Torpedoed by Arcep on its speeds, Free Mobile strikes back and proves the opposite – Freebox universe

Following the publication of Arcep’s annual survey on the quality of service of operators’ mobile networks, Free questions the regulator’s test protocol and tends to prove that its speeds are much higher than advertised. .

A good progression, but a disagreement on the protocol used which is expensive in terms of speeds to the latest arrival on the mobile market in France. In his new annual survey In evaluating the quality of service of metropolitan mobile operators, Arcep is clear: Orange is once again the big winner on a multitude of indicators. For its part, Free Mobile remains far behind in terms of speeds for all technologies combined, in particular on 5G, the speed of which appears even lower than its 4G. There remain sparsely populated areas, where Xavier Niel’s operator appears to be the first alternative operator.

But then, why such differences in speed between certain telcos at a time when their networks, in particular 4G, are reaching the end of their deployment? Does Orange really offer speeds twice as fast as Free Mobile in 2G / 3G / 4G and 4 times higher in 5G as shown by Arcep in its measurement campaign conducted by its QoSi service provider? For the former troublemaker, this is very far from the case: “In real use, Free Mobile subscribers have high speeds in 4G, and even more in 5G”, he assures. To demonstrate this, the operator is now cracking a technical explanation questioning the flow test protocol chosen in this campaign.

The Arcep speed test protocol in the Free Mobile viewfinder

For him there is no doubt, “The Arcep survey shows strong differences in flow rates between operators. These disparities result from the measurement protocol. The download speeds are very different from the most used tools, Speedtest, RNCMobile and NPerf, as in real use ”. The reason for this: the measurement protocol chosen in the survey “Uses a single TCP Cubic connection (single-threaded) and measures the download time of a file before deducing the bitrate” he explains. And to add that “The measurement taken corresponds to the rate at which the congestion control algorithm self-regulates its speed on a single connection”.

Without getting too technical, Free believes that its flow rates are greatly impacted, because this 2008 protocol “is based on packet loss but can add delays and degrade performance ”. On the other hand, the TCP BBR, of 2016, which becomes the majority insures the operator, “Uses estimates of bandwidth and round trip time and approaches the theoretical optimum”.

For its part, the Free Mobile network has been optimized for use with several connections (multithreaded), “Use case of mobile subscribers”, he explains, while the regulator perceives on the contrary the “mono-connections”, namely using a single flow between the mobile and the server where the file is hosted, as the majority use on the Internet.

This is wrong, according to the operator. On the contrary, mobile applications and uses today use several TCP connections and often one or more connections for video (QUIC / UDP). For exemple, “services like Wetransfer, iCloud, Netflix or Youtube and almost all websites use between 2 and 200 TCP connections associated with 1 to 10 connections for video content ”, says Free Mobile, even allowing himself a small spade, because “Even the Arcep site” is concerned.

“Free Mobile speeds are similar or superior to those of the competition”

In other words, the protocol used by Qoosi in its survey for the regulator is not currently representative of the use cases of mobile users in France. To prove it, the Iliad subsidiary today unveils flow measurements for 4 operators carried out from November 1 to 9 by an independent company and Arcep service provider, at 50 points in Paris.

Free Mobile unveils independent study with another measurement protocol

This time, the regulator’s measurement protocol is used with servers using TCP BBR and not TCP Cubic and the measurement application opening one or more TCP connections. Result, in this configuration, “Free Mobile speeds are similar or superior to those of the competition”. Today is a necessity for Free: “the next test campaigns will have to evolve to be closer to the realities in the field”.