计算机科学
资源配置
资源管理(计算)
资源(消歧)
调度(生产过程)
分布式计算
实时计算
方案(数学)
钥匙(锁)
计算机网络
作者
Chao Sun,Rongxi He,Chengzhi Song
摘要
We propose LT-MFA-RMCSA, a multidimensional fragmentation-aware resource allocation scheme for the routing, modulation, core, and spectrum assignment (RMCSA) problem in dynamic space-division multiplexing elastic optical networks (SDM-EONs). Following a filter-then-evaluate architecture, the scheme first removes infeasible allocations according to distance-adaptive modulation and mutual inter-core crosstalk constraints and then ranks the remaining candidates using seven metrics that characterize intra-path resource states, adjacent-link spatial alignment, and release-time temporal coordination. A Likert-scale weighting strategy is integrated with the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to balance fragmentation mitigation against path-resource consumption. Simulations on NSFNET and COST239 under both uniform and non-uniform traffic profiles show that LT-MFA-RMCSA consistently achieves the lowest connection and bandwidth blocking probabilities among the evaluated schemes while maintaining favorable residual-spectrum conditions. At the target blocking probability of 10 −3 , it supports approximately 8.08%–10.62% more traffic than the independently generated hop-count-based heuristics across the two topologies under uniform traffic. The scheme also maintains sub-second average processing time under the evaluated implementation, supporting its applicability to online dynamic provisioning.
科研通智能强力驱动
Strongly Powered by AbleSci AI