We all do better when we work together?

This paper evaluates the impact of a RD signal on traffic crossing the mobile network through a set of experiments in a simulated 5G network scenario built with ns-3. In our experiments, we compare a scenario with no RD treatment (i.e. a single best–effort EPS bearer) with scenarios with RD treatment (i.e. separate EPS bearers to carry RD–partitioned traffic) with honest and cheating users. Our objective is twofold. On the one hand, we want to explore the suitability of RD as a way to harmonise the LTE and Internet QoS models and on the other hand, we aim at providing data to inform protocol design as well as operations-related discussion on the notion of exposing a 1-bit, clear-text rd signal from endpoints to the network path when the use of end-to-end encrypted protocols would otherwise prevent flow classification based on DPI. Specifically, the question we want help answer is whether the gain in terms of end users' QoE and radio spectrum efficiency is enough to justify making room for such signal. All the experiments are fully documented and the tooling used is made available as open-source to ensure their reproducibility.