RFpoweramplifiers(PAs)areusuallyconsideredasmemorylessdevicesinmostexistingpredistortiontechniques.Nevertheless,inwidebandcommunicationsystems,PAmemoryeffectscannolongerbeignoredandmemorylesspredistortioncannotlinearizePAseffectively.AfteranalyzingPAmemoryeffects,anovelpredistortionmethodbasedonwaveletnetworks(WNs)isproposedtolinearizewidebandRFpoweramplifiers.Acomplexwaveletnetworkwithtappeddelaylinesisappliedtoconstructthepredistorterandthenacomplexbackpropagationalgorithmisdevelopedtotrainthepredistorterparameters.Thesimulationresultsshowthatcomparedwiththepreviouslypublishedfeed-forwardneuralnetworkpredistortionmethod,theproposedmethodprovidesfasterconvergencerateandbetterperformanceinreducingout-of-bandspectralregrowth.
