Modified hidden semi-Markov model for modelling the flat fading channel

Fading phenomena impact the performance of wireless communication systems. We propose the modified hidden semi-Markov model (MHSMM) for modeling the flat fading envelope process. The properties of the envelope process are dominated by the physical fading processes and speeds of the mobile terminal. Thus, the statistics of the fading process may be non-stationary, due to different fading conditions over some time durations. The MHSMM incorporate these time-variant statistics of the envelope process in a single model, which facilitates computations of the envelope probability density function and the autocorrelation function. We provide a parameter estimation scheme to estimate the MHSMM parameters. We demonstrate the proposed parameter estimation scheme by using simulated and experimental data. The simulation is based on the GSM system parameters often encountered in practical conditions. We conducted two experiments, including cooperating and non-cooperating channel disturbances. Based on these data, the MHSMM is compared with the amplitude-based finite-state Markov chains model and the hidden Markov model. The results verify the advantages of the MHSMM and the effectiveness of the associated parameter estimation schemes. Index Terms--Fading channel model, modified hidden semi-Markov model, envelope probability density function, autocorrelation function, flat fading.