Theoretical Issues And Applications Of Mechanism Design In Service Overlay And Cognitive Radio Networks

Computer networks have developed rapidly in recent years coupled with the socialization of their behavioral patterns, i.e. distributed individuals in networks decide on their own actions based on their strategies. Such strategic distributed systems become more and more important as network service model reforms. Incentive and mechanism design are two primary research topics of this field. The incentive of a system is an important problem which perplexes many scientists as well as supervisors. Meanwhile, a new methodology of mechanism design can ensure that the price quoted by the service provider reflects the real cost of the service. Mechanism design borrows ideas from economics and game theory in order to describe the strategic agents and provide incentives to them, leading to optimization of the whole system generated from maximization of each selfish agent's profit in the system. In this research, we apply mechanism design to varieties of strategic distributed systems. We aim to solve an important but pendent theoretical problem, viz. behavioral pattern of capacity-aware agents in service overlay networks. We also design mechanisms for some practical applications, including task allocation in distributed service overlay designed for service agents, resource auction on web 2.0, Spectrum selection of nodes in cognitive radio network. The results are applicable to resource allocation in computer networks to maximize social benefit.We first study mechanism designs of resource management in service overlay, where services are provided by strategic agents. Usually, resources in distributed systems are limited. However, the current mechanism design does not take the capacity of agents into consideration. Traditionally, the Vickrey-Clarke-Groves (VCG) mechanism has been the only method to design protocols so that each strategic agent will follow the protocols for its own interest to maximize its benefit. We show that the VCG mechanism is not truthful anymore when the capacity of agents is limited. Thus, we have designed, based on non-uniform prices, a new capacity-aware mechanism which subsidizes the service agents so that each agent maximizes its profit if it truthfully reports its cost. Mechanisms for two widely used pricing models are designed and evaluated.Then, we turn to mechanism designs of spectrum sharing in cognitive radio (CR) wireless networks, which is one of the main challenges of open spectrum usage. Therefore, Game theory has been exploited for analysis and design of CR spectrum access schemes. However, most of the current designs make the assumption of cooperative behaviors among all the users, because non-cooperative spectrum sharing will lead to worse performance. In this paper, we specifically study how to conduct efficient distributed channel allocation in selfish CR wireless networks. We assume that each secondary user will incur a cost for sharing the channel which serves for primary users. We adapt and apply the famous VCG mechanism to this problem. The main contributions of ours are two-folds. First, for game theory, we show that the VCG based mechanism can be fully applicable for CR spectrum sharing problem just as it is successfully deployed for routing optimization problem. Second, for CR wireless networks, we present an efficient algorithm which will lead to a better performance. Analysis and discussion are provided.