Transmitter-TOR Frequency Response Calibration for Wideband DPD

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional systems struggle to accurately model the frequency response of transmit observation receivers (TORs) for wideband transmit signals, leading to performance issues in transmitters due to varying frequency responses over bandwidth, which delays adaptive compensation for nonlinearity in power amplifiers.

Innovation Solution

A Wiener-Hammerstein model is trained by connecting a nonlinear component between the transmitter and TOR, combining linear filters and quasi-memoryless nonlinearity to create a three-box model that compensates for the frequency responses of both the transmitter and TOR, allowing for inverse modeling to equalize frequency responses during operation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a simple delay and gain adjustment component is used to model the transmitter-TOR path, then the device complexity is reduced, but the measurement precision deteriorates for wideband signals due to varying frequency response

Engineering Contradiction:
Improvemodel complexityVSAvoidfrequency response modeling accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the parameters of the modeling approach by transitioning from a simple delay and gain model to a frequency-dependent model that accounts for varying frequency response across the bandwidth. This involves introducing frequency-selective filtering parameters to accurately capture the TOR's frequency response characteristics for wideband signals.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent segments the frequency response modeling into multiple frequency bins or subbands, allowing independent characterization of the frequency response at different frequencies. This segmentation enables accurate modeling of wideband signals by treating each frequency component separately rather than using a single averaged model.

Inventive Principle:
Principle #1Segmentation

2Device complexity

If the frequency response of the TOR is not accurately modeled, then the device complexity is reduced, but the transmitter performance deteriorates due to slower adaptation of the digital predistorter

Engineering Contradiction:
Improvemodeling system complexityVSAvoidadaptation speed
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent implements a feedback mechanism where the accurately modeled frequency response information is fed back to the baseband processor to guide the adaptation of the digital predistorter. This feedback loop enables the system to quickly converge to optimal predistortion parameters by providing accurate error signals that reflect the actual frequency-dependent characteristics of the transmit path.

Inventive Principle:
Principle #23Feedback

3Ease of operation

If conventional modeling approaches are used for wideband signals, then the ease of operation is maintained, but the transmitter performance deteriorates due to residual InterModulation Distortion

Engineering Contradiction:
Improveoperational simplicityVSAvoidInterModulation Distortion
Core Design Contradiction:
Ease of operationVSObject-generated harmful factors

Solution Approach 1:

The patent applies preliminary frequency response modeling and compensation before the signal undergoes nonlinear amplification. By pre-characterizing the frequency-dependent losses and phase shifts in the transmit path using wideband calibration signals, the system can pre-compensate for these effects in the digital domain, thereby reducing the residual IMD that would otherwise be generated by the power amplifier.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS9036732B2Modeling transmitter and/or transmit observation receiver frequency response and utilization thereof
Publication Date: 2015.05.19 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • US9036732B2 patent drawing
  • US9036732B2 patent drawing
  • US9036732B2 patent drawing

AI summary

Systems and methods for training, or calibrating, a model of a frequency response of a transmitter and/or a model of a frequency response of a transmit observation receiver coupled to an output of the transmitter are disclosed. In one embodiment, in order to train a model of the frequency response of the transmitter and/or a model of the frequency response of the transmit observation receiver, a nonlinear component is connected between an output of the transmitter and an input of the transmit observation receiver. A combined model for the frequency response of the transmitter, a nonlinear characteristic of the nonlinear component, and the frequency response of the transmit observation receiver is then trained. Preferably, once the combined model is trained, the nonlinear component is disconnected for normal operation of the transmitter and the transmit observation receiver.