Abhishek Sinha

Abhishek Sinha 

Abhishek Sinha
Assistant Professor
Dept. of Electrical Engg., IIT Madras
Email: abhishek.sinha@ee.iitm.ac.in
Office: ESB-212-A

Research Summary

My primary research interest broadly lies at the intersection of Probability theory, Stochastic control, Information theory, and Learning Theory, where I seek to develop new analytical tools for understanding and enhancing the performance of networked communication and control systems of wide varieties. During my Ph.D., I worked with Prof. Eytan Modiano at MIT, where we studied the multicasting and the broadcasting problems from a network control standpoint and proposed provably optimal and decentralized algorithms targeted for wireless networks. This is the first known throughput-optimal policy for solving a flow problem of such generality. My thesis committee included Prof. Leandros Tassiulas, Prof. David Gamarnik, and Prof. Eytan Modiano.

I interned with Bell Labs at Murray Hill, NJ, in the summer of 2016, where I worked with Matthew Andrews and Prasanth Ananth on optimal resource allocation problems. I spent the summer of 2014 as a research intern at Microsoft, Redmond, working with Jie Liu at Microsoft Research and David Maltz's group at Windows Azure. There I worked on control and optimization of anycast based load management systems.

I obtained my Master's degree from the Electrical Communication Engineering department at the Indian Institute of Science, located in the beautiful city of Bangalore, where I worked with Prof. Anurag Kumar. Some of the highlights of my present and past research works are given below:

  • Our group is currently investigating the fundamental performance limits of the Network Caching algorithms from an online learning perspective. We have some exciting new results and algorithms pertaining to this problem.

  • Our group is also investigating the problem of optimal online scheduling to minimize the Age-of-Information in 5G wireless networks. We have designed a set of robust algorithms, which are oblivious to the underlying statistics of the channel (non-stationary regime), yet operate close to the fundamental limits.

  • In my doctoral work, I designed an exciting new Max-Weight type dynamic routing and scheduling policy, called Universal Max-Weight. UMW provides a unified solution to the general network-flow problem and settles several open and fundamental questions in this field.

  • During my Ph.D., along with my colleagues, I designed a scheduler for minimizing the Age of Information with throughput constraints. This work recently won the Best Paper Award in IEEE INFOCOM 2018, Honolulu, HI.

  • At IISc, I also collaborated with others on solving the problem of optimal deployment of relays on a line for the multi-relay channel from an Information-theoretic perspective.