Evaluation of soft-bit error sequence generators at the output of the decoding process

Soft decision decoding algorithms are widely used in modern wireless systems; convolutional and turbo codes are usually adopted as the inner scheme thanks to their capability to correct symbol errors at low SNR values. Such algorithms can achieve high coding gains using soft decoding, and modern digital hardware technology enables efficient and low cost practical implementations. We apply the experience gained in previous work, concerning the simulation of bit error processes (Costamagna, E. et al., Proc. IEEE, vol.90 p.842-59, 2002), to implement soft-bit generative models based on hidden Markov chains and chaotic attractors. Both the input and the output of the demodulation process of a GSM-GPRS and a 3GPP UMTS transceiver are observed, developing our earlier analysis (Costamagna et al., IEEE 58th VTC, 2003), and the quality of the soft-bit sequences generated for the input is evaluated comparing the sequences obtained at the output of the demodulator when simulated or target sequences are supplied at the input. Moreover, the deep significance of some statistical features exhibited by the sequences in order to describe their error burst behavior is briefly discussed.