Reading acceleromenter data is quite easy on the Iphone but using them effectively to simulate a virtual joystick is a different story. I see that accelerometer data are quite noisy and the average module of each axis accelleration in a rest position is not comparable. I wonder if the most effective way to use accelleromenter is to look at the derivative of each measurement to identify the moment at which there is a change in direction. Is there anyone that can suggest an effective way and/or strategy to use accellerometer to simulate a joystick? Thanks, Luca