PPM Revisited with New Idea on Escape Probability Estimation