AIS – Fundamentals and Limitations

AIS does not detect vessels.
It repeats what vessels choose to say about themselves.

Introduction

AIS arrived on bridges as a supplement. It has become, in practice, a primary reference. On many watches the AIS overlay on ECDIS is the first thing consulted and the last thing questioned. That inversion — from supplement to foundation — happened without any corresponding change in how the system generates its data.

A radar return exists because energy reflected off a physical object. An AIS target exists because a transmitter broadcast a data packet. The difference is fundamental and too often forgotten. One is evidence. The other is testimony.

Testimony can be wrong. It can be stale. It can be absent. It can be deliberately false. Every operational decision made on the basis of AIS data carries these possibilities, and the system provides no internal mechanism to distinguish truth from error. That responsibility falls entirely on the watchkeeper — and the watchkeeper must understand exactly where the data came from, how it was entered, and how far it can be trusted before it reaches the screen.

Contents

• 1. What AIS Is — and What It Is Not
• 2. Class A and Class B — the Operational Divide
• 3. Data Fields: Sensor-Derived, User-Entered, and the Gap Between Them
• 4. Reporting Intervals and Refresh Behaviour
• 5. Range, Propagation, and the Illusion of Coverage
• 6. AIS Is Not a Collision Avoidance Tool
• 7. Spoofing, Switching Off, and the Grey Space
• 8. Integration with ECDIS and Radar — the Circular Data Problem
• 9. Closing Reality

1. What AIS Is — and What It Is Not

The Automatic Identification System is a transponder-based system operating on two dedicated VHF frequencies — AIS 1 (161.975 MHz) and AIS 2 (162.025 MHz) — using Self-Organising Time Division Multiple Access (SOTDMA) to manage slot allocation among multiple users without a central controller. Each unit broadcasts a structured data packet at intervals determined by the vessel’s dynamic state, and simultaneously receives packets from other units within VHF range.

The system was designed for vessel traffic management and maritime safety. It was adopted under SOLAS Chapter V and made mandatory for vessels on international voyages of 300 GT and above, and all passenger ships regardless of size. IMO Resolution A.1106(29) sets out the performance standards.

AIS is not a sensor.

It does not measure range. It does not measure bearing. It does not detect anything. It receives self-reported data from other participants and feeds that data to the bridge display chain. If a vessel does not carry AIS, or carries it but has it switched off, or carries it but has entered incorrect data, AIS has nothing to show — or shows something wrong. There is no echo. There is no return. There is only a broadcast, or the absence of one.

This distinction matters because human behaviour on the bridge has shifted. A generation of officers now exists for whom an AIS target is more instinctively trusted than a radar contact. The AIS target has a name, a call sign, a destination, a speed readout to one decimal place. It looks authoritative. It looks precise. Precision is not accuracy.

2. Class A and Class B — the Operational Divide

Two classes of AIS transponder exist. The difference is not merely technical — it governs what data is transmitted, how often, and with what priority.

Class A units are mandated for SOLAS vessels. They transmit at 12.5 watts, use SOTDMA for slot access, and broadcast a full dynamic data set including rate of turn. Reporting intervals are governed by speed and manoeuvring state, ranging from every two seconds when altering course at speed, to every three minutes when at anchor. Class A units can be interrogated by shore stations and other vessels. They transmit and receive on both AIS frequencies simultaneously.

Class B units are voluntary-fit, found on smaller commercial craft, fishing vessels, and recreational boats. They transmit at 2 watts (some newer Class B+ units at 5 watts), use Carrier Sense TDMA (CSTDMA), and broadcast a reduced data set — no rate of turn, no navigational status, no ETA, no draught. Reporting intervals are longer: typically every 30 seconds above 2 knots, every three minutes below. In congested waters, where slot contention is high, Class B transmissions yield to Class A. A Class B target may simply stop appearing when the channel is busy.

That last point deserves emphasis.

In exactly the conditions where traffic density makes AIS most valuable, Class B targets are most likely to be suppressed.

The practical consequence is this: in a busy estuary or TSS approach, the AIS display may show every large commercial vessel and none of the small craft. The watchkeeper looking at a clean AIS overlay sees order. The radar, if properly tuned, tells a different story.

