To report a botnet PRIVATELY please email: [EMAIL PROTECTED]
----------
Hi All,

As part of a project for a visualization class, I'm working on a  
visual tool for exploring data collected from a variety of realtime  
blocklists (RBLs), a couple of spamtraps, and a few other  
miscellaneous sources, and was hoping to get some feedback or  
suggestions on my ideas. My hope is that correlations in the data can  
be readily visually explored and that through the exploration,  
information about botnets may fall out. This is a very different  
approach from using, say, NMAP to monitor specific ports or a  
honeypot to trap a bot and monitor comms with C&C. Instead of  
determining whether a specific network is hosting bots or the  
behavior of a single type of bot, I'm effectively trying to take  
metadata about broader internet abuse and use it to attempt to  
observe more general behavior consistent with botnets. Of course, I  
have an enormous amount of data, and all data simply represent a  
snapshot in time, so any picture I develop will only be valid for  
that time period.

One possible visualization I've considered includes plotting RBL  
issues on a world map after geocoding the IP's and playing an  
animation of the development of issues over some period of time in  
which the user is interested. As well, the user could stop the  
animation at any time, zoom on a region and by mousing over an issue  
or block of issues, get more detailed information. Another feature  
I'm interested in implementing is a dynamic query
that updates the visual representation of the data in realtime as a  
result of the application of some interactive filters.

Some of the dimensions I have for analyzing this data are

RBL Data:
Issue Type - (Spam source, open relay, open proxy, etc.)
Timestamp (opened)
Timestamp (closed)
AS Name/Number
Source IP

Spam Data:
Source IP
Domain given in ehlo message
Message size
URL's contained in message

Misc Data:
BGP Routing Update logs
DNS and Reverse DNS snapshots
IPs I've scrubbed from this lists archives

One question I have is how many of the issues being reported by RBLs  
are actually coming from bots? I don't really know the answer, but I  
suspect that bots are responsible for a large portion of internet  
misbehavior. However, the extent to which I can establish patterns  
that might qualify machines as bots from the data I have is still  
very much an open question. I think using visual techniques should  
speed the process, but it will depend heavily on choosing appropriate  
visualizations and incorporating the right information.

Anyway, I just thought I'd check in and see if this sounds  
interesting to anyone, or if you guys might have suggestions about  
interesting ways of slicing the data, visualizing the data, or  
hypothesizing and testing relationships between different dimensions  
of the data.

Thanks,
Aaron
_______________________________________________
To report a botnet PRIVATELY please email: [EMAIL PROTECTED]
All list and server information are public and available to law enforcement 
upon request.
http://www.whitestar.linuxbox.org/mailman/listinfo/botnets

Reply via email to