¹The University of New South Wales, Australia

---

Abstract—This paper proposes a new approach to modelling and controlling Internet end-to-end loss behaviours. Rather than select the model structure from the loss observations as being done previously, we construct a new loss model based on the TCP congestion control mechanisms. Thus, the model can explicitly reflect the correlation between end-to-end loss observations and network flow level activities. Besides simulation, the model has been tested in both wired and wireless Internet environments. The result shows that, unless the losses due to the transmission errors are excessive e.g. in some lossy wireless channels, the model can correctly capture end-to-end loss behaviours not only in terms of average rates but also in terms of loss patterns i.e. loss and good runlengths. This implies a good connection between the model structure and network flow level activities, which makes the model attractive for assisting network traffic control.

Index Terms—Internet Loss Modelling, Network Traffic Control, Markov Model, Markov Decision Processes

[PDF]

Cite: Vinh Bui, Weiping Zhu, Ruhul Sarker, "Modelling Internet End-to-End Loss Behaviors: A Congestion Control Perspective," Journal of Computers vol. 2, no. 8, pp. 18-25, 2007.