<![CDATA[The deployment of 5G New Radio (NR) networks requires substantial infrastructure investment, posing challenges for emerging markets. This empirical study explores coverage optimization by retrofitting existing 4G LTE Base Transceiver Stations (BTS) using machine learning-enhanced Self-Organizing Network (SON) algorithms. Over six months, 6,895 drive test measurements were collected across 28 Telkomsel BTS sites in Padang, Indonesia, enabling before-and-after optimization analysis. Paired t-test results showed significant improvements in coverage quality: the service area with excellent SINR increased from 54.5% to 58.1% (p < 0.001, Cohen’s d = 0.27), while maintaining 99.3% RSRP compliance (≥ -92 dBm, 95% CI: 97.8%–100%). Automatic Cell Planning (ACP) effectively identified parameter configurations that enhanced performance without additional infrastructure cost, reducing deployment expenses. However, 5G NSA deployment remained limited to only 2.17% of measurements, which is valid for early-stage insights but restricts generalization to full 5G deployment scenarios. Therefore, the findings primarily apply to NSA overlay on existing 4G infrastructure rather than full 5G standalone deployment. This underscores ongoing economic and technical challenges in emerging markets’ 5G rollout. Despite this, the study provides strong empirical evidence supporting infrastructure reuse via SON-based optimization as a cost-effective way to improve coverage and quality. These results offer valuable guidance for operators and policymakers aiming to accelerate 5G adoption while managing costs in similar regions. Future work should expand validation as 5G NSA and standalone (SA) deployments grow and investigate integration with advanced AI-driven network management techniques]]>
Copyrights © 2025
