Study ANS by classifying each navigation source by how its error behaves over time, then practice deciding which source to trust in short written scenarios. GNSS errors stay roughly bounded with corrections, INS errors grow with time, and the FMS arbitrates between them using estimated navigation performance. Master that error-behavior logic before memorizing equipment standards, and rehearse decisions like coasting through a GNSS outage or reacting when estimated performance exceeds a required limit.
Bounded versus unbounded error: the distinction that organizes the whole syllabus
GNSS position errors stay roughly bounded once corrections are applied, while inertial position errors grow with time. That single contrast explains coasting limits, update requirements, and how integrity monitoring is designed.
Inertial navigation is self-contained: it integrates accelerometer and gyro outputs into velocity and position. Its weaknesses are systematic rather than random. An accelerometer bias becomes a growing velocity error, and a gyro drift compounds into attitude and position error that accelerates the longer the system runs without an external fix. Alignment quality at startup also seeds the drift. This is why inertial performance is described in drift rates rather than a fixed accuracy figure.
Satellite navigation behaves differently. Ranging errors from satellite clock and ephemeris data, ionospheric delay, and multipath fluctuate, but augmentation and receiver filtering hold them within a defined envelope rather than letting them accumulate. The practical consequence: an inertial system can navigate at high update rate between fixes, but every hour without a fix widens its uncertainty, while GNSS can be re-sampled indefinitely yet can be blocked, jammed, or give an undetected fault. Everything else in the syllabus builds on this asymmetry.
Trace it with a simple example: if an inertial system comes out of alignment with a 0.03 nm position error and drift grows roughly 1 nm per hour squared, error after 10 minutes is about 0.06 nm, after 30 minutes about 0.28 nm, and after one hour about 1.0 nm, while a corrected GNSS solution might hold near 0.05 nm. The point where the inertial curve overtakes the flat GNSS line, a few minutes in, marks the start of the coasting-uncertainty regime.
| Source | Dominant error source | Error over time | External dependence | Typical role |
|---|---|---|---|---|
| GNSS | Satellite clock/ephemeris, ionosphere, multipath | Bounded with corrections | Satellite visibility, signal quality | Long-term absolute position |
| INS/IRS | Accelerometer bias, gyro drift, alignment error | Grows, roughly with time squared | None while operating; needs initial alignment | Short-term high-rate position and attitude |
| Baro/radio altimetry | Pressure setting, terrain clearance | Bounded but biased | Atmosphere or terrain below | Vertical reference and aiding |
| Terrain-referenced nav | Map age, radar altimeter noise | Bounded if map matches terrain | Stored terrain database | Position check in GNSS-denied areas |
What a navigation filter actually does when GNSS and INS disagree
A navigation filter blends high-rate, drifting inertial output with low-rate, bounded satellite fixes by estimating the inertial sensor errors themselves, not by simply averaging the two position solutions.
The core idea is complementary filtering. The inertial solution is excellent over seconds and poor over hours; the GNSS solution is excellent over minutes but noisy and intermittent. A Kalman-type filter takes GNSS position as a measurement, compares it with the inertial prediction, and attributes the difference to estimated sensor errors such as gyro drift and accelerometer bias. Those error estimates feed back to correct the inertial solution, so the combined output is smooth at high rate and stays anchored over the long term.
This explains behavior that looks mysterious if you only memorize component descriptions. During brief satellite dropouts the position output does not jump, because the filter coasts on corrected inertial data. After a long outage, the first GNSS fix can cause a visible position snap, because accumulated drift is suddenly corrected. When you study filter behavior, focus on inputs, states, and what the output does during and after a measurement gap rather than deriving the equations.
Scenario: a survey UAV flying a routine inspection route passes under a bridge structure and loses satellite reception for several minutes. A plausible mistake is treating the displayed position as continuously reliable because it updates smoothly. The better decision is to recognize the display as inertial coasting, recall that drift grows with time squared, and expect the largest error exactly when the vehicle emerges, so the operator should plan re-acquisition and a verification fix rather than resuming the route immediately. It matters because smooth-looking output masks growing uncertainty.
Integrity, not just accuracy: RAIM, FDE, SBAS, and GBAS as distinct concepts
Accuracy describes how close a fix is to truth; integrity is the system ability to warn when a fix is unreliable. RAIM and FDE provide receiver-level monitoring, while SBAS and GBAS deliver corrections plus integrity from outside the receiver.
Keep four terms separated. Availability asks whether enough satellites are visible for a solution. Accuracy describes the expected error size. Integrity adds a timely alarm when the solution should not be trusted. Continuity asks whether the service will stay available through an operation. RAIM works inside the receiver by checking consistency among redundant satellite measurements; FDE goes a step further by identifying an inconsistent satellite and excluding it, keeping a solution available. SBAS broadcasts wide-area corrections and integrity from geostationary satellites; GBAS uses a local ground station to serve a small area with very precise corrections.
These concepts map onto decisions. A receiver can be accurate on average yet lack the redundancy needed to detect a faulty satellite, so integrity protection depends on geometry as much as signal quality. That is why integrity monitoring can be predicted in advance: given a planned time and location and expected satellite positions, you can check whether monitoring would have been available.
Scenario: a crew plans a satellite-based procedure for a departure window during a known solar-activity disturbance. The plausible mistake is checking only whether the receiver shows a position solution on the day. The better decision is to run an integrity availability prediction for the specific route and time beforehand, and if the prediction is unfavorable, plan an alternate means of navigation for that window. It matters because a solution can look healthy right up to the moment monitoring is lost.
