Sparse placement of electronic switching nodes for low blocking in translucentoptical networks
Sparse placement of electronic switching nodes for low blocking in translucentoptical networks
复制标题
DOI：
--
复制
发表时间：
2002-12
影响因子：
--
通讯作者：
G. Shen;W. Grover;T. Cheng;S. Bose
中科院分区：
文献类型：
--
作者：
G. Shen;W. Grover;T. Cheng;S. Bose
我们使用相对较少的交换节点，波长转换和电子再生是可能的光网络设计的性能进行评估。放置最少此类节点以达到给定阻塞概率的简单启发式算法是基于经过每个节点的最短路径路由的排名频率。这种策略被发现是有效的，在设计一个半透明的光网络与稀疏的电子开关的位置，执行非常接近的阻塞，一个不透明的光网络。此外，我们应用一个新的二维Dijkstra算法的路由和波长分配在由此产生的半透明光网络。仿真结果表明，在考虑再生前最大透明距离约束的情况下，基于启发式算法的稀疏电子开关半透明光网络比完全透明光网络具有更低的阻塞率.此外，当根据启发式放置开关时，光路阻塞可以接近具有显著更少的总电子开关的完全不透明网络的光路阻塞。在我们的研究结果中，光路阻塞低，与完全不透明的网络的情况下，得到了电子开关选择性地放置在大约三分之一的节点上平均。该启发式算法还可以很好地对抗随机搜索电子开关节点的有效子集，并且比基于组合穷举搜索并且仅限于假设固定最短路径路由的先前最优方法执行得更好。
We evaluate the performance of optical network designs using relatively few switch nodes at which wavelength conversion and electronic regeneration is possible. A simple heuristic for placing the fewest such nodes to reach a given blocking probability is based on the ranked frequency of shortest-path routes transiting each node. This strategy is found to be efficient in designing a translucent optical network with sparse electronic switch placement that performs very close in blocking to that of an opaque optical network. In addition, we apply a new two-dimensional Dijkstra algorithm for routing and wavelength assignment in the resultant translucent optical network. Simulation results indicate that a translucent optical network with sparse electronic switch placement based on the heuristic has much lower blocking than a fully transparent optical network when the constraint of the maximum transparent distance before regeneration is also considered. Moreover, when the switches are placed according to the heuristic, lightpath blocking can approach that of a fully opaque network with significantly fewer total electronic switches. In our results, lightpath blocking as low as that with the fully opaque network case was obtained with electronic switches selectively placed at approximately one node in three on average. The heuristic also performs well against random searching for an effective subset of electronic switch nodes and performs better than a prior optimal method that is based on a combinatorially exhaustive search and that is limited to assuming fixed shortest-path routing.