4.1. eLoran

The GLA have long recognised the dangers associated with an over-reliance on GPS [2,3,4] as well as the many aspects of GPS vulnerability. e-Navigation, the future digital maritime architecture, emphasises the need for robust and resilient PNT in order to reduce the impact of human error and to improve safety, security and protection of the marine environment. The GLA have identified the strategic benefits of having two satellite navigation systems (e.g. GPS and Galileo) as well as the importance of system diversity using eLoran.

Enhanced Loran (eLoran) is needed to mitigate the well-known vulnerabilities of GNSS, thereby securing critical infrastructure and allowing users to retain the safety, security, and economic benefits when GPS, and other GNSS services, are disrupted. The GLA, supported by the UK Department for Transport (DfT) have invested in a 15-year contract for the provision of an eLoran service from a site in Cumbria. In addition, the US government has recently selected eLoran as its national backup to GPS [5,6,7]. eLoran is an all-in-view navigation system employing pulsed groundwave radio transmission with a centre frequency of 100kHz. The times of arrival of the signals are measured by the user's receiver and the pseudoranges are used in a positioning algorithm in much the same manner as that employed by GPS. This trial presented an unrivalled opportunity to investigate the real-life performance of eLoran under conditions of GPS service denial.

The eLoran receiver used for the trial had an integral GPS receiver, which meant that the unit was able to calculate stand-alone GPS, Differential GPS using Eurofix corrections, calibrated Loran using GPS, Addition Secondary Factor (ASF) corrected Loran and stand-alone Loran. For the trial, the receiver operated in Eurofix corrected GPS, while GPS was available, and then reverted to calibrated Loran and finally ASF corrected Loran.

While GPS was available the eLoran output was periodically calibrated by the GPS solutions. This was effectively the same as continuously using the GPS position as a “ground truth” and computing the offset of the eLoran positions from it. eLoran pseudoranges were adjusted according to the position offset. When GPS was not available, the integrated output was derived from Loran using calibrated Loran pseudoranges. Calibration of the Loran ranges was done every time a new Loran measurement was available (e.g. every 5 seconds). The calibration value was determined based on a minimum set of 5 calibration points, hence there was some 30 seconds before a calibrated eLoran position was first output. Over the set of available calibration measurements a weighted average calibration value was computed. When GPS was unavailable the last calibration values were employed. Under GPS denial conditions it depends on how long the GPS receiver still reports positions before it affects the eLoran calibration. Typically, many GPS receivers do not indicate if the position is wrong due to interference and this was witnessed during the trial.

Since the effect of GPS jamming on the GPS calibrated mode of operation was not known, ASFs were recorded during one of the control runs, when GPS jamming was not enabled, and the ASF data was stored in the receiver. Once GPS becomes unavailable, or the latest GPS calibration becomes invalid, the receiver will automatically revert to using ASF corrected Loran for its positioning output.

By using data collected during the control runs of the trial, it was possible to determine the positioning accuracy performance of Loran in the area of the trials. This was achieved by comparing calculated calibrated Loran positions against differential-GPS as provided by Eurofix. Figure 3 shows a scatter plot of the comparison of the Loran positions against the differentially corrected GPS positions. The red circle indicates a 95% position accuracy of 8·1m. It was only possible to assess the accuracy of calibrated Loran in this manner as the eLoran receiver will only revert to the other positioning modes when GPS is not available and, since the receiver was moving, DGPS was required to provide the true position.

Figure 3.

Scatter plot showing the performance of calibrated Loran when compared to Differential GPS. Loran was accurate to 8·1m (95%).

Figure 4 shows eLoran derived position data as the vessel navigated through the jamming region under control conditions, when the jamming unit was not operating (left-hand plot), and when the GPS jamming unit was operating (right-hand plot). In both cases one can clearly see the route taken between the waypoints. From the right-hand plot of Figure 4 it is possible to see that the eLoran receiver was able to provide a useable position output throughout the jamming trial. The only discrepancy between the two plots is a slight wave to the trace. This is due to the vessel being navigated using parallel indexing which is significantly less accurate than GPS and eLoran. Clearly the performance of eLoran is not affected by GPS jamming, as one would expect.

Figure 4.

Two plots of eLoran data displayed in Google Earth™. The left-hand plot is the reported positions during a control run when there was no GPS jamming. The right-hand plot shows the reported positions during a run with GPS jamming switched on. The red lines indicate the approximate boundaries of the main lobe of the GPS jamming signal. eLoran was not affected by GPS jamming.

