Studies on the Quality of Service of Broadband Access Lines 2012 and 2013
The Bundesnetzagentur conducted a study on the service quality of broadband access lines from June to December 2012.
The quality of Internet access service was evaluated by use of an integrated measuring concept that used a combination of two components:
- A measurement platform (consisting of monitoring units at 26 sites throughout Germany and several server systems which served as, among other things, counter test points for the data measurements) conducted measurements in a fully-controlled measuring environment.
- The (upload and download) data transfer rate of fixed Internet access services was however measured as part of the measurements conducted by end customers. For this, the Bundesnetzagentur invited end customers between June and December to measure the data transfer rate of their Internet connection using special web-based software. The fundamental accuracy of the values obtained using the software application was monitored on an on-going basis by randomly comparing the values generated by the two methods. The results are based on a total of 226.543 valid measurements gathered between July and December 2012.
As a first step, the study examined the parameters which decisively influence the quality of Internet access service for the end customer, namely
- the actual data transfer rate of the connection and
- the traffic management in the concentration network and core network.
The actual data transfer rate depends on the individual end customer’s Internet access connection and was therefore measured by individual end customer software measurement via the website. By contrast, the traffic management is influenced by the network design and service profile settings of the provider and requires detailed measurements. Thus, these parameters where measured by use of the measurement platform. Traffic management aspects investigated are the temporal distribution of the actual data transfer rates, transfer times and the usability of standard applications.
As a second step, the study examined the question whether the data transfer rate changes when, in the case of bundled products consisting of Internet, VoIP and IPTV, other products are used at the same time as the Internet connection. To determine this specific test measurements were conducted on the measurement platform: The data transfer rate of the Internet connection was first ascertained and then compared with the data transfer rate that can be achieved when VoIP and/or IPTV are used at the same time that data is being transmitted.
Following this, the question was addressed how end users can be put in a position where they themselves can reliably check the performance of their own broadband access. Selected technical methods for end-user measurements were compared with one another.
The results of the study are available now:
The study also examined the question whether there is evidence that data packets are handled differently on a systematic basis depending on the application, source, destination or content (question of net neutrality). The tests being conducted in this connection are still on-going. The findings from these activities will be taken account of in the work of BNetzA on net neutrality.
In 2013, the Bundesnetzagentur repeated its 2012 nationwide measuring campaign between 1 July 2013 and 31 December 2013. During this campaign users could test whether and the extent to which their actual broadband access speed deviated from the contractually agreed speed. The new campaign aimed to determine whether and the extent to which improvements had taken place over the previous year.
At total of 375,412 end-customer measurements were conducted for the 2013 study; of these measurements 153,216 valid individual measurements were included in the analysis. Only those measurements were deemed to be valid which, for example, the customer had indicated had been conducted in an optimal test environment.
The results of the statistical monitoring conducted for this study show that the sample largely approximates the parent population in terms of the parameters Technology, Provider and Geographical Distribution (federal states as well as urban, semi-urban and rural regions).
In comparison to the first study, the 2013 study had been improved so that additional values could be read out for certain subsets of routers allowing for special analyses.
It could thus be determined whether the measurements for this subset had in fact been conducted via WLAN even though the users had indicated that they had performed the measurements in an optimal test environment without WLAN. Also, it was found that there was parallel data traffic ‒ albeit only to a limit extent ‒ during all measurements conducted for the subset.
Although the results from the subgroup cannot be directly applied to the total sample, they do enable an assessment of the magnitude of the effects (use of WLAN, parallel data traffic) on the measurement results of the total sample.
The overall analysis did not include the results from the special analyses. This was done in order to ensure methodological consistency with the 2012 study and avoid statistical bias.
The results of the study are available now: