I work in the field of optimization methods, control theory, and networked systems. I am also interested more broadly in operations research, learning for decision and control, mechanism design, game theory, planning under uncertanity, and multi-agent coordination.
- Modeling and and control of networked systems: Network analysis and modelling is challenging when we have time-varying, multi-layered systems with complex, stochastic dynamics. By leveraging techniques from systems theory and graph signal processing, I have developed analytically-tractable methods to model, predict, and control processes on time-varying networks. These methods are broadly applicable to several phenomena including the spread of epidemics, flight delay propagation, cascading failures in infrastructure networks, and the propagation of rumors on social networks. My contributions are organized along three themes, and are listed below.
- Optimization under uncertanity: Vehicle scheduling in the presence of temporal uncertainty is typically posed as a stochastic optimization problem. While this approach is well-suited for maximizing efficiency, it often leads to schedules that are not equitable. I have developed a multi-stage sequential optimization approach to obtain equitable solutions without compromising efficiency. This approach gently biases and guides the stage-wise optimization to incorporate fairness, as opposed to imposing fairness as a hard constraint to prevent loss of efficiency. The problem of equity and fairness have gathered significant attention from industry and government, including our collaborators at Airbus and NASA. I have tailored and implemented my approach for three applications described below.
- Privacy and security: I have worked on a few problems related to of privacy and security.
- Risk assessments framework for UAV communications [Tech report]
- Differentially private outlier detection [ACC'21]
