This provider has features allowing you to control the behaviour of background-tracking, striking a balance between accuracy and battery-usage. In stationary-mode, the plugin attempts to decrease its power usage and accuracy by setting up a circular stationary-region of configurable stationaryRadius. iOS has a nice system Significant Changes API, which allows the os to suspend your app until a cell-tower change is detected (typically 2-3 city-block change) Android uses LocationManager#addProximityAlert.
When the plugin detects your user has moved beyond his stationary-region, it engages the native platform's geolocation system for aggressive monitoring according to the configured desiredAccuracy, distanceFilter and locationTimeout. The plugin attempts to intelligently scale distanceFilter based upon the current reported speed. Each time distanceFilter is determined to have changed by 5m/s, it recalculates it by squaring the speed rounded-to-nearest-five and adding distanceFilter (I arbitrarily came up with that formula. Better ideas?).
(round(speed, 5))^2 + distanceFilter
is calculated as the square of speed-rounded-to-nearest-5 and adding configured #distanceFilter.
(round(speed, 5))^2 + distanceFilter
For example, at biking speed of 7.7 m/s with a configured distanceFilter of 30m:
=> round(7.7, 5)^2 + 30
=> (10)^2 + 30
=> 100 + 30
=> 130
A gps location will be recorded each time the device moves 130m.
At highway speed of 30 m/s with distanceFilter: 30,
=> round(30, 5)^2 + 30
=> (30)^2 + 30
=> 900 + 30
=> 930
A gps location will be recorded every 930m
Note the following real example of background-geolocation on highway 101 towards San Francisco as the driver slows down as he runs into slower traffic (geolocations become compressed as distanceFilter decreases)
Compare now background-geolocation in the scope of a city. In this image, the left-hand track is from a cab-ride, while the right-hand track is walking speed.
NOTE: distanceFilter is elastically auto-calculated by the plugin: When speed increases, distanceFilter increases; when speed decreases, so does distanceFilter.