Dual-header optical burst switching: a new architecture for WDM burst-switched networks

Abstract

Absfract-In this paper we introduce a new signalling architecture called Dual-header Optical Burst Switching (DOBS) for next generation burst-switching optical networks. Using DOBS, the functional offset size of eveq burst on a given link can he set to the same size without the use of fiber delay Line buffers.This al1ows DOBS to realize lower burst-scheduling complehity, lower ingress delay, higher throughput and better fairness than conventional single-header OBS systems. We present a new burst5cheduling algorithm called FreeChannel Queue scheduling that requires only O(1) time to execute and that achieves optimal performance in constant-offset DOBS systems. Using simulation, we find that the blocking probability of a 16-channel DOBS system is 50% lower than that of a similar LAW-VF JET OBS system. We also show that DOBS achieves better fairness than JET OBS with respect to burst length and with respect to the residual path length of bursts.

Novel Network Architecture for Optical Burst Transport

“…In case no offset time between control header and data burst is used at all, the three basic scheduling schemes have the same performance [DGSB01, DG01,BS05b]. RFD schemes only yield superior performance over RLD schemes in the presence of voids which are generated by scheduling requests with large offset time differences.…”

“…However, some proposals queue control packets in core nodes for a short time either to schedule bursts in batch mode [CEBCS03, CBD + 03, KA05] or to serve the request queue in a different order, e.g., to account for the actual data burst arrival order [CP02,TMC03,BS05b] or class of service [YZV01,LL02]. Theoretical performance bounds and a complexity analysis of this class of approaches are presented in [LQXX04].…”

“…-(Re)scheduling of control packets: Service differentiation can be provided by not strictly scheduling bursts on a first-come first-serve basis but according to their QoS class [YZV01,BS05b]. However, this causes additional and non-deterministic processing delays for low priority burst control packets and requires more complex reservation control modules.…”

Contour-Based Priority Scheduling in Optical Burst Switched Networks

“…Therefore, after the burst is scheduled, the selected wavelength needs to be communicated between adjacent routers in order for the routers to configure the optical interconnects properly. This additional signaling overhead is similar to the overhead in split processing [3] and is less than the overhead in dual-header OBS [13] since the configuration packets are only communicated between the adjacent OBS routers and are transparent to the ingress edge router. One possible drawback of the separation of the burst information recording phase (with the original BHC arrival) and the wavelength notification phase (with the burst configuration packet) is that downstream nodes may waste some bandwidth on the bursts that are discarded at upstream nodes.…”

“…One possible drawback of the separation of the burst information recording phase (with the original BHC arrival) and the wavelength notification phase (with the burst configuration packet) is that downstream nodes may waste some bandwidth on the bursts that are discarded at upstream nodes. However, studies have shown that such effect on the overall performance is negligible [13]. One alternative signaling protocol is to hold the BHC until the burst is scheduled at ∆ 2 time units ahead of the burst arrival time and release the BHC at that time.…”