Multi-stage adaptive predistortion of HPA saturation effects for digital television transmission

This paper presents a new structure for adaptive predistortion of the memoryless, nonlinear saturation effects causedby High Power Amplifiers (HPA). Timely compensation for HPA distortions is critical for cost-effective prevention of the cliff effect during terrestrial transmission of digital broadcast television. The new structure results from a two-stage approach: The forward model is identified first from measured data, and then the inverse to the forward model is computed. Replacing the analog system by the HPA forward model in the second stage eliminates measurement noise and analog system delays, resulting in faster adaptation and less solution bias. Additionally, block processing reduces noise in the forward modeling stage. In the inverse modeling stage, the use of synthetic data and a closed form expression for the gradient result in more efficient convergence and more accurate solutions. Simulation results using measured HPA data demonstrate an average 5 dB improvement in SINAD for standard SNR operating ranges.