On the Design and Analysis of Incentive Mechanisms in Network Science

Title of dissertation: ON THE DESIGN AND ANALYSIS OF INCENTIVE MECHANISMS IN NETWORK SCIENCE Yang Gao, Doctor of Philosophy, 2014 Dissertation directed by: Professor K. J. Ray Liu Department of Electrical and Computer Engineering With the rapid development of communication, computing and signal processing technologies, the last decade has witnessed a proliferation of emerging networks and systems, examples of which can be found in a wide range of domains from online social networks like Facebook or Twitter to crowdsourcing sites like Amazon Mechanical Turk or Topcoder; to online question and answering (Q&A) sites like Quora or Stack Overflow; all the way to new paradigms of traditional systems like cooperative communication networks and smart grid. Different from tradition networks and systems where uses are mandated by fixed and predetermined rules, users in these emerging networks have the ability to make intelligent decisions and their interactions are self-enforcing. Therefore, to achieve better system-wide performance, it is important to design effective incentive mechanisms to stimulate desired user behaviors. This dissertation contributes to the study of incentive mechanisms by developing game-theoretic frameworks to formally analyze strategic user behaviors in a network and systematically design incentive mechanisms to achieve a wide range of system objectives. In this dissertation, we first consider cooperative communication networks and propose a reputation based incentive mechanism to enforce cooperation among selfinterested users. We analyze the proposed mechanism using indirect reciprocity game and theoretically demonstrate the effectiveness of reputation in cooperation stimulation. Second, we propose a contract-based mechanism to incentivize a large group of self-interested electric vehicles that have various preferences to act coordinately to provide ancillary services to the power grid. We derive the optimal contract that maximizes the system designer’s profits and propose an online learning algorithm to effectively learn the optimal contract. Third, we study the quality control problem for microtask crowdsourcing from the perspective of incentives. After analyzing two widely adopted incentive mechanisms and showing their limitations, we propose a cost-effective incentive mechanism that can be employed to obtain high quality solutions from self-interested workers and ensure the budget constraint of requesters at the same time. Finally, we consider social computing systems where the value is created by voluntary user contributions and understanding how user participate is of key importance. We develop a game-theoretic framework to formally analyze the sequential decision makings of strategic users under the presence of complex externality. It is shown that our analysis is consistent with observations made from real-word user behavior data and can be applied to guide the design of incentive mechanisms in practice. ON THE DESIGN AND ANALYSIS OF INCENTIVE MECHANISMS IN NETWORK SCIENCE