Peter Cochrane's Blog: Wi-fi London

Written on BA285 whilst flying London to San Francisco. Polished in my Santa Clara hotel and dispatched to silicon.com from an open LAN provided for visitors at a Silicon Valley company.

During the past few weeks I have completed a wi-fi survey of London in order to get some measure of the node density across a 4km-diameter area centred on Liverpool Street Station. As with my prior UK rural survey - see my earlier blog - this didn't involve any special equipment or antennas, only a standard laptop with a software wi-fi sniffer.

Each time I have travelled to London I hired a cab, sat in the back with my laptop and monitored the wi-fi signals as I travelled. As far as is possible in London, I arranged to head north, south, east and west in straight(ish) lines and collected data for a distance of approximately 2km in each direction. In addition, I mapped a few radial and haphazard routes to get a further measure of the true density.

So, what did I find? A sample scan page (with node identifiers and Mac Addresses blanked) is included below for 41 wi-fi nodes of which 14 are open and likely to be accessible for anyone wishing to use them. On this particular 2km run more than 400 nodes in offices, hotels and residences were logged.

Because of the nature of the buildings in London it is impossible to estimate with any precision just how many homes and offices were involved on any route. Suffice to say, on some roads I saw more than 10 different wi-fi signals in as many strides.

What of the results for the entire experiment? Including the transversal routings, the wi-fi density seemed more or less constant with an average of close to 27 per cent of all nodes being open access. I never found a location with open wi-fi density of less than 20 per cent, and never greater than 85 per cent. This high end seemed to be due to the hotels and coffee shops.

So just how do you estimate the total density across the entire area spanning 4km or so? Well, on the basis of no really hard information on road lengths and office/population density, I have assumed a uniform distribution of 200 wi-fi nodes per km along all the major roads moving out from Liverpool Street Station (and radially) for 2km.

According to my map, a very conservative estimate would assume 35 such roads. At a minimum the evidence gathered to date therefore suggests a minimum of 14,000 wi-fi nodes with at least 3,700 operating in an open mode.

So how accurate are my results? Not very! Given the insensitivity of my test equipment - just a laptop - the many high-rise buildings in the area I tested and the labyrinth nature of the roads and alleys, I would guess that a factor of two error is very likely. So the real numbers could be around 30,000 active nodes with around 8,000 open, or even higher.

I rest my case! For all the readers who have emailed me complaining that they cannot see or find wi-fi and have suggested I live in a different world, I think there is now sufficient evidence to say that London and the UK has quite a significant wi-fi density.