Modeling of wireless channel behavior for fast simulations: a soft-bit level perspective

In this work, two ways of generating soft-bit received sequences are proposed. In particular, chaos based models and hidden Markov models, which have been well used in recent years to build hard generative error sequences, are trained by soft-bit sequences sampled at the output of a link level simulator of a GSM-GPRS transceiver. Modern wireless standards make use of inner codes, like convolutional or turbo ones, that perform better with soft decision decoding techniques than with hard ones. For this reason, a fast generative model is needed in order to run outer decoding and upper layer fast simulations.