Home // International Journal On Advances in Networks and Services, volume 12, numbers 1 and 2, 2019 // View article
Authors:
Toshihiko Kato
Xiaofan Yan
Ryo Yamamoto
Satoshi Ohzahata
Keywords: Unidirectional Packet trace; Round-trip Time; Autocorrelation; Lomb Periodogram; Congestion Control.
Abstract:
Network operators often attempt to analyze traffic in the middle of their networks for various purposes. In such traffic analysis, the estimation of Round-Trip Time (RTT) is indispensable. Primarily, the RTT estimation is performed by consulting the relationship between a request and its response, such as a data segment and the associated ACK segment. However, in the middle of Internet, it is common that a network operator monitors traffic only in one direction. In such a case, an operator is required to estimate RTT from unidirectional packet traces. So far, several methods have been proposed for RTT estimation from unidirectional traces. Our previous paper showed a result that adopts the Lomb periodogram method to various TCP traces using different congestion control algorithms. In this paper, we show the RTT estimation using the autocorrelation based method and the Lomb periodogram method from unidirectional TCP traces, collected through Ethernet or wireless LAN, using different congestion control algorithms, i.e., TCP Reno, CUBIC TCP, TCP Vegas, and TCP Veno. As a result, the autocorrelation based method could not estimate RTT correctly, the Lomb periodogram method provided reasonable estimation, and the results by the Lomb periodogram method are not accurate enough for subtle analysis, such as congestion window estimation.
Pages: 1 to 9
Copyright: Copyright (c) to authors, 2019. Used with permission.
Publication date: June 30, 2019
Published in: journal
ISSN: 1942-2644