A smartphone uses more power the further away it is from a mobile base and the more obstacles are in its way as it reaches for a signal.

Additional power use by other activities could be factored out with algorithms, the researchers found.

Members of the computer science department at Stanford University created an app designed to collect data about power consumption.

Yan Michalevsky, Dan Boneh and Aaron Schulman, from the, along with Gabi Nakibly said their malicious app has neither permission to access the GPS nor other location providers.
"We only assume permission for network connectivity and access to the power data.

"These are very common permissions for an application, and are unlikely to raise suspicion on the part of the victim."
Activity such as listening to music, activating maps, taking voice calls or using social media all drain the battery but this can be discounted due to "machine learning", the report says.

"Intuitively the reason why all this noise does not mislead our algorithms is that the noise is not correlated with the phone's location," it says.

"Therefore a sufficiently long power measurement (several minutes) enables the learning algorithm to 'see' through the noise."

The tests were carried out on phones using the 3G network but did not measure signal strength as that data is protected by the device.

The team seem to have texted their Apps on Android, but looking at the report  there is nothing to suggest that the same attack vector would not work on other operating systems.