Running a demo

The simulator plays every room computer of a small ward: 5 residents in rooms 11 to 14, room 13 with two beds, 29 sensors. The backend cannot tell it from real hardware.

cd edge/simulator
uv run simulator_gui.py   # buttons: a card per resident, a panel per room, the devices table
uv run simulator.py       # or the terminal; Tab completes commands

Only one simulator can run against a broker at a time; a second one is refused.

Scenarios

Scenario	Window	Terminal	What you should see
Stops breathing	Apnea on a resident card	apnea 1	No breathing after ~10 s, escalated after 30 s more
Low oxygen	Low SpO2	lowspo2 4	Low oxygen
One odd signal	Noisy	noisy 2	Nothing: one sign is not enough
A fall	Fall	fall 1	Possible fall for that resident only
Call bell	Bell	bell 5	A critical bell alert
Crying in a room	Cry on a room panel	cry 3	One soft Check every bed for the room
A sensor dies	Crash in the devices table	crash radar-002	Sensor offline, never No breathing
A new sensor	Plug in	plug radar 14	A pending radar in the control room

In the terminal, the number is the resident (1 to 5), not the room. After a scenario, Back to normal (normal 1) and wait about 20 seconds before running it again.

Let the ward run three minutes before the first scenario: the personal baselines need history.

The laptop microphone

cd edge/micsensor && uv run mic_sensor.py --room room12

Stay quiet for the first three seconds while it learns the room. Approve mic-laptop in the control room; a loud clap or call raises Check every bed for room 12. It measures loudness, not words.

Before a live demo

1. Stop every simulator and the server; reset the broker: docker compose rm -sfv mosquitto && docker compose up -d mosquitto.
2. Start the server, then exactly one simulator; approve all sensors, then room 13’s by hand (two beds).
3. Wait three minutes; check the station shows All clear and Live.
4. Keep the laptop lid open and the charger in: a sleeping laptop makes every sensor look offline.