Java Development

[tkroll]

Hello,

When using the LocationListener, I get lat/lng coords that vary slightly, sometimes greatly,

when the device is stationary, especially with low intervals.

What are the best practices/techniques to make sense of this data? Should I just pick an

arbitrary fix at the interval I require? Or should I somehow average the coordinates?

Thanks

[sonicboomboy]

you may use Criteria object to deal with gps fix:

http://www.blackberry.com/knowledgecenterpublic/livelink.exe/fetch/2000/348583/800332/800703/How_To_...

also this one:

http://supportforums.blackberry.com/t5/Java-Development/Best-practices-for-designing-GPS-application...

[tkroll]

I'm familiar with these documents, thanks. They do not address my question.

Regardless of the criteria, there is a certain amount of "jitter" between fixes,

even when not moving. My question is how to deal with this when calculating

a stationary fix.

[sonicboomboy]

personally what I do in my locationUpdated method:

 public void locationUpdated(LocationProvider provider, Location location) {
        try {
            if (location.isValid()) {
                satellites = Double
                        .valueOf(
                                getNextToken(location
                                        .getExtraInfo("application/X-jsr179-location-nmea"))[7])
                        .intValue();

                if (satellites > 0) {

                    QualifiedCoordinates c = location.getQualifiedCoordinates();
                    currentLongitude = c.getLongitude();
                    currentLatitude = c.getLatitude();
                } 

            } else {
                notifySateliteListener(0);
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
    }

 So I presume I use just basic stuff: checking if location is valid and getting qualified coordinates. Are you doing in similar way?

[peter_strange]

In my experience, jitter is common and to be expected, you do need to deal with the problem of dud values in the data supplied. 

I would ignore any results that have fewer than 4 satellites, I have even seen significant jitter with 4 and sometimes 5 satellites but locations with 6 satellites seem pretty solid. 

One approach if you can't wait for a reading with 6 satellites, is to take a number of readings and average these. 

Another is to ignore changes that are within the accuracy supplied.  Remember that the accuracy value is a 'confidence' measure, not an error measurement. 

[tkroll]

This is the exactly the kind of info I was looking for. I'm just starting with the GPS stuff

and want to forumulate my plan of attack before I start coding.

Can I collect some approaches here and have everyone comment?