<![CDATA[Urban GPON FTTH planning must jointly manage feeder length and limited optical cores. This study develops a two-stage planning framework for a real GPON FTTH network in Medan, Indonesia. A road-constrained graph and Genetic Algorithm (GA) with a set-cover decoder are used to select and order closures; a closure-level pre-splitter power-margin rule then screens locations for splitter 1:2 deployment. This coordinated route-splitter evaluation is the study's main contribution because it links spatial routing with core-use decisions rather than treating them separately. The GA used 80 individuals, 100 generations, tournament selection, ordered crossover (0.90), and swap/segment-shuffle mutation (0.15). The resulting 7.02-km feeder covered all 29 service areas through nine closures and was 6.4% and 15.4% shorter than the two existing routes (7.50 and 8.30 km). Seven closures met the 2.45-dB pre-splitter margin rule; applying splitter 1:2 only at those locations reduced core requirements from 216 to 121 (43.98%), which fits the available 144-core feeder capacity and leaves 23 cores for expansion. The framework provides a repeatable decision-support method for urban FTTH planning; field power measurements and cost constraints should be included before construction.]]>
Copyrights © 2026
