A simple ON/OFF logarithmic model for frame-level errors in wireless channels applied to GSM

An accurate modelling of errors at link layer in wireless networks is especially relevant as it is indispensable for evaluating system performance and for tuning communication protocols at higher layers. In this paper, a simple algorithm to model frame-level errors in wireless channels has been developed. Such a model has been tested, analyzing an experimental GSM link layer frame error trace. The developed model is based on a two state semi-Markov process, where the holding time of each state is characterized by a discrete logarithmic distribution, whose parameter has been estimated from the data trace using the maximum likelihood method. The obtained results in terms of frame error rate (FER) and autocorrelation function have been compared with the GSM trace and with another model already available from the literature (i.e., the extended ON/OFF model). The numerical results show that the proposed method is able to capture first and second order statistics (FER and autocorrelation) with similar accuracy to the extended ON/OFF model, but with minor computational complexity.