Technology-independent topology design heuristics for point-to-multipoint optical access networks

Next Generation Access (NGA) networks offer enormous bandwidth and low latency, mainly due to the exploitation of optical transmission. Deploying optical fiber in the access network, however, requires a huge investment, therefore optimization of the physical infrastructure plays an important role. In the recent years, algorithmic access network design became viable, mainly due to the existence of digital maps and Geographic Information System (GIS) databases. In the previous work, we have proposed technology-dependent, scalable heuristics for Passive Optical Network (PON) and Active Ethernet (AETH) network design. In this paper, we present a novel technology-independent solution based on the Simulated Annealing (SA) metaheuristics for tree-based point-to-multipoint optical access network topologies. The newly proposed heuristics deliver competitive results, within 5-10% of the theoretical optimum, even for scenarios with up to thousands of demand points. The key for scalability is the concept Voronoi-diagrams applied for demand point clustering and evaluation within the Simulated Annealing scheme.