Andrew B. Whinston and Dale O.Stahl
Award Number: IRI-9509914
Previous funding from NSF was utilized in a stream of research that formulated the theoretical constructs underlying the economic model for resource management in computer networks. We developed a series of simulations to test the model under various scenarios and further refine it, culminating in a well-tested economic resource management mechanism. The first stage developed an economic model of decentralized computing where resources would be allocated in an incentive compatible manner, by differentiating access based on priorities. Thereafter, a simulation of a public network such as the Internet was developed for studying the feasibility of the model and to investigate equilibria. Once the basic simulation environment was developed, various pricing mechanisms were prepared and tested. Further, various response predicates were developed and tested to monitor performance and utilization of resources. As a crucial requirement for a practical network resource management system which used demand information, demand forecasting methods which depended only on observed user behavior were developed and successfully employed in pricing. The performance of the network was further tested under various possible market structures. Through extensive simulation runs, it was shown that optimal congestion pricing can result in enormous savings for the network economy, enhance performance and capacity utilization, lead to feasible equilibria, manage congestion with minimum computational complexity and provide better incentives for new capacity investments and rational usage.