Decentralized Resource Allocation and Scheduling via Walrasian Auctions with Negotiable Agents

Auctions are well-studied mechanisms to solve decentralized resource allocation problems. In situations where complementarity exists among resources, combinatorial auctions may be utilized. Unfortunately, combinatorial auctions often incur hefty computational requirements, in that the underlying winner determination problem is known to be NP-hard. Although a number of efficient implementations are proposed (e.g., see Sandholm [11]), combinatorial auctions are still not very computationally efficient.In this paper, we are concerned with solving decentralized resource allocation and scheduling problems; and assuming feasible solutions exist, we ask how we might achieve convergence and good quality solutions under a tight computational budget.