Abstract:
In this letter an intuitive measurement based approximation for packet switched network traffic in backbone links is presented. It consists of a sum of independent non-identical Bernoulli sources. The resulting sum has a Poisson binomial distribution which can be approximated to a skewed Gaussian distribution at the asymptote. It can be further approximated to a binomial distribution for predicting congestion by estimating the number of users of the network. The results are verified using samples of traffic data from 42 academic and research institutions.
Citation:
Premaratne, U., & Premarathne, U. S. (2020). A Sum of Bernoulli Sources Approximation for Packet Switched Network Traffic in Backbone Links. IEEE Communications Letters, 24(1), 141–145. https://doi.org/10.1109/LCOMM.2019.2947060
