Scaling a software application is a well-understood challenge. But scaling a system that bridges physical hardware with cloud infrastructure introduces a unique set of constraints: power stability, network jitter, and the ever-present risk of "offline" operation.
When we built the IoT Attendance System, the goal wasn't just to record "check-ins." It was to create a robust, enterprise-grade appliance that remains reliable even when the world around it isn't.
The Edge: ESP32 Firmware Architecture
At the heart of the device is an ESP32, a powerful dual-core microcontroller with integrated Wi-Fi. However, "integrated Wi-Fi" doesn't mean "always connected."
I designed the firmware with a State-Machine Architecture to handle connectivity gracefully:
- Phase A (AP Mode): If no Wi-Fi is found, the device becomes a Wi-Fi hotspot with a captive portal, allowing users to configure network settings from their phones.
- Phase B (Web Setup): Once connected, a secure local web interface handles device registration and admin credentials.
- Normal Operation: The device switches to a low-power polling mode, monitoring RFID and Biometric (Fingerprint) sensors.
Solving the "Offline" Dilemma
In many parts of the world, internet stability is a luxury. If the cloud is unreachable, the attendance system must still work.
We implemented a Local-First Persistence Model:
- SD Card Logging: Every scan is immediately written to a local
scans.log file on an SD card with a valid NTP timestamp.
- Sync-Back Logic: When the Wi-Fi connection is restored, a background process (using FreeRTOS semaphores) batches and pushes these offline scans to the cloud API.
- Local User Cache: The device periodically downloads a "known users" list. This allows the OLED display to show a person's name (e.g., "Welcome, Shifaul!") even during an internet outage.
Cloud Integration: Secure & Scalable
The firmware communicates with a Node.js/Express backend through a secure REST API. To ensure security at the edge:
- MAC-Based Auth: Initial registration uses the device's unique MAC address.
- JWT Bearer Tokens: Once registered, the device uses a persistent access token for all scan data.
- Remote Commands: The cloud can push commands (REBOOT, FACTORY_RESET, UPDATE_CACHE) to the device via the response body of standard polling requests, creating a lightweight "push" mechanism without the overhead of MQTT.
Hardware Diagnostics
To prevent field failures, I included a Hardware Diagnostic Summary on boot. The system checks the electrical state of all SPI pins (RFID, SD Card) and DHT sensors, logging brownout reasons or watchdog resets to the serial console.
Serial.println("REBOOT REASON: ");
esp_reset_reason_t reason = esp_reset_reason();
switch (reason) {
case ESP_RST_BROWNOUT: Serial.println("BROWNOUT (Power Issue)"); break;
case ESP_RST_TASK_WDT: Serial.println("Watchdog (Task Hang)"); break;
// ...
}
Conclusion
The IoT Attendance System is a bridge between the physical and digital worlds. By prioritizing local persistence and designing for failure at the edge, we've built a system that is more than the sum of its parts. It's a reliable enterprise appliance that proves IoT can be as stable as any "traditional" software stack.
Innovation happens at the intersection of bits and atoms.
Want to dive deeper into the firmware? Check out the IoT Attendance project details.
