My research is along the following themes.

Epidemics on Networks: Modeling, Estimation and Control

Our focus is on rigorously understanding the dynamics of spreading processes (such as infectious diseases, opinions, etc.) on networks. We borrow tools from network science, dynamical systems, optimization, game theory and signal processing for modeling, estimation and control of such processes.

We are interested in investigating the following problems.

  • How to infer the prevalence of infectious diseases from testing data and use it as feedback for deploying optimal non-pharmaceutical interventions (such as lockdown, contact tracing, isolation, etc)?
  • How does individual protective behavior evolve and influence the spread of epidemics?
  • How do behavioral biases influence individuals while taking protective measures (such as self-quarantine, vaccination, etc)?

Representative Publications:

External Collaborators: Philip Paré and Shreyas Sundaram (Purdue University), Saverio Bolognani (ETH Zurich), Vaibhav Srivastava (Michigan State University).

Funding: Joint DST-NSF Indo-US Research Grant by IDEAS, ISI Kolkata jointly with Prof. Philip Paré from Purdue University, USA.

Stochastic (Data-Driven) Optimization and Control

Optimization problems with uncertainty are at the heart of many engineering disciplines. In particular, stochastic optimal control, model predictive control and estimation heavily rely on solving optimization problems with uncertain objectives and constraints.

We are currently investigating new techniques (and their applications) that do not impose any assumptions on the knowledge or nature of the probability distribution of the uncertainty, but rather relies directly on the available data to compute (approximate) optimal solutions of such problems.

Representative Publications:

Funding: ISIRD Grant, IIT Kharagpur

External Collaborators: Ashish Cherukuri (University of Groningen), Bala Kameswar Poolla (NREL), Saverio Bolognani and John Lygeros (ETH Zurich).

Game Theory for Security of Network Systems

Modern engineered systems are highly interconnected and interdependent. Different types of faults, and targeted attacks exploit the underlying network of interconnections to spread throughout the system. In this context, our research is focused on the following problems.

  • What is the impact of decentralized decision making on the security of large-scale networks?
  • How do behavioral biases influence the security investment decisions of human users, and what is the network-wide impact of such decisions?
  • Are certain classes of networks inherently resilient, and if so, what is the underlying structure of such networks?

Representative Publications:

In Press: Purdue News.

Equilibrium and Incentives under Behavioral Biases in Multi-Agent Systems

Technology has enabled increasing human interaction with engineered systems in recent years. In particular, humans actively interact with shared infrastructure, and make decisions that influence their operation. For instances, smart phones and Internet-of-Things have led to proactive participation of end-users in transportation and energy systems. Thus understanding and influencing their behavior is essential for controlling the operation of such systems. We investigate the following questions in this context.

  • How to quantify the impacts of decentralized and human decision-making on shared systems?
  • How to design dynamic incentives to truthfully elicit the preferences and privately-held information from end-users?
  • Do psychological biases cause humans to respond to economic incentives in counter-intuitive ways, and if so, how to mitigate them?

Representative Publications:

In Press: Purdue News.