Computationally efficient table based modeling for digital RF transmitters

In this paper a lookup table (LUT) based model, suitable for describing the nonlinear behavior of digital radio frequency (RF) power amplifiers (PA), is introduced. Futhermore, the LUT model is compared to nonlinear auto-regressive (ARX) models with respect to performance and computational complexity. This is accomplished by modeling a digital RF transmitter with both techniques and comparing the results to measurements. Both models are identified in time domain using short and generic excitation sequences. Having access to such models is essential when designing digital RF transmitters in order to cope with undesired nonlinear distortion. Only if sufficiently accurate models are available, linearization methods like predistortion can be developed without complicated and costly hardware in the loop setups. This paper addresses the advantages and weakpoints of those two methods based on measurements of Triquint´s TGA4954-SL as the device under test (DUT).