GNSS-denied navigation for USVs: jamming, spoofing and short gaps

Jamming takes the GNSS fix away. Spoofing does more harm, because the autopilot is handed a confident, wrong position and has no reason to question it. Both matter to USVs doing mine countermeasures (MCM) and route survey, ISR picket work, and coastal and harbour patrol. Where jamming is deliberate, one outage can last for hours.

Three navigation approaches cover the field. Jamming resilience comes first, with integrity monitoring behind it, because deliberate interference is the defence problem. Gap bridging answers the short outages under bridges, in canals and beside structures. Jamming resilience and gap bridging both fuse their sensors in an inertial unit and give the autopilot a single standard navigation feed. Integrity monitoring can be added to either. The hull and the autopilot stay as they are.

Which approach for which situation

Under a bridge or beside a quay wall the sky is the problem, and an anti-jam antenna does nothing for it. Gap bridging is the answer in those places.

Sometimes the fix is lost only because the RTK correction link has dropped. GNSS still works then, and only centimetre accuracy goes. That loss falls outside GNSS denial. A PPP correction service and raw logging for post-processing (PPK) handle it, and gap bridging includes both.

Jamming resilience

This approach is aimed at waters with documented deliberate jamming, which runs from the Baltic and the Black Sea through the Eastern Mediterranean to the Red Sea and the Gulf. Jamming in those waters runs for hours and frequently comes with spoofing. The antenna keeps GNSS usable across those hours. For the minutes when it cannot, the inertial unit and DVL carry position and heading, using a north-seeking unit. IMCA Information Note 1730 names the same ingredients for areas with jamming.

Every figure above is a maker figure for the grade. System test pending. The 4- and 7-element CRPAs fitted in our configurations do not protect L5 or E5a. Integrity monitoring belongs to this approach (see below).

While the seabed is within the DVL's reach, bottom lock holds and the maker figures apply. Past that reach the DVL falls back to water track, and the drift is then set by a current nobody can see. With a 0.1 m/s current the error grows by roughly 360 m in an hour.

Export licensing applies to this approach.

GNSS integrity monitoring

A spoofed fix looks normal, so it is the hardest failure to catch. Integrity monitoring is designed to cross-check GNSS in three ways and to isolate the fix when they disagree. It comes with jamming resilience and can be added to gap bridging, so it can sit on either. It has no inertial unit of its own. An operator can have it integrated and set up for a given boat.

• What the hardware reports: the interference state the receiver gives for each band, and any jamming alarm from a fitted CRPA. A filtered antenna reports nothing.
• Galileo OSNMA is a general method: on receivers that support it, it authenticates the Galileo navigation data on E1-B. The ranging signal itself is not authenticated.
• A cross-check against dead reckoning: the GNSS position and velocity are checked against the track the INS and DVL hold.

The system is designed to drop GNSS from the inertial unit's aiding when the checks disagree, flag the navigation feed as degraded (which widens its accuracy estimate) and re-admit GNSS only after a period of agreement with the inertial track. IMCA cautions that automatic fallback may mask spoofing, which is why every switch is configured to alert the operator.

GNSS gap bridging

Gap bridging is meant for boats that lose GNSS for a few seconds up to a couple of minutes, at bridges, in culverts, beside structures and along quays. In the design, the inertial unit and the Doppler velocity log (DVL) carry the position across the gap. Once the feed is integrated with the autopilot (integration available), the vessel can keep steering and hold its line. The positions good enough for survey still come from PPK post-processing.

These are maker figures for the grade. System test pending. Export status is confirmed per destination.

Optional filtered anti-jam antenna. A fixed null at low elevation, with out-of-band filtering, rejects weak local interference such as vehicle jammers near ports and roads. Satellites low in the sky are attenuated as well.

Expected hold. With DVL aiding at sea, the maker figure for a MEMS inertial unit is about 1 % of distance travelled, for outages of up to 5 minutes. The DVL's velocity is specified at ±1 % or worse. System test pending.

Heading. The dual-antenna GNSS receiver supplies heading, and the gyros in the inertial unit keep it through a short gap. A MEMS unit cannot find north by itself, which is why gap bridging remains an answer for short outages only.

Navigation output

Fusion takes place in the inertial unit, and a single standard navigation feed goes from it to the autopilot. The feed's accuracy estimate widens as an outage goes on. Two paths carry it to an autopilot: MAVLink GPS_INPUT for ArduPilot Rover (boat mode), NMEA GST for PX4. Once integrated (integration available), either one can tell a dead-reckoned position apart from a GNSS fix.

Integration is possible with ArduPilot Rover (boat mode) and PX4, with ROS- and MOOS-based stacks, and with autopilots that take NMEA 0183.

Where these approaches stop

• Without the seabed in range, the DVL has nothing to lock to. It tracks the water instead and the current sets the drift. In deep open water we make no DVL claim.
• Sand-bed rivers rule out the DVL, since a bed that moves biases its velocity.
• Aeration under a planing hull breaks DVL lock. The A125 measures to 9 m/s (about 17 kt), and at planing speed a layer of air bubbles is drawn under the hull, which scatters and absorbs the DVL's sound. We make no DVL claim for planing hulls.
• Over a long outage, the antenna carries the load. Beyond DVL range, inertial error keeps growing with time.
• Short outages are where gap bridging belongs. The data needed for survey still comes from PPK.
• Ionospheric scintillation at high latitude defeats every antenna, and only multi-frequency receivers and inertial bridging help.
• Integrity monitoring can discard a bad fix; it cannot repair GNSS. A spoof that never leaves the inertial solution's error bounds goes unnoticed.
• Fully enclosed water is outside all three, such as tunnels or the space under wharf decks. Robotic total stations and SLAM are the tools for that work.
• Every figure here is a maker datasheet specification for a component grade. System test pending. Real performance depends on the hull, the sensor mounting and the lever arms.

The configurations, with their specifications, are on the USV navigation page.

Sources