The Asynchronous Backtracking Family

In the last years, the AI community has shown an increasing interest in distributed problem solving. In the scope of distributed constraint reasoning, several asynchronous backtracking procedures have been proposed for finding solutions in a constraint network distributed among several computers. They differ in the way they store failing combinations of values (nogoods), and in the way they check the possible obsolescence of these nogoods. In this paper, we propose a unifying framework for asynchronous backtracking search. We discuss the choices that can be made to obtain a correct and complete algorithm. These choices can lead to already known procedures, or to new algorithms. Our framework permits to better understand the basic steps of these procedures, and to highlight their differences and similarities. We present original techniques that can be added to these algorithms to improve their behavior and to better fit the features of distributed networks. Finally, experiments permit to assess the relative performances of the different versions of the algorithms, and to show the benefit of using some of the proposed improvements.