3. Data Fields: Sensor-Derived, User-Entered, and the Gap Between Them

AIS data falls into three categories: static, dynamic, and voyage-related. Understanding which fields come from sensors and which come from human input is essential to knowing which can be trusted.

Sensor-derived (dynamic) data

Position: Fed directly from the ship’s GNSS receiver. Generally reliable to within the accuracy of the connected GPS — typically a few metres. Failures tend to be obvious: a target appearing inland, or jumping erratically. More insidious is a degraded fix that drifts slowly, placing the target 50 or 100 metres from its true position without triggering any alarm.

Speed over ground (SOG): Derived from successive GNSS positions. Reliable under normal conditions but always SOG, never speed through the water. In strong tidal streams, SOG and the vessel’s actual movement through the water diverge significantly. AIS provides no mechanism to distinguish between the two.

Course over ground (COG): Also GNSS-derived. Reliable at speed. At low speeds or when stationary, COG becomes unstable and can swing wildly — a vessel at anchor may show COG values rotating through 360 degrees. This is not an error in the conventional sense; it is a limitation of deriving course from position deltas when those deltas are below the noise floor of the receiver.

Rate of turn: Class A only. Fed from the gyrocompass or a dedicated ROT sensor. When connected and calibrated, it is useful. When not connected — which is not uncommon — the field transmits as “not available.” Some units transmit a default zero, which is worse than transmitting nothing.

Heading: Fed from the gyrocompass via the heading sensor interface. This is true heading — the direction the bow is pointing, independent of the vessel’s track over the ground. It is one of the most operationally significant AIS fields and one of the most frequently missing, particularly on smaller vessels where the gyro feed to the AIS unit has not been connected or has failed without detection. When heading is absent, the AIS symbol on ECDIS reverts to a circle or a generic triangle. The vector shown is COG only. In a cross-current or crosswind, the displayed vector may differ from the vessel’s aspect by 20 degrees or more.

The absence of heading data removes any ability to infer aspect from AIS alone.

User-entered (static and voyage-related) data

MMSI, IMO number, call sign, vessel name, ship type, dimensions: Entered once during installation or commissioning. Generally correct — but not always. Dimension offsets (the position of the GPS antenna relative to the hull) are frequently entered incorrectly. A vessel whose antenna is 10 metres aft of the bow but entered as amidships will appear on an ECDIS overlay with a positional offset that matters at close quarters, in confined waters, or when assessing CPA.

Navigational status: Manually selected by the OOW. This field is notorious. Vessels at anchor broadcasting “Under Way Using Engine.” Vessels making way broadcasting “Not Under Command.” Vessels constrained by draught broadcasting “At Anchor” because no one changed it on departure. The field is updated by human memory and human diligence. Both fail regularly.

Destination and ETA: Free-text entry. Subject to truncation, abbreviation, misspelling, and neglect. A vessel bound for Felixstowe may show “FELIX,” “FELIXTOW,” “GBFXT,” or a destination three ports ago. ETA fields are routinely days out of date. These fields are useful for VTS coordination. They are not useful for collision avoidance.

Draught: Manually entered. Rarely updated after initial entry.

The pattern is clear. Sensor-derived fields are broadly reliable within known limitations. User-entered fields are only as reliable as the last person who touched them. Any decision that depends on navigational status, destination, or draught being correct is a decision built on hope.

4. Reporting Intervals and Refresh Behaviour

Class A reporting intervals are speed- and manoeuvre-dependent:

• At anchor or moored: every 3 minutes
• SOG 0–14 knots, not changing course: every 10 seconds
• SOG 0–14 knots, changing course: every 3⅓ seconds
• SOG 14–23 knots, not changing course: every 6 seconds
• SOG 14–23 knots, changing course: every 2 seconds
• SOG >23 knots, not changing course: every 6 seconds
• SOG >23 knots, changing course: every 2 seconds

Class B (standard CS): every 30 seconds when SOG >2 knots, every 3 minutes when SOG ≤2 knots. Class B+ (SO): intervals approaching Class A but still longer in practice.

