Ken Lo wonders whether data about coverage and signal strength could be crowd-sourced to provide a more accurate mobile phone “coverage checker”.

Each of the big five networks currently provides their own coverage checker tool. Enter your postcode and you’ll be told whether you can get a signal where you live. Some coverage checkers make a prediction on the signal strength and tell you how many “bars” of signal you’ll get (O2, Orange, T-Mobile and Vodafone) whereas other coverage checkers claim to predict whether or not you’ll be able to get signal indoors (Three and Vodafone).

My experience is that there are huge problems with these coverage checkers. Pretty much every network says I’ll be able to get a 3G signal in Central London where I live. In reality when I ran tests with two of the networks, one network struggled to maintain a 3G signal (it would often revert to 2G) whereas another network would often give no signal at all. In another twist, on one network downloads over the 3G connection would sometimes be slower than downloads over a 2G connection (maybe due to network congestion or poor signal quality).

There are several problems with network-provided coverage checkers. Firstly, networks have an incentive to slightly over-state their coverage. After all, if you tell a potential customer they won’t be able to get a signal where they live, they won’t become a customer of your network. Secondly, there are a lot of variables which contribute to signal quality: whether you’re indoors or outdoors, which phone you use, what kind of building materials are used for your house, etc. Today’s coverage checkers don’t take factors such as phone and building materials into account.

Proposal: Crowd-source signal strength statistics

Metallica at Rock Werchter 2009 ♫♪
Creative Commons License photo: crsan

The problem with coverage checkers got me thinking. Wouldn’t it be great if we could have one central independent website where you could compare signal strength across different networks objectively? You could enter your postcode and the website would show you other people’s real-world experiences getting signal from each of the networks. You’d be able to do a quantitative side-by-side comparison of other people’s experiences rather than relying on predictions and computer models.

The collection of data can be crowd-sourced to smartphone users. Users would be able to install an application which would report a location using GPS and measure signal strength using APIs built in to Android. Such an application could even be configured to run in the background reporting signal strength information as users travel around the country (subject to privacy concerns, data protection, etc).

With this collection of real-world experimental data, such a coverage checker tool can finally take into account factors such as the effects of buildings on reception and the ability of different phones to maintain a signal. The effect of buildings on reception would be immediately seen in the real-world indoor coverage data: densely-populated areas with old Victorian buildings might experience poorer coverage compared to other less densely populated areas – we would see this in our data whereas we might not in computer models.

How practical is it to crowd-source coverage data?

Vodafone experimented with creating a crowd-sourced coverage map in June last year. Unfortunately it never took off – with too few people submitting coverage data, the map is next to useless for anybody who wants to find out what Vodafone reception is really like. One of the problems with the Vodafone system is that users need to manually tweet coverage information and nobody really wants to go to all of that effort.

With an application to automate the whole task of determining your location, signal strength and reporting it back to a central server, more people might be willing to contribute. As people go about their everyday lives shopping in the high street, commuting to work, etc. the application can continue to measure and report this information in the background. Each application user could generate hundreds of real-life signal strength measurements meaning that the map will have plenty of measurements.

The home/landline broadband industry is currently successfully using an application called isposure which does something similar.

Your thoughts…

I’d like to get this idea out to the community of mobile phone users, developers and networks. Do you think a crowd-sourced coverage checker is a good idea? Would it work in practice? Drop us a comment below and let us know what you think… we’d love to hear from you.

Your Comments 3 so far

We'd love to hear your thoughts and any questions you may have. So far, we've received 3 comments from readers. You can add your own comment here.

  • I have been doing some recent experimenting with an Android app called Cellumap by RadioRaiders that fits the bill perfectly. It tracks the signal strength and other detail as you move around and uploads the detail to a coverage map at http://www.cellumap.com

    The upload to the coverage map can be single-shot under the user's control, or it can be set to "Auto Send" where it will periodically send the cell info based on distance travelled – useful for tracing a journey.

    • Hey Danny,

      Thanks for sharing that – it looks like a really interesting application & website! Unfortunately there doesn't yet seem to be any data for the UK (I checked Central London for all the networks) – or maybe I was using it incorrectly. Also looks quite patchy in the US (where I looked it only had data on some of the highways). So I guess this just highlights how difficult it is to build this tool: the amount of data to build the map coupled with the fact that it's unlikely you'll get data in rural areas, etc. Seems to me that the highway data is probably fairly decent quality but a thousand data points on a highway and none at home is no good!

      Cheers

      Ken

  • Love this idea, Ken. As you say, the trick is getting it to work – which means bringing it to scale with enough users to get significant data for it to be reliable.

Leave a Reply