Deep Learning and Optimization Approaches in Environmental Weather Monitoring and Prediction System Using IoT and Multi-Model Progressive Dense Self-Attention: A Review

Abstract

Environmental weather monitoring and prediction have become critical components in addressing global climate change, disaster management, agriculture optimization, and smart city planning. Traditional numerical weather prediction (NWP) models, while effective, often struggle with high computational complexity and limited capability in modelling nonlinear spatiotemporal dependencies. The emergence of Internet of Things (IoT) technologies combined with deep learning (DL) and optimization techniques has revolutionized the accuracy, scalability, and responsiveness of weather prediction systems. This paper presents a comprehensive review of advanced deep learning architectures such as Long Short-Term Memory (LSTM), Convolutional Neural Networks (CNN), Transformer-based models, and attention-driven architectures, integrated with IoT-based environmental monitoring systems. Recent advancements highlight the effectiveness of hybrid models such as CNN-LSTM and attention-based recurrent networks in capturing both spatial and temporal dependencies in weather datasets. Additionally, optimization algorithms including metaheuristic techniques (e.g., Golden Jackal Optimization) and hyperparameter tuning strategies significantly enhance model performance and convergence. The review also explores the role of multi-model progressive dense self-attention mechanisms, which enable efficient feature extraction from heterogeneous IoT sensor data, improving prediction accuracy and robustness. Furthermore, challenges such as data heterogeneity, scalability, energy efficiency, and real-time processing are discussed alongside emerging solutions like federated learning and edge computing. The study concludes by identifying future research directions, emphasizing the need for explainable AI, energy-efficient models, and integrated IoT–AI frameworks for sustainable environmental monitoring systems.