光分组在核心节点处的冲突解决问题是实现全光组播分组交换的关键。本文提出了一种新型的解决光组播冲突的节点结构,用于冲突解决的部分包括输出共享的网络编码模块和光纤延迟线(FDL)环形反馈共享缓存(FDL-LSFB)模块。网络编码模块将异或网络编码作为冲突解决方式,利用全光异或门将冲突组播进行网络编码并且改变编码后分组波长,从而避免波长冲突。而FDL-LSFB模块由子交换矩阵和FDL缓存组连接成环状,且冲突组播可从任意子交换矩阵进出缓存模块,使用少数的FDL可以提供大容量的光缓存、提高FDL利用率。针对FDL-LSFB模块冲突组播调度问题,为减小缓存时延,设计了最小缓存长度级联控制算法(MLCBST),进而提出冲突光组播的总调度策略。仿真结果表明,本文提出的组播节点结构和调度策略在降低丢包率(PLP)和减小缓存时延方面具有明显的效果。
Optical packets' contention resolution in core node is a key to realizing the all-optical multicast packet switching. A novel node architecture is proposed to resolve contention for opticaI multicast in this paper. In the proposed architecture, output shared network coding module and Fiber Delay Line Loop Shared Feedback Buffering (FDL-LSFB) module are designed in the node for contention resolution. As for the network coding module, XOR network coding is used as the first contention resolution method. All-optical logic XOR gates are used to encode conflicting multicasts and exchange the carrier wave length of the encoded multicast to avoid wavelength contention. The FDL-LSFB module, consisting of sub-switches and FDL buffer groups,is designed as the second contention resolution method. In FDL- LSFB module, FDL buffer groups are connected by sub-switches that guarantee the conflicting multicasts to enter or leave the FDL-LSFB module from any sub-switch. The FDL-LSFB module uses only a few FDLs and provides large buffering memories to resolve the multicast time contention with high FDL uti- lization. Moreover, the Minimum length Cascaded Buffer Scheduling Technique (MLCBST) is brought forward to schedule the FDL-LSFB module. Based on the MLCBST,the total multicast scheduling strat- egy for optical multicast is designed for the proposed node architecture. The simulation results show that the proposed node architecture and the scheduling strategy have an obvious effect in reducir~ packet loss probability (PLP) and decreasing the average delay.