DMTCS Proceedings, Fifth Colloquium on Mathematics and Computer Science

Font Size:  Small  Medium  Large

Analysis of an algorithm catching elephants on the Internet

Yousra Chabchoub, Christine Fricker, Frédéric Meunier, Danielle Tibi


The paper deals with the problem of catching the elephants in the Internet traffic. The aim is to investigate an algorithm proposed by Azzana based on a multistage Bloom filter, with a refreshment mechanism (called shift in the present paper), able to treat on-line a huge amount of flows with high traffic variations. An analysis of a simplified model estimates the number of false positives. Limit theorems for the Markov chain that describes the algorithm for large filters are rigorously obtained. The asymptotic behavior of the stochastic model is here deterministic. The limit has a nice formulation in terms of a M/G/1/C queue, which is analytically tractable and which allows to tune the algorithm optimally.

Full Text: GZIP Compressed PostScript PostScript PDF

Valid XHTML 1.0 Transitional