tor-node-map

See it here: http://cdetrio.github.io/tor-node-map/

Uses tor relay information from https://onionoo.torproject.org/. Plots relay nodes as bubbles (circles) with area proportional to their observed bandwidth. The bubbles are plotted on a kinetic.js canvas, overlayed on top of a d3.js map. I did it this way (kinetic/d3 mashup) for better performance. On my machine, panning/zooming more than a few hundred d3 svg elements is too slow. But it can comfortably handle a few thousand canvas sprites. Kineticjs provides event handlers on canvas sprites, d3 does not. Firefox needs an offsetX/Y event property kludge, so relay selection works a little bit better in Chrome.

One issue to be fixed is when multiple relays have the same geoip latitude/longitude coordinates. Where this happens, you will see lots of white as several bubbles of different sizes are placed at the exact same spot (each bubble has a thin white outline stroke). I'm not yet sure how to deal with this.

There's also a group of bubbles in the top-left (0,0), these are relays with ip addresses not in the geoip database.

To fetch latest relay details from onionoo: wget https://onionoo.torproject.org/details?type=relay -O onionoo-relay-details.json

To run a local webserver: python -m SimpleHTTPServer

inspired by jordan-wright/tormap http://raidersec.blogspot.com/2013/09/mapping-tor-relays-and-exit-nodes.html

Incentive for Tor Relays - proof-of-relay / proof-of-bandwidth

Currently, Tor relay bandwidths are estimated by active measurements, performed by a set of "bandwidth authorities" (bwauths). The bandwidth authorities vote to form a "consensus bandwidth" (listed as "consensus_weight" in onionoo) value for a relay.

LIRA is an incentive scheme for Tor relays - think of it as "TorCoin". LIRA proposes using an opportunistic decentralized bandwidth measurement, namely EigenSpeed, instead of active measurements by bandwidth authorities. In a decentralized EigenSpeed measurement, each node would communicate to others its estimate of other nodes' bandwidths. The result would be each node maintaining its own large sparse EigenTrust-like matrix.