Controllable Markov Jump Processes. II. Monitoring and Optimization of TCP Connections
- 2 Downloads
The article considers the practical application of the analysis and estimation of the controlled Markov jump process states by continuous, discrete, and counting observations in the development of state monitoring algorithms for network connections operating under the Transmission Control Protocol (TCP). A specific feature of the applied problem is the physical heterogeneity of the channel providing the TCP connection under study: along with the wired section, there is a wireless “last mile” of the channel. The current state of the entire connection cannot be directly observed, and there is just indirect statistical information in the form of a flow of acknowledgements of successful packet transmission, as well as packet loss counting processes and timeouts. In this part of the work, not only the controlled stochastic dynamic observation system was used for the mathematical description of the TCP New Reno connection but also the developed high-precision algorithm for tracking this connection state according to the available statistical information. The numerical examples make it possible to define causes of channel losses, such as congestion in the wired section or signal attenuation in the wireless section, and as a result modify the TCP algorithm so as to significantly increase the bandwidth.
This work was supported by the Russian Foundation for Basic Research (grant no. 16-07-00677).
- 1.D. Kurose and K. Ross, Computer Networking: A Top-Down Approach, 7th ed. (Pearson, Upper Saddle River, NJ, 2016).Google Scholar
- 5.A. V. Borisov, A. V. Bosov, and G. B. Miller, “Modeling and monitoring of VoIP connection,” Inform. Primen. 10 (2), 2–13 (2016).Google Scholar
- 6.G. Haßlinger and O. Hohlfeld, “The Gilbert-Elliott model for packet loss in real time services on the internet,” in Proceedings of the 14th GI/ITG Conference on Measurement, Modelling and Evaluation of Computer and Communication Systems MMB, Dortmund, Germany, 2008, pp. 269–283.Google Scholar
- 9.G. P. Basharin, Lectures on Mathematical Theory of Teletraffic (Ross. Univ. Druzhby Narodov, Moscow, 2004) [in Russian].Google Scholar
- 12.J. Domanska, A. Domanski, T. Czachorski, and J. Klamka, “Fluid flow approximation of time-limited TCP/UDP/XCP streams,” Bull. Polish Acad. Sci.: Tech. Sci. 62, 217–225 (2014).Google Scholar
- 22.S. Mascolo and G. Racanelli, “Testing TCP westwood+ over transatlantic links at 10 gigabit/second rate,” in Proceedings of the 3rd International Workshop on Protocols for Fast Long-Distance Networks FLDNET05, Lyon, France, 2005.Google Scholar
- 26.A. V. Borisov, G. B. Miller, and A. I. Stefanovich, “Controllable Markov jump processes. I. Optimum filtering based on complex observations,” J. Comput. Syst. Sci. Int. 57 (6) (2018, in press).Google Scholar
- 29.E. B. Dynkin, Markov Processes (Fizmatgiz, Moscow, 1959; Springer, Berlin, Heidelberg, 1965).Google Scholar
- 30.J. C. Cox and V. Smith, Recovery Theory (Sov. Radio, Moscow, 1967) [in Russian].Google Scholar
- 31.A. A. Borovkov, Asymptotic Methods in Queueing Theory, Probability and Mathematical Statistics (Fizmatlit, Moscow, 1980; Wiley, New York, 1984).Google Scholar
- 32.http://www.isi.edu/nsnam/ns/.Google Scholar