What's the most battery-efficient approach of using LocationClient to periodically get updates?

Question

I am thinking about having two separate alarms to gather a user's location data every hour, one that goes off every 59 minutes to "connect" the client and a second to actually get the location and then subsequently disconnect the client.

In terms of battery life, is there anything else I should consider doing if getting the user's location will be the primary drain of the app? Or, is there a different approach to having two alarms? I originally only had a single alarm, but performing a (!mLocationClient.isConnected) then connect check does not give the client enough time to connect.

Thanks for your insight.

The two alarms would go off something like this:

private int PERIODIC_UPDATE = 60000*60;  //gets location and disconnects every hour
private int PERIODIC_RECONNECTION_UPDATE = 60000*59;  //connects 1 minute before getLocation call

    Timer toReconnect = new Timer();
    toReconnect.schedule(new TimerTask() {

        @Override
        public void run() {
            mLocationClient.connect();
        }
    }, 5000, PERIODIC_RECONNECTION_UPDATE);

    Timer theTimer = new Timer(); 
    theTimer.schedule(new TimerTask(){
        @Override
        public void run() {
            try {
                if(!mLocationClient.isConnected()) {
                    mLocationClient.connect();
                    //This will not have much affect because cannot so quickly, will remove. 
                }

                Location theLocation = mLocationClient.getLastLocation();
                if(theLocation!=null) {
                    checkPostLocation(theLocation); 

                    mLocationClient.disconnect();
                }
            } catch (Exception e) {
                e.printStackTrace();
            }   
        }}, 5000, PERIODIC_UPDATE);

Answer 1

See the "Receiving Location Updates" section on the Android developer docs for a detailed discussion of this topic with the new Fused LocationProvider:

http://developer.android.com/training/location/receive-location-updates.html

This gives you the option to register an Intent with a LocationListener to the Fused LocationProvider that is automatically triggered by the internal Google Services framework when it is considered "efficient" to do so. I would trust that this framework has a much greater potential to optimize power usage, since it has a lot more knowledge of what else is going on in the system vs. an app-registered Timer.

Here are your options for registering a listener with different power priority levels, which will result in different levels of battery drain, as listed in the above docs:

• PRIORITY_BALANCED_POWER_ACCURACY - Used with setPriority(int) to request "block" level accuracy. Block level accuracy is considered to be about 100 meter accuracy. Using a coarse accuracy such as this often consumes less power.
• PRIORITY_HIGH_ACCURACY - Used with setPriority(int) to request the most accurate locations available. This will return the finest location available (and the greatest potential for energy drain).
• PRIORITY_NO_POWER - Used with setPriority(int) to request the best accuracy possible with zero additional power consumption. No locations will be returned unless a different client has requested location updates in which case this request will act as a passive listener to those locations.

Answer 2

Do you actually need to track the user?

If it's just about UI, then use getLastKnownLocation(PASSIVE_PROVIDER) and you should get something semi-accurate assuming they used location services on their phone somewhere else.

If you need to actually triangulate the user, realize the different providers use different battery. Passive < Network < GPS.

The more you locate the user, the more battery with GPS taking the most battery and time.

Start the service by intent one a schedule, 1 hour or whatever, only one service necessary. Only live for a maximum of 1 minute (or less), listen on all Location providers. After the minute or accuracy is good enough, you save the result and shut down the service.