CN118075077A - OFDM digital communication system and method based on reinforcement learning - Google Patents

Abstract

An OFDM digital communication system and method based on reinforcement learning belongs to the technical field of digital communication. In an OFDM digital communication system, a modulation calculation module comprises an M-level modulation calculation submodule, an M-level modulation calculation submodule comprises 2 ^M‑m modulation basic units, and each modulation basic unit comprises two input ends and an output end; 2 ^M‑m modulation basic units of the 1 st-stage modulation calculation sub-module are input with 2 ^M mapping symbols in parallel, and the output of the modulation basic units of the M-1 st-stage modulation calculation sub-module is used as the input of the modulation basic units of the M-stage modulation calculation sub-module, and the rotation factors are sequentially utilized to perform operation until the M-stage modulation calculation sub-module sequentially and serially outputs OFDM symbols to be transmitted, wherein m=1, 2 … and M. In the invention, when information is transmitted, the modulation module directly outputs serial information; when receiving information, the decoding module directly outputs serial information without parallel-serial conversion, thereby simplifying the OFDM digital communication system and further saving the cost.

Description

An OFDM digital communication system and method based on reinforcement learning

Technical Field

The invention relates to an OFDM digital communication system and method based on reinforcement learning, belonging to the technical field of digital communication.

Background technique

For example, a Chinese invention application with publication number CN103595685A discloses a SIM-OFDMOFDM digital communication method. In the Chinese invention application, when sending information, the bit information to be transmitted is subjected to serial-to-parallel conversion, IFFT conversion and parallel-to-serial conversion; when receiving information, the received information is subjected to serial-to-parallel conversion, FFT conversion and parallel-to-serial conversion. That is, when sending information, the information after IFFT conversion needs to be subjected to parallel-to-serial conversion to be converted into serial output information, and when receiving information, the information after FFT conversion also needs to be subjected to parallel-to-serial conversion to be converted into serial output information, which increases the complexity of the OFDM digital communication system.

Summary of the invention

In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide an OFDM digital communication system and method based on reinforcement learning. When sending information, serial information is directly output by the modulation module; when receiving information, the decoding module also directly outputs serial information, and there is no need to perform parallel-to-serial conversion, thereby simplifying the OFDM digital communication system and saving costs.

In order to overcome the shortcomings of the prior art, the present invention provides an OFDM digital communication system based on reinforcement learning, which includes a sending device, the sending device includes a speech recognition module, an encoding module, a first storage module and a modulation module, the speech recognition module recognizes speech information as a text string, and includes a self-attention mechanism model; the encoding module encodes the text string to generate a bit information sequence; the modulation module includes a modulation segmentation module, a modulation symbol mapping module and an OFDM symbol generation module, the modulation segmentation module divides the bit information sequence into a group of K bits B={b ₀ … b _k … b _K-1 } to obtain multiple groups of bit groups and provides them to the modulation symbol mapping module; the modulation symbol mapping module maps the bit information sequence into a mapping symbol sequence {X(n)}, n=0,1,…, The OFDM symbol generation module includes a modulation code bit reverse reading module, a modulation control counting module, a modulation rotation factor generation module and a modulation calculation module. The code bit reverse reading module reads 2 ^M mapping symbols X(q) from the mapping symbol sequence {X(n)} in binary code bits and stores them in the first storage module. The first storage module outputs the 2 ^M mapping symbols X(q) input by the code bit reverse reading module in parallel. The modulation rotation factor generation module includes an M-level modulation rotation factor generation submodule. The m-th level modulation rotation factor generation submodule generates a rotation factor under the control of the modulation control counting module. , m = 1, 2, ..., M, M is a positive integer greater than or equal to 2; p = 0, 1, 2, ..., 2 ^M -1; q = 0, 1, 2, ..., 2 ^M -1;

The modulation calculation module includes M-level modulation calculation submodules, and the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, each of which includes two input terminals and one output terminal; the 2 ^Mm modulation basic units of the first-level modulation calculation submodule input 2 ^M mapping symbols X(q) in parallel, and the output of the modulation basic unit of the m-1-th level modulation calculation submodule is used as the input of the modulation basic unit of the m-th level modulation calculation submodule, and the rotation factors are used in turn. The calculation is performed until the M-th level modulation calculation submodule sequentially outputs the OFDM symbols to be transmitted .

