Two adaptive AQM algorithms for quantitative Differentiated Services

The DiffServ Assured Forwarding (AF) service provides a scalable solution to QoS guarantee. Currently, AF only provides qualitative differentiation of delay or loss between classes of service, but not quantitative guarantee. By studying the quantitative behaviour of the steady state operating point for RIO, we propose two adaptive RIO algorithms for per-hop QoS provisioning in AF service. These two algorithms, ARIO-D and ARIO-L, work by dynamically adjusting the RIO parameters so that the packet delay and the packet loss are, respectively, kept at their target level. Simulation results show that they can provide both high link utilization as well as stable and differentiated delay or loss for different AF classes on a single router.