A forward-only recursion based HMM for modeling burst errors in digital channels

This paper presents a last and efficient algorithm to model error the burst patterns that occur in many communication channels. As an application, we characterise the indoor wireless channel using our algorithm. Our proposed modeling algorithm is a modified hidden Markov modeling (HMM) technique that uses the forward-only algorithm proposed by Tan et al. Our algorithm is better than the existing standard HMM algorithms in four ways. Firstly it does not perform any computation during the length of any burst of observation symbols. Secondly unlike the existing algorithms, our method provides uniquely identifiable models. The third advantage is that the use of the forward-only recursion allows us to avoid the backward recursion variable and thus save on memory. Last of all, our method also distributes the computation evenly over each observation sequence burst. The indoor wireless channel model that we develop is based on Rappaport´s (1991) line-of-sight (LOS) multipath model. The channel exhibits statistically complex fading characteristics for which our method yields easy-to-use models that are derived from more general concepts