Improve fairness with adaptive RIO for assured service in DiffServ networks

RIO is the primary queue management mechanism used in DiffServ networks. However, it is analytically proved that it is necessary to adaptively adjust the maximum dropping probability of OUT packets to ensure per-flow fair share of excess bandwidth. in this paper, we propose and implement an enhanced version of RIO, called Adaptive RIO (ARIO), to improve the fairness among different TCP flows. Simulation results show that ARIO can effectively improve per-flow fairness of RIO. It is also observed that ARIO can also improve the throughput utilization of the bottleneck link in different communication environments.