A Cooperative Multi-Agent Approach to Constrained Project Scheduling

Abstract

Different approaches to a problem generally produce distinct algorithms that have significantly different performance characteristics in varying problem environments. In most instances, a single algorithm is chosen to be applied throughout the solution process of a given problem, thus losing the opportunity to exploit the desirable features of other methods. An Asynchronous Team (A-Team) is a software organization that facilitates cooperation amongst multiple algorithms so that together they produce better solutions than if they were acting alone. This paper explores the feasibility of applying the A-Team approach to the practical problem of resource-constrained project scheduling with cash flows (RCPSPCF). This problem investigates the scheduling of precedence-related and resource-constrained activities to maximize the Net Present Value of the expenses incurred and payments made over the duration of a project. This is a complex combinatorial optimization problem which precludes the development of optimal schedules for projects of practical size. We embed several simple heuristics for solving the RCPSPCF within the iterative, parallel structure of A-Team which provides a natural framework for distributed problem solving. Preliminary results on several randomly generated project networks show that the combination of multiple, simple heuristics perform very well when compared to many single pass, complex optimization-based heuristics proposed in the literature.