platform may require a full realignment on the ground, with entry of an exact position: the operation is not without impact on a short turnaround, and it should be planned for as soon as the event occurs, not at start-up.
The report only has value if it is usable. A note reading « GPS loss in cruise » feeds no mapping. What is useful: the position obtained by an independent means, the time in universal time, the level, the duration, the systems affected, and the exact wording of the messages. These elements feed into the European occurrence reporting scheme, and it is from their aggregation that the notices to navigators consulted before the next departure are born. Exchanges between operators on the FL330 professional forums usefully complete this circuit without ever replacing it.
What is decided before start-up
- Notices to navigators concerning satellite disturbances, along the route and at the planned airfields, including en-route alternates.
- Prediction of integrity availability for the approaches envisaged, and its consequence on the choice of runway and minima.
- An alternate equipped with a non-satellite approach, with verification that the ground guidance there is in service and not under maintenance.
- Task-sharing agreed at the briefing: who cross-checks, who communicates, who isolates, and the threshold for giving up.
- Fuel planning: the extra required for conventional routing and raised minima is decided at departure, where it only costs a decision.
What is left for you to check, and the grey area
Three checks are worth doing outside a live situation, on your own type. First: is there a documented means, accessible to the crew, of deselecting the satellite update, and at what level does it act — receiver, hybridisation, flight management system? On certain architectures it does not exist on the flight deck, and the handling of the spoofing case is changed entirely by that. Next: does your abnormal procedure documentation address a false position, or only the loss of position? Many texts were written for denial and remain silent on deception. Finally: what criterion does your operator use to qualify a terrain alert as not relevant, and who has the authority to declare it so?
The grey area, for its part, will not be closed by careful reading: no general text fixes the margin beyond which a hybrid position must be declared unusable. The threshold remains a matter of judgement, framed by the performance required for the segment underway: the same position error is tolerable on an oceanic route and disqualifying on approach. For anyone preparing a type rating or a line check, the shift is notable: the subject has moved out of the « avionics » chapter and into threat management, and the question bears on the order of actions and the nature of the cross-check, not on the definition of the system. The FL330 advanced training courses address the case from that angle.
The applicable regulation, the manufacturer’s manuals, and your operator’s approved documentation take precedence over this article. In case of a discrepancy between what is described here and your approved procedures, the latter prevail.
Questions fréquentes
How can a jamming attack be told apart from spoofing in flight, without dedicated equipment?
Jamming produces an absence: no more position solution, simultaneous loss messages on several receivers, a reversion to inertia. Spoofing produces a silent error: the position remains displayed and plausible, but the computed wind becomes aberrant, estimated times jump, and terrain or proximity alerts trigger for no reason. The sorting rule is therefore simple: no position at all is denial; a position that is present but contradicted by an independent cross-check is deception.
Why does autonomous integrity monitoring not flag a spoofing attack?
Because it works by consistency between redundant measurements, and its failure model is the failure of one or two satellites within an otherwise healthy constellation. A spoofing attack presents an entirely consistent constellation pointing to a false position: the residuals stay small and integrity is declared available. The flag means the measurements are mutually consistent, never that the signals are authentic.
Is altimetry affected by a GNSS disturbance?
The displayed altitude and flight level come from static pressure and the barometric setting: they do not depend on the satellite and remain valid. On the other hand, functions that use a geometric altitude of satellite origin, notably the terrain function and certain fields of the position report, can produce erroneous indications. Vertical separation remains ensured by the pitot-static chain.
Should GNSS be deselected as soon as there is doubt, or should confirmation be awaited?
When spoofing is suspected, early isolation is preferable where the type’s approved documentation provides for it and makes it accessible to the crew. Since the inertial position is continuously corrected by the satellite, every minute of coupling to a false position further degrades the reference, and removing the input later does not clean the state already estimated. Under simple jamming, there is nothing to isolate: the reversion happens on its own and one monitors the growth of the error.
Does control see an aircraft undergoing spoofing correctly?
Not necessarily. Surveillance based on the onboard position report broadcasts the position derived from the satellite receiver: if it is false, control’s picture is false too, and credibly so. Cross-checking with control retains value when it relies on an independent radar observation, which is better requested explicitly rather than assumed.
What must the report contain to be genuinely usable?
The position obtained by a means independent of the satellite, the time in universal time, the flight level, the duration of the phenomenon, the list of systems affected, and the exact wording of the messages displayed. These elements feed into the European occurrence reporting scheme, then the notices to navigators and the maps of disturbed zones. A report that merely mentions a GPS loss allows no mapping and therefore serves no one.

Leave a Reply
You must be logged in to post a comment.