Location-based services on mobile devices used to be the preserve of app-happy shoppers and geo-social networking. Now, their usefulness is being explored for more serious – even mission critical – situations.
On the mobile phone as a 'platform', location is everywhere. It may not be king, but pretty much any app that aims to provide a service now includes a location-based element, whether it is to help you find a nearby restaurant or minicab, or simply to push local promotions your way. The space-satellite-based Global Positioning System (GPS) is the key enabler. Even low-end smartphones now include not just a GPS chip, but can also locate themselves by scanning for local mobile basestations and Wi-Fi access points.
In serious business though, location's use has been limited. It is big in logistics; virtually all buses and most commercial vehicles (such as vans and lorries) will now have GPS locators fitted to help with service-tracking and route optimisation. But there is a whole group of applications where location information could make a huge difference to the good, but where GPS is in practice either too inaccurate, too slow, or simply not available.
This was illustrated in 2007, when four Warwickshire firefighters died in a large smoke- and flame-filled vegetable packing factory. This tragedy occurred not because the burning building collapsed, but because the firefighters ran out of oxygen before they could find their way out of the stricken premises. It is unlikely that a satnav or smartphone would have helped save them – they were inside a steel-framed and clad building where 3G signals would have been weak, and GPS contact non-existent.
Even when smartphone users are outdoors, and when lives are not immediately at risk, GPS can prove inadequate. For example, the architectural concrete canyons found in dense urban environments can cause problems, with the receiver unable to 'see' enough of the sky to lock on to enough GPS satellites. In built-up areas where a lock-on is possible, the situation is better now since the US turned off the deliberate fuzzying of civilian GPS in 2000, but your position fix can still be tens of metres out.
Two overlapping areas of research are now coming to the rescue. The first is alternative location technologies, ranging from counting how many steps you have taken to measuring subtle shifts in the geomagnetic field. The second is the use of sophisticated algorithms to merge location data from multiple sources, thereby using different location technologies to correct each other's errors.
Schemes like these are now widely used. For example, a car satnav system will constrain its options by assuming that you are on the closest road and following your programmed route, not in the middle of the field next to it. A smartphone will take data from the network to help lock onto satellites (assisted GPS, which can in many cases much improve the start-up performance, or time-to-first-fix (TTFF), of a GPS-based positioning system), and can also try to locate itself according to which cell basestations and Wi-Fi access points it can see. The nature of radio waves still means, however, that this may not be accurate enough to avoid accidentally shooting your own comrades on the battlefield, to navigate through a darkened building, or to report the location of an emergency by mobile phone.
"With fixed-line phones we can get the location from BT, but it is more challenging with IP phones and mobiles," explained Olaf Baars, chief fire officer at Royal Berkshire Fire and Rescue Service, speaking at the October 2013 Cambridge Wireless event 'Locations and Beyond 2013: Knowing where you will be tomorrow (means so much more)', which was sponsored by CSR, and produced in partnership with Silicon South West and the IET. "Even when mobile phones have GPS, there are GPS-denied environments – complex structures such as shopping malls and airports, or buildings with multiple basements. Structures often attenuate wireless too, for example due to the use of metal-backed boards in building."
Even when lives are not at stake, "we spend 80 per cent of our time inside buildings, so we will probably want to enable a lot of location-based services for indoor use", says Hamid Ahmadi, VP/chief innovation officer at wireless semiconductor specialist CSR (formerly Cambridge Silicon Radio), adding that there are several challenges here. For instance, access points are designed for communication, and not for location, so it is important to ask yourself how much dedicated positioning infrastructure you are willing to build, and whether you really need position or merely proximity. Ahmadi adds: "Sometimes, just knowing you're in the building, or on the second floor, is enough".
The other issue is maps. "Google and others spend a lot of money on maps outdoors, but indoors is different," he continues. Sometimes, drawings or maps exist in some form – Google has done some indoor mapping, for instance, "but how will you bring them into your infrastructure? Who pays for them? Crowdsourcing is possible, but it means parsing a lot of unstructured data – the problem with crowd-sourcing is you get more noise than data – and again there are privacy issues".
These are early days, but it seems obvious that where LBS data services exist to serve public interiors – shopping malls, say, or large visitor attractions – that that information should be available for other parties, such as emergency services, should the need arise; and that that requirement be built into the software's specification from the outset.
