Development of Decision-Making for Selection Splitter Distribution Points for the Optical Fiber Networks

Abstract

The appropriate selection of the distribution point on the internet for the fiber-optic network (FTTx) is an important step that affects the efficiency of Internet service. The purposes of this research were to (1) develop a decision-making system for selecting the Splitter Distribution Points for fiber optic networks, (2) compare data classification methods, and (3) analyze the satisfaction evaluation results of system. The sample group consisted of 100 people, using a specific sample selection method. The research methods were web-based applications for deciding on the selection of the Splitter Distribution Points for fiber-optic networks through K-Nearest Neighbors: K-NN. The feature used consisted of Loss Bandwidth Package Dist and Class. The statistics applied in the study were means and standard deviations. The results of the research were as follows: the development of a decentralized decision-making system for fiber-optic networks made it an effective tool for selecting Splitter Distribution Points in the department. The research used K-Nearest Neighbors (K-NN) technique for decision-making in selecting Splitter Distribution Points model from three comparing methods of data segmentation, which were K-Nearest Neighbors (K-NN), Neural Network, and Support Vector Machine. The result of the experiment showed that K-Nearest Neighbors (K-NN) had the highest accuracy at 97.88%. The results of the system user satisfaction evaluation were divided into 3 areas: the operators’ satisfaction toward imported data was found at the highest level ( = 4.65, S.D. = 0.45); the workers' satisfaction toward the work processes of the system was perceived at the highest level ( = 4.68, S.D. = 0.51); and the satisfaction of the executives toward the results was observed at the highest level ( = 4.77, S.D. = 0.42).