<![CDATA[The research conducted a PRISMA-based systematic review through Scopus, IEEE Xplore, Web of Science, and ScienceDirect, using a rigorous 10-year recency filter to assess peer-reviewed papers implementing physical sensor telemetry. A qualitative data extraction matrix was used to exclude any theoretical models and identify resource management and socio-economic scaling structure gaps. From these findings, a key challenge emerged concerning the interaction between data and algorithms - namely, that highly sophisticated reinforcement learning and computer vision algorithms are very susceptible to practical implementation issues in terms of algorithmic bias and latency due to physical sensor drift and changes in lighting conditions. In quantifiable terms, despite the efficiency gains realized via optimized AIoT systems in saving up to 30-40% of water and reducing fertilizer runoff, the efficiencies rapidly decreased once the hardware had degraded in its environment. Moreover, the literature has a strong socio-economic skew, with over 80% of state-of-the-art algorithms being developed exclusively for large enterprises. In addressing this digital divide, this research validated an open-source AIoT model based on affordable microcontrollers (such as ESP32), together with Edge-ML (TinyML) and multi-tenant cloud networks. In practical terms, agricultural cooperatives have the capability to adopt this modular approach by consolidating the cost of cloud computing, deploying low-cost local sensor nodes, and performing automated validation through edge computing to ensure that crops are not subjected to excessive irrigation or fertilization.]]>
Copyrights © 2026
