Basically the idea is to
let a user to record a pattern of some repetitive event coresponding to some
physical or non-physical entity, upload it to servers (say, Google's), apply an
expert algorithm over the uploaded pattern to get the characteristics of the
physical or non-physical entity. The algorithm is to be stored on the servers,
but it can be downloaded and run on the client, say, in
To illustrate the idea:
Imagine someone has bought an air conditioner. And he wanted to monitor its working parameters from time to time. A new airconditioner has a certain vibration pattern. Any even potential failure would show in the changed vibration pattern. So he would get his mobile device, attach a vibration sensor to the air conditioner, measure the pattern, send it to the servers. The algorithm would compare the pattern with the patterns supplied by the air conditioner manufacture, and the expert statement would be generated about the health of the air conditioner motor and compressor parts. The sensors can be universal to be attached to many different devices, or stationary, that is specific to a device and placed (or even built-in) in the device. The communication between a handheld and the sensor can be wireless, infrared or of any other kind.
The pattern can be of any physical or non-physical origin: vibrational, sound, electromagnetic etc, etc (even behavioral). A proper pattern can talk a lot about the underlying processes.
The scope of the entities to be monitored is virtually unlimited, just to name a few:
- Vehicles: cars - many points to monitor, bicycles, scateboards;
- Home applience,
- Industry-related set-ups: wind or steam tourbines, any kinds of engines, compressors etc;
- Bridges, roads under the repetetive load;
- Human body - a lot of ties with the specific charecteristics;
- etc, etc;