Title :
A new approach to the statistical analysis of HMM modelled bursty channels
Author :
Santos, Marcos V. ; Pinto, Ernesto L. ; Grivet, Marco
Author_Institution :
Univ. Fed. do Rio de Janeiro, Brazil
Abstract :
A novel methodology for obtaining statistics of a hidden Markov model (HMM) associated to a burst error process is presented which is based on the CCITT definition of a burst error. Its fundamental step is the construction of a properly chosen classical Markov chain associated to the HMM which is shown to give rise to the derivation of recursive expressions for the burst statistics of interest. This methodology is applied to derive several burst error statistics related to the CCITT definition of burst, which have been addressed by computer simulation in other works. Some numerical examples are presented and illustrate the effectiveness of the presented methodology in the derivation of expressions specially suitable for quick evaluation.
Keywords :
channel coding; error correction codes; error statistics; hidden Markov models; multimedia communication; protocols; statistical analysis; CCITT; HMM; Markov chain; burst error process; bursty channel; computer simulation; hidden Markov model statistics; international telephone & telegraph consultative committee; recursive expression; statistical analysis; Analytical models; Error analysis; Error correction; Hidden Markov models; Multimedia communication; Performance analysis; Physical layer; Protocols; Statistical analysis; Statistical distributions;
Conference_Titel :
Communications, 2004 IEEE International Conference on
Print_ISBN :
0-7803-8533-0
DOI :
10.1109/ICC.2004.1312553