In order to overcome the shortcomings of the prior art, the present invention also provides an OFDM digital communication method based on reinforcement learning, which converts a bit information sequence into an OFDM symbol sequence through a modulation module, wherein the modulation module includes a modulation segmentation module, a modulation symbol mapping module and an OFDM symbol generation module, wherein the modulation segmentation module divides the bit information sequence into a group of K bits B={b ₀ …b _k …b _K-1 } to obtain multiple groups of bit groups and provides them to the modulation symbol mapping module; the modulation symbol mapping module maps the bit information sequence into a mapping symbol sequence {X(n)}, n=0, 1, …, The OFDM symbol generation module includes a modulation code bit reverse reading module, a modulation control counting module, a modulation rotation factor generation module and a modulation calculation module. The code bit reverse reading module reads 2 ^M mapping symbols X (q) from the mapping symbol sequence {X (n)} in binary code bits and stores them in a first storage module. The first storage module outputs the 2 ^M mapping symbols X (q) input by the code bit reverse reading module in parallel. The modulation rotation factor generation module includes M modulation rotation factor generation submodules. The m-th modulation rotation factor generation submodule generates rotation factors respectively under the control of the control counting module. , m=1,2,…,M, M is a positive integer greater than or equal to 2; p=0,1,2,…,2 ^M -1; q=0,1,2,…,2 ^M -1;

The modulation calculation module includes M-level modulation calculation submodules, and the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, each of which includes two input terminals and one output terminal; the 2 ^Mm modulation basic units of the first-level modulation calculation submodule input 2 ^M mapping symbols in parallel, and the output of the modulation basic unit of the m-1-th level modulation calculation submodule is used as the input of the modulation basic unit of the m-th level modulation calculation submodule, and the rotation factors are used in turn. The calculation is performed until the M-th level modulation calculation submodule sequentially outputs the OFDM symbols to be transmitted .

Compared with the prior art, the OFDM digital communication system and method based on reinforcement learning provided by the present invention have the following beneficial effects:

1. When sending information, the modulation module can generate OFDM symbol Y(p), saving a lot of hardware costs of oscillators and modulators; when receiving information, the decoding module can obtain the mapping symbol X(q) sent by the transmitter, saving a lot of hardware costs of oscillators and decoders;

2. When sending information, the modulation module directly outputs serial information; when receiving information, the decoding module directly outputs serial information without the need for parallel-to-serial conversion, thereby simplifying the OFDM digital communication system and saving costs.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a block diagram of an OFDM digital communication system based on reinforcement learning provided by the present invention;

FIG2 is a block diagram of a speech recognition module provided by the present invention;

FIG3 is a block diagram of a modulation module provided by the present invention;

FIG. 4 is a block diagram of the decoding module provided by the present invention.

Detailed ways

It should be noted that the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The advantages and features of the present invention and methods for achieving the advantages and features will become clear with reference to the accompanying drawings and the embodiments described in detail below.

However, the present invention is not limited to the embodiments disclosed below, and can be implemented in a variety of different forms. This embodiment is only used to make the disclosure of the present invention more complete and to fully inform the scope of the invention to ordinary technicians in the technical field to which the present invention belongs. The present invention is only defined by the scope of protection claimed in the invention.

Although "first", "second", etc. are used to describe various elements, components and/or parts, these elements, components and/or parts are not limited by these terms. These terms are only used to distinguish one element, constituent element or part from other elements, constituent elements or parts. Therefore, it is obvious that within the technical spirit of the present disclosure, the first element, first element or first part mentioned below may also be the second element, second element or second part, and the terms used in this specification are only used to describe the embodiments and are not intended to limit the present disclosure.

