Suryanarayana Sankagiri
I am a postdoctoral researcher in the School of Information and Communication Sciences at EPFL in Lausanne, Switzerland. I work in the Information and Network Dynamics (INDY1) group led by Prof. Matthias Grossglauser.
I currently work on probabilistic choice models. In particular, I am studying recommendation systems that learn from comparisons. I am also interested in generating realistic embeddings from comparison data.
Broadly speaking, I am interested in modeling, designing, and analyzing engineering systems using probability and algorithms. I enjoy learning new theoretical concepts and explaining what I know to others.
Overview of Ph.D. Research
In July 2022, I graduated with a Ph.D. in Electrical and Computer Engineering from the University of Illinois at Urbana-Champaign I was affiliated with the Coordinated Science Lab. My advisor was Prof. Bruce Hajek. In my Ph.D., I worked on blockchains, studying them from a theoretical viewpoint.
Blockchains are peer-to-peer systems that maintain an ever-growing, tamper-proof ledger. This ledger forms the basis of many applications, ranging from cryptocurrencies to NFTs. All blockchains are powered by an underlying consensus protocol, using which the peers stay in agreement about the ledger. The design and analysis of secure, efficient consensus protocols (layer 1) has been an active area of research recently. In parallel, there is ongoing research into developing tools that can help blockchains scale to the speed and volume required of today's web-based services (layer 2). Last, but not the least, developers are actively building application platforms suited for blockchains (layer 3). My research has spanned all three layers.
A major thrust of my research was on analyzing the security of blockchains under adverse network conditions. In particular, I focused on the longest-chain protocol, the backbone of cryptocurrencies like Bitcoin and Ethereum.
My first work on this theme studies the longest-chain protocol under a communication model where delays are random, i.i.d., and potentially unbounded. This work establishes the protocol's security under more general and realistic network conditions.
A second work develops a an add-on gadget that can reinforce the security of the longest-chain protocol under arbitrary network delays. The work reveals nuances in the blockchain CAP theorem, an impossibility result in distributed consensus.
I have also worked towards improving the efficiency of blockchains. My focus has been on the following two directions:
Improving storage efficiency: A stateless client, built using cryptographic accumulators, vastly reduces the storage requirement of blockchain users without compromising in security. Our work highlights a principled design for such clients and improves the efficiency of an open-source project, UTREEXO.
Improving transaction throughput: Payment channel networks are a layer-2 solution to improve the transaction throughput and reduce fees on blockchains. Currently, I am developing new routing and pricing protocols in payment channel networks. I'm also studying fundamental limits of routing in such networks.
On the applied front, I have worked on two projects, both of which arise in the context of supply chain management. In both these projects, a blockchain serves as a replacement for a trusted intermediary among distrustful or competing agents.
Collaborative shipping: We built a platform which helps individual shippers bundle together shipments traveling over the same segment.
Inventory pooling: We designed a multi-agent reinforcement learning framework for independent businesses to control a joint inventory.
Education
Ph. D., Electrical & Computer Engineering, University of Illinois at Urbana-Champaign (2018-2022)
M.S., Electrical & Computer Engineering, University of Illinois at Urbana-Champaign (2016-2018)
B.Tech., Electrical Engineering, Indian Institute of Technology Bombay (2012-2016)
Awards and Honors
zkCapital Paper of the Week recognition for 'The Longest-Chain Protocol Under Random Delays'.
Rambus Fellowship, ECE Department, University of Illinois
Mavis Future Faculty Fellowship, Grainger College of Engineering, University of Illinois
List of Teachers Ranked as Excellent: as a TA for 'High-Dimensional Geometric Data Analysis', Fall 2019, University of Illinois
Nominated for the 2016 departmental undergraduate colloquium series, EE, IIT Bombay
Best poster award at the 2013 Undergraduate Research Symposium, IIT Bombay
Publications
Ph.D.
S. Sankagiri and Bruce Hajek
Routing and Flow-Control in Payment Channel Networks
In Preparation
S. Sankagiri, Shreyas Gandlur, and Bruce Hajek
The Longest-Chain Protocol Under Random Delays
Accepted for Publication in Stochastic Systems, 2023
S. Sankagiri, Xuechao Wang, Sreeram Kannan, and Pramod Viswanath
Blockchain CAP theorem allows User-Dependent Adaptivity and Finality
Financial Cryptography, 2021
Bolton Bailey and S. Sankagiri
Merkle Trees Optimized for Stateless Clients in Bitcoin
Workshop on Trusted Smart Contracts, Financial Cryptography, 2021
Internship
Kameshwaran Sampath, Sai Koti Reddy Danda, Ken Kumar, Krishnasuri Narayanam, Pankaj Dayama, and S. Sankagiri
Spot collaborative shipping sans orchestrator using Blockchain
IEEE International Conference on Blockchain, 2020
M.S.
Bruce Hajek and S. Sankagiri
Community recovery in a preferential attachment graph
IEEE Transactions on Information Theory, November 2019
Bruce Hajek and S. Sankagiri
Recovering a hidden community in a preferential attachment graph
IEEE International Symposium on Information Theory, 2018
B.Tech.
Vinutha T. P., S. Sankagiri, Kaustuv Kanti Ganguli, and Preeti Rao
Structural Segmentation and Visualization of Sitar and Sarod Concert Audio
International Society for Music Information Retrieval, 2016
Vinutha T. P., S. Sankagiri, and Preeti Rao
Reliable tempo detection for structural segmentation in sarod concerts
National Conference on Communications, 2016