Due to an existing combination of software, collection devices, and computer environments that require growing quantities of energy to run, monitoring and measuring forest ecosystems is a challenging task. A unique technique for employing artificial intelligence and machine learning to monitor soil moisture has been created by the University of Maine‘s Wireless Sensor Networks (WiSe-Net) lab. This method could be used to improve measurement across the vast forest ecosystems of Maine and beyond.
In both agricultural and forestry ecosystems, soil moisture is a crucial factor, especially in the recent drought conditions of recent Maine summers. Researchers, foresters, farmers, and other people who want to keep an eye on the health of the land may not be able to do so because commercial soil moisture sensors are expensive and use a lot of power. This is true even though there are strong soil moisture monitoring networks and large, freely accessible databases.
The University of Maine’s WiSe-Net created a wireless sensor network that employs artificial intelligence to learn how to be more power-efficient when monitoring soil moisture and processing the data, along with researchers from the Universities of New Hampshire and Vermont. The National Science Foundation provided funding for the study.
Ali Abedi, who led the recent study and is a professor of electrical and computer engineering at the University of Maine, says, “AI can learn from the environment, predict the quality of a wireless link, and calculate how much solar energy is coming in. This allows a robust, low-cost network to run longer and more reliably with less energy.”
When compared to current industry norms, the software produces power-efficient systems at a lower cost for large-scale monitoring as it gains experience using the network resources that are accessible to it.
Aaron Weiskittel, the director of the Center for Research on Sustainable Forests, and WiSe-Net also worked together to make sure that all hardware and software research is supported by science and is suited to research needs.
According to Weiskittel, “Soil moisture is a fundamental driver of tree development, yet it varies swiftly, both daily and seasonally.” “We haven’t been able to effectively monitor at scale. In the past, we employed pricey sensors that were unreliable and collected data at set intervals—every minute, for instance. This sensor definitely opens the door for future applications for researchers and practitioners alike because it is more affordable, more reliable, and has wireless capabilities.
The research was published in Springer’s International Journal of Wireless Information Networks on August 9, 2022.
Although the system developed by the researchers focuses on soil moisture, the same concept could be used for other types of sensors, such as ambient temperature, snow depth, and more. Additionally, the networks may be scaled up with more sensor nodes.
“Different sampling rates and power levels are required for the real-time monitoring of various variables. Instead of sampling and delivering every single data point, which is not as efficient, an AI agent may learn these and change the data collection and transmission frequency accordingly, “said Abedi.
