Hybrid ML-MP Digital Predistortion for MIMO Power Amplifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital predistortion (DPD) methods for massive multiple input multiple output (M-MIMO) systems face challenges in handling the nonlinearities of power amplifiers, particularly in hybrid analog-digital (HAD) beamforming, due to high computational complexity, limited applicability, and dynamic traffic effects, leading to significant out-of-band emissions and signal degradation.
Innovation Solution
A combination model utilizing both a machine learning (ML) model and a memory polynomial (MP) model is employed for DPD, where training of the ML and MP models is alternated periodically to compensate for power amplifier nonlinearities in HAD beamforming M-MIMO systems, enhancing linearization performance under static and dynamic traffic conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a memory polynomial (MP) model is used for digital predistortion in M-MIMO systems, then DPD performance is improved, but computational complexity increases due to the necessity of having DPD in each transmitter chain
Solution Approach 1:
The system divides the M-MIMO transmitter into multiple independent transmitter chains, each with its own power amplifier and DPD module. By segmenting the DPD implementation to operate independently in each chain rather than as a centralized complex system, the computational burden is distributed and reduced while maintaining effective DPD performance for each PA.
Solution Approach 2:
The patent applies DPD selectively to each individual transmitter chain rather than attempting to compensate for all nonlinearities in the entire M-MIMO system simultaneously. This partial action approach focuses computational resources on local DPD tasks in each chain, reducing overall complexity while achieving sufficient linearization performance.
2Measurement precision
If an MP model is used for DPD, then significant modeling performance is achieved in single antenna systems, but applicability is limited due to validity only within a narrow power range
Solution Approach 1:
The DPD system is designed to dynamically adapt to changing power conditions in each transmitter chain. By making the DPD model dynamic and responsive to real-time operating conditions rather than static, the system maintains accurate modeling performance across a wide power range, overcoming the narrow validity limitation of traditional MP models.
Solution Approach 2:
The patent changes the operating parameters of the DPD system by implementing separate DPD modules in each transmitter chain that can independently adjust their parameters based on local power conditions. This allows the system to maintain optimal DPD performance across varying power levels by adapting parameters locally rather than being constrained by a fixed narrow-range model.
3Use of energy by moving object
If complex power amplifier architectures such as multiband and multimode PAs are used, then energy efficiency is improved, but nonlinearity compensation becomes difficult to achieve with desired linear gain
Solution Approach 1:
The system segments the complex multiband/multimode power amplifier architecture into separate transmitter chains, each with its own DPD module. This segmentation allows each DPD module to independently compensate for nonlinearities in its associated PA without being affected by the complexity of other bands or modes, making nonlinearity compensation achievable despite the complex PA architecture.
Solution Approach 2:
The DPD module acts as an intermediary between the input signal and the complex power amplifier. By placing DPD in each transmitter chain as a mediator that pre-distorts the signal before it enters the PA, the system compensates for the difficult-to-manage nonlinearities of multiband/multimode PAs, enabling reliable linear gain despite their complex architecture.
Data Source
AI summary
Embodiments described herein relate to methods and apparatuses for performing digital predistortion, DPD, to provide a transmit signal, x̆(n) wherein the transmit signal, x̆(n), is for deriving one or more amplifier signals, x̆m(n), for driving one or more power amplifiers, wherein the one or more power amplifiers are associated with a respective one or more antenna elements. A method comprises: receiving a first signal, x(n); inputting the first signal, x(n), into a combination model, wherein the combination model comprises a machine learning, ML, model and a memory polynomial, MP, model; and outputting the transmit signal x̆(n), from the combination model.