TCN-DPD: Parameter-Efficient Temporal Convolutional Networks for Wideband Digital Predistortion

-- Digital predistortion (DPD) is essential for mitigating nonlinearity in RF power amplifiers, particularly for wideband applications. This paper presents TCN-DPD, a parameter-efficient architecture based on temporal convolutional networks, integrating noncausal dilated convolutions with optimized activation functions. The increasing demand for high-efficiency, wideband communication systems has intensified the need for effective power amplifier (P A) linearization techniques. As modern wireless systems push toward broader bandwidths and higher efficiency requirements, digital predistortion (DPD) has become indispensable for maintaining signal quality while allowing P As to operate in their efficient nonlinear regions [1]. The Generalized Memory Polynomial (GMP) model has long been the industry standard for P A linearization [2].