2026
WardWatch
Camera-free indoor positioning for hospital wards over BLE, an ESP-NOW mesh and MQTT
- Position accuracy
- ~0.45 m
- Zone detection
- 85%
- Leader re-election
- < 2 s
- BLE
- ESP-NOW
- MQTT
- Kalman filter
- Trilateration
- Node.js
Summary
WardWatch tracks patient movement in a ward without cameras, aimed at dementia patients prone to wandering. A BLE wearable is ranged by anchor nodes, zone nodes estimate position, and the result travels over an ESP-NOW mesh to a gateway that publishes to a dashboard.
Problem
CCTV and manual monitoring raise privacy concerns, need constant supervision, and do not give continuous real-time location.
Approach
Anchors smooth RSSI readings with a Kalman filter; zone nodes estimate position with weighted least-squares trilateration; a mesh with leader election routes data to a Pico W gateway, which deduplicates and publishes over MQTT to a Node.js dashboard.