[tor-dev] Relay diversity master thesis

[Robin Descamps]

Hello,

I already sent this message to the metrics team, but they advice me to address it to the dev team, which seem to be more relevant.

I realise this year a master thesis, in the Université catholique de Louvain in Belgium, about measuring the utility brought to the Tor network diversity by adding a new relay, according to its configuration. I added to this message my master thesis plan, as well as a poster that presents a summary of the key elements.

May I ask you advices/feedback about this master thesis plan? Since I would like this project to bring a real contribution to the Tor development, I want to make sure that all the steps I will perform are useful and/or worth it.

The master thesis plan: https://drive.google.com/open?id=1XEOSS29owavKJ_cJJAVaPiJe34Ez6XXx
The poster: https://drive.google.com/open?id=1BlF2U-Kexyz6ihVSqvsVHv4PUsvXATc4

Thanks,
Robin Descamps

[teor]

Hi Robin,

Sorry it's taken a while for someone to respond to your email.
Many of us have been on leave from the start of December until this week.

Post by Robin Descamps
Hello,
I already sent this message to the metrics team, but they advice me to address it to the dev team, which seem to be more relevant.
I realise this year a master thesis, in the Université catholique de Louvain in Belgium, about measuring the utility brought to the Tor network diversity by adding a new relay, according to its configuration. I added to this message my master thesis plan, as well as a poster that presents a summary of the key elements.
May I ask you advices/feedback about this master thesis plan? Since I would like this project to bring a real contribution to the Tor development, I want to make sure that all the steps I will perform are useful and/or worth it.
The master thesis plan: https://drive.google.com/open?id=1XEOSS29owavKJ_cJJAVaPiJe34Ez6XXx
The poster: https://drive.google.com/open?id=1BlF2U-Kexyz6ihVSqvsVHv4PUsvXATc4

Have you considered relay bandwidth capacity, measured bandwidth,
consensus weight, or bandwidth authorities in your plan?

When using the Tor path selection algorithm, relay consensus weight has
a big impact on the paths selected by clients.

At the moment, relay consensus weight is a function of relay bandwidth
capacity, and geographic location. For a map of consensus weights, see
"Consensus Weight versus Bandwidth" on:

https://atlas.torproject.org/#map


Have you considered relay operators or relay families?
In particular, operators that could perform end-to-end correlation?

https://nusenu.github.io/OrNetStats/


Have you considered the relay's Operating System?
Are you aware that the Tor network has historically been a Linux
monoculture, and 90% of relays still run Linux?

https://nusenu.github.io/OrNetStats/
https://torbsd.github.io/blog.html


Have you considered the Tor version that the relay is running?

https://nusenu.github.io/OrNetStats/


Recently, someone created a website that gave badges for different
kinds of relay diversity. But I can't remember what it was called.


I've also cc'd nusenu, who has done some work in this area.

T

--
Tim Wilson-Brown (teor)

teor2345 at gmail dot com
PGP C855 6CED 5D90 A0C5 29F6 4D43 450C BA7F 968F 094B
ricochet:ekmygaiu4rzgsk6n
xmpp: teor at torproject dot org
------------------------------------------------------------------------

[grarpamp]

Post by teor

Post by Robin Descamps
May I ask you advices/feedback about this master thesis plan?
The master thesis plan: https://drive.google.com/open?id=1XEOSS29owavKJ_cJJAVaPiJe34Ez6XXx
The poster: https://drive.google.com/open?id=1BlF2U-Kexyz6ihVSqvsVHv4PUsvXATc4

In particular, operators that could perform end-to-end correlation?
Have you considered the relay's Operating System?

If considering as yet non tor daemon, non measured, non consensus voted
things like operators and OS, then you should extend research into similar
meta parameters about the relays themselves such as datacenter hosted vs
cable/dsl/fiber "home" relays, country locations, opposing legal jurisdictions,
operation by "known" or "trusted" operators / entities or not, by
working / fake / no
contact info, by any PKI Web Of Trust asserted among operators, funding
sources, employer / corporate / political / other affiliations,
statistical analysis
of historical relay "presence" on the network (add/drop/uptime, nicknames,
movement, versions, bulk turnups, correlation groups, etc), and many more
possible metas that people should think up and add to this list.

That research then followed by development of third party subscription
lists of categorized / ranked relays the user or tor daemon may further
pluggably select from when choosing nodes to path through.

There have been posts on tor-relays@ and tor-talk@ that mention more
about these sorts of meta parameters. AFAIK, no one has done any
research into them or their potential impact / benefits, whether to particularly
affected, or for plain preferential choice users, or to the network as a whole.
So the chance of a first good paper in the area awaits whoever does that
meta analysis project.

[xpost for open project oppurtunity]