http://www.rootmetrics.com/uk/compar...-october-2013/

EE overall best performance.

O2 launched 4G service in the Glasgow area, and its speeds were comparable to the speeds found in May. Vodafone’s 4G launch, meanwhile, resulted in significantly improved speeds compared to the speeds we found during our last visit to Glasgow:

O2 came last for Mobile data, despite having 4G! beaten by Three, Vodafone and EE, average download speed on O2 half that of Three! A tiny 15k file on O2 or Vodafone with 4G takes on average 3 times longer than Three, despite O2 having 4G.

EE and Three best for calls in the Glasgow area with few failed calls. All networks did well with SMS. O2 and Vodafone the least reliable networks on the network reliability index.

And thats with very few customers on O2/Voda's 4G service...

Guys in our office with O2 4G are seeing speeds roughly the same as MBNL DC speeds. Totally pointless to pay extra for with them unless you need the extra upload. Depth isnt there yet to even justify getting it for coverage improvement either.

What confuses me is why downloading a tiny 15k file takes 3 times longer on O2 and why average speeds across the area are only 3Mb/s, making it the slowest network in the area despite having 4G! ..and they want people to pay extra for this?

Does Wavejock post here anymore? last I remembered he bought an S4? I think? it would be interesting to see what the 4G speeds are from anyone in Glasgow on any of the 4G networks.

Proxys maybe causing the issues... But the fastest speed test (and tethered downloads) in Edinburgh via O2 4G is http://www.speedtest.net/iphone/692914630.png
Many were around 15-20...

One thing I did notice was Root Metrics area covered a much larger area than O2's 4G coverage, so if average speeds are only 3Mb/s, then there must a be a lot of areas that just have GPRS surrounding Glasgow?

Most of mine were 50-55Mbps except for one site that was around the speed of yours - once I actually left my home area which supposedly has an outage on 4G already.

Latency has been pretty good though - 37ms to London, beating EE's LTE speeds.

That's always been the case, O2's 3G coverage dies outside cities, although they do now have some 3G900 along the motorway between Edinburgh and Glasgow.

Mind you - these reports often come out quite a while after the testing is actually done - OFCOM's reports usually lag the actual start of the testing period by 9-12 months for example, coming out 3-6 months after the end of the half-year run.

Remember Rootmetrics' London report said O2's service was second only to EE's (not surprising given EE's "Up to 300Mbps" 40Mhz LTE vs O2's 75Mbps 10Mhz) and faster than Vodafone's. Incidentally their London report also specifically quoted LTE speeds in addition to "blended" 3G+4G speeds:


Where's the equivalent for this article?

In fact now that I look at it, the article doesn't actually mention testing 4G speeds at all, unlike the London tests. In fact, with average "blended" speeds of 22.7Mbps in London and 13.2Mbps in Glasgow on EE, despite running the same network at the same speeds, it looks like the Glasgow test isn't including any 4G results at all. The London report also does specifically say things likee "This first look at London’s 4G services bodes well for the new providers and, more importantly, for people who chose to sign up.", again not mentioned in the Scotland reports.

The Glasgow report also says "Vodafone’s average download and upload speeds were the fastest we’ve recorded by the operator in any market we’ve tested to date. [9.5 mbps download]" yet their own London report shows Vodafone's download speed to average 20.8Mbps on 4G and 16.2Mbps on 3G+4G So they're actually contradicting their own previous releases.

Aha, just reading the methodology and they admit they didn't use a 4G phone for most of the tests. Of the four networks tested, they used an outdated, non-DC-HSPA 3G phone for three of them. Also most of the tested area is rural, and has no 4G coverage from any network. Less than 10% of the covered area actually has any 4G at all.

The edge of the "Glasgow area" they've mapped is so far outside the city of Glasgow that if you applied the same scale to London you'd be 150 miles away and have Sheffield, Cardiff and Southampton included in the "London area".

This is more of a test of south-central Scotland than Glasgow.

I know for a fact that Vodafone are using only 800Mhz spectrum to get the coverage we need, but when they start using the higher spectrum we should see speed increases.

Makes no different what 4G spectrum they're using if the testers are not using a 4G phone.

This hardly seems fair.

If they're using different kit in the cities, how can we compare like for like? Especially using a non-DC phone. For goodness sake!

I'm beginning to seriously doubt the validity of root metrics methodology. Who is behind them/pays for them? Is there an agenda here, or is it just *very* badly organised?

There was an LTE version of the S3 wasn't there? I think that was a typo though and they meant S4. That would actually make EE look worse than O2, so it doesn't at all back up the theory of some designed EE bias. Despite the fact (if it's not a typo) the only handset that has DC / LTE used for for O2 measurement, O2 STILL came last on mobile data.

The Glasgow area they used is the extended map area according to Eurostat, which includes quite a wide area of commuter areas.

Phones used were as below

When measuring performance in Glasgow, the following devices were used: Samsung Galaxy S® III (EE, Three, and Vodafone) and the Sony Xperia™ Z (O2).

I think you'd need to presume its the I9305 Galaxy (LTE version) which is DC-HSDPA.

Indeed, the phone used for O2 was also the only LTE phone listed. O2 incidentally doesn't use DC-HSPA much, and most handsets don't support inter-band aggregation anyway (3G900 + 3G2100). There was also an LTE version of the S3 but it wasn't generally available in-stores in the UK - where they claim to have got their phones.

