RFID Access Control

SPI to the reader, TCP/IP over WLAN to the server, with a lean text protocol on TCP in between: the client reports the UID it read, the server answers with the decision, and the display and status LEDs mirror it straight away. In normal operation the server decides: the client sends the UID, the answer comes back as ALLOW or DENY, readable in the terminal on the left. Take that connection away and it becomes the only genuinely interesting question in the project.

The server (left) checks the lockout window and access list and logs every decision; the ESP32 client (right) tries WLAN and the server first, and only falls back to the local list once that fails. On connection loss the client falls back to OFFLINE, checks a list in its own flash and retries the connection in the background, which makes ONLINE and OFFLINE two operating states with automatic reconnect. A lock that stops working without a network isn't a lock, even though that costs a deliberately accepted inconsistency: a permission revoked moments earlier can still be honoured offline. It isn't a security system either, since a MIFARE card's UID clones trivially and the link stays unencrypted. It's finished all the same, an exercise in system integration from bus through network and state logic to the enclosure.

A ten-minute walkthrough recorded for the supervising professor, stepping through every feature the system handles.