Consumer-grade location has all historically been based on GPS or cell signal strength, "and that can be rather inaccurate in urban settings", says Andy Thurman, the CEO of mesh network and location technology developer Omnisense. "If you give permission, Google then tracks you past every Wi-Fi access point and maps you from that, it's pretty good – but not quite as real-time as what we're doing. Performance is adequate for most consumer services though, where the bigger aim is to make it seamless to the user."
Other places where accurate location data could save lives include in mines and similar subterranean environments, and aboard ocean-going vessels. Thurman says: "At sea, it's tracking your crew in the engine room of a cruise ship, say, so you know everyone's evacuated before you pump in gas to kill a fire. In mining it's mainly for collision avoidance between vehicles and people. The signals can only propagate through the tunnels – they can't go through rock – but that means the bounce paths can't go miles off track. That's where you need multiple hops – a mesh network – in order to get the data out." The bounce paths are what Omnisense's wireless beacons use to communicate and locate each other, he notes.
A further option is opportunistic signals – spatially variable metrics that can be repurposed, says Dr Rob Harle, a researcher in the Computer Laboratory at University of Cambridge. As well as Wi-Fi and Bluetooth, light, heat, and the local magnetic field can all be highly repeatable. The challenge is that they require laborious surveying, plus they can change over time.
Dr Harle is therefore working on ways to combine opportunistic signals and the 'mobile phone's 'compass' with pedestrian dead-reckoning – basically, using the phone's accelerometer to count your steps. The problem is that dead reckoning accumulates errors over time, so he applies algorithms derived from robotics to correct these errors. In effect, each subsequent walk – around an office, a supermarket, or wherever – can recalibrate the basic Wi-Fi map.
Wheres and wherefores
In Transport for London's bus-tracking scheme, lateral thinking led to another opportunistic signal: bus drivers are only supposed to open the doors at a bus stop, and the Transport for London control system knows where its 19,000 bus stops are, so it can use the doors opening to narrow down a bus's location.
"The system transmits the vehicle location'to base every 30 seconds, it also transmits every time the doors open," says Simon Reed, head of Transport for London's Technical Services Group. "There is a thing now about how much more information we can get off. For example, the weight of the bus is a proxy for the number of people onboard, we also get the number of Oyster cards scanned."
Oyster is Transport for London's contactless smartcard ticketing system, introduced in July 2003. More than 43 million Oyster Cards have been issued. It'now accounts for more than 80 per cent of all public transport journeys that take place in the capital.
Of course, what you do with the location data is just as important as how you gather it. "A consumer device may know where you are – that's useful for pushing services to you – but in our applications it's more situational awareness and command and control, so there is a central repository," explains Omnisense's Thurman.
"I cannot stress enough that without an accurate data supply, we would be finished," agrees Reed. "The ability to accurately know a service is coming is very important, it's also an incentive to the operators to run according to their contract – the operators get the same information we have, they can see their vehicles and talk directly to the drivers, for example to speed up or slow down the vehicle to maintain service frequency."
The introduction of visual and audible next-stop announcements across its bus fleet in 2009 was a big location-based technology win for Transport for London. "Making that ubiquitous has helped greatly," says Reed.
"A lot of the stuff we do is for service control," he adds, "but we also use the same data to performance-manage the network and inform passengers – as well as the countdown signs on bus stops, it's essential for the audio location signal which has made the bus network so much easier to use."
Reed adds that Transport for London also publishes more than 50 data sources, typically by transport medium, all of them free for anyone to use and analyse. "The last time we looked, there were 60 apps using our bus data alone, and over 190 apps doing things for passengers," he reveals. "Some apps merge our data sources to improve your travel experience and planning – we would not merge sources and do syntheses ourselves due to budget constraints."
And all of this is based on the slimmest of data feeds. "When the system was put together in 2006 data was expensive, so the vehicle only transmits 52 bytes every 30 seconds, although we also collect more data as it passes certain Wi-Fi points around London," Reed explains.
"Technology has to have a business use. Sure, we could livestream video from the bus, say... but that's not useful. However, 52 bytes every 30 seconds – now that is useful. Then it's what you can do with the data and give back to people, and what they can do with it after that."