I might be beating a dead horse, but I still don't quite see what you hope to accomplish by doing this. There's nothing wrong with the pngs - it's how the signal is calculates so that it knows which png to pick. If you modify the lower-signal pngs to match the ones with more signal, how do you expect to know the difference between when you really have signal and when you don't? I'm certainly not trying to be mean or anything, but I just want to caution you that it's designed the way it is for a reason. That being said, this is a good exercise for learning how to modify things in framework-res. Just remember to make a nandroid backup, first

. I'll let evomattnc take it from here. He's doing a good job explaining it.