The explosive growth of the Internet of Things (IoT) has contributed to the large-scale deployment of wireless sensor devices in various domains, resulting in the generation of massive volumes of data. Extracting meaningful insights from these data streams requires advanced computational techniques capable of identifying hidden patterns. Machine Learning (ML) and Deep Learning (DL) models have been extensively explored to analyse sensor data and derive actionable knowledge. This paper presents a comprehensive review of recent advancements in ML and DL techniques tailored for IoT-enabled wireless sensor networks. The study critically evaluates the current research landscape, identifying key achievements, persistent challenges, and potential future directions. The work provides an end-to-end review that spans data preprocessing, feature extraction, model selection, anomaly detection, and prediction, while critically analysing the suitability of these approaches for resource-constrained IoT devices. The review uniquely integrates emerging paradigms such as Explainable AI, federated learning, and edge computing, offering insights into privacy-preserving and low-latency data analytics at the network edge, an area underexplored in existing literature. Additionally, the paper examines edge intelligence and distributed learning architecture to guide the design of future intelligent IoT-WSN systems. This review systematically analyzes machine learning and deep learning techniques applied in IoT and wireless sensor network environments, highlighting prevailing algorithmic trends, deployment constraints, and performance trade-offs. The analysis reveals that hybrid ML–DL approaches, when carefully optimized for computational cost, energy efficiency, and latency, consistently outperform standalone models in resource-constrained IoT edge scenarios while maintaining acceptable predictive accuracy.
Egwuche et al. (Thu,) studied this question.