In this specification, as long as there is no special mention in the sentence, the singular also includes the plural. The use of "comprising" and/or "consisting of..." in the specification does not exclude the existence or addition of one or more other structural elements, steps, actions and/or elements of the mentioned structural elements, steps, actions and/or elements.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those commonly understood by ordinary technicians in the field to which the present invention belongs. In addition, terms defined in commonly used dictionaries shall not be interpreted ideally or excessively unless they are clearly defined.

In addition, the same reference numerals are given to the same or similar structures, and duplicate descriptions are omitted.

FIG1 is a block diagram of the OFDM digital communication system based on reinforcement learning provided by the present invention. As shown in FIG1 , the OFDM digital communication system based on reinforcement learning provided by the present invention includes a sending device, which includes a speech recognition module, an encoding module, a first storage module and a modulation module. The speech recognition module recognizes speech information as a text string, which includes an automatic attention mechanism model; the encoding module encodes the text string to generate a bit information sequence; the modulation module includes a modulation segmentation module, a modulation symbol mapping module and an OFDM symbol generation module, the modulation segmentation module divides the bit information sequence into a group of K bits B={b ₀ … b _k … b _K-1 } to obtain multiple groups of bit groups and provides them to the modulation symbol mapping module; the modulation symbol mapping module maps the bit information sequence into a mapping symbol sequence {X(n)}, n=0,1,…, The OFDM symbol generation module includes a modulation code bit reverse reading module, a modulation control counting module, a modulation rotation factor generation module and a modulation calculation module. The code bit reverse reading module reads 2 ^M mapping symbols X(q) from the mapping symbol sequence {X(n)} in binary code bits and stores them in the first storage module. The first storage module outputs the 2 ^M mapping symbols X(q) input by the code bit reverse reading module in parallel. The modulation rotation factor generation module includes an M-level modulation rotation factor generation submodule. The m-th level modulation rotation factor generation submodule generates a rotation factor under the control of the modulation control counting module. , m=1,2,…,M, M is a positive integer greater than or equal to 2; p=0,1,2,…,2 ^M -1; q=0,1,2,…,2 ^M -1; , is the subcarrier angular frequency, is the baseband signal symbol period.

In the present invention, the modulation symbol mapping module includes 2 ^M mapping symbols X(q)}, each mapping symbol corresponds to a different combination of K bits of information, and the expression of the qth mapping symbol is: .

In the present invention, the modulation code bit reverse reading module reads 2 ^M mapping symbols X(q) in reverse order from the mapping symbol sequence {X(n)} in binary code bits according to the following table, for example:

Decimal order Binary Order Bit code reverse reading order binary order Bit code reverse reading order decimal order 0 000 000 0 1 001 100 4 2 010 010 2 3 011 110 6 4 100 001 1 5 101 101 5 6 110 011 3 7 111 111 7

.

In the present invention, the modulation calculation module includes M-level modulation calculation submodules, the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, each modulation basic unit includes two input ends and one output end; the 2 ^Mm modulation basic units of the first-level modulation calculation submodule input 2 ^M mapping symbols X(q) in parallel, the output of the modulation basic unit of the m-1-th level modulation calculation submodule is used as the input of the modulation basic unit of the m-th level modulation calculation submodule, and the rotation factors are used in sequence. The calculation is performed until the M-th level modulation calculation submodule sequentially outputs the OFDM symbols to be transmitted .

When sending information, the present invention can generate OFDM symbols Y(p) through the above-mentioned modulation module, which can be obtained through calculation, and does not require a large number of hardware composed of oscillators and modulators, thereby saving a large amount of hardware costs composed of oscillators and modulators; at the same time, the modulation module directly outputs serial information without the need for parallel-to-serial conversion, thereby simplifying the OFDM digital communication system and further saving costs.

The transmitting device also includes a DAC converter and a transmitting radio frequency module. The DAC converter converts the OFDM symbol sequence {Y(p)} into analog information. The transmitting radio frequency module modulates the information provided by the DAC converter to a high frequency and performs power amplification, impedance matching, and then converts it into electromagnetic waves through a transmitting antenna and sends it into space.

