Measurement and modelling of the temporal dependence in packet loss

M. Yajnik, Sue Moon, J. Kurose, D. Towsley
1999 IEEE INFOCOM '99. Conference on Computer Communications. Proceedings. Eighteenth Annual Joint Conference of the IEEE Computer and Communications Societies. The Future is Now (Cat. No.99CH36320)  
Understanding and modelling packet 10SSin the Internet is especially relevant for the design and arudysis of delay-sensitive multimedia applications. In this paper, we present analysis of 128 hours of endto-end unicast and multieast packet loss measurement. From these we se-Iectedl 76 hours of stationary traces for further analysis. We consider the dependence as seen in the antocorrelation function of the originat loss data as well as the dependence between good run lengths and loss run
more » ... The correlation thnescale is found to he 1000wM or less. We evaluate the accuracy of three models of increasing complexity: the Bernoulfi model, the 24ate Markov chain model and the .k-th order Markov chain model. Out clf the 38 trace segments considered, the Bernoulti model was found to be accurate for 7 segments, the 2-state model was found to be accurate for 10 segments. A Markov chain model of order 2 or greater was found to be necessary to accurately model the rest of the segments. For the ease of ad?ptive apptieations wh]ch track loss, we address two issues of on-line loss estimation: the required memory size and whether to use exponential smoothing or a sliding window average to estimate average loss rate. We find that a large memory size is necessary and that the stkting window average provides a more accurate estimate for the same effective memory size.
doi:10.1109/infcom.1999.749301 dblp:conf/infocom/YajnikMKT99 fatcat:zmtntev2lvgztkbqj7drzramta