RF power amplifier behavioral modeling using wavelet multiresolution

The polynomial-type structure is widely used for modeling Power Amplifiers (PA) in wireless applications, but it may present some limits when the adjacent channel response is modeled. We propose a new technique that consists of decomposing the signal in different resolution levels by means of the Stationary Wavelet Transform. The technique is applied with a waveform corresponding to an OFDM with a 8 MHz bandwidth. Computer simulations confirm the new technique improves the Normalized Mean Square Error (NMSE) about 4.5 dB and 6.24 dB in Adjacent Channel Error Power Ratio (ACEPR).