The Problem: Water Waste in Rural Agriculture
In Colombia, 20% of the population lives in rural areas and depends on agriculture for their livelihoods. Yet agricultural water use globally accounts for approximately 70% of all freshwater consumption — and much of that water is wasted through inefficient irrigation. For Colombian farmers, water waste isn't just an environmental problem; it's an economic one that directly threatens household food security and income.
A team of Colombian engineering students took on this challenge with a sophisticated combination of IoT hardware, wireless communications protocol, and machine learning algorithms — building an open-source solution that any farmer with basic hardware access could deploy.
Engineering Approach: Zigbee, MQTT-SN, and Gradient Boosted Trees
The student team's solution used three core technologies working together:
- Zigbee wireless protocol: A low-power, low-cost mesh networking standard ideal for agricultural sensor networks where battery life matters and distances between sensors can be large.
- MQTT-SN (Message Queuing Telemetry Transport for Sensor Networks): A lightweight protocol designed for constrained IoT devices, enabling efficient communication between field sensors and a central control node without draining battery power.
- Gradient boosted trees and random forests: Machine learning algorithms trained on historical soil moisture and weather data to predict when irrigation is actually needed — eliminating guesswork from the watering schedule and preventing both over- and under-irrigation.
The result was a fully automated system that monitors soil conditions, applies machine learning to predict optimal irrigation timing, and activates drip irrigation hardware only when needed. Crucially, the entire system was built as open-source software and hardware, allowing other communities to replicate or adapt it without licensing costs.
Project Outcomes
Four students participated in the project, taking it from concept to working prototype. Beyond the technical deliverable, the project created a dedicated research team focused on IoT and machine learning applications in agriculture, and an open-source codebase available to other engineering teams across Latin America.
The Food and Agriculture Organization of the United Nations has identified precision agriculture as a key strategy for feeding a growing global population while reducing environmental impact. Student-built, open-source systems like this demonstrate that precision agriculture doesn't require expensive proprietary technology.
Why This Model Works
The Colombian project succeeded because it combined local knowledge with modern engineering tools. The students understood which crops were most water-sensitive, which fields were most at risk of water stress, and which communication technologies were available and affordable in the rural context. This local knowledge — combined with skills in machine learning and embedded systems — produced a solution that external experts would have been unlikely to design.
Learn more about the technologies used in this project: IoT for Development and AI for Social Impact.