In 2014 two ships collided in a busy German waterway. But when investigators analysed the GPS data, it showed the vessels were never less than 500m apart. Was RF interference to blame?
(With thanks to the Resilient Navigation and Timing Foundation for bringing this story to my attention.)
In September 2014, two coasters navigating Germany’s busy Kiel Firth collided in the early hours of the morning, both sustaining superficial damage.
Any collision between ships can damage the environment, making it critical to investigate the causes of such incidents. The collision of the MV FRANCISCA and MV RMS BREMEN thus triggered a full investigation by the German Federal Bureau of Maritime Casualty Investigation (BSU).
The resulting accident report does not seek to attribute blame for the collision: its purpose is purely to uncover information that could prevent future accidents. However, it does establish the contributing factors to the accident in exhaustive detail - and that makes it very interesting reading for those of us in the GNSS community.
The ships that passed in the night
In an investigation like this, AIS data plays a key role. The Automatic Identification System uses GPS to establish the position of vessels at sea and in port, enabling navigators and port authorities to track their own position and other ships’ positions on an electronic chart display and information system (ECDIS).
By examining historical AIS data, investigators can plot ships’ relative positions at the time of the incident to try to establish what happened.
But in this case, a study of the AIS data must have astonished the BSU’s investigators. It showed unequivocally that the two coasters did not collide, but rather passed each other safely.
As the report notes:
“The AIS recordings of the vessel traffic service indicate that the two vessels clearly passed each other. An electronic chart of the manufacturer and type TRANSAS 400 was on board both vessels. Recordings of them also indicate that the vessels passed each other.”
This couldn’t be true, as both vessels had sustained damage in the incident, and crews on both ships confirmed that a collision had occurred.
The obvious conclusion is that there were errors with either the GPS data or the GPS equipment on one or both ships.
The BSU passed the suspect data to the Federal Waterways Engineering and Research Institute (BAW), which “confirmed that no collision would have occurred according to the data pool, which indicated that the vessels would have passed each other at too wide a berth.”
A detailed analysis of the data found the following:
The scene of the collision was not correlated with the time. The collision occurred at 0211.At that time, according to the AIS data, the two ships were far apart
The closest distance the two ships’ GPS antennas were from each other was shown to be 566m
A GPS error compounded by missing radar data and imprecise timing
This is strongly indicative of a GPS error of some sort, especially as when the ships were moored together at a pier in Scheerhafen, Kiel, following the accident, the AIS data showed no anomalies and both ships’ positions were displayed accurately.
There were other factors, too. Firstly, the BSU were unable to correlate the AIS data with radar data on ship locations, as the Vessel Traffic Service in the port of Travemünde, which collects both AIS and radar data, deletes its radar data after a month.
Additionally, it wasn’t possible to determine the exact timings of the events leading up to the collision, because ships only transmit AIS data in seconds. It’s up to monitoring stations receiving the AIS transmission to generate a full timestamp down to the microsecond level, but this didn’t happen because the Joint Control Centre of the Waterway Police of the Coastal States didn’t have a precise time server installed. (The BSU notes in its report that this has since been rectified.)
This demonstrates how important it is to assign precise timing data to critical events and transactions, especially where they are likely to play a part in determining liability or responsibility.
Was RF interference to blame?
So what caused AIS to misreport the position of one or both of the ships involved? We can’t be sure, but the closeness of the collision to land infrastructures means GNSS interference could be a factor.
In 2010, the STAVOG project undertaken by the General Lighthouse Authority (GLA) in Harwich showed that a GPS jammer on land could affect positioning and navigation systems on ships approaching and leaving the port.
Disturbingly, that project showed that depending on its power and its location relative to the ship, a jammer doesn’t always fully block GPS reception, but can instead cause some GPS receivers to miscalculate the vessel’s position.
In the STAVOG instance, a low-power jammer on board the ship caused bridge systems to misreport position by up to 270m, with no alert or alarm being triggered.
Image courtesy of the STAVOG project
The Kiel Firth accident report doesn’t rule out RF interference as a factor, stating instead that “it was not possible to find any evidence of interference caused by […] intentional manipulation”. However, that conclusion was arrived at by examining data from “monitor stations located further away from the scene of the accident”. If the interference was particularly localised, more distant stations would not necessarily have recorded anything amiss.
If RF interference did play a role in this incident, it wouldn’t be the first time it has disrupted operations in or near busy ports. In January 2016, the U.S. coastguard warned mariners to be on the lookout for GPS disruption, following an incident in the previous year when vessels leaving an unspecified non-U.S. port experienced a total loss of GPS signal for 6 nautical miles. The blackout rendered their ECDIS and other bridge systems (including systems for collision avoidance) non-operational until the signal returned. And in 2015, automated crane operations at a container port on the east coast of the United States were brought to a halt for seven hours due to GPS disruption.
Detecting RFI in the environment
The increasing reliance on GPS data for maritime operations (both onshore and at sea) suggests that port authorities should consider monitoring the RF environment in and near the port.
Analysing the environment for interference to GPS and other GNSS systems will help authorities to determine whether there is any risk to vessels entering and leaving the port - and to include it or rule it out as a contributing factor when incidents like the BREMEN - FRANCISCA collision occur.
A good way to do that is to employ a network of detector stations that continuously monitor those frequencies for interference events. These can be analysed and correlated against the time of unusual anomalies being observed, and, in the case of a collision like this one, can provide invaluable evidence for investigators.