ReproTrack — IoT Heat Detection for Cattle and Water Buffalo

In extensive livestock farming, reproductive efficiency is the main economic driver. Every "open day" — the period a female stays non-pregnant beyond the optimal window — costs the producer between €3.50 and €5.00 per day in upkeep with no return.

ReproTrack is where my two degrees converge: Animal Production Engineering and Computer Engineering.

The problem

Current monitoring systems (smart collars and ear tags) were designed for housed dairy cows and fail in two critical scenarios:

• The water buffalo blind spot. Heat in buffalo (Bubalus bubalis) is mostly silent, happens at night, and often while the animals are submerged in lagoons or mud, where standard sensors suffer complete signal attenuation.
• The cost barrier. Fitting 100% of the females with active collars (€100–150 per animal) demands an upfront investment that is prohibitive for mid-sized farms.

The innovation: asymmetric architecture

Instead of monitoring every female individually with expensive hardware, ReproTrack inverts the market's logic:

• On the bull (active node): a single rugged device on the chest. It combines accelerometry (IMU), on-device edge processing, and BLE + LoRaWAN radios.
• On the females (passive nodes): sealed, static BLE beacons costing €2–3 each, placed at the base of the tail.

Rather than inferring heat from activity patterns, the system validates the direct physical mounting event.

How it works

1. Idle. The bull's device stays in deep sleep; only the IMU watches motion at low frequency.
2. Detection. When the bull mounts, the sensor registers a sharp vertical tilt on the Z axis and fires a hardware interrupt that wakes the MCU.
3. Validation. The MCU turns on the BLE receiver for 20 seconds and reads the female's tag. If RSSI exceeds −50 dBm, close proximity is confirmed and false positives are discarded.
4. Storage. The event [bull_ID, female_ID, timestamp] is encrypted and written to non-volatile Flash.
5. Asynchronous transmission. Once the animal is back on dry land, the LoRa radio dispatches the buffered packets to the farm gateway (868 MHz), with no cellular network costs.
6. Processing. The server stores the history and notifies the farmer of the optimal artificial insemination window using the AM-PM rule.

The store-and-forward protocol guarantees no data is lost while animals are in water or out of coverage.

Technical stack

Hardware and firmware

• SoC with native BLE and ultra-low power draw (ESP32-WROOM / Nordic nRF52840)
• Semtech SX1262 LoRa module (European 868 MHz ISM band)
• LIS3DH accelerometer with hardware interrupts
• Firmware in C++ (Arduino / ESP-IDF)
• Industrial polymer housing sealed with epoxy resin for genuine IP68 rating

Cloud and application

• REST/WebSocket API in Node.js (TypeScript)
• PostgreSQL with TimescaleDB for time-series data
• Responsive PWA dashboard in React and Tailwind CSS

Economic impact

Annual savings follow N × ΔD × C, where N is herd size, ΔD the average reduction in open days (estimated at 15), and C the daily cost of a non-pregnant animal (€4.00).

For a typical 100-head farm in the Viseu region, this translates into €6,000 per year in direct savings, with the investment paying for itself in under 8 months.

Current status and roadmap

The project is in advanced ideation and engineering prototyping:

• ✅ Hybrid network architecture (BLE + LoRaWAN) designed
• ✅ Mounting-detection algorithm structured
• ✅ Dashboard prototype built
• ⏳ First PCB batch manufacturing and IP68 watertightness lab testing
• ⏳ Real-world pilot trials (cattle herd in Viseu, buffalo herd in Ribatejo)