A study of the impacts of pricing and user budget constraints on user and network behavior in packet networks

The objective of this research is to study the impacts of pricing and user budget constraints on user and network behavior in packet networks. In the first part of this research, it is assumed that users have access to unlimited amount of funds in the budget and they have different values for different types of applications. The users send the traffic to the network if the perceived value of the application's traffic is higher than the price announced by the network. We consider networks to be under single and multiple administrative domains (AD). In the case of networks under a single AD, the welfare implications of flat and congestion pricing schemes are examined. The steady state and transient behaviors of price and traffic are also analyzed under flat and congestion pricing schemes. In the case of networks under multiple ADs, each network can change its technical and economic strategies when there are perceived gains in profits or reductions in costs. When a network changes its technical and economic strategies, it affects the traffic and revenue of other networks. The incentives of a network to change its behavior and the resulting implications and consequences of this behavior on other networks are analyzed in terms of revenue flows among the networks.;In the second part of this research, it is assumed that users have access to limited amount of funds in the allocated budget over an interval of time. The users make requests for connection establishment for different types of applications. These applications differ in terms of their traffic profiles and QoS requirements. A new request for a connection establishment is accepted or rejected based on the current level of funds in the budget, the advertised network price, and the type of the application. The question is how should the limited amount of funds in the budget be allocated to different types or classes of user applications. We develop budget management policies in order to allocate the limited amount of funds in the budget to support different classes of user applications such that their QoS requirements are met. Development of these policies raises several questions that need to be answered: "Given a traffic characterization of the underlying applications, is it possible to derive the budget required to meet the QoS requirements of different classes of applications?" If so, what policy should be adopted in order to support high priority traffic classes without severely impairing the service provided to low priority traffic classes? We show that such policies can be used to provide acceptable levels of support to different classes of user applications and evaluate the effectiveness of these policies.