Granted Patent B2

US 12,261,629 · App. 18/639,673 · Granted Mar 25, 2025

Inventors: Nikola Alic (La Jolla, CA); Andreja Radosevic (Los Angeles, CA)

Assignee: NTWINE, LLC

H03M13/6331G06N3/04H03M13/1197H03M13/13H03M13/27H03M13/2957H04L25/03203H04L25/03834H04L27/2628

Abstract

Claims (40)

1. A method comprising:

providing an input signal comprising symbols;

encoding the symbols using an error control code encoder in a transmitter of a communication system to produce encoded symbols, wherein the error control code encoder appends redundant information onto the symbols;

interleaving the encoded symbols using a first interleaver in the transmitter to produce interleaved symbols;

intentionally introducing memory into the interleaved symbols in the form of inter-symbol interference using a pulse-shaping filter in the transmitter to produce pulse-shaped symbols;

transmitting the pulse-shaped symbols to a receiver of the communication system over a physical channel with noise;

receiving the transmitted pulse-shaped symbols using a receiving filter in the receiver to produce a received filtered signal; and

processing the received filtered signal using a neural network to detect and retrieve the encoded symbols;

wherein the neural network is trained with positive mappings between training signals output from a training transmission channel of a training communication system and decoded training signals as well as negative mappings between the training signals output from the training transmission channel and erroneous decoded signals that are known to contain errors.

2. The method of claim 1 , wherein the physical channel with noise comprises a physical memoryless channel with additive white Gaussian noise.

3. The method of claim 1 , wherein the receiving filter is matched to the pulse-shaping filter.

4. The method of claim 1 , wherein the decoded training signals are produced using a method comprising:

receiving transmitted pulse-shaped training symbols using a training receiving filter in a training receiver of the training communication system to produce a training receiving filtered signal, wherein the received pulse-shaped training symbols were transmitted over the physical channel with noise;

equalizing the training receiving filtered signal using a training information-retrieving equalizer in the training receiver to produce equalized training symbols, wherein the training information-retrieving equalizer is a trellis-based equalizer;

deinterleaving the equalized training symbols using a training deinterleaver in the training receiver to produce deinterleaved training symbols;

processing the deinterleaved training symbols using a non-linear function block in the training receiver to produce a non-linear training signal, wherein the non-linear function block transforms the deinterleaved training symbols according to a non-linear function;

decoding the non-linear training signal using a training error control code decoder in the training receiver to produce decoded training symbols; and

iteratively updating the decoded training symbols using a turbo equalization loop in the training receiver to produce updated decoded training symbols after each iteration.

5. The method of claim 4 , wherein the turbo equalization loop in the training receiver comprises iteratively repeating:

equalizing iterated interleaved training symbols using the training information-retrieving equalizer to produce iterated equalized training symbol likelihoods;

deinterleaving the iterated equalized training symbol likelihoods using a deinterleaver in the turbo equalization loop to produce iterated deinterleaved training symbol likelihoods;

transforming the iterated deinterleaved training symbol likelihoods using the non-linear function to produce iterated non-linearly transformed training symbol likelihoods;

decoding the iterated non-linearly transformed training symbol likelihoods to produce updated decoded training symbol likelihoods; and

interleaving the updated decoded training symbol likelihoods using a second training interleaver in the turbo equalization loop to produce an iterated interleaved training signal comprising the updated decoded training symbols; wherein:

the training error control code decoder comprises code that is optimized based on the intentionally introduced memory into the interleaved training symbols, a code rate, a signal-to-noise ratio, and an equalizer structure in the training receiver;

the training communication system is bandwidth constrained; and

the transmitted training signal output from the training transmission channel comprises an information rate that is higher than for an equivalent system without intentional introduction of the memory at the transmitter.

6. The method of claim 1 , wherein the training transmission channel introduces noise into the training signals, and the noise comprises one or more of white Gaussian noise, colored noise, and phase noise.

7. The method of claim 1 , wherein the training signals were produced using a method comprising:

providing a training input signal comprising training symbols;

encoding the training symbols using a training error control code encoder in a training transmitter of the training communication system to produce encoded training symbols, wherein the training error control code encoder appends redundant information onto the training symbols;

interleaving the encoded training symbols using a first training interleaver in the training transmitter to produce interleaved training symbols;

intentionally introducing memory into the interleaved training symbols in the form of inter-symbol interference using a training pulse-shaping filter in the training transmitter to produce pulse-shaped training symbols;

transmitting the pulse-shaped training symbols to a training receiver of the training communication system over the training transmission channel, wherein the training transmission channel adds noise to the pulse-shaped training symbols; and

receiving the transmitted pulse-shaped training symbols using a training receiving filter in the receiver to produce a received filtered signal.

8. The method of claim 1 , wherein the neural network performs functions of one or more of an equalizer, a BCJR equalizer, a de-mapper, an error-control code decoder, an LDPC decoder, an interleaver, a deinterleaver, a turbo equalization loop, a non-linear function block, a coding block, an iterative information processing block, and an impulse response adjustment block.

9. The method of claim 1 , wherein the neural network performs functions of an equalizer in the receiver.

10. The method of claim 1 , wherein the neural network performs functions of an error control code decoder in the receiver.

11. The method of claim 1 , wherein the neural network comprises a recursive NN, a recurrent NN (RNN), a long short-term memory RNN, a gated recurrent unit NN, an Elman NN, and/or a Jordan NN.

12. The method of claim 1 , wherein the encoding the symbols using the error control code encoder further comprises the error control code encoder using an error control code that is a Low-Density Parity-Check (LDPC) code, a turbo code, or a polar code.

Assignments (1)

Continuity (3)

References Cited (50)