4th PhD School on Traffic Monitoring and Analysis (TMA), 2014


Dario Rossi, Chair (tma2014-phd-school@listes.telecom-paristech.fr)

Sabri Zaman, Local chair (sabri.zaman@qmul.ac.uk)

Hamed Saljooghinejad, Local chair&Webmaster (h.saljooghinejad@qmul.ac.uk)


Timur Friedman, Jordan Auge  and Marc-Olivier Buob (Université Pierre et Marie Curie)

Benoit Donnet and Korian Edeline (Université de Liege)

Renata Teixeira and Anna-Kaisa Pitilainen (Inria)

Alessandro Finamore (Politecnico di Torino)

Dates and venue:

April 15th-16th, London UK

Colocated with the 6th TMA Workshop, held the April 14th



Please refer to following URL:


Application is free of charge and entitles participation to the TMA Workshop

Application deadline: March 12th


Room: People’s Palace 2 (PP2), School of EECS at Queen Mary University of London, Mile End Road, London E1 4NS, United Kindom


The 4th PhD school on Traffic Monitoring and Analysis (TMA),  continues the traditional blend of theory and practice initiated by the three former editions (see the TMA portal http://www.tma-portal.eu/cost-tma-action/phd-schools/ ). Talks will be followed by hands on laboratory sessions where PhD students will have the chance to put in practice cutting-edge methodologies they just have been exposed to — aiming at reinforcing the learning process and drive down the cost to start new research work in the TMA domain. This year school will cover both passive and active measurement techniques, with emphasis  on inference of network behavior, possibly at scale, covering the whole value chain of the Internet ecosystem.

In the current Internet, users access  Cloud based applications through Web browsers or smartphone apps. The overall user experience depends on different and independently operated network segments and software tools:  so that the overall path from the household to the Cloud starts from an home network and traverse several autonomous systems, before finally hitting the data center or content distribution node. It follows that to assess the quality of user experience, and troubleshoot issues that potentially arise at any step in the path, a multitude of methodologies are needed.

For example, problems may be local to the user home (e.g., bufferbloat), or in the path (e.g., failing or misconfigured equipment),  or in the datacenter (e.g., faraway Cloud node). Techniques and tools learnt throughout the school will allow students to detect Cloud applications that are using anycast addresses by leveraging multiple vantage points. They will reveal and fingerprint intermediate routers in the paths to the Cloud. They will assess application performance from the edge, either by peeking through the user browser, or by passive inspection of the traffic from network links.

The PhD school will take place right after the TMA conference and school attendees will have the chance to follow the TMA workshop at no cost. Additionally, by means of  poster sessions, will give PhD students a chance to present their ongoing work to TMA attendees and take valuable feedback. Finally, participants can share her/his research with peers and specialized professors during practical sessions with the aim of advancing her/his own project and consolidating a network of practitioners and scholars in the TMA field.



Apr 14th Apr 15th Apr 16th
9h – 10h TMA Workshop Theory I
Achieving scale
Theory III
Passive inference
10h -12h TMA Workshop Lab I Lab III
12h – 14h Student Poster session
during TMA workshop
Student Poster session
during buffet lunch
Student Poster session
during buffet lunch
14h – 15h TMA workshop Theory II
Active Inference
Theory IV
Close to users
15h – 17h TMA workshop Lab II Lab IV
Social Event
  • Active inference: Revealing Middleboxes with Tracebox

    • Benoit Donnet and Korian Edeline (Université de Liege)

  • Achieving scale:  Large scale active measurement from PlanetLab

    • Jordan Auge  and Marc-Olivier Buob (Université Pierre et Marie Curie) 

  • Passive inference: Troubleshooting the Cloud with Tstat

    • Alessandro Finamore (Politecnico di Torino) 

  • Close to users: Performance measurement via Fathom

    • Renata Teixeira and Anna-Kaisa Pitilainen (Inria)


Material for theory lessons and labs will be available during the school. Talks will be recorded in HQ video, and made available offline.  Labs will require you to bring your own device (BYOD). Labs are set up so that either software can be installed on standard Linux installations, with virtual machines for Labs with special requirements.
  • Achieving scale:  Large scale active measurement from PlanetLab

    • Slides: Lecture and Lab Introduction

    • Video: Lecture

    • Further documentation:

    • Software requirement:  virtualbox and the VM

    • Software installation instruction: please be aware that the PlanteLab account that was setup for TMA PhD school was available only during the school, and that for security reasons, it was subsequently canceled. If you want to repeat the Lab, you will need to setup a PlanetLab account on your own (no assistance provided).

  • Passive inference: Troubleshooting the Cloud with Tstat

    • Slides: Lecture and Lab Introduction

    • Video: Lecture

    • Further documentation: the logfile format description, the exercices description and their solution

    • Software requirement:  virtualbox and the hadoop VM

    • Software installation instructions: the virtual box image is preconfigured with all the needed software. To run it

      • you need to use a release of virtualbox (or vmware) supporting 64bit

      • when creating a new image on your virtualbox (or vmware) client, select the HadoopVM.vmdk file

      • it is recommended to set up the vm with at 2GB of memory
  • Close to users: Network measurement from the browser

    • Slides: Lecture and Lab introduction

    • Video: Lecture

    • Further documentation: Fathom API, helper script tmalab.js and Ping examples  from the intro and other relevant pointers

    • Software requirement: Fathom extension, Iperf installers  and Helper javascripts

    • Software installation instructions:

      • Start Firefox browser. If you’d like to keep the Fathom extension away from your usual Firefox profile, start Firefox with firefox –ProfileManager to create a new profile for this lab.

      • Click on the Fathom extension link: the browser will prompt you to install the extension. Once installed, the extension will popup a data collection settings page. You can ignore (close) this. No data is collected nor uploaded by this development version of Fathom.

      • You can manage (enable/disable/remove) installed extensions at Tools -> Addons -> Extensions page in the browser.
Evaluation and School credits

Several laboratories will be offered to students, to allow them to practice with the methodologies and tools presented during the course. A final exam will be present at the end of the school (for students that require it): exam modes are still to be decided, and will be given during the classes (e.g., written test, assignments).

At the end of the school, we will release a certificate that will report the total number of hours of the school (and the outcome of the final examination). You can use that certificate to ask for credits at your institution. In most Universities 1 credit = 5 hours.