The experiment utilized a Yandex TV Station (model YNDX-00092) equipped with a MediaTek MT7663 radio module. The goal was to detect human movement by analyzing Wi-Fi signal fluctuations between the TV and a router.
The developer initially aimed to use Channel State Information (CSI) for high-precision detection but found no public Android API or accessible driver interface to extract these metrics from the MT7663 chip. Attempts to access internal ICAP mechanisms and beamforming profiles via binary protocols did not yield usable real-time channel data.
The final prototype relies on Received Signal Strength Indicator (RSSI) data, which is accessible via the standard Android API. The system monitors signal drops—typically from -30 dBm to -32–35 dBm during movement—and uses a calibration process involving a 30-second background measurement followed by three test passes.
For notifications, the developer implemented the ntfy service, which uses AES-256-GCM encryption for data exchange. To manage API quotas, the system limits movement alerts to one per minute and sends status updates every 30 minutes, with all events logged locally on the TV.
The project source code and a test APK are available on GitHub. While the current setup functions as a motion detector, the developer notes that it does not reliably track static presence, as RSSI levels may return to baseline even if a person remains in the room.
