If Quicksort worst case performance is a problem then use Introsort (https://en.wikipedia.org/wiki/Introsort), which the paper fails to mention.