This trial justifies and confirms the GLAs' eLoran strategy.

4.2. Differential GPS

The GLA operate 14 Differential GPS reference stations, arranged throughout the United Kingdom and Ireland. These stations provide mariners with GPS corrections by which they can increase their positioning accuracy and gain integrity. Each station consists of a reference station (RS) unit and an integrity monitor (IM) unit, which calculate pseudorange corrections and then check the integrity of those corrections. The GLA provide two of each unit within their stations ensuring system availability should one unit fail. To ensure the integrity of the pseudorange corrections, each unit (RS or IM) calculates its own GPS position, therefore one would expect GPS service denial to affect the corrections provided.

The DGPS reference station at Flamborough Head was intentionally disrupted using the jamming unit on a reduced power. The power of the jamming unit was gradually increased until it affected the performance of the reference station. This was observed on the Trimble Beacon Control Software™ (BCS), which is used to control the reference station. Figure 5 shows a screenshot of the BCS software with a number of windows visible. The four windows on the top right of the image give details of the satellites being observed by the two reference stations in the top row (RS1 and RS2) and the two integrity monitors in the bottom row (IM1 and IM2). Within each of these windows the location and health details for each satellite being tracked are shown along with the signal-to-noise (SNR) ratios. Before the GPS jamming signal was enabled, typical SNR values of between 15 and 20 were observed for all satellites. When the power of the jamming signal was increased these SNR values fell and the number of satellites used by the reference station and integrity monitor units reduced until there were no usable satellites. The reference station raised an alarm when the number of satellites fell below the required minimum, showing that GPS jamming does affect the performance of Differential GPS reference stations as expected.

Figure 5.

Screen shot of the Beacon Control Software^TM at Flamborough Head showing that the Signal-to-Noise ratios (SNR) for all satellites on both reference stations and on both integrity monitors (four windows to the top right of the image) are zero and that this has lead to the station issuing an alarm (red area at the bottom left of the screen).

There is great potential for GPS service denial to have serious consequences for maritime radiobeacon differential-GPS service providers and their users. The trial shows that a relatively low power jammer placed near a reference station, or a passing vessel with faulty equipment onboard (e.g. a defective UHF active television amplifier [8]) could result in the disruption of DGPS service provision out to several hundred kilometres from the reference station.

4.3. Synchronised Lights

With the increase of background lighting in ports and port approaches, it is becoming more difficult for mariners to recognise AtoN lights and thereby ensure that they are in the correct position. The GLA are actively looking at methods for making AtoN lights more conspicuous so that they can be easily recognised. One important method is to deploy synchronised lights.

Synchronised lights are conventional AtoNs, however when multiple lights are situated in close proximity, they can be synchronised to a common time source and configured to either flash together or flash in sequence, drawing the attention of the eye. Once they have attracted the eye the aid can be identified by its flash character and referenced to a navigation chart to confirm the vessel's position. Typically, GPS is used as the common timing source in these units as it provides universal time (UTC) at low cost in a conveniently small package. Clearly, lights using GPS for timing will be affected by GPS service denial. The trial was set up with the GPS jamming unit set on reduced power to limit the jamming area. The lights used in the trial were configured to illuminate on power-up rather than wait until dusk, and two tests were performed. The first test followed the scenario that the lights are already on and synchronised when GPS is jammed. The second scenario was for GPS to be already jammed before the lights are powered-up. The particular lights used in the trial synchronise their internal oscillator to GPS and then resynchronise with GPS every 20 minutes. If the units fail to re-synchronise then they rely on their internal oscillator to keep time.

For the first scenario, the lights were powered and allowed to fully synchronise, at which point the GPS jamming signal was enabled. The power of the GPS jamming unit was increased until a hand-held GPS unit situated alongside the lights failed to acquire any GPS satellites. The lights were left to operate for over an hour, during which time they could not re-synchronise, and over which time it was not possible to observe any loss of synchronisation. The lights were then switched off for 30 minutes before the second scenario started.

For the second scenario, the GPS jamming unit was enabled on the same power as before and then the lights powered. After 30 minutes the lights had not synchronised and it was clear that GPS jamming prevented synchronisation; at this point the lights were flashing out of phase.

Clearly the results here are that the effect of jamming depends on whether the lights have been able to synchronise or not. If the lights have been able to synchronise then they are reasonably resilient to jamming signals, with manufacturers stating that lights can remain synchronised for several hours before any noticeable effects are seen. This latter statement remains to be confirmed although eLoran would also be a suitable timing source that is unaffected by GPS interference.

