A Combinatorial Auction Negotiation Protocol for Time-Restricted Group Decisions

Abstract

This paper focuses on multi-agent contract formation by automated negotiation. Commonly individuals are not willing to share information or cooperate and negotiation protocols may give way to unwanted strategic behavior. Socially beneficial contract agreements require a lot of negotiation time. Furthermore, possible interdependencies of contract items lead to complex contract spaces which restrain contract agreements. Therefore, we propose a novel negotiation protocol applying combinatorial auctions for contract formation which consider interdependencies and yield a rapid decision rights allocation. Additionally, this market-based approach utilizes Vickrey-Clarkes-Groves-mechanisms which may lead to truthful preference uncovering and information sharing through bids. However, combinatorial auctions have a computational drawback: winner determination is $\mathcal{NP}$-hard. In simulation experiments, two approximation algorithms as well as an optimal computation are tested in comparison with an established negotiation protocol. The results show that our protocol yields an effective solution and requires very short run time.