针对多粒度应用，开发了一种用于流量疏导的波群模型，并引入一种基于此模型的新型智能交换结构．该光交换结构提供了独特的区分粒度到相应隧道进行有效处理的方法．此外，还讨论了控制层粒度分离时采用的两个关键的动态算法模块．仿真结果显示这种特殊的通道分离方法有效提高了处理动态连接请求时每个光路径的平均信号通道质量和阻塞性能．

For multi-granularity application,both a lambda-group model used in traffic grooming and a new intelligent switching fabric based on the new model were presented. The optical switching fabric presented a distinctive approach of dividing granularities into specific tunnels for effective optical treatment. In addition,two key dynamic algorithm modules of configuration for granularity separation in the control layer were discussed. Simulation results show that the method of particular channel partition can greatly improve the average channel quality and the blocking performance along every optical path for dynamic connection requests.