Inferring statistics of flow from aggregated-flow measurement

摘要

For a statistical perspective of global traffic flows on a large network, we have proposed a way to infer unobservable statistics of each flow from measurements of aggregated-flows (sums of flows), which are easily measured at some links (router interfaces) in the network. In this paper, to relax some limitations in the previous work, we discuss some improvement: i) determining minimum rates of flows, ii) indirectly inferring statistics of some flows from those of other flows, iii) rounding (quantization) by an adequate bin size, and show simulation results, which indicate effectiveness of our improvement.