FIG2 is a block diagram of the speech recognition module provided by the present invention. As shown in FIG2 , the speech recognition module includes a control component, a control module, a display module, a speech acquisition module and a language model, wherein the control component is configured to be operated by a user so as to input the user's intention into the control module, and the control module selects a text from the first storage module according to the instruction formed by the control component and outputs it to the display module for display; the user reads the displayed text according to the display of the display module, so that the speech acquisition module acquires the user's voice and inputs the acquired voice into the language model; the language model is trained using the user's voice acquired by the speech acquisition module and displays the recognized text on the display module; the self-similarity of the text input from the first storage module and the text recognized by the language model is compared, if the self-similarity is greater than or equal to the set threshold, the training of the language module is stopped, and the trained language model is applied to the OFDM digital communication system, if the self-similarity is less than the set threshold, the language model continues to be trained using the voice of the read displayed text.

In the present invention, the language model includes a self-attention mechanism model, and the feature value of the i-th text input in the first storage module is , the feature value of the i-th text output by the language model recognition is , then their correlation for:

,

In the formula, is the eigenvalue The transformation matrix, is the eigenvalue The transformation matrix of , T represents transpose, and i is a positive integer greater than or equal to 1.

The present invention, through the above-mentioned speech recognition module, can achieve the goal that different people use different speech to train the language model, thereby improving the accuracy of speech recognition for different people.

In the present invention, the OFDM digital communication system based on reinforcement learning also includes a receiving device, which includes a receiving radio frequency module and an ADC converter. The receiving radio frequency module senses spatial electromagnetic waves through a receiving antenna, and then obtains the analog information sent by the sending device through small signal amplification, frequency conversion and decoding. The ADC converter converts the received analog information into an OFDM symbol sequence {Y(p)}.

The receiving device also includes a decoding module and a second storage module. The decoding module includes a decoding segmentation module, a decoding code bit reverse reading module, a decoding control counting module, a decoding rotation factor generation module, a decoding calculation module and a symbol demapping module. The decoding segmentation module segments the received OFDM symbol sequence to obtain OFDM symbol groups, each group including ^2M OFDM symbols Y(p); the decoding code bit reverse reading module reads ^2M OFDM symbols Y(p) from the OFDM symbol group in binary code bits and stores them in the second storage module, and the second storage module outputs the ^2M OFDM symbols Y(p) input by the decoding code bit reverse reading module in parallel; the decoding rotation factor generation module includes M decoding rotation factor generation submodules, and the m-th level decoding rotation factor generation submodule generates rotation factors respectively under the control of the decoding control counting module. ,m=1,2,…,M; p=0，1， 2，…，2 ^M -1.

The decoding calculation module includes M-level decoding calculation submodules, and the m-th level decoding calculation submodule includes 2 ^Mm decoding basic units, each decoding basic unit includes two input terminals and one output terminal; the 2 ^Mm decoding basic units of the first-level decoding calculation submodule input 2 ^M OFDM symbols in parallel, and the output of the decoding basic unit of the m-1-th level decoding calculation submodule is used as the input of the decoding basic unit of the m-th level decoding calculation submodule, and the rotation factors are used in turn. The calculation is performed until the M-th level decoding calculation submodule sequentially outputs the mapping symbols sent by the sending device in series. ; The symbol demapping module will map the symbol Demap into bit groups, q = 0, 1, 2, …, 2 ^M -1.

When receiving information, the present invention can obtain the mapping symbol X(q) sent by the transmitter through a decoding module, and the decoding module can obtain it through calculation, without the need for a large number of hardware composed of oscillators and decoders, thus saving a large amount of hardware costs composed of oscillators and decoders; at the same time, the modulation module directly outputs serial information, and there is no need to perform parallel-to-serial conversion, thereby simplifying the OFDM digital communication system and further saving costs.

The receiving device also includes a synthesis module, which combines the bit information into text. If the synthesis module is a speech synthesis module, the speech synthesis module synthesizes speech according to the bit information group using a language model.

