Decentralized Multi-Project Scheduling via Multi-Unit Combinatorial Auction

In industry, many problems are considered as the Decentralized Resource-Constrained Multi-Project Scheduling Problem (DRCMPSP). Existing approaches encounter difficulties in dealing with large problems while preserving information privacy of project agents. In this paper, we propose a novel approach to solve DRCMPSP based on the multi-unit combinatorial auction, which can efficiently solve the problem without violating information privacy. It adopts a greedy resource allocation strategy with fixed resource cost to simplify computation required for project (bidder) and auctioneer agents. In addition, a bid modification step is incorporated to allow project agents to better utilize resources. Analysis and empirical results indicate that our approach outperforms state-of-the-art decentralized approaches in minimizing average project delay, and scales well to large problems with thousands of activities from tens of projects.