Characterization of bursty channels using Markov models
S. Srinivas, K. Sam Shanmugan · 2002
The modeling of error burst characteristics of communication channels is considered. Two techniques, hidden Markov modeling (HMM) and Fritchman characterization, are discussed. While the Fritchman method yields a model faster, it is assumed that the error burst (and error free run) lengths are statistically independent. The HMM algorithm is computationally intensive. A modified HMM algorithm is presented. It is computationally very efficient and better suited for channel models requiring more than one bad state.>