Original URL: http://www.theregister.co.uk/2011/03/10/rae_gps_report/
Chicken Little report: Sat-nav dependency spells DISASTER!
Quango pushes 1940s tech as backup for GPS
Analysis A heavyweight UK tech body has just issued a report claiming that growing dependence on satellite navigation systems poses serious economic and safety risks to society. There's some truth in the report, but unfortunately it verges on scaremongering at times and appears to have been unduly influenced by organisations which can't be regarded as impartial.
So what does the report from the Royal Academy of Engineering (PDF/984KB) have to say?
Much of it covers well-trodden ground that many Reg readers will be familiar with. Satnav receivers (almost always using the US military Global Positioning System satellite constellation, GPS, at the moment) are now ubiquitous. Quite apart from car dashboard units, GPS receivers are now found in phones, tracking bugs, aircraft, ships, drones, automated combine harvesters etc etc.
Somewhat fewer of us are aware that GPS also offers a very precise timing service – it naturally has to, as the ability to measure very small intervals of time is required in order to work out one's position using GPS signals. These days a lot of hardware makes use of the GPS timing signals for purposes unrelated to navigation: for instance to synchronise the precise clocks often needed in digital communication systems, or to put exact time stamps on financial transactions. Using GPS is often cheaper than installing a highly accurate timepiece.
All this, according to the RAE report, introduces vulnerabilities. GPS signals received on the ground are weak, which means they can easily be interfered with – by natural phenomena such as solar storms, accidentally by faulty equipment, or on purpose. This could mean serious economic disruption or in some cases injuries and deaths.
So what should we be doing about that – in the opinion of the report's distinguished authors?
Again, it's mostly common sense. Users shouldn't develop an over-reliance on GPS – or satnav generally, once complementary services such as Europe's Galileo or Russia's GLONASS expand their coverage. Designers should build systems so that a loss of signal won't be catastrophic: quite a lot of communications hardware apparently has no option for timing other than GPS, for instance, and this should be rectified.
So far, so uncontroversial. But then we get onto the question of backup systems. The report says:
The provision of a widely available [navigation and timing] service as an alternative to [satellite services such as GPS] is an essential part of the national infrastructure. It should be cost effective to incorporate and free to use. Ideally it should provide additional benefits, such as availability inside buildings ... We are encouraged by progress with eLORAN [enhanced LORAN] in this context.
LORAN is a nav system dating from the 1940s, using long-range lower frequency radio signals from ground stations. It is wildly inaccurate compared to GPS or other satnav methods, but it was widely used at sea in the pre-GPS era as it was often the only electronic navigation aid available. The UK had a different system called Decca.
Here in the UK (and Ireland), Decca was formerly run by an organisation known as the General Lighthouse Authorities (GLAs), an alliance of the organisations set up to run lighthouses in the mainland UK (Trinity House and the Northern Lighthouse Board) and Ireland (the Commissioners of Irish Lights).
Trinity House tends to set the lead for the GLAs. It is an interesting organisation: partly an arm of government funded by a special tax (the "light dues" paid by ships visiting English ports), partly a private-sector offshore contractor and pilot bureau, and partly a seafarers' charity. The GLAs as a whole are in the nature of a sector-straddling quango, at least partly dependent on government funding to exist – but not really under any serious government supervision and operating in many ways like a private firm.
Decca was always highly unreliable and inaccurate (your correspondent, as a Royal Navy navigating officer at sea in the early 1990s, was furnished with GPS only occasionally and thus had to make occasional reluctant use of Decca). As soon as GPS got into widespread use, even back in the old days when the civilian GPS signal's accuracy was purposely degraded by the Americans, Decca was dead. It was finally switched off in 2000.
At first the GLAs thought they might find a replacement for their lost Decca portfolio in operating a new network of ground stations delivering a service known as "differential GPS". This was a workaround which eliminated the deliberate inaccuracy built into the civil GPS signal: a ground station, knowing its own position, would calculate the error in near-real-time and transmit corrections on a radio channel. Thus a dGPS receiver in range of a GLAs station would obtain something close to the full accuracy the system could deliver, an error of a few metres at most, instead of the 100-150m accuracy the US defense department wanted to offer to people not in possession of a military P-code receiver and accompanying crypto keys.
The GLAs' free-to-air dGPS network got running in 2002, enabling a fairly seamless transition from Decca for the organisation. Unfortunately dGPS had already been rendered obsolete for all but specialist applications, as President Clinton had ordered that the civil GPS signal no longer be degraded two years earlier.
A dGPS receiver is still somewhat more accurate than an unassisted GPS – perhaps a single metre of error – but hardly anybody needs such accuracy right now. By the time they do, say a decade or two from today, Galileo and the new GPS Block III satellites will offer 1m accuracy without any ground-station assistance. The case for maintaining the GLAs' free-to-air dGPS ground stations at public expense is already shaky, and soon enough it will be untenable.
