Pricing communication networks: optimality and incentives

Network pricing is a cross-disciplinary research area, which requires deep under- standing of both networking technology and microeconomics. The goal of network pricing is to achieve satisfied network performances by allocating the scarce resource to satisfy different users’ qualities of services while keeping in mind the incentives of different network entities. Proper design of pricing schemes is indispensable to the operation and management of communication networks. In this thesis we divide network pricing into four categories: static optimization-oriented pricing, dynamic optimization-oriented pricing, static profit-driven pricing, and dynamic profit-driven pricing. The first one is well studied in the literature, and our focus will be on the latter three categories. For each category, we illustrate the key design challenges and insights through a concrete networking example.First, we investigate the issue of static profit-driven pricing. We consider a revenue maximization problem for a monopolist service provider, and discuss how to set incentive-compatible prices to induce proper allocation of limited resources among different types of users. We capture the interaction between the service provider and users through a two-stage Stackelberg game with both complete and incomplete information. With complete information, we study three pricing schemes: complete price differentiation, partial price differentiation, and no price differentiation. We characterize the trade-offs between the performance and complexity of different schemes. With incomplete information, we show that it is still possible to realize price differentiation, and provide the sufficient and necessary condition under which an incentive compatible price differentiation scheme can achieve the same revenue as the best scheme with complete information.Then we investigate the issue of dynamic profit-driven pricing. We consider a general resource allocation and profit maximization problem for a cognitive virtual mobile network operator. Dynamics of the cognitive radio network include dynamic user demands, unstable sensing spectrum resources, dynamic spectrum prices, and time-varying channel conditions. In addition, we also consider multiuser diversity and imperfect sensing technique so that the network model is more realistic. We develop a low-complexity on-line control policy that determines pricing and resource scheduling without knowing the statistics of dynamic network parameters. We show that the proposed algorithm with dynamic pricing can achieve arbitrarily close to the optimal profit with a proper trade-off with the queuing delay.We later investigate the issue of dynamic optimization-oriented pricing. We consider a node-capacitated multicast network with time-varying topology. By utilizing network coding, we design a dynamic pricing scheme that can achieve arbitrarily close to maximum network utility in a distributed fashion, while maintaining network stability. Moreover, we show that this algorithm is incentivecompatible, i.e., no matter what role a node plays in the network, the algorithm guarantees that the node has a non-negative profit. This result has practical importance for constructions for node-capacitated networks with multiple individual users (e.g., P2P networks), since it provides the proper incentives for individual nodes to join, stay, and contribute as relays in the network even if they have no interested contents.The results developed in this thesis highlight the importance of pricing in communication networks. Specifically, our results show that pricing can be used as an effective tool to achieve optimal network performances while providing proper incentives for all network entities. This not only helps us better understand network pricing, but also gives us insights on the design of network pricing schemes.