Transport layer and congestion control in the future Internet: from control to anarchy

The transport protocol layer needs to evolve. Although we have witnessed these last years of the birth of new congestion controls, where some of them have been rapidely deployed, there still exists no universal transport protocol solution. Several studies agree that the transport layer must be changed and new challenges involved by new networking technologies and usage overtake the original congestion collapse problem. Even the OSI layer is now discussed and some researchers propose to re-slice this model while others emphasize the obsolescence of the congestion collapse problem or consider fairness as a criterion unrealistic and impractical. In this habilitation thesis, we propose to explore several transport protocol solutions, mainly for the best-effort area, ranging from QoS networks to the anarchical networks. We aim at investigating approaches to change the transport layer without having to convince operating system designers or to deploy core network mechanisms.