Static and voyage-related data is broadcast every 6 minutes for Class A, regardless of dynamic state.

The critical operational point: between updates, the AIS target on screen is a prediction, not a measurement. The display extrapolates the last known position using the last known SOG and COG. If a vessel alters course between updates — and a 10-second gap at 14 knots covers 72 metres — the displayed position is wrong until the next packet arrives. At 3-minute intervals for anchored vessels or Class B, the extrapolation error can be substantial.

Radar updates continuously with every antenna revolution — typically every 2 to 4 seconds. AIS does not. The smoothly moving AIS target on the ECDIS screen is, between reports, a mathematical ghost.

5. Range, Propagation, and the Illusion of Coverage

AIS operates on VHF, and VHF propagation is essentially line-of-sight. Typical ship-to-ship range is 20 to 30 nautical miles, depending on antenna height. In practice, ducting can extend range significantly — AIS targets appearing at 60 or 70 miles are not uncommon in certain atmospheric conditions. Conversely, in heavy rain, or where antenna installations are poor, range can contract below 15 miles.

The problem is not short range. The problem is inconsistent range. A target that appeared at 40 miles may disappear at 25 miles as propagation conditions change, then reappear at 18 miles. The watchkeeper who saw it vanish may assume it altered away. It did not. It is still there, still closing, and the VHF path simply failed.

Satellite AIS and shore-based AIS networks extend coverage for traffic monitoring purposes but do not change the ship-to-ship reality. The bridge receives only what its own antenna can hear.

AIS coverage is not surveillance. It is a peer-to-peer exchange limited by physics, power, and the assumption that the other party is transmitting.

6. AIS Is Not a Collision Avoidance Tool

This needs to be stated plainly because it is stated in the regulations, stated in the performance standards, stated in every competent training programme — and ignored on a significant number of bridges every single day.

IMO Resolution A.1106(29) is explicit: AIS is an additional source of navigation information. It does not replace any navigational system, including radar. COLREGS make no reference to AIS. Rule 7 requires the use of “all available means” to determine risk of collision, and specifically names radar. AIS is implicitly included under “all available means” but carries no special status.

The reasons are structural, not bureaucratic.

AIS does not see vessels that are not transmitting. Small craft, naval vessels, vessels with failed equipment, vessels that have deliberately switched off — all invisible to AIS. A collision avoidance strategy that depends on AIS depends on every potential threat broadcasting correctly. In open ocean, with SOLAS traffic, this assumption is tolerable. In coastal waters, in approaches, near fishing grounds, or in regions where regulatory enforcement is weak, it is dangerous.

AIS-derived CPA and TCPA values are calculated from SOG and COG. They do not account for speed through the water, set and drift, or a vessel’s actual heading and aspect. A vessel with a beam current will show a COG-based vector that diverges from its heading-based aspect. The CPA alarm may be clear while the physical situation is not.

A CPA calculated from self-reported data that is updated every 10 seconds is not the same thing as a CPA derived from continuous radar tracking.

ARPA calculates CPA from measured range and bearing, updated every antenna revolution. AIS calculates CPA from broadcast position and velocity, updated at the reporting interval. Both have errors. But ARPA errors are measurement errors — they can be bounded and understood. AIS errors are input errors — they can be anything, including deliberate.

7. Spoofing, Switching Off, and the Grey Space

AIS can be spoofed. False targets can be injected into the VHF channel by anyone with modest technical capability and inexpensive hardware. This is not theoretical. Documented instances exist in multiple regions, including state-level spoofing where hundreds of vessel tracks have been manipulated simultaneously. Vessels have been shown at positions they never occupied, following routes they never sailed.

The implications for navigation are obvious. The implications for traffic assessment, for VTS decision-making, and for search and rescue coordination are worse.

AIS can be switched off. SOLAS mandates continuous operation, but the regulation includes a provision allowing the master to disable AIS if they judge it necessary for safety or security. In practice, this provision covers a vast grey space. Vessels engaged in sanctions evasion routinely go dark. Vessels approaching areas of piracy may disable AIS on security grounds, and not unreasonably. Fishing vessels in some regions switch off to conceal fishing positions from competitors, or to operate in areas where they should not be.

