The Internet's congestion control landscape is evolving. Since being first introduced on 2016, BBR has been adopted by websites on the Internet at an unprecedented rate. In this study, we aimed to measure and identify the congestion control algorithms deployed by the 20,000 most popular websites on the Internet. Our results show that in three short years since first being introduced BBR has been adopted by about 18% of the websites and is estimated to command about 40% of the downstream traffic.

The Great Internet TCP Congestion Control Census
SIGMETRICS 2020
Ayush Mishra, Xiangpeng Sun, Atishya Jain, Sameer Pande, Raj Joshi, Ben Leong

The rapid adoption of BBR brings to attention an interesting question about the future of the Internet's Congestion Control Landscape. Can we expect everyone to run BBR on the Internet at some point? Given that most decisions to switch to BBR seem to be driven by performance (primarily throughput), we investigate how BBR's throughput advantage over CUBIC will evolve as more flows on the Internet start running BBR - and what impact this will have on the CCA composition of the Internet.

Conjecture: Existence of Nash Equilibria in Modern Internet Congestion Control
APNet 2021
Ayush Mishra, Jingzhi Zhang, Melodies Sim, Sean Ng, Raj Joshi, Ben Leong

Are we heading towards a BBR-dominated Internet?
IMC 2022
Ayush Mishra, Tiu Wee Han, Ben Leong

We can expect the heterogeneity of the Internet's congestion control landscape to only worsen with the widespread adoption of QUIC. QUIC is a userspace transport stack that is set to be the standard with HTTP3. While QUIC offers us the opportunity to rethink a lot of our standard transport mechanisms, its ease of customization can prove to be a double-edged sword when it comes to congestion control. It significantly lowers the barrier to implementing new CCAs as well as modifying exisiting ones. We therefore risk Speciation in the standard congestion control algorithms that have offered us stability guarantees for the last 30 years. This project aims to understand this speciation and suggest mitigation strategies to avoid this in the future.

Understanding Speciation in QUIC Congestion Control
IMC 2022
Ayush Mishra, Sherman Lim, Ben Leong