On The Sensitivity of RF Transmitters´ Memory Polynomial Model Identification to Delay Alignment Resolution

In this letter, the effects of delay alignment between the input and output baseband waveforms on the performance of a radio frequency transmitters´ memory polynomial based model are studied. A 100-W average power transmitter is characterized; various delay values are applied to align the input and output data; and, a model is derived for each delay value. The models´ performances, evaluated for these delay values both in time and frequency domains, demonstrate that the memory polynomial model is sensitive to delay overestimation but not to delay underestimation. It is established that a delay underestimation by up to one sampling period does not affect the performance of the identified model. This overcomes the need for the signal oversampling required for high-resolution delay alignment. Consequently, the computational complexity of the digital signal processing algorithm employed for delay estimation and alignment is considerably reduced.