Authors: Associate Professor Shilpa K
Abstract: Agricultural practice in India has evolved from knowledge-intensive traditional systems to technology-supported approaches that combine sensing, communication, cloud services, and data-driven analytics. This review examines how Internet of Things (IoT) infrastructure and Machine Learning (ML) can be combined for smart agriculture monitoring and decision support. The discussion covers historical agricultural practices, IoT architecture, field sensing, soil and environmental monitoring, irrigation support, crop-condition assessment, and ML techniques reported for agricultural prediction. The reviewed approaches use variables such as soil moisture, soil temperature, pH, humidity, rainfall, nutrient status, weather observations, and remotely sensed information. ML methods reported in the literature include decision trees, support vector methods, random forest, Cub-ist, K-means, regression models, gradient-boosted trees, extreme learning machines, and convolutional neural networks. The review also reorganizes the technology stack into a practical monitoring cycle in which sensors acquire observations, a gateway transfers the data, analytics interpret the measurements, and recommendations are returned to the farming process. The synthe-sis indicates that IoT provides the observation and communication layer, while ML can convert heterogeneous measurements into predictions or recommendations. The paper identifies opportunities for integrated, affordable, and continuous monitoring while also highlighting constraints related to connectivity, cost, data quality, interoperability, and adoption.
