Previous dynamic resource allocation methods (Separate Path Select, SPS) for survivable networks are always separated into two steps, that is, to determine an optimal primary path first and then select a backup path f...Previous dynamic resource allocation methods (Separate Path Select, SPS) for survivable networks are always separated into two steps, that is, to determine an optimal primary path first and then select a backup path for the primary path. But SPS cannot achieve a good network performance. In this paper we analyze the drawback of the traditional SPS technique and propose an efficient shared protection method called M _ SPS. Instead of rejecting an arrival call immediately when a backup lightpath for the optimal primary lightpath cannot be found, M _ SPS continue to try the sub-optimal one until a legal primarybackup pair can be selected. We apply M SPS to CERNET_ Like and NSFNET networks. The results show that M _ SPS improves the network performance remarkably.展开更多
文摘Previous dynamic resource allocation methods (Separate Path Select, SPS) for survivable networks are always separated into two steps, that is, to determine an optimal primary path first and then select a backup path for the primary path. But SPS cannot achieve a good network performance. In this paper we analyze the drawback of the traditional SPS technique and propose an efficient shared protection method called M _ SPS. Instead of rejecting an arrival call immediately when a backup lightpath for the optimal primary lightpath cannot be found, M _ SPS continue to try the sub-optimal one until a legal primarybackup pair can be selected. We apply M SPS to CERNET_ Like and NSFNET networks. The results show that M _ SPS improves the network performance remarkably.