Mae'r cynnwys hwn ar gael yn Saesneg yn unig.

UK fixed-line broadband performance, May 2011 - The performance of fixed-line broadband delivered to UK residential consumers

27 Gorffennaf 2011

Executive Summary

Ofcom's work on broadband speeds

1.1 The UK broadband market has been transformed in the period since Ofcom first researched fixed-line broadband speeds in 2008 and continues to change apace. The deployment and take-up of faster broadband services, the increasing popularity of mobile broadband and the greater use of internet applications and services using more bandwidth have all contributed to this transformation. Ofcom has taken a number of steps to ensure the market remains dynamic and competitive:

  • Publishing this research bi-annually allowing consumers to see how broadband speeds vary across different providers, technologies and time periods and supplementing this with similar research on mobile broadband performance.
  • Introducing and updating the voluntary Code of Practice on broadband speeds which ensures that consumers get clear information on the speeds available on their particular line.
  • Advising consumers through new guides on how to improve their broadband speeds and get the best deal on their broadband service.
  • Creating a UK broadband map which allows consumers to see at a glance how their region compares across a range of metrics with other areas.
  • Ensuring a clear regulatory framework for the deployment of superfast broadband infrastructure.

1.2 The UK fixed broadband market currently stands on the cusp between current generation broadband services and next generation services which offer much faster speeds. We estimate that by July 2011 superfast broadband (i.e. broadband with a headline speed of 30Mbit/s or higher) was available to around 57% of UK homes and continuing investment from providers means that availability continues to grow week-by-week. As this change progresses, Ofcom will continue to act to ensure that consumers have the information they need to make choices about the right broadband service for them.

UK broadband speeds increasing in 2011

1.3 The results of our latest research conducted in partnership with SamKnows show that in May 2011 average actual broadband speeds in the UK were 6.8Mbit/s, a 0.6Mbit/s (10%) increase since our last round of testing in November/December 2010.

1.4 This increase in average speeds is the result of consumers increasingly moving to faster broadband services: 47% of UK broadband connections had a headline speed above 10Mbit/s in May 2011, compared to 42% in November 2010, 24% in May 2010 and 8% in April 2009.

Speeds for ADSL services varied widely and were typically much lower than advertised 'up to' speeds

1.5 Over 75% of current UK residential broadband connections in the UK are delivered via ADSL, using the copper telephone line between the local telephone exchange and the customer's premises. A characteristic of ADSL broadband is that performance degrades due to signal loss over the length of the telephone line. This means that the speeds available to different customers vary significantly, with those with shorter line lengths (i.e. who live closer to the exchange) typically able to achieve higher speeds than those with longer line lengths.

1.6 Therefore, headline 'up to' speeds have little meaning for services delivered via ADSL. We found that the average download speed received for 'up to' 20Mbit/s or 24Mbit/s ADSL packages was 6.6Mbit/s, and 37% of customers had average speeds of 4Mbit/s or less.

1.7 It is therefore important that customers have information about the speed their line is actually capable of in order to inform their decision about which broadband service is right for them. In 2008 Ofcom and most of the UK's leading ISPs (collectively representing over 90% market share of residential connections) agreed a voluntary Code of Practice which requires signatories to provide an estimated line speed before a sale is completed; an update to this Code comes into force on 27 July 2011 which includes provision for customers to exit their contracts if actual speeds are significantly below the estimate provided. In addition, some ADSL providers have moved away from advertising broadband on the basis of 'up to' speeds and now advertise their service on the basis that they will deliver the fastest speed that the line can support, promoting the fact that they will provide a line speed estimate to prospective customers.

1.8 Because of the wide range in performance, advertising guidelines are currently in place which restrict how theoretical speeds may be used in advertising, requiring for example that speeds are prefixed with 'up to' and that caveats are included which detail that actual performance varies with location. However, theoretical 'up to' speeds continue to be widely used in advertising and we consider that these have the potential to mislead customers. The CAP/BCAP committees of the Advertising Standards Agency are currently concluding a review into the use of 'up to' in advertising and are expected to announce their conclusions in late summer; Ofcom's public response to the consultation recommended that when speed is used in broadband advertising it should be based on a 'typical speed range' (TSR), which should have at least equal prominence to an 'up to' speed.