5.1. GPS and Differential GPS Receivers

Three additional receivers were installed on the trial vessel, two of which were typical marine grade differential GPS receivers, the third being a more expensive dual-frequency surveying receiver (configured to operate on GPS L1 only). Data in the form of NMEA sentences were recorded from each receiver throughout the jamming trial. It should be noted that due to a lack of space on the vessel's mast, antennas for the three receivers were installed on the handrail of the main deck, which meant there was a certain amount of sky obscuration due to the vessel's superstructure.

Over the course of the dynamic trials the receivers were monitored and all of them lost GPS lock. The two differential receivers maintained lock on the medium frequency broadcast from the nearby Flamborough Head DGPS reference station; however as the reference station was also affected by the jamming signal, there were no corrections to apply and their position solution was derived from stand-alone GPS. When processing the recorded data from the three receivers, the NMEA GPRMC (recommended minimum content) sentence was used as this provides the reported position, speed and time. This sentence also provides an indication of the validity of the data, setting or clearing a single bit flag. The decision to set or clear the data valid flag is one that is made by the receiver. When processing the recorded data from the various receivers only data declared valid was used, which resulted in the two typical marine grade receivers providing erroneous positions as they entered and exited the jamming region. The magnitude of the position error varied, with some small errors, but with others several tens of kilometres away from the true location. Figure 6 provides the reported positions from one receiver, plotted on Google Earth™. The left-hand plot is from the control run where the jamming unit was disabled; the right-hand plot is from a run where the jamming unit was enabled and erroneous data was observed. The colour of each reported position is an indication of the reported vessel speed at that moment, with blue positions indicating a speed of less than 15 knots; yellow positions indicating a speed of between 16 and 50 knots; orange positions indicating a speed of between 51 and 100 knots and red positions indicating a reported speed of greater than 100 knots.

Figure 6.

Google Earth™ plots of recorded positions identified as valid. Left: No GPS jamming unit active. Right: GPS jamming unit active. The colours are used to represent the reported vessel speed with blue <15knts, yellow <50knts, orange <100knots and red >100knts.

Figure 6 (Right), shows that the number of erroneous positions was significant, with the majority of positions coloured red, indicating the reported speed was greater than 100 knots (the greatest reported speed was over 5000 knots). Clearly, if this data was being used as input to a navigation system, whether it was an autopilot or simply an electronic chart the implications would be serious. The results shown were typical from the two marine grade receivers, although it was noted that the effect of jamming was more severe when sailing North with the vessel superstructure between the jamming unit and the GNSS receivers' antennas. Therefore, it may be presumed that the jamming signal was attenuated due to the shadowing effect of the vessel's superstructure and the “moment of indecision”, that period of time when the strength of the jamming signal was comparable with that of the GPS satellites, was greater and resulted in an increased number of erroneous positions. The more expensive survey grade receiver did not provide any erroneous data positions, rather opting to provide no position information when experiencing interference from the jamming unit; clearly this is the preferred situation.

The GPS receivers onboard Pole Star were also affected by the jamming signal and also reported inflated speeds, albeit to a smaller degree. The reported position on the vessel's ECDIS wandered around and the reported speed also increased above the maximum speed of the vessel. However, the vessel's receiver did stop providing position information quite quickly once the vessel had passed into the jamming area. The implications of providing erroneous positions can be severe and can greatly affect the safety of the mariner and those around them.

As noted in Section 4.1, the reported position from eLoran was not affected during GPS jamming and the receiver was able to function as normal, as one would expect. In order to calculate the magnitude of error, an erroneous position from one of the typical marine grade receivers is compared to that from the eLoran receiver for the same moment in time, in Figure 7.

Figure 7.

Google Earth™ plot of valid GPRMC data from one of the typical marine grade receivers with the comparison of an erroneous GPS position (red circle) against the eLoran position (green square) for the same time. The GPS position is reported as being inland 22Km West from the true eLoran position. (Red lines indicate main lobe of the jamming unit and position colours indicate reported speed: blue <15knts, yellow <50knts, orange <100knots and red >100knts).

