Performance of probabilistic approaches to Steiner tree problem

The rectilinear Steiner tree problem is of essential importance to the electronic design automation. While many different heuristics have been derived to deal with this problem, there is room for improvement of algorithms in terms of the speed and/or solution quality. This paper proposes several algorithms, which represent a class of probability-based approaches for Steiner tree designs, and shows their performance compared with the state-of-the-art heuristic.