The Ebtic senior researcher Dr Nayef Al Sindi at the Khalifa University campus in Abu Dhabi. Dr Al Sindi and his team are working on an indoor locating system. Silvia Razgova / The National
The Ebtic senior researcher Dr Nayef Al Sindi at the Khalifa University campus in Abu Dhabi. Dr Al Sindi and his team are working on an indoor locating system. Silvia Razgova / The National
The Ebtic senior researcher Dr Nayef Al Sindi at the Khalifa University campus in Abu Dhabi. Dr Al Sindi and his team are working on an indoor locating system. Silvia Razgova / The National
The Ebtic senior researcher Dr Nayef Al Sindi at the Khalifa University campus in Abu Dhabi. Dr Al Sindi and his team are working on an indoor locating system. Silvia Razgova / The National

GPS-style tracking system may help those lost indoors


  • English
  • Arabic

ABU DHABI // Imagine if a firefighter rushing to save victims from a burning building did not have to paw blindly through the smoke to find them; if he could save critical minutes by checking a tracking system to find exactly where they were.
Or imagine if a visitor overwhelmed by the Dubai Mall could find himself on a map on his mobile and get directions to the store he wanted.
An effective indoor-positioning system (IPS) would offer so many important and profitable uses that engineers worldwide are competing to come up with the best ones.
Now researchers at the Etisalat BT Innovation Centre (Ebtic) in Abu Dhabi are preparing to file a patent for what they say will be one of the strongest techniques.
"The value of indoor positioning, or indoor localisation, is really an untapped market and it has been limited by the technology," says Dr Nayef Al Sindi, a Bahraini senior researcher on the project.
Unlike GPS, or global positioning systems, IPS cannot rely on satellites because their signals cannot penetrate walls.
Effective IPS must cover sweeping indoor spaces - airports, malls, conference centres and hospitals - quickly, affordably and accurately, ideally to the nearest metre.
The Ebtic team proposes to do that by using, as other IPS systems do, the Wi-Fi internet routers in most buildings. This gives them an immediate, low-cost and comprehensive network of "access points".
Next they will identify "location fingerprints" based on the signals picked up by individual devices, such as mobile phones and laptops.
Many systems are based on the "received signal strength" (RSS), or the total power of the signals arriving at a device from the various access points, be it by a direct route or "multi-path", having bounced off walls and other objects before arriving weakened and delayed.
The strength of the signals received by the phone is then cross-checked with a detailed map of signal strengths across the area covered. The point with the fingerprint most closely matching what the device is picking up is identified as the spot where the person is standing.
But different locations can have the same total power and the RSS can be distorted if people or objects are in the way. It may appear weaker and incorrectly indicate a longer distance between the router and the phone.
To avoid that error, Ebtic engineers have devised a way of getting more specific information from the multi-path data, using the strength of individual signals received over time.
Unlike RSS, multi-path differs for each location depending on its distance from access points and the layout of obstacles.
"The traditional technique just adds all these signals and you get one number, the power," says Dr Al Sindi.
But using the multi-path data reveals the history of the rays as they arrive at the receivers.
"It can tell you so much about where you are," Dr Al Sindi says. "If I'm sending a signal to you now, one is bouncing off the table and another is bouncing off the wall, et cetera.
"But if you move to the corner, the way the signals bounce and reach you is very different."
Plotted over time, the multi-path looks like a crooked line - a little like the jagged teeth of a key.
A phone 10 metres from a router could receive many rays almost immediately, while others bouncing off walls would arrive very slightly later. Its multi-path line would spike quickly and then taper.
With objects in the way, the rays would be received more slowly. In that case, the multi-path line would have no initial peak but a range of "mountains" of similar height.
And while the RSS measurement looks only at the total signal received - in effect, the area under the line on the graph - the multi-path method looks at the exact shape of the line.
The next hurdle for the Ebtic team is to fine-tune the algorithm they use to calculate locations. They will test and tweak their technique over the next year before building a prototype.
Among the factors to consider are how many access points, pre-mapped location fingerprints and multi-path histories they will need.
Tied to these decisions is the question of how to account for interference. All of the people walking around a mall, for example, will disrupt the path of the Wi-Fi signals.
One option may be to have the system collect a large number of multi-path readings, then throw out the minority of those that look abnormal.
This raises a separate problem: the more data is collected the more accurate the fingerprint, but also the more processing power and time needed to calculate the location.
This can create time lags that place people behind their actual location or drain their smartphone batteries faster.
The Ebtic team say they have figured out how to gather more accurate information and store and compute it faster. They do this by converting the multi-path data into quantities according to an index.
"We have found a way around dealing with a massive database, basically. And also providing more robust fingerprints," Dr Al Sindi says, declining to share specifics until the patent is approved.
Over the coming months they will create a trial system at the Khalifa University campus in Sharjah, with the help of an Emirati graduate student, Nuha Al Khambashi.
The goal, Ms Al Khambashi says, is to "try to develop a new algorithm for localisation and deploy it here in the UAE".
Collecting more specific data is a valuable step for IPS, says Dr Muhieddin Amer, an electrical engineer at the Rochester Institute of Technology Dubai.
"It could be a good approach . to extract more information from the received signals," Dr Amer says of the Ebtic project.
Different types of signals used in other IPSs offer advantages but also trade-offs. Ultrasound gives a stronger signal but is more prone to interference. Ultra Wideband, which emits shorts pulses at multiple frequencies, provides a lot of signals that are easy to differentiate. Hybrid IPSs draw data from more than one type of signal.
But these alternatives may require costly new equipment or installation. A major benefit of wireless is the access points are already there.
"These nodes are everywhere," he says. "Why not use them?"
 
chuang@thenational.ae