3 ms·

I would be too. As for the ants- they would have to be giant ants in my point of view. Able to sniff if something out of the thousands of somethings is wrong. A

by dbz 17y ago

I would be too. As for the ants- they would have to be giant ants in my point of view. Able to sniff if something out of the thousands of somethings is wrong. And if they are all equal detectors, then everything would get a strong trail or no trail. And wouldn't those trails build up bigger and bigger over the years until something not a threat becomes "scented" into a giant one? AND if all of the ants have different strengths, then one and will miss a threat and wont put a very important scent marking on it... uhg. I guess I am just a pessimist when it comes to this approach of protection. It seems like it will take up space and memory and wont be particularly effective even though real life ants seem to survive a-okay

krisneuharth 17y ago
I work with similar systems in another domain. You would use different scents to mark different indicators of an attack. To prevent build up you would evaporate the scents over time, possibly at different rates. Each ant is a simple agent that is equal in capability although it is possible to have different specializations for some ants. The real power of swarming systems is their simplicity.
- dbz 17y ago
  That makes sense- but I guess what I failed to say was that all of the ants need to go over the same information, which is basically just a system scan taking a longer amount of time and not at once, which would just lower available ram- and I can use a LOT of ram at times.
  - krisneuharth 17y ago
    I see what you are saying. I think the value of the swarm here is the rapidly adaptive nature of the system when there are so many agents. I don't really know what the overhead is for this approach compared to a normal virus scanner. I looked around on the Wake Forest CS site trying to find a paper describing their approach but I didn't come up with much except for a more detailed article and a list of publications by the researcher. http://www.wfu.edu/wowf/2009/20090921.ants.html http://www.wfu.edu/wowf/2009/20090921.ants.html http://www.cs.wfu.edu/~fulp/ewfpub.html#securityPublications http://www.cs.wfu.edu/~fulp/ewfpub.html#securityPublications I imagine that the grad students mentioned in the article will be documenting this research as part of their theses. I am interested to see what they have to say, my company has looked at similar ideas in intrusion detection a little in the past. EDIT: I found this paper at the PNNL site: http://www.pnl.gov/main/publications/external/technical_reports/PNNL-17868.pdf http://www.pnl.gov/main/publications/external/technical_repo...

by dbz 17y ago

krisneuharth 17y ago

I work with similar systems in another domain. You would use different scents to mark different indicators of an attack. To prevent build up you would evaporate the scents over time, possibly at different rates. Each ant is a simple agent that is equal in capability although it is possible to have different specializations for some ants. The real power of swarming systems is their simplicity.

dbz 17y ago
That makes sense- but I guess what I failed to say was that all of the ants need to go over the same information, which is basically just a system scan taking a longer amount of time and not at once, which would just lower available ram- and I can use a LOT of ram at times.
- krisneuharth 17y ago
  I see what you are saying. I think the value of the swarm here is the rapidly adaptive nature of the system when there are so many agents. I don't really know what the overhead is for this approach compared to a normal virus scanner. I looked around on the Wake Forest CS site trying to find a paper describing their approach but I didn't come up with much except for a more detailed article and a list of publications by the researcher. http://www.wfu.edu/wowf/2009/20090921.ants.html http://www.wfu.edu/wowf/2009/20090921.ants.html http://www.cs.wfu.edu/~fulp/ewfpub.html#securityPublications http://www.cs.wfu.edu/~fulp/ewfpub.html#securityPublications I imagine that the grad students mentioned in the article will be documenting this research as part of their theses. I am interested to see what they have to say, my company has looked at similar ideas in intrusion detection a little in the past. EDIT: I found this paper at the PNNL site: http://www.pnl.gov/main/publications/external/technical_reports/PNNL-17868.pdf http://www.pnl.gov/main/publications/external/technical_repo...