Recent Advances in Enhancing Air Pollution Detection Accuracy and Quality Monitoring Using Pyramidal Convolution Split-Attention Networks and IoT: A Systematic Review

Abstract

Air pollution has emerged as a critical global environmental and public health challenge, contributing to millions of premature deaths annually. Accurate monitoring and prediction of air quality are essential for effective environmental management and policy-making. Traditional monitoring systems rely on static stations, which are often expensive and lack spatial resolution. The integration of Artificial Intelligence (AI) with Internet of Things (IoT) technologies has significantly enhanced air pollution detection and forecasting capabilities by enabling real-time, distributed, and data-driven monitoring systems. Recent advances in deep learning, particularly convolutional neural networks (CNN), recurrent neural networks (RNN), and attention-based architectures, have demonstrated superior performance in capturing complex spatiotemporal patterns in air quality data. Emerging models such as pyramidal convolution and split-attention networks further improve feature extraction by focusing on multi-scale representations and adaptive feature weighting. These architectures enhance prediction accuracy and robustness in heterogeneous IoT environments. This paper presents a systematic review of AI-based IoT air pollution monitoring systems between 2020–2023. It highlights recent trends, comparative performance, and challenges, including data quality issues, sensor calibration, computational complexity, and scalability. The study concludes by identifying research gaps and future directions toward efficient, interpretable, and scalable air quality monitoring systems.