Multi-agent technology for scheduling and control projects in multi-project environments. An Auction based approach

Most existing methods for scheduling projects are based on centralized or hierarchical decision making using global models. So far, these methods have not produced the desired results, especially in highly dynamic multi-projects environments. In this study, we investigate a new method based on multi-agent systems and a combinatorial auction mechanism to allocate resources for the projects tasks. In combinatorial auction the bidders demand a combination of dependent goods with a single bid. We consider basically two types of agents, projects and resources. Projects are the bidders and resources are the goods. An auction mechanism based on Lagrangianbased decomposition is designed to achieve efficient allocation resources. Our approach allows manage some traditional aspects of multi-project environments: dynamic addition of resources or projects, changes of the resources capabilities, allocation flexibility, etc.