The Berkeley NetSys Lab

intermesh

Autonomous driving advances rely on ML models whose compute needs exceed edge devices. While the cloud offers scale, networks are treated as bottlenecks, not part of control. We present TURBO, a cloud-augmented framework that jointly optimizes control and bandwidth across car and cloud, guaranteeing safety and improving accuracy by up to 15.6 %pt.

html for http://netsys.cs.berkeley.edu/

Serverless for all computation

BESS: Berkeley Extensible Software Switch

CS 168 Fall 2022 routing project