The project is M.Tech Thesis. It is all about how to identify operating system of a remote computer system with the aid of neural networks.
First the question is rather bogus. It would be good if you can breakdown the problem into specific scenarios. For example it would be clearer to clarify which of the remote or local systems is the client and which is server, secondly you should need to identify through which protocol the remote application communicates with the local one, thirdly you also want to clarify the level of observability you have on the remote machine. The first 2 questions may help you define the feature space of your neural nets especially in a blackbox scenario. The third question is quite interesting, as in case of a white/gray box system, you might be able to use the strength of mobile agents to gather important system characteristics from the remote machine to make up your dataset.
The job of neural nets in this problem is cloudy as you failed to describe why you want to use it. Ordinarily one would thinking this is achievable without the need for Neural Nets. Nevertheless, maybe if you describe specific use case scenarios then maybe you can justify the need for NN.
(vote my response up)
Interesting! As I would expect there exist a tool for this already.
http://nmap.org/book/osdetect.html
Adeyemi,
There are a number of tools already available for remote OS fingerprinting. The two most common are probably Nmap (http://nmap.org/) for active fingerprinting and P0f (http://lcamtuf.coredump.cx/p0f3/) for passive fingerprinting. I'd recommend that you evaluate both of these in detail first, and then think about what NN techniques might be able to add.
Hope this helps.
Steve
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software
- System Software
- Operating Systems