Most of my research has been in the field of data communication networks — on performance modeling, analysis and optimization questions for the Internet as well as wireless and sensor networks. At a high level, a unifying theme that characterizes much of my research activities in this context is the design and analysis of solution approaches that can optimize usage of communication-critical resources (bandwidth and energy)in networks, with the goal of attaining maximum global network utility or fair sharing of resources in the network. A significant part of this research is focused on developing and analyzing network algorithms that can be implemented in a decentralized manner, through local coordination, often only requiring limited information about the network topology and conditions. This also includes studying related questions in resource pricing and network economics related to use of shared bandwidth among self-interested agents.

In recent years, I have also applied some of the key control and optimization techniques developed for communication networks to energy flow networks with the goal of attaining better spatio-temporal load and energy supply distribution through demand-response strategies.

Most recently, I have been investigating the design of efficient inference networks for AIML tasks, and associated query routing questions. I have been also investigating training, hyper-parameter optimization and inference using distributed AIML modelssuch as in federated learning systems.

A distinguishing feature of my work is the use of linear, non-linear and stochastic optimization and control techniques, to study these network algorithm design/resource allocation questions. Since my work mostly focuses on distributed systems/agents which may sometimes have strategic behavior, some of my work also has a game-theoretic flavor.