As for it being a typo - the same devices are listed in the Edinburgh report. If it was indeed the S4 I see no reason why they should have picked a different handset just for O2 - it would have been fairer to use the same handset for all networks.


That's the thing - they title the report Glasgow, you name the thread Glasgow, yet people from 90% of the areas wouldn't dream of describing themselves as "from Glasgow" or "in Glasgow". The area far exceeds what any sensible person would describe as Glasgow (city), Greater Glasgow or even the Lanarkshire.

Yes the Eurostat area is more centered around all areas where people commute to Glasgow, which obviously is much larger than Glasgow itself. But most people associate the term "Glasgow" as referring to the city Glasgow, which this report most certainly is not. Again, the area covered is mostly rural Scotland, and I find it quite misleading to refer to it as "Glasgow".

Who cares about so much nitpicking as to the areas defined, there's a map on the report under methodology. The area they choose is an extended urban area, and in any case all operators are compared against the same area.

A lot of people care, when they see Glasgow they assume it means Glasgow. I bet you most people would be surprised when they saw the map and found actually 90% of it is not Glasgow. The data then becomes unrepresentative if not completely irrelevant to the 95% of the people actually living in Glasgow if most of the results are not from the area they are interested in.

The same thing occurred for me when I looked at the Edinburgh map and saw why the results were completely unrepresentative for me and most people I knew. Like most people in the Edinburgh area I actually live in Edinburgh, yet it appears less than 10% of their tests were actually done in Edinburgh if the distributions were as geographically fair as they claim.

Do you think people in London would consider the London test useful and relevant if it included Southampton and Cardiff and Sheffield within the "London" area?

I think it's good that they do the extended urban area as defined by Eurostat as those are all the commuter areas and surrounding towns were people who work in Glasgow live. It shows up operators like O2 that don't have coverage of 3G the minute you step out of a town or city boundary sure.

Whilst not perfect, Root Metrics does provide a good comparison of different networks within the defined area of the report. I know some will always try and nitpick, pick apart and argue things to death, but as a general indication it is one good resource of many. Several resources seem to report common trends, which indicates to me that they must be fairly accurate.

That's true but again, that's south-central Scotland, not Glasgow. The vast majority of the area's population live in Glasgow, not the surrounding potential commuter belts. According to Eurostat's data, over two thirds of the population live within the city boundaries, which is just 10% of the tested area.

The remaining 90% of the tests are only relevant to 31% of the population. Not all those people work in or commute to Glasgow either, and many commute to Edinburgh or Carlisle or Dumfries or Livingston and so on.

It would be nice for, y'know, the 69% of people who lived in the city if the Glasgow report actually just covered Glasgow.

I bet people that live in Glasgow travel the local area using their phone though and want mobile coverage on those commuter routes and surrounding towns.

As I say, you can see links between different data sources, Speedtest.net stats on average network speeds, Ofcom reports on operators breaching their obligations, opensignalmaps, broadbandgenie and others. It is generally EE and Three always coming up trumps for mobile data, and overall network performance with O2 and Vodafone lagging behind on call quality, data coverage and speeds.

You'd be surprised how few of them actually do. Many people, particularly in the deprived suburbs of Glasgow city itself have never even been to the city centre, let alone 50 miles outside of the city.

The majority of people living and working in the city spend most of their time within the city and/or their own locality. It's only a few hardcore travelers that spend any significant amount of time outside their home area or traveling.

I did spend several years working for a Glasgow transport metrics company recording and analyzing travel habits for Transport Scotland and the rail companies mind you. So I might just have a clue.

As I understand it the amount of tests in any area are based on the population density in that area, I.e. Even if the tests cover the Glasgow LUZ, they will be heavily weighted towards where people are, I.e. Within the urban areas.

In general Root Metrics use the extended urban areas to see what the outer urban coverage is like around a location as well as the city. This obviously disadvantages operators who cannot be bothered to invest in data coverage in areas outside of main towns and cities. A lot of people live in the areas and towns outside of main cities, so I think it is important that those areas are measured.

That's good if it's true. They don't mention it in their methodology though, they state:


Doesn't say anything about weighting or selecting areas according to population density.

Incidentally now I think about it they're missing a big trick with their "goodput" testing for data but not for voice. They only count a call failure if it is blocked completely or dropped completely, without accounting for the different networks' treatment of "cut outs" and timeouts for dropping a call. A common problem I hear on EE is calls going silent but not actually dropping, which counts as "good" in their results even if it's impossible to hold a conversation.

I believe they are just randomly chosen.

http://rootmetrics.com/uk/methodology/

Like I said though 90% of the area is rural. Yes, certain networks can't be bothered with rural 3G, and while it might be "nice" they cover that area it would be nice for the 69% of people that live within 10% of the area to not have to deal with the added noise of results that are not relevant to them.

The Glasgow tests actually cover areas further away from Glasgow than the City of Edinburgh is.

I'm not disagreeing that more tests = good, what I'm trying to say is it should be broken down between Glasgow/Greater Glasgow (urban) and Glasgow's surrounding commuter areas (the 90% of it that's rural).

No-one really expects the same sort of results at this guy's house than here