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

2025-06-13

AI Summary

To address the challenge of nonlinear distortion compensation in broadband radio-frequency power amplifiers (RF PAs), this paper proposes TCN-DPD, a lightweight temporal convolutional network for digital pre-distortion (DPD). TCN-DPD introduces a novel parameter-scalable, non-causal dilated TCN architecture, integrated with an optimized Swish variant activation function, enabling efficient end-to-end temporal modeling and DPD. With only 500 parameters, it achieves state-of-the-art (SOTA) performance; remarkably, it still outperforms existing methods with merely 200 parameters. Evaluated on the DPA_200MHz dataset, TCN-DPD attains adjacent channel power ratios (ACPR) of −51.58 dBc (left) and −49.26 dBc (right), error vector magnitude (EVM) of −47.52 dB, and normalized mean square error (NMSE) of −44.61 dB—demonstrating substantial improvements in wideband dynamic-range linearization efficiency. The model has been integrated into the OpenDPD framework.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Non-convex Optimization
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

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. Evaluated on the OpenDPD framework with the DPA_200MHz dataset, TCN-DPD achieves simulated ACPRs of -51.58/-49.26 dBc (L/R), EVM of -47.52 dB, and NMSE of -44.61 dB with 500 parameters and maintains superior linearization than prior models down to 200 parameters, making it promising for efficient wideband PA linearization.

Problem

• Mitigate RF power amplifier nonlinearity in wideband applications
• Propose parameter-efficient temporal convolutional networks for DPD
• Achieve superior linearization with minimal model parameters

Innovation

• Parameter-efficient temporal convolutional networks
• Noncausal dilated convolutions integration
• Optimized activation functions for DPD