Thursday 22 January 2009

Regulation (even self-regulation) of advertised mobile broadband speeds will be a difficult trick to pull off

Mobile broadband service providers in the UK are set to sign up to a new voluntary code of conduct over advertised headline speeds for their services (see article in The Guardian). I'm sceptical.

It is true that up to now advertised speeds have been something of a joke. There is no way that MNOs can make reliable claims of 1.8, 3.6 or 7.2 Mbit/s when local environmental conditions (e.g. topography, distance from basestation, thickness of walls, contention) will have a massive impact on user experience. Arguably, environmental issues have a greater impact than the variety of HSPA that happens to be implemented in the local base station. In the UK even though MNOs are at different stages with the HSPA roll out (i.e. some at only 1.8Mbit/s while others provide 7.2Mbit/s), there is no consensus over which one provides the best speeds. Clearly local factors dominate in determining the user experience.

DSL providers implemented a code of practice in 2008 in the face of a threat of legislation from Ofcom. However, DSL speeds are far more predictable than mobile broadband. Any guidance from mobile operators on mobile broadband speeds will either be inaccurate or so complicated as to render them useless.

Users will have to get used to the fact that "up to" really does mean "up to". Any other definition would be a nonsense. However, MNOs do have a responsibility to ensure that they are saying "up to" and that the claimed speeds are at least possible. In 2008 Vodafone went to market with a headline rate of 7.2Mbit/s when that rate was limited to a selection of UK airports and roughly the area covered by the London congestion charge. In October the Advertising Standards Authority told them to stop it. Clearly the mobile industry needs to avoid the situation where an MNO can deploy a single LTE basestation and claim to have 43MBit/s, or even "up to 43Mbit/s". But, as the Vodafone example demonstrates, the ASA is more than capable of dealing with these kinds of issues.

6 comments:

  1. Maybe one way to get round this is for MNOs to all publish average actual data rates so that people can then compare? It's a bit similar to a regulator publishing the % of dropped calls

    ReplyDelete
  2. It's an idea, but I don't think than publishing a mean national speed will give much indication of what the actual user experience will be. Ideally I'd like to know what speeds I (as a user) would get. The only way to find that out is to trial a couple of suppliers and see who's best. So the best solution for the end user is for MNOs to offer a good opportunity to try-before-you-buy.

    ReplyDelete
  3. Thinking further on this, the logistics of measuring average speed would also be a nightmare. Do you do an average per base-station? Or pick a random selection of locations and measure speeds for all of the operators? Neither would really be that appealing. The first because it's open to operators gaming the system to ensure they have the highest average, potentially at the expense of, for instance, rural coverage. The second because the logistics involved would be painful and because it would inevitably be unfair to one operator or another.

    ReplyDelete
  4. This is easy to solve. Make available speedtest app for most popular phones. Use the phone's GPS to get coordinates, run a speed test and upload results to Google Maps. Rinse and repeat, instant user generated coverage and actual measured speed map.

    ReplyDelete
  5. Actual implementation, although it seems without the installable phone app.

    http://evdomaps.com/

    Google is your friend

    ReplyDelete
  6. I was thinking something along the lines of:

    MNO measures the total mobile BB bandwidth and divides it by the number of active users at that time

    You could do it over the day and pick the minimum number (i.e. busy hour). It's not perfect, but it's give a simple number to compare

    I'd expect the results to be south of 500kbit/s in the UK

    ReplyDelete