IoT Cattle Health Monitor using Smart Collar with Farm Dashboard

Dairy farmers usually notice a sick animal only after visible symptoms appear — by which time treatment costs more and milk yield has already dropped. Cattle mask early illness well, but two signals change first: body temperature rises and normal movement (grazing, rumination motion) falls. The engineering gap is a low-cost way to watch those two signals continuously on every animal without manual checks. This project builds exactly that: a collar-worn ESP32 node with a skin-contact temperature probe and a motion sensor, transmitting readings over LoRa to a farm gateway, with a dashboard that flags fever and abnormal inactivity per animal. Everything is demonstrated on a real collar prototype plus a working farm dashboard, so the sensing-to-alert chain is fully visible.

1. The collar's DS18B20 probe measures skin-contact temperature while the MPU6050 samples acceleration; firmware converts motion into a 0–100 activity index.
2. Every 30 seconds the ESP32 packs temperature, activity index, battery level and collar ID into a LoRa packet and transmits it to the farm gateway.
3. The gateway ESP32 receives packets, checks each animal against the fever (39.5 °C) and low-activity thresholds, and serves the farm dashboard over Wi-Fi.
4. The dashboard renders one card per animal with live temperature, activity bar, battery and last-packet age, plus a selectable 24-hour temperature chart with the fever band marked.
5. When a reading crosses a threshold, the dashboard raises an alert banner, changes the animal's card status, and appends a timestamped entry to the event log.
6. The farmer checks the watch list each morning; the buyer-run calibration procedure documents how the fever threshold and activity baseline are tuned for the demo herd.

Tech stack:

• ESP32 (collar node + LoRa gateway)
• DS18B20 temperature probe · MPU6050 IMU
• LoRa radio (SX1276/Ra-02 class)
• 18650 cell with charging module
• Arduino IDE (C/C++ firmware)
• HTML/CSS/JavaScript companion dashboard
• 3D-printed collar enclosure

• [Smart collar hardware] Rugged 3D-printed enclosure on a fabric neck strap carrying the ESP32, temperature probe, motion sensor, LoRa radio and rechargeable cell — a wearable students can actually handle and demonstrate.
• [Body-temperature sensing] DS18B20 probe in skin contact reports body temperature with ±0.5 °C datasheet accuracy; fever flag at a configurable 39.5 °C threshold.
• [Activity index] The MPU6050 motion sensor is converted into a 0–100 activity score, so reduced grazing or rumination movement shows up as a falling trend before visible symptoms.
• [LoRa collar-to-gateway link] Each collar transmits its readings every 30 seconds to a farm gateway — long range on a farm with no Wi-Fi or mobile data needed at the collar.
• [Per-animal farm dashboard] The companion dashboard shows every animal as a card with temperature, activity, battery and last-packet time, plus a 24-hour temperature chart and an event log.
• [Fever and inactivity alerts] Temperature above 39.5 °C raises a fever alert; sustained low activity adds the animal to a watch list, both logged with timestamps.
• [Battery monitoring] Collar battery percentage is reported with every packet so dead collars never silently drop out of the herd view.

• Working smart collar prototype (enclosure, ESP32, sensors, LoRa radio, battery)
• LoRa gateway node with firmware
• Companion farm dashboard web app (per-animal cards, charts, alerts, event log)
• Arduino firmware source for collar and gateway
• Wiring diagram and 3D enclosure print notes
• Project report PDF (background, sensor theory, LoRa link design, methodology, results)
• PPT presentation for final review
• Viva Q&A preparation document (temperature sensing, IMU activity scoring, LoRa basics, threshold tuning)

• The probe measures skin-contact temperature, an approximation of core body temperature — it is an early-warning aid, not a veterinary diagnostic device.
• LoRa range of 1–2 km is a design target for open farm terrain; buildings and hills reduce it, and the buyer verifies range on their own site.
• Battery life of 3–5 days is a design target; heavier packet rates shorten it.
• No GPS or geofencing is in the build — the system tracks health signals per collar ID, not location.
• The collar is a student prototype enclosure, not a chew-proof, weather-sealed commercial tag.