Flag state enforcement of continuous AIS operation is inconsistent. Port state control can check that AIS is fitted and functioning at the time of inspection but cannot easily verify that it was operating throughout the preceding voyage. The legal obligation exists. The enforcement mechanism is weak.

There is also the category of unintentional silence: equipment failure, antenna damage, power supply problems. A vessel whose AIS fails in the engine room switchboard is invisible to every other AIS user. If radar watch has been allowed to degrade — as it has on too many bridges — then the vessel is effectively undetected until visual range.

The ethical dimension is straightforward. AIS is a cooperative system. It works only if participants participate. Every vessel that goes dark makes the system less reliable for every other vessel. The master who switches off to avoid piracy has a defensible case. The master who switches off to avoid detection while transferring cargo at sea does not. The system cannot distinguish between the two.

8. Integration with ECDIS and Radar — the Circular Data Problem

Modern bridge systems overlay AIS targets on ECDIS and correlate them with radar/ARPA tracks. When it works correctly, this provides a rich fused picture: a radar echo with an associated AIS identity, giving name, type, destination, and AIS-derived vectors alongside ARPA-derived vectors.

When it works incorrectly, it produces something worse than either system alone.

The first problem is association error. The radar echo and the AIS target must be correlated by the system or the operator. When two targets are close together — in a traffic lane, in a convoy, at anchor — the system may associate the wrong AIS identity with the wrong echo. The watchkeeper sees a named target on the radar overlay and assumes the identity is correct. The name belongs to the vessel 200 metres away. Decisions based on that identification — including VHF calls by name — are directed at the wrong ship.

The second problem is more insidious: circular data. If the ECDIS display merges AIS and radar into a single fused target, and the operator uses that fused display as the primary situational awareness tool, then the known limitations of AIS (update interval, SOG/COG-only vector, user-entered fields) are buried inside a display that looks like confirmed, cross-referenced truth. The fusion conceals the weakness. The operator cannot tell, at a glance, whether the CPA value beside a target name came from ARPA measurement or AIS broadcast — or from some algorithmic blend of both.

When AIS position and radar position disagree — as they do in strong current, with antenna offset errors, or with degraded GNSS — the fused display must choose. Some systems average. Some prefer one source. Some alert the operator. The operator’s response depends on understanding what each source actually represents. If that understanding is absent, the alert is dismissed, and the average is accepted.

Data fusion is only as trustworthy as the weakest input. If that input is unchecked, the fusion is worse than separation — because it hides the disagreement.

The third problem is skill erosion. Officers who have grown accustomed to AIS-labelled targets on ECDIS do not always maintain the ability to build a traffic picture from raw radar returns alone. When AIS is lost — through equipment failure, through spoofing, through an area of poor coverage — the fallback is radar and visual watch. If radar interpretation skills have atrophied, the fallback does not exist.

The best practice remains the oldest: keep AIS and radar mentally separate. Use AIS to inform. Use radar to measure. Use the eyes to confirm. If any two disagree, trust the sensor over the testimony, and trust the eyes over both.

9. Closing Reality

AIS transformed maritime situational awareness. It gave names to radar echoes, made traffic patterns visible to shore authorities, and enabled a level of fleet monitoring that was previously impossible. None of that is in dispute.

But AIS is testimony, not evidence. It tells the bridge what other vessels say about themselves — their position, their speed, their status, their intentions. It does not verify any of it. It cannot detect the vessel that does not transmit. It cannot correct the navigational status that was never updated. It cannot compensate for the dimension offset that was entered wrong at commissioning and has never been checked since.

The system’s greatest danger is its appearance of completeness. A screen full of labelled, vectored, identified targets looks like total awareness. It is not. It is a partial, self-reported, intermittently updated picture with known gaps, known errors, and unknown manipulations.

A radar return with no AIS correlation is not a lesser contact. It is, in some respects, a more honest one. It says: something is there. It does not pretend to know what.

AIS tells you what the world wants you to see. Radar tells you what is there. The difference, on the wrong night, is everything.