Daiwa EuroTelcoblog No. 47: Tuesday, 15th June, 2004 - Measuring Wi-Fi

Way back in February 2003 we devoted a section of our sector piece to the potential of the free networks "movement" to create a patchwork of Wi-Fi hotspots dense enough to be meshed together, in effect forming an ad hoc alternative local loop in certain areas. As we have documented before and since, the efforts of entrepreneurial outfits like (www.locustworld.com) and (www.4g-systems.de) have been focused on this issue, with some considerable strides made to date. Assessing just how dense the "network" is, of course, highly problematic, in light of the fact that the various databases of commercial hotspots and open nodes have traditionally had little overlap, and the market is also growing very rapidly. Efforts by UK free networking forum Consume, documented in the excellent report by Julian Priest highlighted here a couple of weeks back, appeared to show that the actual deployment of Wi-Fi nodes per square kilometer in Greater London is on the order of 15.5 (or an estimated 19,451 nodes within the M-25 orbital motorway), much higher than the level calculated by the group as the critical threshold for a viable meshed network (1.25 per square kilometer). Of the 1,525 nodes actually detected by aerial surveying, 45% displayed SSIDs (identifiers seen by Wi-Fi users seeking to access a certain node) classified as "default". In other words, it is highly likely that as many as 45% of nodes surveyed were simply plugged in with no additional configuration, probably including security settings. The result, as observed by Julian Priest, is that a large proportion of nodes, whether by design or through ignorance, are in effect "open nodes" freely accessible to other Wi-Fi users.

We are now halfway through the fourth annual WorldWide War Drive, organized by WiGLE.net (Wireless Geographic Logging Engine) and running from 12th June to 19th June. In this global effort, contributors from around the world upload location information on Wi-Fi nodes logged during reconnaisance drives (WAR drives). The stats are being compiled in real time on the site (https://wigle.net/gps/gps/GPSDB/stats/), and as at this writing, the total number of nodes documented worldwide is 1.32m (and has increased by 2,000 in the past hour that I have been monitoring the site). This is interesting, as the number cited in Julian Priest's report (date-stamped 31st March, 2004) was 778k, and other posts on the WiGLE site show only 900k as recently as 22nd April. In other words, the number of Wi-Fi nodes catalogued by the initiative appears to have grown by 70% in under three months, and by around 47% in just seven weeks.

What is interesting is that the Consume survey showed a number of nodes in Greater London more than four times higher than that included in the WiGLE database at the time of the survey, which suggests that even this rapidly growing catalogue of nodes is vastly under-reporting the number of hotspots deployed worldwide. This impression is reinforced by the European map compiled by the initiative (https://wigle.net/images/europe.png) which shows an apparently hugely disproportionate number of nodes in the Benelux as compared to other regions (Italy and Spain appear to be mostly uncharted territory). We think this may reflect higher adoption in the Benelux, but perhaps also a greater level of geek activist involvement in node cataloguing. A couple of other startling facts of interest are found on the site. Firstly, the percentage of nodes displaying default SSIDs is 31.8% of the total, lower than in the Consume survey, but still remarkably high. More interestingly and perhaps alarmingly, the ratio of nodes without WEP (the default security protocol for 802.11 networks) switched on is an amazing 50.8%, and the three years of data displayed in graphic format shows that the proportion of WEP to non-WEP nodes is essentially unchanged over the three years.