We examine competition between two Internet Service Providers (ISPs), where the first ISP provides basic Internet service, while the second ISP provides Internet service plus content, i.e., enhanced service, where the first ISP can partner with a Content Provider to provide the same content as the second ISP. When such a partnering arrangement occurs, the Content Provider pays the first ISP a transfer price for delivering the content. Users have heterogeneous preferences, and each in general faces three options: (1) buy basic Internet service from the first ISP; (2) buy enhanced service from the second ISP; or (3) buy enhanced service jointly from the first ISP and the Content Provider. We derive results on the existence and uniqueness of a Nash equilibrium, and provide closed-form expressions for the prices, user masses, and profits of the two ISPs and the Content Provider. When the first ISP has the ability to choose the transfer price, then when congestion is linear in the load, it is never optimal for the first ISP to set a negative transfer price in the hope of attracting more revenue from additional customers desiring enhanced service. Conversely, when congestion is sufficiently super-linear, the optimal strategy for the first ISP is either to set a negative transfer price (subsidizing the Content Provider) or to set a high transfer price that shuts the Content Provider out of the market.

Congestion-dependent pricing is a form of traffic management that ensures the efficient allocation of bandwidth between users and applications. As the unpredictability of congestion prices creates revenue uncertainty for network providers and cost uncertainty for users, it has been suggested that forward contracts could be used to manage these risks. We develop a novel game-theoretic model of a multiprovider communication network with two complementary segments and investigate whether forward contracts would be adopted by service providers. Service on the upstream segment is provided by a single Internet service provider (ISP) and priced dynamically to maximize profit, while several smaller ISPs sell connectivity on the downstream network segment, with the advance possibility of entering into forward contracts with their users for some of their capacity. We show that the equilibrium forward contracting volumes are necessarily asymmetric, with one downstream provider entering into fewer forward contracts than the other competitors, thus ensuring a high subsequent downstream price level. In practice, network providers will choose the extent of forward contracting strategically based not only on their risk tolerance, but also on the market structure in the interprovider network and their peers' actions.

Congestion-dependent pricing is a form of traffic management that would ensure the efficient allocation of bandwidth between users and applications. As the unpredictability of congestion prices creates revenue uncertainty for network providers and cost uncertainty for users, it has been suggested that longer-term financial agreements such as forward contracts could be used to manage these risks. In a network managed by a single service provider, long-term forward contracts are beneficial for both the provider and the users. We investigate whether forward contracts would be adopted by multiple service providers in a future Internet with congestion-dependent pricing. We develop a novel game-theoretic model of a multi-provider communication network with two complementary segments. Service on the upstream segment is provided by a single Internet Service Provider (ISP) and priced dynamically to maximize profit, while several smaller ISPs sell connectivity on the downstream network segment, with the advance possibility of entering into forward contracts with their users for some or all of their capacity. We show that the equilibrium forward contracting levels are necessarily asymmetric, with one downstream provider entering into fewer forward contracts than the other competitors, thus ensuring a high subsequent downstream price level. In practice, network providers will choose the extent of forward contracting strategically based not only on their risk tolerance, but also on the market structure in the interprovider network and their peers' actions.

The problem of sharing bandwidth in a communication network has been the focus of much research aimed at guaranteeing an appropriate quality of service to users. This is particularly challenging in an environment with a great diversity of users and applications, which makes it difficult, if not impossible, to tightly constrain user attributes and requirements. This motivates shifting the burden of rate allocation from the network to the end-systems. We propose a decentralized scheme for user adaptation and study its dynamics. The proposed scheme uses congestion prices as a mechanism for providing both feedback and incentives to end-systems.

This paper analyzes competition between two Internet service providers (ISPs), either or both of which may choose to offer multiple service classes. In the model analyzed, a social planner who maximizes the total benefit from network usage and a profit maximizing monopolist will both form multiple service classes; but two networks competing to maximize profits will not. The reason is that a competition effect always outweighs a segmentation effect. Networks wish to offer multiple service classes in order to increase user benefits and hence charge higher prices. In doing so, however, they effectively increase the number of points in the service quality range at which they compete. Consequently, in any equilibrium competitive outcome, both ISPs offer a single service class. The analysis has particular implications for the Paris Metro pricing (PMP) proposal, which is considered in depth in this paper, since it suggests that PMP may not be viable under competition.