Long sequences of error gaps derived from chaotic generators optimized for short ones in mobile radio channels

Chaotic generators derived from Lorenz attractors have been already introduced to model the error gap process in fading transmission channels. Although the parameters of the generators have to be optimized observing short sequences, long runs are often needed during simulations at system level. However, only very preliminary investigations have been reported in the literature on the statistical properties of long sequences generated during long runs of generators already optimized for short ones. This is a very important topic when considering the suitability of chaotic models to match correlation and long range dependence features of the target sequence, and to supply input error processes to system level and network simulators exploiting their fast bit generation rate. A deeper analysis is performed in this paper, showing a major cause of the uneven behavior of sequences of different lengths.