Funnily enough, the GLAs have now switched from trying to augment satellites to saying that they are unreliable and that some kind of publicly funded backup nav-and-timing service – provided by ground signal stations – is essential. They already have a government deal to run a central LORAN transmitter at the famous (in radio circles) Anthorn station in Cumbria – also the location of the UK's very-low-frequency transmitter, used for communications with submerged nuclear submarines.
There's a major problem with LORAN, though – it is very inaccurate. Ordinary LORAN-C only tells you where you are to within several hundred metres or worse: it is basically useless for anything but long-range maritime or air navigation (often no good even for that as it offers no coverage far out in the ocean). It can't as it stands act as a backup for GPS in bringing ships into harbour or planes down to runways; it would be entirely pointless on a car dashboard, in a tracking device etc.
The GLAs really don't like to mention the accuracy of the "enhanced LORAN" – eLORAN – system they are pushing now. No figures are given in the RAE report.
GPS jamming trial caused chaos on ship's bridge
The fact is that accurate eLORAN is more a goal than a reality. It might prove capable one day of meeting a US Coastguard requirement calling for "8-20m" accuracy, but this hasn't been demonstrated: and such performance is marginal at best for most modern GPS applications. It is unacceptable for future ones that will call for sub-1m accuracy, such as lane control on the motorway and so forth.
The GLAs prefer to talk about a trial they did in which they used differential eLORAN – that is LORAN enhanced by the use of a nearby ground station sending error corrections, as with dGPS. Naturally this is right up their street. As the GLAs' document The Case for eLORAN tells us:
Initial differential eLoran trials conducted at Harwich in April 2006 have demonstrated horizontal positioning accuracies better than 9m with 95 per cent confidence using modern, miniaturised eLoran receivers. This puts eLoran on the same basis as single frequency GPS or Galileo: each requires differential corrections to guarantee meeting the International Maritime Organisation's future accuracy requirements for port approach and restricted waters.
This is disingenuous at best: in no way does <9m with differential assistance put eLORAN "on the same basis" as satnav for accuracy. Given the same conditions, differential GPS or Galileo would be accurate to within a metre, an order of magnitude better than eLORAN. Both systems promise sub-metre accuracy unassisted by ground station in future.
It's pleasing to see that the RAE report stopped short of outright recommending eLORAN as a mandatory backup for satnav, but it was still mentioned a lot: and the GLAs managed to slip into the appendices another recent trial of theirs in which they carried out GPS jamming using borrowed Ministry of Defence kit against one of their ships, NLV Pole Star, off Bridlington.
This apparently caused dangerous chaos on the Pole Star's bridge:
When Pole Star entered the jamming zone, numerous alarms sounded on the bridge over a period of approximately 10 minutes. These alarms were all linked to the failure of different functions to acquire and calculate their GPS position, which included: the vessel’s DGPS receivers, the AIS transponder*, the dynamic positioning system, the ship’s gyro calibration system and the digital selective calling system ... In the situation where a crew was not expecting this level of system failure then the distraction caused by so many alarms sounding at once could have had a significant effect ...
Some vessels have integrated bridge systems, which enable automatic execution of a passage plan on autopilot. If this system is operating at a time that jamming occurs, then the vessel's course and heading may change without informing the crew, potentially leading to extremely hazardous consequences.
Although the Pole Star's crew was expecting GPS failure, problems were experienced. The vessel's Electronic Chart Display & Information System (ECDIS) was not updated due to the failure of the GPS input, resulting in a static screen. ECDIS is the normal mode of positioning on board Pole Star (with paper chart backup) and during the periods of jamming some crew members became frustrated when trying to look at the ECDIS. This resulted in the monitor being switched off!
There are several questions raised by this trial, such as the ability of a vessel's crew to quickly revert to traditional means of navigation and also the extent to which they are able to navigate with these means. Given the greater reliance on satellite navigation, in particular GPS, these skills are not being used daily and are no longer second nature.
QED, then – we definitely need eLORAN! And differential eLORAN stations everywhere run by the GLAs, because eLORAN on its own isn't much good!
Actually, it seems that what we need is properly designed bridge hardware and properly trained watchkeeping officers. As a bridge watchkeeping officer myself from 1993 to 2001, I have been in the kind of situation described above far too many times: the Royal Navy minehunters I was handling gradually acquired nearly all the kit described above in early military forms (therefore highly unreliable forms).
