Constructing Frequency Modulation-Broadcasting Map Based on Semi-Supervised Clustering

Abstract

The continuous evolution of current and future wireless networks across land, sea, air, and space places increasing demands on broadcasting coverage capabilities. Existing models are difficult to fully adapt to radio propagation scenarios in China. They are generally less effective than machine learning models in capturing the complex characteristics of radio channel propagation. To support the industrial upgrading of Frequency Modulation (FM) broadcasting and enhance its quality, this study presents the construction of a high-quality FM Broadcasting Map (FM-BM) using a semi-supervised clustering method. The core construction idea is to determine the spatial distribution pattern of the optimal model, model adaption map, by clustering based on the typical model predicted path loss of the Irregular Terrain Model, ITU-R P.1546, and ITU-R P.2001 and combining them with measured path loss in the Beijing area, and to accurately map the FM-BM accordingly. This study fully considers the influence of channel propagation characteristics and terrain features in constructing FM-BM to achieve dual improvements in classification refinement and prediction accuracy. Especially in the prediction of path loss of broadcasting signal propagation, compared with the three classical models, the method proposed shows significant superiority, with the prediction accuracy improved by 48.96%. This study not only provides a novel and efficient solution for the construction of FM-BM but also lays down the technical support and reference for upgrading FM broadcasting technology and improving coverage efficiency.