designing sleep monitors prototype 1

A few years ago, I was looking to improve my sleep. Measuring it would be a start. However I was not satisfied with any sleep monitors on the market.


I needed a non invasive, no "additional daily routine" solution, ready to trade accuracy for maximum convenience.
So back in late 2010 I started work on my first prototype, contactless sleep monitor based on actigraphy.

A year before I had stumbled upon one of Steve Tarzia's projects: he developed a screensaver detecting movement using a computer speakers and microphone in the same fashion as a sonar.
He is kindly providing the source here so I adapted his code, loaded on a netbook to detect and count movements throughout a night sleep.
The echos probes are sent on the 20 kHz range, unaudible for most humans and it runs on a netbook so the noise of the fan is minimum, 2 conditions required to avoid disrupting sleep.
This is the result:

The higher blue line denotes a threshold calculated based on average echoes.
Unfortunately it is quite hard to count accurately the movements as there is a lot of "noise", although in this case it worked quite well, it triggered the alarm 10min before the time set as it sensed movement and it assumed I was in a light phase of my sleep.

Unsatisfied with this "software" approach, issues with signal/noise and requiring to set the netbook every evening, I started working with microcontrollers and sensors, that leads to my second prototype described in the next article.