| The flat-rate pricing scheme of the current Internet, based on the best-effort service model, has resulted in moral hazard problem on the consumer side and congestion problem on the network side. Given the congestion, there has been wide dissatisfaction with the perceived performance and therefore, customers' demand for high-quality premium service has stimulated the development of various protocols to provide differentiated QoS(Quality-of-Service) levels for the future Internet.; Given this context, we develop a canonical model for a single node of two-class (premium class/economy class) integrated-services network, with the following objectives in mind: First, to technically introduce premium-class service, in addition to best-effort service; Second, to implement usage-based pricing scheme for the efficient utilization of network resources; Third, to develop ISP(Internet Service Provider)'s pricing and capacity-planning strategies.; In this dissertation, we first develop a stochastic network model and perform steady-state analyses to derive the relevant QoS parameter for each class. With these parameters as valuable inputs, we develop a demand model describing customers' choice behavior. Then we integrate these two models and solve fixed-point problems numerically, given the static price vector. We also develop systematic ways to check the stability of a fixed point in equilibrium and derive convergence conditions. Ultimately, we find an optimal price vector maximizing an ISP's revenue. Finally, we apply our models to real-world problems in the Internet service market, considering monopoly situations as well as duopoly situations in which some game experiments are conducted. |
