Content analysis of random cell injection in ATM networks

This paper presents an analysis of random bitstream content segmented into fixed-length "cells" similar to those used in asynchronous transfer mode (ATM) networks. This bitstream is generated by randomly interleaving cells from a source with deterministic character with those of white-noise character. The autocorrelation function of this bitstream is derived using a hidden Markov model (HMM). Simulations are presented for a range of injected cell to valid data cell ratios. Noticeable changes in content can be observed for ratios as low as one random cell to one hundred valid data cells.