Stochastic optimization in digital pre-distortion of the signal

TL;DR

This paper evaluates modern stochastic optimization methods for digital pre-distortion in wireless communication, introduces a new testing framework, and demonstrates improved performance and efficiency over existing methods.

Contribution

It identifies the most effective stochastic optimization method for DPD, proposes a quasi-online testing framework, and validates improvements in real-life conditions.

Findings

Maximum 7% improvement in depth in standard regime

Halved working time with 3-6% depth improvement

Adamax outperforms other methods in online regime

Abstract

In this paper, we test the performance of some modern stochastic optimization methods and practices in application to digital pre-distortion problem, that is a valuable part of processing signal on base stations providing wireless communication. In first part of our study, we focus on search of the best performing method and its proper modifications. In the second part, we proposed the new, quasi-online, testing framework that allows us to fit our modelling results with the behaviour of real-life DPD prototype, retested some selected of practices considered in previous section and approved the advantages of the method occured to be the best in real-life conditions. For the used model, maximum achieved improvement in depth was 7% in standard regime and 5% in online one (metric itself is of logarithmic scale). We also achieved a halving of the working time preserving 3% and 6% improvement in depth for the standard and online regime, correspondingly. All comparisons are made to the Adam method, which was highlighted as the best stochastic method for DPD problem in paper [Pasechnyuk et al., 2021], and to the Adamax method, that is the best in the proposed online regime.