부산대학교 도서관

Point-to-multipoint (P2MP) coherent pluggable transceivers leveraging digital subcarrier multiplexing (DSCM) techniques are gaining momentum among telecommunication operators as a technology that can potentially provide broadband connectivity in metropolitan area networks (MANs) with flexible bandwidth allocation at reduced costs. This technology allows trees to be built where central metro (hub) nodes are interconnected with remote access/aggregation (spoke) nodes with dedicated and flexible bandwidth slices, enabling better resource management, simplified architectures, and reduced costs. This paper proposes a modified version of AI/ML clustering techniques that concentrates on the same P2MP tree as those low-rate leaf nodes with uncorrelated daily traffic patterns in a cost-effective manner, thus allowing 10%–18% tree reductions. This, when compared with fixed P2P transceivers, translates into potential CAPEX savings between 40% and 50% in the different scenarios here considered.