A bridge watchkeeping officer who can't cope on passage without an electronic chart and automated collision warning is an unskilled monkey hardly worth paying: an "integrated" bridge that sounds multiple alarms for a single problem is not integrated at all (certainly none that I ever had charge of were, but one copes all the same). An ECDIS system which doesn't continue running an automated dead-reckoning plot (and display a growing circle of uncertainty) on losing GPS is badly designed. The ship's log – the speed readout, not the records book – and its compass could and should also serve as a simple safeguard against the "Hazardously Misleading Information" problem, where satnav or another electronic nav aid gives locations which are plausible but wrong.
How on earth did sailors ever manage without GPS and AIS*? We used to use this amazing thing called radar
Quite frankly in coastal situations it will not be eLORAN that saves a ship when GPS goes out – it will be radar and the Mark I eyeball, both essential anyway to avoid the large numbers of smaller vessels and other hazards which carry no AIS transponders and are not marked on charts. Radar will also keep you off the mud if you can navigate at all.
Nonetheless the RAE report's authors write:
At sea in fog or at night, jamming could cause collisions between ships or with obstructions ...
That's perilously close to being movie-plot stuff. We contacted Richard Ploszek at the RAE, who helped write the report, to see what he had to say.
"The emphasis is on could," he told the Reg. "We feel that safety margins are being eroded."
Radar and basic seamanship will get you in and out of harbour, too, in any visibility conditions short of freakish. Sure, plenty of ordinary longhaul bridge officers – and even some captains and/or specialist navigating officers, both merchant and naval – would struggle, especially in really thick fog, but they don't normally take their ships in and out of unfamiliar ports: expert local harbour pilots do, and they won't have any trouble without GPS.
Fully automated berthing (where you might not get the chance to call in a pilot if the system failed) calls for an extremely manoeuvrable ship, most likely fitted with bow thrusters and/or omnidirectional Voith-Schneider thruster propulsion. A captain who can't get something like that alongside the jetty safely without help from automated positioning, fog or no fog, shouldn't be in his job.
Far out at sea where you can't use radar to navigate, it takes long enough to get anywhere that you can simply run on dead-reckoning and wait for GPS to come back on – or get out your trusty sextant once the weather is clear enough as people used to do, and as all sea officers are still trained to do. Don't bother turning on the eLORAN, though, not far out at sea. It doesn't reach there.
As for non-maritime uses for eLORAN, it just isn't accurate enough - certainly not without loads of differential assistance stations all over the place, and not even then for metre-accuracy applications of the future. It might serve as a means of detecting a gross error in a satnav position, but in most cases you could do this much more easily using the local cell tower network as a cross-check. A large number of GPS-enabled devices already include mobile comms (or are mobile comms) and this will only increase. Basic cell location based on a single tower is inaccurate – often as inaccurate as LORAN, in fact – but you can do a lot better by triangulation using several cells rather than just one, and plenty of work has already been done in this field.
Still though, there is the matter of timing. Those local cell towers are often dependent on GPS time signals to stay in sync, remember.
Well, sort of. There is actually a government-supplied free backup time signal already: the "MSF" (nobody really knows what this stands for) radio station which has long been run by the National Physical Laboratory, now co-located with the new UK LORAN transmitter at Anthorn. So in fact nobody needs to be dependent on GPS for accurate radio time signals.
The RAE's Ploszek suggested to the Reg that the MSF signal isn't good enough to stand in for GPS, saying "at 60Hz, I'd suspect that it isn't going to offer enough precision".
But the GLAs - who would surely tell us if MSF is no good - write in The Case for eLORAN:
MSF is used by many industries to back up GPS as the primary means of timing. Users include telecommunications networks ... The need for the MSF will be reviewed in 2010, at which time a decision will be taken on its possible replacement by LORAN in 2017. There is a window of opportunity to have eLORAN take over the provision of timing signals from the MSF.
In any case, cutoff of timing signals is hardly an occasion to panic that the mobile data network or other important digital comms will come crashing down. The GLAs also admit that cell towers will keep on working for days before loss of timing signals will cause them to fail. Local jamming or interference would have to go on for days at a time before having any effect on comms. Certainly, as the GLAs say, "a long-term catastrophic loss of the core GPS constellation" would have severe effects, but not a total wipeout - if the GLAs are to be believed, at least some of our comms network can switch to MSF backup for timing.
Then, it seems quite plausible that soon enough highly accurate clocks will be easy to install, so largely removing the whole timing issue:
Chip Scale Atomic Clocks [CSAC] are just emerging from the R&D labs with first commercial deliveries recently announced ... This represents a clear paradigm shift in technological innovation and this is a technology to watch since it promises an ability to withstand outages for many months.
'We're not saying the sky is about to fall in'. Yes you are, a bit
Despite all this, the RAE report also says:
Erroneous GPS signals in an urban area could cause road accidents whilst disrupting the dispatch and navigation of emergency vehicles and causing their communications systems to fail [several days later, it would seem].