Cable and fibre services were faster than ADSL and much closer to advertised speeds

1.9 Virgin Media's 'up to' 10Mbit/s cable service delivered average speeds of 9.5Mbit/s, 95% of advertised 'up to' speeds and significantly faster than the average speeds delivered by ADSL services with headline speeds of 'up to' 20Mbit/s or 24Mbit/s (Figure 1.1). Virgin Media's 'up to' 20Mbit/s cable service averaged 18.2Mbit/s (91% of advertised speeds), its 'up to' 30Mbit/s service 31.0Mbit/s (103% of the 'up to' speed) and its 'up to' 50Mbit/s service 48.4Mbit/s (97% of the advertised 'up to' speed).

1.10 We found that download speeds for BT's 'up to' 40Mbit/s FTTC service (which is currently available to around a fifth of UK homes) averaged 33.8Mbit/s (85% of the advertised 'up to' 40Mbit/s speed).

Download speeds fell during peak times but by more for some ISPs than others

1.11 Download speeds can fall during peak periods as a result of capacity constraints on ISPs' networks (caused by simultaneous users sharing the same bandwidth). We found that in May 2011, speeds measured during the peak weekday hours of 8pm to 10pm (which on average were the hours in the week when speeds were slowest) were 90% of the maximum speeds delivered (typically during 'off-peak' hours such as 12am to 6am) and 98% of the average speeds delivered over a 24-hour period.

1.12 Among the 20Mbit/s and 24Mbit/s ADSL packages, O2/Be had the lowest average slowdown during peak periods, with speeds in the period 8pm to 10pm averaging 95% of the maximum speed. Karoo (the ISP of Kingston Communications, the incumbent in Kingston-upon-Hull) had the greatest slowdown, with download speeds in the peak period averaging 78% of maximum speeds.

1.13 There was little difference in levels of slowdown among the super-fast ISP packages covered in the research, although Virgin Media's 'up to' 30Mbit/s suffered the least slowdown, with speeds in the peak period being 96% of maximum speeds.

Metrics other than download speed also determine broadband performance

1.14 Download speed is typically the most important single metric in determining broadband performance and, along with price, continues to be the most important metric in the advertising of broadband. However, there are of course many other metrics that determine the overall performance of a broadband connection. As it is not always the case that the technologies and providers that deliver the best performance on download speeds always deliver the best performance on other metrics it is important that consumers understand the impact other metrics have on performance and choose the service that best meets their needs.

1.15 Upload speeds are particularly important for users looking to share large files, use real-time two-way video communications and for some games. As use of such services has increased, so too has the focus on upload speeds, which has become more of a source of differentiation in recent months, with BT's FTTC service offering advertised upload speeds of 'up to' 10Mbit/s (and 'up to' 2Mbit/s), and Virgin Media has increased the upload speeds associated with its cable services. Our results showed that BT's 'up to' 40Mbit/s download and 'up to' 10Mbit/s upload FTTC service delivered average upload speeds of 8.8Mbit/s, significantly higher than any other service we measured. Virgin Media's 'up to' 50Mbit/s delivered average upload speeds of 4.1Mbit/s, with its 'up to' 30Mbit/s package averaging upload speeds of 2.7Mbit/s and its 'up to' 20Mbit/s package averaging upload speeds of just over 1.5Mbit/s. Among the ADSL services, O2's 'up to' 20Mbit/s package was significantly faster than the others with average upload speeds of just over 1Mbit/s.

1.16 Our research included measurements of the time it took to download an Ofcom test web page which was hosted on SamKnows' servers. In general, packages with faster download speeds were also quicker at downloading web pages Virgin Media's cable services and the BT's FTTC service all delivered pages faster than any of the ADSL services, indicating that even for basic web internet functions there are benefits to higher speed services.

