MQTT Cold Chain Logger using ESP32

Vaccines lose potency, and food and blood products spoil, when they drift outside their required temperature window — the WHO-recommended 2–8 °C for most vaccines — during storage or transport. The failure is usually invisible: a fridge door left ajar overnight, a delivery van parked in the sun, a power cut nobody logged. Manual twice-daily thermometer checks miss all of it. Continuous electronic logging is the established answer in the pharmaceutical world, but commercial loggers are expensive black boxes that teach a student nothing. This project builds the same capability from open parts: an ESP32 with calibrated digital temperature probes, MQTT telemetry to a live dashboard, local SD-card buffering for connectivity gaps, and instant excursion alerts. It demonstrates the full IoT loop — sense, publish, visualize, alert, recover — on hardware cheap enough for a student bench.

1. The two DS18B20 probes are placed at the monitored points (e.g. inside the cool box and near its door); the ESP32 reads both over the 1-Wire bus each sampling interval.
2. The firmware timestamps each reading via NTP, packages probe ID, temperature and battery voltage into a JSON payload, and publishes it over MQTT with QoS 1.
3. The web dashboard, subscribed over MQTT WebSockets, appends each reading to the live chart inside the shaded 2–8 °C safe band and updates the statistics panel.
4. If a reading falls outside the configured safe window, the firmware immediately publishes to the alert topic and the dashboard raises a visible excursion flag with the timestamp.
5. If Wi-Fi or the broker is unreachable, readings are written to the SD card's CSV log instead; on reconnect the backlog publishes in order and the dashboard backfills the chart.
6. Between intervals the ESP32 enters deep sleep, waking on its timer — the current budget in the report shows how the sampling interval trades freshness against battery life.
7. A front-panel button forces an immediate reading and publish, useful for demonstrating the full sense-to-dashboard loop live in front of examiners.

Tech stack:

• ESP32 DevKit (Wi-Fi + Bluetooth, deep-sleep capable)
• 2x DS18B20 waterproof digital temperature probes (±0.5 °C per datasheet)
• PubSubClient MQTT library with QoS 1 publishing
• Mosquitto local broker or public HiveMQ broker for demos
• Web dashboard (MQTT over WebSockets, live charting)
• MicroSD card module for offline CSV buffering
• NTP time synchronization for trustworthy timestamps
• Arduino IDE / PlatformIO firmware with configurable thresholds

• [Dual DS18B20 temperature probes] Two waterproof digital probes (±0.5 °C accuracy from –10 °C to +85 °C per the Maxim datasheet) monitor, for example, fridge interior and door-ambient, so gradients and door-open events are both visible.
• [MQTT telemetry] Every reading is published as a timestamped JSON payload over MQTT (QoS 1) to a topic hierarchy like coldchain/unit01/temperature, using either the public HiveMQ broker for demos or a local Mosquitto broker.
• [Live web dashboard] A browser dashboard subscribes over MQTT WebSockets and plots both probes against a shaded 2–8 °C safe band, with current values, min/max/24 h statistics and connection status.
• [Excursion alerts] The moment either probe leaves the configured safe window, the firmware publishes an alert message and the dashboard flags it; the alert topic is ready to wire into email/SMS gateways as an optional customization.
• [SD-card offline buffering] When Wi-Fi is unavailable, readings are appended to a CSV log on the SD card with NTP-synced timestamps and bulk-published on reconnect, so no gap goes unrecorded.
• [Deep-sleep power design] The ESP32 sleeps between configurable sampling intervals (default 60 s), giving multi-day operation from a small Li-ion cell — the report includes the duty-cycle current calculation.
• [Configurable thresholds and intervals] Safe-band limits, sampling interval and MQTT topics are set in a config header (and documented), so the same logger adapts from vaccine fridges to food transport demos.

• Assembled cold chain logger prototype (ESP32, probes, SD module, battery, enclosure)
• ESP32 firmware (Arduino sketch) with MQTT, SD buffering, deep sleep and alert logic
• Web dashboard source (subscribe, chart, statistics, excursion flags)
• Mosquitto broker setup notes plus public-broker demo configuration
• Circuit schematic and wiring diagram with probe placement guide
• Project report PDF (background, cold chain standards, firmware design, power budget, test procedure)
• PPT presentation for final review
• Viva Q&A preparation document (MQTT QoS, 1-Wire, NTP, deep sleep, cold chain norms)
• Setup and demo guide (broker setup, dashboard launch, demonstrating an excursion safely)

• The logger depends on Wi-Fi for live telemetry; without connectivity it records locally to SD card but the dashboard and alerts go stale until reconnect — the report documents this behavior explicitly.
• The public HiveMQ broker is convenient for demos but is a shared public service: it must not carry sensitive or production pharmaceutical data, and broker availability is outside the project's control.
• DS18B20 accuracy is ±0.5 °C per the datasheet; the logger is an academic monitoring prototype, not a pharma-validated instrument, and the report states it cannot replace certified cold chain qualification.
• Timestamps need NTP at boot: without internet at startup, logged times fall back to relative millis until the first successful sync, which the firmware flags in the CSV.
• Battery life is a design-target calculation from the duty cycle, not a measured multi-day field test; the buyer verifies runtime on their own cell per the included procedure.
• The 2–8 °C default suits vaccines; frozen goods (–18 °C and below) are outside the DS18B20's sweet spot and outside this project's scope.