I recently saw a tweet from the Gottardo lab about their new software "Probabilistic Inference for Nucleosome Positioning with MNase-Based or Sonicated Short-Read Data":
I realise this is not an informed suggestion, but i rate their work very highly, e.g. PICS for ChIP-seq analysis.
Old question, but I've recently tried PING 2.0 and am also trying DANPOS. Though the installation was a little hairier for DANPOS (a set of python scripts that depends on python package rpy2, which requires an oldish version of R), the method is easier for me to understand. Also works for comparison of nucleosomes in different samples, whereas PING is more designed to find nucleosomes in an individual sample.