Speaker: Shaileshh Bojja Venkatakrishnan
Faculty host: Prof. Javad Ghaderi
Abstract: In recent years, complex large-scale networked systems and applications have emerged over the Internet, to serve increasing user demand for content, storage, computation, privacy and security. For example, cryptocurrencies such as Bitcoin are becoming popular as secure decentralized payment systems with millions of active users; data centers are massively scaling up to house hundreds of thousands of servers, while providing very high bandwidth and low latencies. At these scales, it becomes critical to operate at high efficiencies as the price of wasted resources can be significant. In my work, I have explored designing effective network-level algorithms and protocols as a principal way of achieving high efficiencies in such systems.
In this talk, I will discuss a number of fundamental algorithmic challenges that arise in emerging networked systems, and present (near-)optimal solutions for them. The first part of the talk will focus on privacy and scalability challenges in Bitcoin’s peer-to-peer network and other decentralized payment systems. Although designed for payments, we show how ideas familiar to data communication networks can be used to tackle core algorithmic problems in these networks. In the second part of the talk I will briefly discuss how recent advancements in machine learning can be leveraged for learning powerful data-driven algorithms for networking applications. I will finish the talk with important open questions and directions.