Case Studies
Solar-Powered Wireless Sensor Mesh for Precision Farming
An agri-tech startup with a validated concept but a failing first-generation node design needed a complete ground-up redesign before their runway expired.
The Challenge
An agri-tech startup had raised seed funding around a precision soil monitoring concept but significantly underestimated the engineering complexity of deploying wireless sensor networks in agricultural terrain. Their first-generation nodes drew too much current to survive winter months on solar power alone, had an effective mesh range of under 400m in crop cover, and drifted out of calibration after three months in the field.
The original firmware contractor had delivered code that worked in the lab but had never been characterised against real solar harvesting curves or in the RF environments that exist in crop cover. The startup knew the nodes were failing but lacked embedded and RF expertise to diagnose root causes.
Ankh was engaged with eleven months of runway remaining. A ground-up redesign was the only viable path. Speed of execution was not optional.
Our Approach
Rapid Prototyping & RF Characterisation
Ankh began with a four-week prototyping sprint to characterise RF propagation across three representative field types: open wheat, dense corn, and orchard. The data showed that 915 MHz sub-GHz radios with directional mesh topologies consistently outperformed 2.4 GHz solutions at ranges beyond 300m in crop cover. LoRa was selected as the primary radio PHY, with a proprietary mesh routing protocol layered on top for guaranteed delivery and TDMA-based power gating. Designing to datasheet assumptions alone would have repeated the first contractor's mistakes.
Hardware & Power Architecture
The sensor node was redesigned around an ultra-low-power MCU with hardware-accelerated sensor interfaces. A multi-cell solar harvesting circuit with maximum power point tracking and LiFePO4 battery chemistry was specified for winter operation in northern latitudes. Soil sensors were individually calibrated against NIST-traceable references and given a polynomial correction curve burned to non-volatile memory at the factory, eliminating the field calibration drift that had plagued the first-generation design.
Firmware & AI/ML Pipeline
Firmware implemented an adaptive duty cycle that adjusted transmission intervals based on battery state of charge and a time-of-day solar forecast derived from latitude and season. On the backend, Ankh's software team built a machine learning pipeline to detect anomalous sensor drift patterns in deployed nodes — flagging units for recalibration before data quality degraded. The model achieved 94% accuracy identifying pre-failure drift signatures ahead of any farmer-visible degradation.
Field Pilot Execution
A 200-node pilot was deployed across three farms over a single growing season. Installation was designed to be executable by farm staff with no electronics background — a constraint that shaped enclosure design, connector choices, and the installation guide from the first prototype. Mesh uptime was tracked continuously against the 99% target throughout the pilot.
Results
The 200-node pilot achieved 99.1% mesh uptime over 8 months — above the 99% target. No node required battery replacement during the pilot period. The ML drift detection system flagged 11 nodes for recalibration before any data quality issue was visible to farm operators.
The startup closed a Series A round six weeks after pilot completion, with the validated field performance data as the centrepiece of their investor deck. Ankh's ability to execute a ground-up redesign in eleven months without compromising field reliability was the factor that made the fundraise possible.