5.2. Automatic Identification System

The Automatic Identification System (AIS) provides information on the vessel's identification, position, course and speed, as well as destination, estimated time of arrival and other information. It can provide this information from ship to ship or ship to shore. AIS transponders exchange information over two VHF frequencies using Self Organised Time Division Multiple Access (SOTDMA), which requires a common timing source, for which AIS uses GPS. Therefore, GPS denial will not only affect the vessel's reported position and heading, but also the synchronisation of data between AIS transponders. Although AIS transponders would primarily use GPS for slot timing, they can still function by using a base station for synchronisation. However, as long as they rely on GPS for position, GPS service denial could render AIS useless.

When entering the jamming region, Pole Star's AIS unit provided an audible alarm when it lost GPS. From that point it was not able to calculate its own position and although it was receiving information from surrounding vessels it was not able to calculate a range or bearing. The result of this was that the data presented on the Minimum Keyboard Display (MKD) had limited use, although corrupted AIS data could still be overlaid on the vessel's radar display. By observing the radar display, one could see one of the more significant effects of GPS jamming, one with hazardous consequences. Figure 8 shows a photograph of Pole Star's radar display. On the radar there is a crosshair (marked in the red square), which highlights the radar return for a nearby vessel. That vessel's reported position via AIS is considerably further to the north (vessel number 33458, highlighted in the red circle), clearly showing the effect of jamming on the GPS receiver onboard this vessel, which is resulting in an erroneous AIS reported position.

Figure 8.

Photo of NLV Pole Star's radar with AIS overlay. Note the yellow crosshair (red square) indicates the radar return of vessel “33458” whose AIS position is reported to be some distance to the North (red circle).

The effect of GPS jamming can also be observed in Figure 9, which was provided by the Maritime and Coastguard Agency (MCA) and was recorded from the AIS base station in Flamborough. The figure shows a series of snapshots of the maritime traffic off the East coast of Flamborough, with the left-hand image preceding the right-hand image by a few seconds.

Figure 9.

AIS traces as observed from the MCA station at Flamborough. Note the recorded trace of NLV Pole Star as she passed through the GPS service denial region. The trace not only extends further north and south than the true passage, it also shows the vessel crossing the land. The location of Dutch Progress is also marked as her position changes due to erroneous GPS input to her AIS.

Both images show the reported position of the trial vessel NLV Pole Star, which was being tracked and one can see the green trace over the peninsular as a result of several runs between the two defined waypoints. The image was taken while the GPS receiver that provided Pole Star's AIS position was in the “period of indecision” and is wandering around. These images give an indication of the effect of GPS jamming on other vessels in the vicinity. On these AIS plots, vessels are identified by green triangles, the orientation of which provides an indication of the vessels' respective headings (the direction in which the vessels are pointing). The course-over-ground of each vessel is indicated by the blue dashed line (the direction in which the vessels are moving), with the length of the line indicating the vessel's speed. Most vessels are seemingly not affected and their reported position data looks normal – the vessel's heading and course-over-ground agree with each other. However, if one compares the vessel highlighted in the two images, the Dutch Progress, she is clearly being affected by GPS jamming. In the left-hand image, the vessel is South of Pole Star and reporting that she is travelling in a South Easterly direction, although the reported course over ground is North East. However if one then compares the reported position in the right-hand image, taken a few seconds later, her location has moved North by some way and one sees that she is reported to be inland while showing a heading of South East. Her course-over-ground is in the opposite direction and with a significant speed. This is a result of the vessel's GPS reporting erroneous positions and as such the vessel's reported position jumps around randomly.

AIS enables vessels and land based infrastructure to build up an image of the marine traffic around them and as observed, in times of GPS jamming, whether intentional or unintentional, the effect is to distort this image and to introduce hazardously misleading information. While larger vessels and port infrastructures combine reported AIS data with radar returns, the result is that two positions are given with uncertainty over which position to believe.

5.3. Communication systems

Vessels may employ several radio-communication systems including analogue (short-range maritime VHF and long-range maritime HF) and digital (cellular telephone and satellite systems). GPS service denial may affect digital communications systems if it is used as a source of accurate timing for data slots, or is used to provide a position. Digital Selective Calling (DSC) systems use GPS to provide a positional input should an alarm be activated. DSC allows mariners to transmit their position, via digital data modulation of maritime band VHF, to other DSC units in case of an emergency, in order to request assistance.

Because of the implications of testing an emergency system, the emergency capability of the DSC unit onboard Pole Star was not activated, however the unit did issue an audible alarm when it lost GPS. Having not been able to actively test the DSC unit, it is not clear whether the integral GPS receiver would report an erroneous position or not. Clearly the implications could be significant if it did provide an erroneous position.