Exploring the Internet Using Public Discovery
Objectives
-
Visualize internet connections in a logical mesh view, unconstrained by geography
-
Experiment with
Gephi
layout algorithms
-
Find a good tool for zoom/pan/scroll of large images
Methods
-
Traceroute to both random and selected IP addresses (same database as the
geographic view
)
-
Import extracted edge list with Gephi
-
Use algorithmic layout to obtain a
moderately
tangle-free topology
Challenges
-
Obtaining a layout that wasn't a tangled or jumbled mess; still needs refinement!
Findings
-
Multiple paths can be involved in reaching any particular node if it's not too distant. This is especially apparent since these routes were traced over several days and times of day. Highly parallel paths are visible in the dark central ring representing my ISP's regional infrastructure.
-
Gephi is a very useful tool for exploring relationships in a large variety of datasets. A principle challenge for the would-be data scientist can be selecting a layout which suits the data being depicted. The process of obtaining a good layout can take some time, but that layout can then be exported to use for realtime browsing.
-
The layout featured here was obtained using YifanHu's Multilevel layout with a set of parameters determined experimentally, in particular a quadtree max level of 16.
-
Leaflet is a powerful map tool with features comparable to Google Maps and therefore quite underutilized in this context.
-
A strong resemblance to
these diagrams of neural networks
is apparent.
See Also
The neural network diagram is courtesy of
this publication
.