FIG3 is a block diagram of the modulation module provided by the present invention. The modulation module shown in FIG3 takes M=3 as an example, but the present invention is not limited to M=3, and M can be any positive integer. As shown in FIG3, the modulation module includes a modulation segmentation module, a modulation symbol mapping module, and an OFDM symbol generation module. The modulation segmentation module divides the bit information sequence into a group of K bits B={b ₀ b ₁ b ₂ } to obtain multiple groups of bit groups and provides them to the modulation symbol mapping module; the modulation symbol mapping module maps the bit information sequence into a mapping symbol sequence {X(n)}, n=0,1,…, The OFDM symbol generation module includes a modulation code bit reverse reading module, a modulation control counting module, a modulation rotation factor generation module and a modulation calculation module, wherein the code bit reverse reading module reads 8 mapping symbols X(q) in binary code from the mapping symbol sequence {X(n)} and stores them in the first storage module, and the first storage module outputs the 8 mapping symbols X(q) input by the code bit reverse reading module in parallel; the modulation rotation factor generation module includes M modulation rotation factor generation submodules, and the mth modulation rotation factor generation submodule generates a rotation factor under the control of the control counting module. , m=1,2,3; p=0,1,2,…,7; q=0,1,2,…,7;

As shown in FIG3 , each modulation basic unit is composed of a modulation multiplier and a modulation adder, wherein the signal input to the first input terminal of the modulation basic unit is provided to the first input terminal of the modulation adder, the signal input to the second input terminal of the modulation basic unit is multiplied by the rotation factor and then provided to the second input terminal of the modulation adder, and the modulation adder is used to add the signals input to the first input terminal and the second input terminal and provide the input signal or the OFDM symbol to be transmitted to the modulation basic unit of the next level modulation calculation submodule through its output terminal. .

As shown in Figure 3, M = 3, {X(q)} = {X ₁ (0), X ₁ (4), X ₁ (2), X ₁ (6), X ₁ (1), X ₁ (5), X ₁ (3), X ₁ (7)};

The modulation calculation module includes 3 levels of modulation calculation submodules, and the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, where m=1, 2, 3;

The first-level modulation calculation submodule includes a first modulation basic unit C ₁₁ , a second modulation basic unit C ₁₂ , a third modulation basic unit C ₁₃ and a fourth modulation basic unit C ₁₄ . The first input terminal of the first modulation basic unit C ₁₁ inputs X ₁ (0), and X ₁ (0) is input to the first input terminal of the first modulation adder P ₁₁ . The second input terminal of the first modulation basic unit C ₁₁ inputs X ₁ (4), and X ₁ (4) is combined with the rotation factor The signals are multiplied in the first modulation multiplier _M11 and then input to the second input terminal of the first modulation adder _P11 . The first modulation adder _P11 is used to perform addition operation on the signals input from the first input terminal and the second input terminal.

The first input terminal of the second modulation basic unit _C12 inputs _X1 (2), and _X1 (2) is input to the first input terminal of the second modulation adder _P12 ; the second input terminal of the second modulation basic unit _C12 inputs _X1 (6), and _X1 (6) and the rotation factor In the second modulation multiplier M ₁₂ is multiplied and then input to the second input terminal of the second modulation adder P ₁₂ , the second modulation adder P ₁₂ is used to add the signal input from the first input terminal and the second input terminal;

The first input terminal of the third modulation basic unit _C13 inputs _X1 (1), which is input to the first input terminal of _the third modulation adder _P13 ; the second input terminal of the third modulation basic unit _C13 inputs _X1 ( ₅ ), which is combined with the rotation factor In the third modulation multiplier M ₁₃ is multiplied and then input to the second input terminal of the third modulation adder P ₁₃ , the third modulation adder P ₁₃ is used to add the signal input from the first input terminal and the second input terminal thereof;

The first input terminal of the fourth modulation basic unit _C14 inputs _X1 (3), _which is input to the first input terminal of the fourth modulation adder ( _P14 ); the second input terminal of the fourth modulation basic unit _C14 inputs _X1 (7), _which is combined with the rotation factor In the fourth modulation multiplier M ₁₄ is multiplied and then input to the second input terminal of the fourth modulation adder P ₁₄ , the fourth modulation adder P ₁₄ is used to add the signal input from the first input terminal and the second input terminal thereof;

