Integrating volterra series model and deep neural networks to equalize nonlinear power amplifiers

Patent metadata

Field	Value
Publication number	US-11855813-B2
Application number	US-202217947577-A
Country	US
Kind code	B2
Filing date	Sep 19, 2022
Priority date	Mar 15, 2019
Publication date	Dec 26, 2023
Grant date	Dec 26, 2023

Abstract

The nonlinearity of power amplifiers (PAs) has been a severe constraint in performance of modern wireless transceivers. This problem is even more challenging for the fifth generation (5G) cellular system since 5G signals have extremely high peak to average power ratio. Nonlinear equalizers that exploit both deep neural networks (DNNs) and Volterra series models are provided to mitigate PA nonlinear distortions. The DNN equalizer architecture consists of multiple convolutional layers. The input features are designed according to the Volterra series model of nonlinear PAs. This enables the DNN equalizer to effectively mitigate nonlinear PA distortions while avoiding over-fitting under limited training data. The non-linear equalizers demonstrate superior performance over conventional nonlinear equalization approaches.

First claim

The invention claimed is:

1. A distortion-compensating processor, comprising:
at least one automated processor configured to decompose a non-linearly distorted signal derived from an information signal, received from a channel having a channel non-linear distortion into a truncated series expansion of at least third order with memory comprising a series of terms, each term representing incremental non-linearity order and associated delay;
an adaptive multi-layer feedforward deep neural network comprising a plurality of hidden layers, and at least one dropout layer, receiving as inputs the series of terms, and producing an equalized output signal; and
an output port configured to present the equalized output signal,
the multi-layer feedforward deep neural network being trained with respect to the channel non-linear distortion associated with communication of a series of symbols using training data comprising the series of terms, to equalize the signal,
the multi-layer feedforward deep neural network being configured to receive the respective terms associated with incremental non-linearity orders and associated delay values, and to selectively produce the equalized output signal, representing the information signal wherein the channel non-linear distortion is reduced.

2. The distortion-compensating processor according to claim 1 , wherein the training data comprises a set of small amplitude training signals to estimate a channel response and a set of large amplitude training signals to estimate a power amplifier non-linearity.

3. The distortion-compensating processor according to claim 1 , wherein the information signal is distorted by amplification by a radio frequency power amplifier and transmission through a radio frequency communication channel, wherein the non-linearly distorted signal is received by the at least one automated processor from a radio receiver.

4. The distortion-compensating processor according to claim 1 , wherein the series expansion of at least third order with memory comprises a Volterra series expansion.

5. The distortion-compensating processor according to claim 4 ,
wherein the terms of the Volterra series expansion are defined by:

y
⁡
(
n
)

=

∑

d
=
0

D

∑

k
=
0

P

b
kd

⁢

x
⁡
(

n
-
d

)

⁢

❘
"\[LeftBracketingBar]"

x
⁡
(

n
-
d

)

❘
"\[RightBracketingBar]"

k
-
1

;

and
the multi-layer feedforward neural network is trained with respect to the channel non-linear distortion to produce:

𝓏
⁡
(
n
)

=

∑

k
=
1

P

∑

d
1

=
0

D

…
⁢

∑

d
k

=
0

D

f

d
1

,
…

,

d
k

⁢

∏

i
=
1

k

r
⁡
(

n
-

d
i

)

,

where:
y(n) represents the non-linearly distorted signal;
z(n) is the equalized output signal;
d is a memory depth parameter;
D is a total memory length;
k is a respective nonlinearity order;
P is a total nonlinearity order;
k is a nonli

Assignees

• Univ New York State Res Found
• The Res Foundation For Suny

Inventors

• Li Xiaohua
• Thompson Robert

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• H04L25/03165 Primary

  using neural networks · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title

• G06N3/08

  Learning methods · CPC title