Pre-Launch Validation of 5G Site Prioritization Decisions Using PCA and Autoencoder Models


Cogen F., Gokcen A., Benli B., Kesik G., Alp S. S.

9th International Balkan Conference on Communications and Networking, Balkancom 2026, Ulcinj, Karadağ, 16 - 19 Haziran 2026, ss.103-108, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/balkancom71095.2026.11605033
  • Basıldığı Şehir: Ulcinj
  • Basıldığı Ülke: Karadağ
  • Sayfa Sayıları: ss.103-108
  • Anahtar Kelimeler: 5G planning, autoencoder, decision support, device readiness, fixed wireless access, planning validation, principal component analysis, site prioritization
  • İstanbul Gelişim Üniversitesi Adresli: Evet

Özet

This paper presents a pre-launch validation framework for assessing whether 5G site-prioritization decisions prepared ahead of the planned April 1, 2026 opening in Türkiye are aligned with latent demand, load, device-readiness, fixed-wireless-access, and commercial-value patterns extracted from anonymized site-level observations. Rather than relying on a single ranking mechanism, the study compares two distinct unsupervised validation models with different inductive biases. The first is a PCA-based scoring method that transforms normalized feature families into a single interpretable priority score derived from the first principal component. The second is an autoencoder-based method that learns a nonlinear low-dimensional latent representation from standardized numerical and encoded planning-context features and identifies sites with extreme multidimensional profiles. Rather than replacing expert judgment, the proposed models are used as a decision-support layer to validate existing planning choices when post-launch 5G ground-truth data are not yet available. Experiments on anonymized operator data show substantial agreement with the existing planning list and consistently surface high-load, high-demand, and high-value sites. The findings further indicate that model-plan disagreements are often associated with strategic and operational constraints, highlighting the complementary roles of data-driven analytics and engineering judgment in early-stage 5G investment planning.