Simpler than FRESH Filter: A Parametrri...

A key challenge in narrowband power line communications (NB-PLC) is the mitigation of impairments introduced by the correlated cyclostationary noise. The frequency-shift (FRESH) filtering approach has been recently proposed to reproduce a cyclostationary NB-PLC noise with characteristics similar to those obtained from field measurements. In this paper, we use a classification of the noise generated by the FRESH filter into three classes to propose a simple method to produce noise samples with statistics similar to those obtained at its output. The approach consists of parametrized spectral and temporal shapings applied to a white Gaussian noise sequence. We validate our proposed method by comparing its generated noise samples with those obtained using the FRESH filter and using measurements in terms of: i) normalized mean-squared error between the cyclic auto-correlations; ii) bit error rate.