Development of 5G/F5G technology leads to massive applications accessing to backbone networks, which requires the backbone networks to be upgraded. Semi-filterless elastic optical network (semi-FEON) is a suitable technology to cheaply and gradually upgrade backbone networks. In semi-FEON, routing, modulation and spectrum assignment (RMSA) problem is one of the key issues. In this paper, we study the dynamic RMSA problem in semi-FEON and propose an RMSA algorithm. The algorithm includes three innovations: a K-shortest-subnet-paths (KSSP) algorithm is designed to search candidate paths in semi-FEON, a load-balancing-least-resources (LBLR) policy is introduced to re-sort the candidate paths, and a maximum-occupied-neighbors (MON) rule is proposed to assign spectrum resources to connection requests in semi-FEON. Simulation results show that the proposed KSSP-LBLR-MON algorithm outperforms the existing works in term of bandwidth blocking probability. Concretely, the improvement ratio is greater than 59.98% and 66.64% in German-Net and Henan-Net, respectively.

5G/F5G技术的发展导致海量应用接入骨干网，需要对骨干网进行升级改造。半无滤波器弹性光网络（semi-FEON）是一种适合廉价且逐步升级骨干网络的技术。在半FEON中，路由、调制和频谱分配（RMSA）问题是关键问题之一。在本文中，我们研究了半FEON中的动态RMSA问题并提出了一种RMSA算法。该算法包括三项创新：设计了K最短子网路径（KSSP）算法来搜索半FEON中的候选路径，引入负载均衡最少资源（LBLR）策略来对候选路径进行重新排序。 ，并提出了最大占用邻居（MON）规则来为半FEON中的连接请求分配频谱资源。仿真结果表明，所提出的KSSP-LBLR-MON算法在带宽阻塞概率方面优于现有算法。具体来说，德国网和河南网的改善率分别大于59.98%和66.64%。