Performance Comparison of Two Algorithms for Task Assignment

In this article we investigate a new algorithm for solving the optimal task assignment problem. The assignment is based on Stone's "throughput" metric with the optimality criteria being the total cost for execution and interprocess communication. Two approaches studied are a new approach based on genetic algorithms and A*, a well known tree search algorithm for solving the same problem. We use the algorithm execution time as a performance criteria for the two algorithms. It is shown that the genetic algorithm techniques are more favorable than A* for larger search spaces while for smaller search spaces A* is preferred.