Betalord wrote: > I'm not very familiar with omnet++ either, I'm using it simply because > we are required to write the simulation in omnet for the seminar.
Fair enough, I'm leaning towards using Omnet to test the TCP-friendliness of the AIMD flow control, and Java for the larger-scale simulations - it should be easier to port my own AIMD implementation from C++ to Java than to write and test a simulated TCP stack in Java. > since I have only a few weeks left I won't be simulating > anything big, probably only basic things like routing and swapping > algorithms on randomly generated networks (that won't be of much use to > you I presume). I'm not sure - my simulations will have to include routing and swapping. One of the advantages of writing in Java is that I can steal code from the real implementation. :-) As for generating the networks, I'm thinking of using some kind of small world model, probably one of these: http://dx.doi.org/10.1073/pnas.012582999 http://nlsc.ustc.edu.cn/BJKim/PAPER/PRE_CLUS.PDF > I would need node to node protocol description though, I couldn't find > it online but I would need it to write this simulation. Is this > description available online at all? As far as I know, the source is the only documentation at this point. How much do you need to know? Success and failure conditions for each type of request would seem to be more relevant than the node-to-node protocol, or am I misunderstanding you? Cheers, Michael
