| Network Tomography can infer diversified network status parameters including topology without the cooperation of internal nodes. Recently, it has been pointed out that link-level temporal parameters - parameters that can reflect the bursty natural of network traffic by characterizing the performance of a link when successive packets pass it - can be obtained by using Network Tomography. However, the methods for infering link-level temporal parameters which have been proposed are failed to reflect the unicast link-level temporal characteristics because they are all multicast-based.Aiming at these deficiencies, a novel unicast-based method for bursty loss temporal parameters estimation has been proposed in this paper. Contributions are:①Based on analyses of prior work concerned unicast link loss inference, we advocate using Multi-Packet Stripe as probing method, aiming to simulate the multicast condition.②Pointing out that multicast-based temporal parameters estimation method can be used in unicast network by using Multi-Packet Stripe.③Based on the combination of Multi-Packet Stripe and the temporal parameters estimation method, we proposed an Multi-Packet Stripe–based unicast bursty loss temporal parameters estimation method, estimated the loss temporal parameters such as successive passage probabilities and mean loss-run.④The feasibility and accuracy of our method have been validated by simulation on NS2.Furthermore, we also developed an unicast-base delay temporal characteristics inference method. We proposed a Multi-Packet Stripe-based method for inferring delay temporal parameters. We simulate the multicast condition by using Multi-Packet Stripe, adapt multicast-based delay temporal parameters estimating method into unicast network, and infer delay temporal parameters, such as probabilities of arbitrary patterns of delay, mean duration of high-delay-runs or low-delay-runs over unicast network. NS2 simulation shows the validity of the proposed method. |
PDF Full Text Request