A Distributed Genetic Algorithm Solution to the Boolean Satisfiability Problem
Mohammad Khatim Hasan, Bala P. Amavasai, J.R. Travis · Staffordshire Online Repository (Staffordshire University) · 2005
This paper attempts to improve the solution of the NP complete Boolean Satisfiability (BSAT) problem by partitioning the task into three sub-tasks and distributing them over an experimental 3-node Distributed Computing System (DCS). A genetic algorithm (GA) has been used to consider multiple feasible solutions. The GA based algorithm is applied to the standard BSAT benchmarks on a single computer and on DCS configuration using non-optimised and optimised executables. The task is coarsely partitioned and distributed over the DCS using the Simple Object Access Protocol (SOAP) technology. The results reveal that the DCS enabled solution exhibits better performance than a single computer configuration for non-optimised GA code. However, no clear correlation could be identified between the single computer and the DCS for the optimised version of the GA search. The main contribution of this investigation is the design of a GA based solution to the BSAT problem for DCS.