Artificial Intelligence Techniques for Enhancing Air Pollution Detection Accuracy and Quality Monitoring Using Pyramidal Convolution Split-Attention Networks and IoT: Trends and Challenges

Abstract

Air pollution has become a critical global issue affecting public health, climate stability, and environmental sustainability. Accurate detection and monitoring of air pollutants such as PM2.5, NO₂, CO, and SO₂ are essential for mitigating their adverse effects. Traditional monitoring systems, although accurate, are limited by high costs, sparse deployment, and lack of real-time adaptability. The integration of Artificial Intelligence (AI) with Internet of Things (IoT) technologies has emerged as a transformative solution, enabling distributed, real-time, and intelligent air quality monitoring. This paper presents a comprehensive review of AI-based techniques, focusing on deep learning architectures such as CNN, LSTM, Transformer models, and pyramidal convolution split-attention networks. IoT devices provide continuous environmental data streams, while AI models enhance detection accuracy by capturing nonlinear spatiotemporal relationships. Advanced attention-based and multi-scale convolutional architectures significantly improve feature extraction and prediction accuracy in complex environmental scenarios. Recent studies highlight the effectiveness of hybrid models and optimization techniques in improving model performance and reducing prediction errors. For instance, attention-based convolutional BiLSTM models and vector map convolution networks have demonstrated superior accuracy in AQI prediction tasks by learning spatial and temporal dependencies simultaneously. Despite these advancements, challenges such as data heterogeneity, sensor calibration, computational complexity, and energy efficiency remain. Emerging solutions such as federated learning, edge computing, and explainable AI offer promising directions for future research. This review explores recent trends (2020–2023), comparative insights, and challenges in AI-driven air pollution monitoring systems.