Design of Network Architectures: Role of Game Theory and Economics

The economics of the market that a network architecture enables has a important bearing on its success and eventual adoption. Some of these economic issues are tightly coupled with the design of the network architecture. A poor design could end up making certain markets very difficult to enable, even if they are in the better interest of society. Theanalysis of these cross-disciplinary problems requires understanding both the technology and the economic aspects. This thesis introduces three major recurring themes in these problems - revenue maximization, welfare maximization and missing markets - and provides enlightening examples for them. It then delves deeper into three problems representative of these three themes and provides a complete analysis and discussion for each of them.First, the thesis studies user incentives for the adoption of femtocells or home base stations and their resulting impact on network operator revenues. The thesis develops a model of a monopolist network operator who offers the option of macrocell access or macro+femtocellaccess to a population of users who possess linear valuations for the data throughput. The study compares the revenues from two possible spectrum schemes for femtocell deployment; the split spectrum scheme, where femtocells and macrocells operate on different frequencies and do not interfere, and the common spectrum scheme, where they operate on the same frequencies (partially or fully) and interfere. The results suggest that the common spectrumscheme that creates heavy interference for the macrocell still performs comparably to the split spectrum scheme for revenue maximization. This suggests that the common spectrum scheme with good interference management may be the pathway to better femtocell adoption.Second, the thesis investigates the impact of the provision of two classes of service in the Internet on the surplus distribution between users and providers. The study considers multiple competing Internet Service Providers (ISPs) who offer network access to a fixed user base, consisting of end-users who differ in their quality requirements and willingness to pay for the access. User-ISP interactions are modeled as a game in which each ISP makes capacity and pricing decisions to maximize his prots and the end-users only decide which service to buy (if any) and from which ISP. The model provides pricing for networks with single- and two-service classes for any number of competing ISPs. The results indicate that multiple service classes are socially desirable, but could be blocked due to the unfavorable distributional consequences that it inflicts on the existing Internet users. The research proposesa simple regulatory tool to alleviate the political economic constraints and thus make multiple service classes in the Internet feasible. The third topic is a problem involving missing markets for cyber-security insurance. The study explains why insurance markets for Internet security fail to take off due to a number of factors including information asymmetry, eciency losses due to network externalities and competition. The interdependent nature of security on the Internet causes a negativeexternality that results in under-investment in technology-based defenses. The research investigates how competitive cyber-insurers affect network security and user welfare. The model explores a general setting, where the network is populated by identical users with arbitrary risk-aversion and network security is costly for the users. The user's probability to incur damage (from being attacked) depends on both his security and the network security. First, the model considers cyber-insurers who cannot observe (and thus, affect) individualuser security. This asymmetric information causes moral hazard. If an equilibrium exists, network security is always worse relative to the no-insurance equilibrium. Though user utility may rise due to a coverage of risks, total costs to society go up due to higher network insecurity. Second, the study considers insurers with full information about their users' security. Here, user security is perfectly enforceable (zero cost). Each insurance contract stipulates the required user security and covers the entire user damage. Still, for a significant range of parameters, network security worsens relative to the no-insurance equilibrium. Thus, although cyber-insurance improves user welfare, in general, competitive cyber-insurers may fail to improve network security