Again, this is scary stuff. The lead author of the report, Dr Martyn Thomas, tells the BBC that "We're not saying that the sky is about to fall in; we're not saying there's a calamity around the corner," but he sort of is, actually.
We put this to Ploszek, who said: "Satnav errors can have effects on the situational awareness of drivers and mariners."
Nonetheless we should probably bear in mind here that GPS gets jammed and/or spoofed all the time already, without any of the disasters that the RAE report predicts:
In the USA, the JLOC network has been established to detect jamming nationwide. They are currently registering thousands of incidents of jamming each day – many of them legitimate use by authorised agents**.
Ploszek adds however that if and when location-based road charging comes in, the scope for jamming will increase and nuisance/outages will rise with it unless the system is set up correctly.
Even so it's hard to imagine how eLORAN would help. We might consider that one of the authors of the RAE report is from the GLAs, with a conflict of interest over eLORAN – which might explain why the document mentions timing-signal vulnerabilities and pushes eLORAN repeatedly but doesn't mention the MSF signal at all. The fact that useful checks and backups for satnav position errors already exist – cell location, inertial etc – is mentioned but only in order to say that eLORAN would be better, at least at sea.
So: this report is definitely incomplete and, bluntly, has strayed into scaremongering territory.
There are surely some grounds for concern at the increasing use of satnav technology - nobody would deny that. As Ploszek and the RAE say, it is certainly unwise to build hardware which relies solely on GPS, whether for position or timing - and it's madness to do so when the kit in question is going to be used as a backup for GPS, as in the case of marine radar (this is apparently quite common nowadays).
But many backups and checks already exist that could be incorporated into equipment today: the case for more government funded infrastructure in general is unproven. The existing eLORAN station, whose signals reach well beyond the shores of the UK, already solves the timing-signal issue for applications where the MSF station is unsuitable: there seems no real reason to use it for navigation too, not when backups like radar, cell triangulation and inertial are already present. One of the primary reasons offered by the RAE and the GLAs for eLORAN positioning on top of timing is that it will permit such applications as road charging and prisoner tracking to work better - but many people would question whether this actually strengthens the case for eLORAN.
People often forget, as well, that GPS wasn't primarily intended for air or surface navigation, or for timing. The primary reason that the USA created it was actually to supplement the inertial guidance of nuclear missiles. As such GPS is a key part of the US nuclear deterrent, which means that the US military and government regarded it – and mostly continue to regard it – as essential to their national survival. It was built, and is maintained, with that in mind.
The chance (for instance) that any future US administration will actually allow GPS worldwide coverage to lapse due to relatively paltry budget issues, as has been speculated in recent times***, is about the same as the chance that the USA will unilaterally disarm itself.
GPS is probably a lot more to be relied upon than people think. ®
*AIS is nowadays mandatory for ships of any size. It transmits the vessel's identity, position, course and speed data on a VHF band, most of which info it works out using GPS. An AIS receiver can thus plot all nearby ships on a display and tell whether any of them pose a collision risk.
You still need eyeballs and radar because small vessels don't carry AIS. A proper radar can also track contacts and work out their closest-point-of-approach (CPA) just as well as AIS can. (It isn't in the report, but Ploszek tells us that the Pole Star's radar was also reliant on GPS timing - we can certainly agree that nobody should be making or buying radars like that for use on ships).
**One interesting item revealed by the report is that so-called "Blue Team jamming" of GPS is actually quite common, both in the USA and UK. We learn:
Blue Team Jamming [where Blue Team is a generic term for "friendly forces"] is deliberate – generally to defeat a perceived threat of covert tracking. It will probably be low power and have a similar impact to criminal jamming. However there would be an impact if they [who?] parked for long periods near critical infrastructure which used GPS timing.
One might suggest that officially sanctioned GPS interference would also be a likely tactic against a perceived threat from improvised drone missiles, relatively easily made by adding an explosive payload to a commercially available GPS-guided UAV. This threat is already much bigged-up in security-fear circles.
***By the US Government Accountability Office. This would seem an impeccable source of information, but actually the GAO is surprisingly erratic at times - for instance it recently, nonsensically, asserted that Craigslist and eBay are terrorist arms bazaars.
Lewis Page was a Royal Navy officer from 1993 to 2004. He served as navigator (and precise-navigation officer, necessary aboard mine countermeasures ships) during the days when civil GPS was degraded and p-code or differential GPS weren't commonly offered to the minewarfare community. As such he often had to use various recondite and unreliable nav technologies including Decca, Hyperfix, Microfix etc. He also spent much time asking people what the hell spheroid and datum their lat/long coordinates were on, usually without getting an answer. Then he would be blamed for the fact that the reported thingy on the seabed could not be found. It was all most unfair.