Artificial Intelligence Methods for Adaptive Regression Learning of Signal Propagation Loss in Cellular Communication Networks: A Review
Year: 2025
Type: article
Abstract: Accurate prediction of signal propagation loss is paramount for efficient planning, deployment, and optimization of cellular communication networks. Traditional methods, ranging from empirical models to deterministic ray tracing, often suffer from limited adaptability, high computational complexity, or require extensive site-specific calibration. The advent of Artificial Intelligence (AI) and Machine Learning (ML) has revolutionized this domain, offering data-driven, adaptive, and highly accurate solutions for predicting continuous values like propagation loss—a classic regression problem. This paper provides a comprehensive review of AI techniques applied to the predictive regression learning of signal propagation loss in cellular networks. We examine the evolution from conventional ML algorithms like Support Vector Machines and Ensemble Methods to advanced Deep Learning (DL) architectures such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Graph Neural Networks (GNNs). The review categorizes existing approaches based on their underlying AI methodologies, discusses their strengths and limitations, and highlights the crucial aspects of data feature engineering and model training. Furthermore, we identify key challenges, including model interpretability, generalization across diverse environments, and computational overhead. Finally, we explore promising future research directions, such as hybrid physics-informed AI models, federated learning, explainable AI (XAI), and the integration of digital twin technology, all of which aim to enhance the robustness, accuracy, and deployability of AI-driven propagation loss prediction for future 5G and 6G cellular ecosystems.
Authors: Isabona, Joseph, Risi, Ikechi
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7220147066