This repository contains an experiment that compares two different methods of measuring round trip time:
- RTTs from TCP timestamps using the method described in "New Methods for Passive Estimation of TCP Round-Trip Times
- RTTs from a square wave, as described in the Proposal for adding a Spin Bit to QUIC
In particular, we want to find out whether the probabilistically adding a measurement header (e.g. timestamp) to some of the packets is able to achieve similar accuracy more efficiently than using a single bit.
To run, first install python-libtrace. Then, install the required modules and start Jupyter:
pip install -r requirements.txt
jupyter notebook
Then, follow the instructions in the notebooks 00_extract_flows.ipynb and 01_compare_rtts.ipynb, respectively.