Europe’s Open Internet Regulation  guarantees that internet service providers will treat all traffic equally and in a non-discriminatory fashion. In France, Arcep is responsible for monitoring compliance with the obligations resulting from this regulation, and so for analysing internet traffic management practices.
Raise consumer awareness of net neutrality, and inform the regulator of possible instances of throttling
For a year now, Arcep has been lending its support to the development of a tool called Wehe, by Northeastern University (Boston, USA). Available on the Apple Store and Google Play app stores, its purpose is to detect potential throttling of internet traffic. This tool is part of Arcep’s data-driven and crowdsourced approach to regulation, and allows every citizen to help protect net neutrality. By running their own tests, users participate in detecting possible cases of traffic throttling, and so assist Arcep with its monitoring duties. In the United States, the tool has already helped reveal throttling of several video apps by several mobile operators. Arcep teams will be presenting this tool to Internet Governance Forum participants on Monday, 12 November at 2 pm, at the Arcep stand.
A collaboration between Arcep and Northeastern University in Boston
Arcep’s teams collaborated with Dave Choffnes’s team of researchers at Northeastern University in Boston. The different stages of the work included ensuring the tool’s accuracy by reducing the number of false positives, developing a new feature for identifying the existence of Deep Packet Inspection rules, responsible for possible traffic throttling or prioritisation, hosting the tool in France (notably on a server hosted by K-Net), translating it and adapting it to the French market, developing a scorecard for tracking statistics in France, and reporting positive tests to Arcep using a model similar to the “J’alerte l’Arcep” reporting platform.
From a technical standpoint, the Wehe app records traffic to see if an operator is throttling or prioritising the traffic generated by any services. The test takes place in two stages. Wehe simulates use of the service so that an operator will treat the traffic the way it would treat real traffic coming from that service. Wehe then replays the same traffic by replacing the identified content (name of the app and servers used, etc.) with randomised bytes, which prevents the operator from identifying the traffic as coming from a specific service. Next, Wehe compares the throughput recorded during those two runs, and repeats the test to rule out the impact of bad network conditions as much as possible, to then establish whether traffic differentiation has occurred. If it proves to be the case, Wehe gives the option of running additional tests to see whether Deep Packet Inspection (DPI) rules are in place, which the operator may have installed to detect the traffic streams to be throttled or prioritised.
Download Wehe :
Statistics on the tests conducted in France can be found here: https://dd.meddle.mobi/StatsFrance.html