Info found via Geeks are Sexy and New Scientist
Using air pressure to generate intrusion alarms is nothing new, but this is an interesting integration. Shwetak Patel at the Georgia Institute of Technology are working on HVAC applications and have discovered that sensors put on the building's central filter have the potential to accurately signal pressure changes in specific places of the building.
It's much more effective with closed rooms, and intruders opening and closing doors but pressure changes are seen regardless of door state changes and even if the HVAC system is on or off. The key use for this technology is usually to automatically control building automation systems but security may be another application. (One good side effect of the increase of security awareness is scientists considering security in their research)
Patel describes this method as an infrastructure mediated sensing approach rather than the usual distributed direct sensing approach. I believe the shift to using infrastructure to sense occupancy and movement is a trend worth keeping up with for many reasons. It's a tempting idea, especially since it would be much more cost effective in theory to place sensors on central filters than wire them up into every room. However - the stats I've seen are not nearly accurate enough to rely on for security intrusion, but the technique is very promising.
"After a lengthy training process, software could use the sensor data to identify which particular door was active in a 10-20 room house 75-80% of the time. It could also detect when a person walked through a particular door 60-75% of the time.
"When a door movement event occurs, the pressure differentials are slightly different at each of the sensors," says (Shwetak) Patel. Those small differences are used to calculate exactly where in the system the pressure shift originated."
The stats are reduced by 20% if the system is not running, so you can see that this is no-where ready for prime-time.Two general notes here:
1. Sensing pressure changes on a central filter and pinpointing the location of the pressure changes is pretty amazing. Although this is not accurate enough yet to use as intrusion detection, maybe it can be a usable back-up on systems that should have a fall-back detection method that's not easy to identify and circumvent?
2. Reverse the motivation here and consider how useful that central point of control and monitoring is. It's no secret that HVAC systems can be one of the most useful methods of deploying chemical and biological agents, but I'm sure smart people with malitious intent can think of many other uses for that central filter location.
