Article Details
  • Published Online:
    April  2025
  • Product Name:
    The IUP Journal of Computer Sciences
  • Product Type:
    Article
  • Product Code:
    IJCS010425
  • DOI:
    10.71329/IUPJCS/2025.19.2.7-22
  • Author Name:
    Abdulmumini Yakubu Musah, Fatimah Binta Abdullahi, Bisallah Ibrahim Hashim, Francisca Nonyelum Ogwueleka and Amina Imam Abubakar
  • Availability:
    YES
  • Subject/Domain:
    Engineering
  • Download Format:
    PDF
  • Pages:
    7-22
Volume 19, Issue 2, April 2025
5G-CPLNet Architecture for Path Loss Prediction
Abstract

Accurate path loss (PL) prediction is essential for optimizing the performance and reliability of 5G wireless networks. This paper introduces the 5G-Convolutional Path Loss Network (5GCPLNet), a hybrid deep learning (DL) model that combines convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and fully connected networks (FCNs). The model leverages the unique strengths of each architecture to effectively capture spatial, temporal, and feature-based patterns, addressing the inherent complexities of 5G signal propagation. Extensive evaluations using synthetic and real-world datasets demonstrate the model’s superior performance, with respect to Mean Absolute Error (MAE) and coefficient of determination (R²), outperforming traditional machine learning (ML) and existing DL models. The findings highlight 5G-CPLNet’s capability to provide precise and reliable predictions, which is crucial for network planning, resource optimization, and maintaining quality of service in diverse 5G environments. This study also discusses potential extensions of the model to handle dynamic network conditions, paving the way for advancements in autonomous network management and next-generation wireless communication.

Introduction

The rapid evolution of fifth-generation (5G) wireless networks has introduced unprecedented challenges in ensuring reliable communication.