The second-level modulation calculation submodule includes a fifth modulation basic unit C ₂₁ and a sixth modulation basic unit C _22. The first input end of the fifth modulation basic unit C ₂₁ inputs the output signal of the first modulation adder P ₁₁ , and the output signal of the first modulation adder P ₁₁ is input to the first input end of the fifth modulation adder P ₂₁ ; the second input end of the fifth modulation basic unit C ₂₁ inputs the output signal of the second modulation adder P ₁₂ , and the output signal of the second modulation adder P ₁₂ is combined with the rotation factor The signals are multiplied in the fifth modulation multiplier _M21 and then input to the second input terminal of the fifth modulation adder _P21 . The fifth modulation adder _P21 is used to perform an addition operation on the signals input from the first input terminal and the second input terminal thereof;

The output signal of the third modulation adder _P13 is input to the first input end of the sixth modulation adder _C22 ; the output signal of the fourth modulation adder _P14 is input to the second input end of the sixth modulation adder _C22 ; the output signal of the fourth modulation adder _P14 is input to the second input end of the sixth modulation adder _C22 ; the output signal of the fourth modulation adder _P14 is input to the rotation factor The signals are multiplied in the sixth modulation multiplier _M22 and then input to the second input terminal of the sixth modulation adder _P22 . The sixth modulation adder _P22 is used to perform an addition operation on the signals input from the first input terminal and the second input terminal thereof.

The third-level modulation calculation submodule includes a seventh modulation basic unit C ₃₁ , a first input end of the seventh modulation basic unit C ₃₁ inputs the output signal of the fifth modulation adder P ₂₁ , and the output signal of the fifth modulation adder P ₂₁ is input to the first input end of the seventh modulation adder P ₃₁ ; a second input end of the seventh modulation basic unit C ₃₁ inputs the output signal of the sixth modulation adder P ₂₂ , and the output signal of the sixth modulation adder P ₂₂ is combined with the rotation factor The signals are multiplied in the seventh modulation multiplier _M31 and then input to the second input terminal of the seventh modulation adder _P31 . The seventh modulation adder _P31 is used to add the signals input from the first input terminal and the second input terminal to output the OFDM symbol to be transmitted. .

FIG4 is a block diagram of the composition of the decoding module provided by the present invention. The decoding module shown in FIG4 takes M=3 as an example, but the present invention is not limited to M=3, and M can be any positive integer. As shown in FIG3, the decoding module includes a decoding segmentation module, a decoding code bit reverse reading module, a decoding control counting module, a decoding rotation factor generation module, a decoding calculation module and a symbol demapping module. The decoding segmentation module segments the received OFDM symbol sequence to obtain OFDM symbol groups, and each group includes 8 OFDM symbols Y(p); the decoding code bit reverse reading module reads 8 OFDM symbols Y(p) from the OFDM symbol group in binary code bits and stores them in the second storage module, and the second storage module outputs the 8 OFDM symbols Y(p) input by the decoding code bit reverse reading module in parallel; the decoding rotation factor generation module includes M decoding rotation factor generation submodules, and the m-th level decoding rotation factor generation submodule generates the rotation factor under the control of the decoding control counting module. , m=1,2,3; p=0,1,2,…, 7;

As shown in FIG4 , each decoding basic unit is composed of a decoding multiplier and a decoding adder, wherein the signal inputted at the first input terminal of the decoding basic unit is provided to the first input terminal of the decoding adder, and the signal inputted at the second input terminal of the decoding basic unit is combined with the rotation factor The multiplication is then provided to the second input end of the decoding adder, and the decoding adder is used to perform an addition operation on the signals input from the first input end and the second input end and provide the input signal or the received mapping symbol to the decoding basic unit of the next level decoding calculation submodule through its output end .

