progwriter
Usage of gravity model for the generation of traffic matices is different from the reference "Simplifying the synthesis of Internet traffic matrices"
Hi, Thanks for sharing code. I found there is something different from the reference "Simplifying the synthesis of Internet traffic matrices" when you generate random traffic with gravity model. Here it is： In the reference, the author said "In fact synthesis turns out to be quite easy. We start by taking Tin(ni ) and Tout(ni ) to be independent, identicallydistributed exponential random variables for i = 1, . . . , N .The TM is then generated using (1)."
However, you generated traffic matrix like this: cpdef TrafficMatrix random_gravity_tm dist = numpy.random.rand(num_nodes, 1) dist = dist / sum(dist)
The dist did not follow the exponential distribution anyway. This is my first puzzle. The second is that how can i generate traffic matrices with diagonal elements of zero. I need to use this matrices to generate traffic flow from source nodes to different destination nodse without node itself traffic in-and-out. Would you please give some explanations? Thanks.
Wayne.
