Mike Smuts: July 2008 Archives

Mobile phone tracking seems to be very popular at the moment. There has for some time been a form of tracking known as triangulation for mobile phones.  This is, in my opinion, one of the most over sold products around today.  It is always sold with a picture of a map and a precise position on a street, giving the impression that it is pin-point accurate. In most industries this would be instantly stopped as it is (and I am being kind here) untrue. Well actually there are places you can get an accurate location but they are very rare, it is a bit like saying that a broken clock is telling the right time twice a day.

The reason I am fixated by this today is that we have several projects underway aimed at the safety of the general public.  These always involve a conversation about accessibility (everyone needs to be able t access the systems) and then the old chestnut pops up "but you can track mobile phones?".

So what does triangulation do? Simply put it measures the relative strength of your mobile phone signal (the one your phone gives off while looking for the nearest mast). If you know the position of the mobile phone masts, and you know how strong the signal starts out from the mast, and how long it takes to fizzle out over distance, you can work out roughly how far away from the mast the phone is.  Get two signals and relative strengths and you can draw a triangle and work out the position - sort of.  The key here is that the measurements are approximate and the further you are from a mast the more approximate they are.  So instead of a pin on a map saying you are on "Main Street" you really get a note saying you are within, say, 2.5 miles of Main Street.  Quite a different result!

We have a video on the web site that shows this in action. But as a subject it is like an urban myth.

The words "Continuous tracking" came into sharp focus again today.  We talk about continuous as an interval of 5-12 seconds between positions. For some reason I am always asked if I have that right. For me this is a bit like asking someone if they have spelt their own name correctly. We work with the smallest interval we can (given the processing the tracking units have to do), all based on the fact that when you track humans rather than vehicles they move in any direction they like and there is no point trying to find someone based on their position 1 or 5 minutes ago.

This philosophy is borne of experience.  Even with a 5-12 second interval, when our duty officer guides someone to find a missing person, we need to predict where they will go next and "head them off".  If this is hard with a 5 second interval imagine what it might be like with 1 minute or longer between known positions!