摘要
Optical Orthogonal Frequency Division Multiplexing (OOFDM) has been proposed as a highly spectrum-efficient modulation technique, which can provide flexible spectrum assignment with fine granularity. In OOFDM-based flexible optical networks, Routing and Spectrum Assignment (RSA) has become a key problem. However, widely used dynamic RSA schemes, such as Fixed Routing (FR) and K-shortest Paths (KSP) routing schemes, are not able to realize route computation based on the link state information, thus leading to poor blocking performance and inefficient resource utilization. To solve this problem, Adaptive Routing (AR) schemes, e.g., the Entire Path Searching (EPS) scheme, have been proposed recently. These schemes have low blocking probability; however, since their computational complexities are factorial, they are not suitable for use in real networks. In this paper, we propose a novel Spectrum-Scan Routing (SSR) scheme in dynamic flexible optical networks. To the best of our knowledge, SSR is the first polynomial-time AR scheme that can realize adaptive shortest-route computation. Simulation results show that our proposed SSR scheme has lower blocking probability and higher resource utilization compared with FR and EPS. Moreover, the worst-case computational complexity of SSR increases linearly with the network scale of the torus topologies, making it applicable to real networks.
Optical Orthogonal Frequency Divi- sion Multiplexing （OOFDM） has been pro- posed as a highly spectrum-efficient modula- tion technique, which can provide flexible spectrum assignment with fine granularity. In OOFDM-based flexible optical networks, Routing and Spectrum Assignment （RSA） has become a key problem. However, widely used dynamic RSA schemes, such as Fixed Routing （FR） and K-shortest Paths （KSP） routing schemes, are not able to realize route computa- tion based on the link state information, thus leading to poor blocking performance and inef- ficient resource utilization. To solve this prob- lem, Adaptive Routing （AR） schemes, e.g., the Entire Path Searching （EPS） scheme, have been proposed recently. These schemes have low blocking probability; however, since their computational complexities are factorial, they are not suitable for use in real networks. In this paper, we propose a novel Spectrum-Scan Routing （SSR） scheme in dynamic flexible optical networks. To the best of our knowledge, SSR is the first polynomial-time AR scheme that can realize adaptive shortest-route compu- tation. Simulation results show that our pro- posed SSR scheme has lower blocking prob- ability and higher resource utilization compared with FR and EPS. Moreover, the worst-case computational complexity of SSR increases linearly with the network scale of the torus to- pologies, making it applicable to real networks.
基金
supported in part by projects of National 863 Program under Grant No.2012AA011301
National 973 Program under Grants No. 2010CB328203, No. 2010CB328205
National Natural Science Foundation of China under Grant No. 61201188
