Why is location technology so Low-Tech? By David Bartlett, Omnisense Ltd
David Bartlett, Omnisens, explores current location technology – issuing a wake up call to the wireless industry.
First of all, let me be clear: GPS is not “low tech”.
Modern GPS receivers achieve astonishingly good performance. Considering that they are measuring the time of arrival of very weak radio signals transmitted from far away; signals that are just 2 MHz in bandwidth; getting reliable positioning of better than 5 metres is brilliant. However, GPS has its limitations: the weak signals don’t penetrate deep into buildings and the narrow signal bandwidth makes for poor tolerance to multi-path in cluttered environments. One only needs to look at findings available from the FCC to realise that the localisation service for emergency calls in the USA, E-911, as implemented now has significant issues. The calls are localised using Assisted-GPS in the first instance, and when that fails they fall back to rather crude secondary techniques. Published data suggests that about 20% -25% of calls fall back to secondary methods because GPS fails to get a fix (because the caller is indoors or similar). The FCC only requires an accuracy of 50 metres 67% of the time within 30s of placing the call, and yet there is no satisfactory fall back solution. For more information about E-911 and the challenges it faces visit http://findme911.org/resources/wireless-e-911-a-crisis-in-location-data/.
GPS is great for outdoors, but indoor localisation poses a number of different, and challenging, problems.
Multi-path is a significant issue and a tough problem to solve. When one looks at current solutions for indoor positioning, most are based on signal strength measurements: iBeacons, most Wi-Fi positioning, RFID etc. A huge amount of research effort is being poured into ever more complicated ways of using signal strength measurements for positioning:
- Finger Printing characterises the signal strength across the operating area so that measurements can be pattern matched against a database to infer the position.
- SLAM (Simultaneous Localisation and Mapping) continues to focus on using signal strength coupled to increasingly sophisticated Bayesian and particle filters.
- Inertial and dead reckoning solutions struggle to use RSSI positioning to calibrate them.
Let’s be honest: signal strength is a terrible measurement for positioning. It is true that signal strength does vary with distance travelled. However, any environmental perturbation has an impact on signal strength. Obscuration by one’s body or the object carrying the radio causes huge unpredictable changes in signal strength. Antenna radiation patterns are not isotropic and therefore orientation of the device (antenna) has an enormous impact. Even radios don’t have precisely controlled transmit powers, nor very accurate RSSI measurements.
So why are we still using RSSI? It is positively Low Tech. Using RSSI is akin to using steam engines in our motor cars, or AM radio to transmit our HD Digital TV pictures, or trying to hit a target blindfold using a bow-and-arrow.
The are far better techniques available.
GPS shows that using time and phase measurements works and is feasible; anyone who suggested that satellite signal strength should be used instead of time would be laughed out of the park. Rightfully so.
Many of the widely used local area radio signals, such as Wi-Fi, have plenty of bandwidth, 20 MHz or more (10 times that of GPS), and could give great time of arrival measurements. So why is the industry still focussed on RSSI for indoor positioning? Well there are a few possible reasons. Firstly very few commercially available radio chips can make measurements of a signal’s time of arrival. Secondly solving the positioning equations using observed time of arrival is not easy.
Frankly these are just excuses. GPS shows that using time of arrival works; now it is time the wireless industry woke up and started seriously introducing observed time of arrival and phase measurement capability into ALL radio chips. Come on, let’s see it happen: then and only then will we have the basis for true ubiquitous indoor positioning.
About The AuthorDavid Bartlett is CTO of Omnisense Ltd, and Co-Champion of the CW L-SIGYou can find him on LinkedIn: http://lnkd.in/djKWTfDFollow Omnisense on twitter: @iSenseOmni
Keep the conversation going on Twitter: @CambWireless
Want to find out more about Cambridge Wireless? Visit http://www.cambridgewireless.co.uk/