For best results, you should use kCLLocationAccuracyBest. What you put in your remote filter depends on what error you are ready to deliver. Basically, you have to make decisions based on accuracy and availability. That is, during periods when the answer is not available with the best accuracy, what will you show?
One approach is to allow the phone to receive less accurate answers and, using the forecast of what happened the last time you had the information with the best accuracy, see if it makes sense to make sense.
That is, suppose I'm running 6 miles per hour north. You plot me by point-to-point, point-B, point-C ... then you will get a low-precision answer (maybe kCLLocationAccuracyNearest100Meters.) Look at the place where it says I’m figuring out, “could I get this spot from point-C, if I continued my current path, making reasonable adjustments for possible speed changes? " If so, then the new point is within limits. (If not, throw it away.) Then project from point-C onto my last known speed and find out where you think I'm probably ballistic. Save it as a ballistic dot-D.
Of course, you use the accelerometer to get some inertial feeling, how did I go, right? So, you can’t know the direction (you don’t know how the phone is pointing), but you can make a reasonable punch from a distance.
Using all this information, speak up the most likely place where, in your opinion, I probably am.
NOTE. When testing, do not just enter coverage areas with a good cell. See how your app works in the hills, away from cell phones. Many people enjoy cycling and running through these areas!
Olie
source share