Absolute quality-of-service differentiation in optical burst switching networks

Abstract

Absolute Quality of Service (QoS) differentiation is an important issue in Optical Burst Switching (OBS) networks. This thesis aims at providing loss rate guarantees for the priority-guaranteed (PG) classes on an end-to-end (E2E) basis while keeping the loss rate of the best effort (BE) class traffic as low as possible. We first present a per-hop absolute QoS differentiation scheme called Reserve-and-Preempt Scheme (RPS). The Analytical model of RPS is provided as well. In order to support networks with multiple PG classes, RPS is extended into a variety of new schemes. Among the proposed schemes, Extended Reserve-and-Preempt Scheme (ERPS) is easier to implement and provides the best performance. On an E2E basis, we apply ERPS over an entire OBS network. To achieve a balance between scalability and unnecessary bandwidth consumption on the intermediate links, we propose optimized priority grouping with single-class constraint (OPT-SC) and optimized priority grouping without single-class constraint (OPT-NC). Furthermore, to cope with the traffic-fluctuation and the load-unbalance problems in OBS networks, we present a novel self-adaptive algorithm, which integrates two priority-level adjustment mechanisms, priority-level upgrading (PLU) and priority-level degrading (PLD). Extensive simulations are conducted to demonstrate the effectiveness of the proposed schemes. Simulation results show that our proposed RPS scheme outperforms the existing schemes. The correctness of the analytical model of RPS is demonstrated by simulation. We also observe that ERPS performs the best among the schemes extended from RPS. Simulation results also demonstrate that the proposed OPG-SC and OPG-NC techniques can reduce the bandwidth consumed by the PG classes and improve the performance of the BE class. Finally, we observe that the proposed self-adaptive algorithm can improve the performance of the PG classes with minimum expense on the BE class. At the same time, it is shown to be self-adaptive to traffic fluctuation.