Journal of Agronomy and Field Innovations

Agricultural irrigation accounts for the largest portion of freshwater consumption globally, mostly utilizing calendar-based or experience-based approaches that do not correspond to specific crop water requirements. The use of wireless soil sensor networks (WSSNs) in conjunction with artificial intelligence (AI)-based modeling has created a way to meaningfully bridge this gap, allowing continuous monitoring and precise scheduling of irrigation activities. The objective of this literature review is to summarize relevant literature on using WSSNs as a base for the development of AI methodologies aimed at the optimization of irrigation process. Studies on the use of machine learning, particularly random forest, XGBoost, support vector regression, and neural networks, show that these techniques could be used for accurate predictions of soil moisture amount and its characteristics. Communication technologies which are based on low-power wide-area protocols, such as LoRaWAN, as well as multi-tiered edge-cloud computing, have intriguingly established themselves as the most desirable technological foundation of WSSN implementations on a large scale, in terms of feasibility in energy consumption, range, and price. The review finds several difficulties, including inconsistencies in the reporting of engineering criteria of the total WSSN system, such as energy consumption and reaction time in total system accuracy, lack of longitudinal multi-season testing of an integrated AI-WSSN installations in practice, as well as cost and infrastructural barriers for small household usage. It can be stated that the implementation of irrigation optimization tools based on wireless soil sensors will be a scientifically tested way toward agricultural water management improvement in terms of water resource management, although with the broad accessibility of such systems depending on their future affordability.