| Name of the measurement | GPS Traceroute |
| Tools used | Traceroute and a python script collecting the GPS coordinates from the metadata stream during the course of the experiment |
| Description | According to our original plan, we aimed at discovering the effect of mobility on L3 routing. The question was to identify if the effect of handovers can also be seen in the IP level network paths. Unfortunately, we only managed to collect GPS logs from a very limited number of mobile nodes that made the deep analysis and drawing conclusions impossible. The GPS logs of the non-mobile nodes were however used for the analysis of geographic properties of operator topologies collected for both PlanetLab and Alexa Top 1K targets. The experiment script first connects to our centralized server and downloads the list of target nodes and then launches traceroute measurements through all the different network interfaces (operators) and the GPS log collection in parallel. After the measurements are finished, the results are automatically uploaded to our central server. |
| Jupyter Dashboard | http://dblab.inf.elte.hu:3000/traceroute |
| Source |
| Name of the measurement | Scamper |
| Tools used | Scamper tool by CAIDA |
| Description | For topology discovery we aimed at using the double-tree implementation of Scamper tool. Scamper enables experimenters to manage their measurements running on distributed nodes in a centralized way in real-time. After deployment the docker container starts the Scamper client that connects to our centralized server where the Scamper server is running. Then the server can recognise that a new node connected that may trigger to execute some measurements. In this scenario, the remote measurements are coordinated and initiated by the Scamper server and the results are also collected by the server. Note that after turning out that only some pieces of the double-tree topology discovery method are implemented in Scamper we decided to use the GPS Traceroute experiment instead. We have published this docker since it may be useful for other experimenters. |
| Jupyter Dashboard | |
| Source |
| Name of the measurement | Youtube |
| Tools used | We used the pytomo tools (https://github.com/LouisPlisso/pytomo) for the measurement. |
| Description | For measurement we used 4 youtube videos. We download a popular short video: ‘First Look at Nontendo Labo’ (https://www.youtube.com/watch?v=P3Bd3HUMkyU) duration: ~2:52, a popular long video: ‘Oscar nominations 2018 announced for the 90th Academy Awards | ABC News’ (https://www.youtube.com/watch?v=jdSUea1CEPc) duration: ~54:50, a non-popular short video: ‘Az M1 híradóban a migránsos videó’ (http://www.youtube.com/watch?v=rjf94G9HFkU) duration: ~3:01 and a non-popular long video: ‘146. Der Trucker .A kamionos. S2 E3’ (http://www.youtube.com/watch?v=w5k7A7vuVic) duration: ~1:00:59. We downloaded from each video the first 120 seconds. |
| Jupyter Dashboard | http://dblab.inf.elte.hu:3000/youtube/ |
| Source | http://dblab.inf.elte.hu:3000/youtube/youtubePcaps |
| Name of the measurement | DiffGen |
| Tools used | |
| Description | |
| Jupyter Dashboard | |
| Source |