1.17 In this report we also look at how variations in various broadband connection metrics affect the end-user quality of experience (QoE) when undertaking a number of online activities. This shows that:

  • For video streaming services, low data rates or high latency or packet loss lead to viewing 'glitches' that render the service annoying or unusable, although typical levels of jitter do not have an effect on these services as they are accommodated within data buffers.
  • With download services and interactive websites the main observable effect with a decreased data rate or higher levels of latency and/or packet loss is that it takes longer to complete a task (again, typical jitter levels do not affect the QoE).
  • With VoIP, performance varies gradually as data rates fall or latency and/or packet loss increases, however, although there are noticeable effects on sound quality, speech remains intelligible and typical jitter levels do not affect the QoE; and
  • Massively Multiplayer Online Games (MMOG)s are sensitive to even the lowest levels of latency and packet loss, with users noticing reduced responsiveness and increases in jitter having a detrimental effect on the QoE.

Conclusion and next steps

1.18 This research report is a representative snapshot of the state of residential broadband performance in May 2011, and we have noted the limitations of the research. The broadband market is changing rapidly as operators are continuing to invest in their networks in order to make faster broadband available. Therefore, the results set out in this report will not necessarily reflect the future performance of networks and providers. Our research programme continues, and we expect to publish our next report in early 2012.

1.19 The results of our research show continuing improvements in average broadband speeds. We also note that the highest speed cable services and fibre-to-the-cabinet services are consistently delivering speeds that are sufficient for virtually all applications likely to be used by residential broadband consumers.

1.20 Nevertheless, most broadband is still delivered via ADSL, over copper lines which were originally designed for phone services and which have been stretched to the edge of their capability in order to provide broadband. An inescapable characteristic of ADSL broadband is that performance is constrained by the length and quality of the copper line.

1.21 It is therefore essential that consumers have accurate information on the speeds available on their line when they make their choice of broadband supplier. An update to the voluntary Code of Practice on Broadband Speeds comes into force on 27 July 2011 which requires that ISPs provide estimates to a common methodology in the form of a range, and gives consumers the right to exit the contract if actual speeds are significantly lower than the range. Most of the UK's largest ISPs are signatories to the Code, and a list is available on the Ofcom website.

1.22 In addition, in our response to the Committee for Advertising Practice (CAP) and Broadcast Committee for Advertising Practice (BCAP)'s public consultation into the use of speed claims in advertising (which closed on 25 February 2011) we recommended the following:

  • if speed is used in advertising it must include a 'Typical Speed Range' (TSR), which should be based on average actual speeds that the 25th to 75th percentile of customers receive (i.e. the inter-quartile range);
  • this TSR must have at least equal prominence to any 'up to' claims made;
  • that if an 'up to' speed is used it must represent the actual speed that a materially significant proportion of customers are capable of receiving; and
  • that any TSR or 'up to' speed used must be based on statistically robust analysis.

1.23 Figure 1.1 shows Ofcom's estimates of the TSRs by technology and headline speed, based on data collected in May 2011.

Figure 1.1 Estimated typical speed ranges (25th to 75th percentile), by technology and headline speed
Current packages Typical speed range
ADSL 'up to' 8Mbit/s 1Mbit/s to 5Mbit/s
ADSL 'up to' 20/24Mbit/s 3Mbit/s to 10Mbit/s
Cable 'up to' 10Mbit/s 9Mbit/s to 10Mbit/s
Cable 'up to' 20Mbit/s 18Mbit/s to 19Mbit/s
Cable 'up to' 30Mbit/s 31Mbit/s to 32Mbit/s
Cable 'up to' 50Mbit/s 49Mbit/s to 50Mbit/s
FTTC 'up to' 40Mbit/s 32Mbit/s to 37Mbit/s


Source: Ofcom based on SamKnows measurement data for all panel members with a connection in May 2011

1.24 Although average speeds are improving, our research, and the data we have collected from operators into line speeds across the UK , finds that for many consumers the speeds available to them via ADSL are not sufficient for a high-quality experience of high-bandwidth services such as internet TV, or for connecting multiple devices to the internet. The continuing investment by BT, Virgin Media and others into next-generation broadband is therefore essential to ensuring that UK consumers can participate fully in the communications and entertainment services enabled by higher speed broadband.