Close

July 2015

A project log for WiFi Sensors

A range of sensors that use WiFi to communicate

michael-otooleMichael O'Toole 07/30/2015 at 01:380 Comments

I've been busy with some long standing projects but have managed to do some testing with inexpensive piezo sensors in order to determine if they are suitable (and reliable) as the main sensing device for the project...

Inertia Sensors:

Analysing Inertia Sensor Signals:

Having established that we will receive some signal data for even the slightest vibration it follows that we must analyse each signal both in amplitude and time (relation to any previous signal)...

A single high amplitude might signal a gross attack (glass breaking or force entry for example) and if above a predetermined level we might generate an alarm straight away... A much more common occurrance might be several small vibration in a short period (gently prying or forcing a window or door without making noise), here the analysis is a bit more tricky, after all we don't want an alarm if it's raining...

On a side note, some advanced alarm might even include a rain sensor, wind gauge and thunder sensors to augment inertia sensor analysis... we could add that too...

Implement criteria in software

On first detecting a signal we measure its amplitude and if above a predetermined level we signal an alarm condition but if below this level we record (count) and start a timer.

Should the signal be below the gross level, we start a timer and count each successive signal up to a predetermined number... If that number is exceed during the sensor time period we trigger an alarm condition...

Setting the sensors sensitivity can also be accomplished with software (during the install process) and this help remove some of the guess work but of course you can also modify sensitivity via the web interface...

A second line of defence:

As all WiFi sensors send data to a central computer it also follows that we can further analyse signals in relation to each other. For example, if several sensors in different locations appear to indicate a break in at the exact same time (within milliseconds) , we can over-ride the alarm as it's almost certainly something natural, thunder for example...

A simplified Approach:

Allowing for the fact that a central computer can do all the analysis, it follows that we could actually simplify the WiFi sensors job and instead simply pass the detection event and the level back to the central computer (after first eliminating extraneous signals locally) and not have to write more complex code...

Notes to add...

Note the sensor also signals battery low using the ADC pin...

Discussions