Following the FMS: sensor priority, estimated performance, and what triggers a downgrade
The flight management system fuses inertial, satellite, and radio inputs, computes lateral and vertical guidance, and continuously compares its estimated navigation uncertainty against the performance a procedure requires.
Conceptually the FMS sits above the sensors. It maintains a best composite position, usually anchored by inertial data updated by satellite and radio fixes, and it keeps a running estimate of position uncertainty, often labeled estimated position uncertainty or actual navigation performance. Procedures carry a required navigation performance value. When estimated performance stays better than required, the system supports the procedure; when estimated performance degrades past the requirement, the displayed navigation class changes and the procedure can no longer be flown as designed.
Study the fallback chain rather than a sensor list. A radio update can restore composite accuracy when satellite data degrades; inertial coasting carries the solution briefly; beyond that the system announces that it can no longer support the required performance. Learn what each annunciation means in terms of which sensors are feeding the solution, because the display shows the consequence, not the cause.
Scenario: during a remote-area leg, estimated navigation performance grows until it exceeds the value required for the planned arrival. The plausible mistake is continuing the arrival because the aircraft has flown it before and the map display still shows a route. The better decision is to recognize that the procedure authorization depends on current estimated performance, descend to a less demanding procedure or a conventional navaid-based arrival, and let the crew work the sensor problem separately. It matters because the route on the display being present is not the same as the guidance being valid.
Reading certification documents by function instead of memorizing numbers
Group navigation certification into three layers: equipment performance standards, installation airworthiness, and operational approval. Learning what each layer does is more useful than recalling document identifiers in isolation.
In the United States, technical standard orders define minimum performance standards for categories of equipment, such as satellite navigation sensors with different levels of integrity monitoring and augmentation, and standards for area navigation systems. An installation approval then addresses how specific equipment is integrated into a specific aircraft, including antennas, power, and failure effects. Operational approval, expressed through operations specifications or letters of authorization, addresses what a particular operator is permitted to do. A unit can be certified, installed, and still not authorized for a given procedure.
For ANS study, build a one-page map of this structure: for each syllabus topic, note which layer governs it and what question that layer answers. When you encounter a standard or document family, attach it to a function, such as the difference between a sensor relying only on receiver-level monitoring and one using space-based augmentation. For current document lists, effective revisions, and any administrative detail about the credential itself, go to the issuer at ncatt.org; treat any catalog description as a scope indicator only, and verify structure and eligibility with the issuer rather than a secondary summary.
Navigation without satellites: what uncrewed systems add to the toolkit
When satellite signals are denied or degraded, uncrewed aircraft lean on relative and environment-referenced sensors: vision, lidar, radar altimetry, ultra-wideband ranging, and terrain-referenced navigation, each with short-range or map-dependence limits.
Visual-inertial odometry fuses camera motion with inertial data to estimate relative movement accurately over short periods but drifts without absolute references and struggles with featureless or fast-changing scenes. Terrain-referenced navigation compares radar altimeter readings with a stored terrain model, giving a bounded absolute check where terrain relief exists but little information over flat water or uniform ground. Ultra-wideband ranging gives precise relative positioning within a small instrumented area, which suits indoor or staged operations. Each substitute answers the same trust question from a different direction.
Notice how these options mirror the crewed syllabus rather than replacing it. Terrain reference is a cousin of radio-altimeter-based systems in crewed aircraft; vision-based methods raise the same alignment, drift, and fusion questions as inertial navigation. Study them by the same error-behavior framework: what bounds the error, what external reference is required, and how the solution degrades when that reference is absent. This lets one mental model cover both the crewed and uncrewed syllabus topics instead of two separate memorization tracks.
A preparation sequence, a plotting exercise, and readiness checks you can score
Study in five passes: error models, sensor fusion, integrity concepts, FMS behavior, then certification and uncrewed topics. After each pass, write the explanation from memory and score it against a fixed rubric.
A realistic adaptable sequence: week one, build the error-behavior table from section one from memory and justify every row; week two, explain fusion by predicting filter output during and after a measurement gap; week three, separate the four integrity terms and predict monitoring availability for an invented route and time; week four, trace an FMS fallback chain and write the downgrade scenario from section four in your own words; week five, map certification layers and extend the error table with uncrewed sensors. Compress or stretch the pacing to your schedule; keep the order, because later topics reuse earlier error models.
Exercise: plot inertial position error against time on graph paper. Assume 0.03 nm initial error after alignment and drift of 1 nm per hour squared, so error is roughly 0.03 plus t squared over 3600, with t in minutes. That gives about 0.06 nm at 10 minutes, 0.28 at 30, and 1.03 at 60. Draw a horizontal line at 0.05 nm for corrected GNSS. Expected observations: the inertial curve starts just below the GNSS line, accelerates upward, crosses it at roughly 8 to 9 minutes, and reaches about 1 nm inside an hour while the GNSS line stays flat. Self-check rubric: two points for correctly labeled axes and accelerating curve shape, two for identifying the crossover as the point where inertial uncertainty exceeds the GNSS envelope, one for stating why the GNSS line is flat. Reaching five of five means the underlying model is set.
Readiness checks before you conclude study: state the bounded versus unbounded distinction in two sentences without notes; explain the difference between fault detection and fault detection and exclusion in one sentence each; trace an FMS sensor fallback in order and name the displayed consequence; write the two scenarios above as decisions with reasons, closed-book; and reproduce the error table with all rows and at least one justification per row. Treat these as learning milestones that confirm concept coverage, not as predictions of any exam result.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.