Computer Science Speaking Skills Talk
In Person - Gates Hillman 8102
MARGARIDA FERREIRA , Ph.D. Student, Computer Science Department, Carnegie Mellon University
Program Synthesis for Reverse Engineering Congestion Control
Congestion Control Algorithms (CCAs) impact numerous desirable Internet properties such as performance, stability, and fairness. Hence, the networking community invests substantial effort into studying whether new algorithms are safe for wide-scale deployment. As new applications arise, operators are continuously innovating, and proprietary CCAs are becoming increasingly prevalent on the Internet, it is becoming harder to analyze them to gain insights into their interactions with other network elements.
In this talk, I will explain how we use techniques from program synthesis to reverse-engineer CCAs from observed traces of their behavior. We developed a synthesis pipeline, Abagnale, that helps users automate the reverse-engineering task. By limiting the output with a Domain-Specific Language (DSL), using partial searches of the output space to inform future search, and accepting approximate answers, we find success in reverse-engineering implementations of well-known CCAs as well as those from a dataset of student-written CCAs from a graduate networking class.
Presented as part of the PLunch Series
Presented in Partial Fulfillment of the CSD Speaking Skills Requirement