When M=3,{Y(p)}={Y ₁ (0), Y ₁ (1), Y ₁ (2), Y ₁ (3), Y ₁ (4), Y ₁ (5), Y ₁ (6), Y ₁ (7)};

The decoding calculation module includes 3-level decoding calculation submodules, the m-th level decoding calculation submodule includes 2 ^Mm basic decoding units, m = 1, 2, 3; q = 0, 1, 2, ..., 7;

The first level decoding calculation submodule includes a first decoding basic unit C ₄₁ , a second decoding basic unit C ₄₂ , a third decoding basic unit C ₄₃ and a fourth decoding basic unit C _44. The first input terminal of the first decoding basic unit C ₄₁ inputs Y ₁ (0), and Y ₁ (0) is input to the first input terminal of the first decoding adder P ₄₁ ; the second input terminal of the first decoding basic unit C ₄₁ inputs Y ₁ (4), and Y ₁ (4) is combined with the rotation factor The signals are multiplied in the first decoding multiplier _M41 and then input to the second input terminal of the first decoding adder _P41 . The first decoding adder _P41 is used to perform addition operation on the signals inputted from the first input terminal and the second input terminal thereof;

The first input terminal of the second decoding basic unit C ₄₂ inputs Y ₁ (2), and Y ₁ (2) is input to the first input terminal of the second decoding adder P ₄₂ ; the second input terminal of the second decoding basic unit C ₄₂ inputs X ₁ (6), and X ₁ (6) and the rotation factor In the second decoding multiplier M ₄₂ is multiplied and then input to the second input terminal of the second decoding adder P ₄₂ , the second decoding adder P ₄₂ is used to add the signal input from the first input terminal and the second input terminal thereof;

The first input terminal of the third decoding basic unit C ₄₃ inputs X ₁ (1), and X ₁ (1) is input to the first input terminal of the third decoding adder P ₄₃ ; the second input terminal of the third decoding basic unit C ₄₃ inputs X ₁ (5), and X ₁ (5) is combined with the rotation factor In the third decoding multiplier M43, the multiplied signal is input to the second input terminal of the third decoding adder P ₄₃ , and the third decoding adder P ₄₃ is used to add the signal input from the first input terminal and the second input terminal thereof;

The first input terminal of the fourth decoding basic unit C ₄₄ inputs X ₁ (3), and X ₁ (3) is input to the first input terminal of the fourth decoding adder P ₄₄ ; the second input terminal of the fourth decoding basic unit C ₄₄ inputs X ₁ (7), and X ₁ (7) and the rotation factor In the fourth decoding multiplier M ₄₄ is multiplied and then input to the second input terminal of the fourth decoding adder P ₄₄ , the fourth decoding adder P ₄₄ is used to add the signal input from the first input terminal and the second input terminal thereof;

The second-level calculation submodule includes a fifth decoding basic unit C ₅₁ and a sixth decoding basic unit C _52. The first input end of the fifth decoding basic unit C ₅₁ inputs the output signal of the first decoding adder P ₄₁ , and the output signal of the first decoding adder P ₄₁ is input to the first input end of the fifth decoding adder P ₅₁ ; the second input end of the fifth decoding basic unit C ₅₁ inputs the output signal of the second decoding adder P ₄₂ , and the output signal of the second decoding adder P ₄₂ is combined with the rotation factor The signals are multiplied in the fifth decoding multiplier _M51 and then input to the second input terminal of the fifth decoding adder _P51 . The fifth decoding adder _P51 is used to perform addition operation on the signals input from the first input terminal and the second input terminal thereof;

The output signal of the third decoding adder P ₄₃ is input to the first input terminal of the sixth decoding basic unit C ₅₂ , and the output signal of the third decoding adder P ₄₃ is input to the first input terminal of the sixth decoding adder P ₅₂ ; the output signal of the fourth decoding adder P ₄₄ is input to the second input terminal of the sixth decoding basic unit C ₅₂ , and the output signal of the fourth decoding adder P ₄₄ is combined with the rotation factor The signals are multiplied in the sixth decoding multiplier _M52 and then input to the second input terminal of the sixth decoding adder _P52 . The sixth decoding adder _P52 is used to perform an addition operation on the signals input from the first input terminal and the second input terminal.

The third-level calculation submodule includes a seventh decoding basic unit C ₆₁ , a first input terminal of the seventh decoding basic unit C ₆₁ inputs an output signal of a fifth decoding adder P ₅₁ , and an output signal of the fifth decoding adder P ₅₁ is input to a first input terminal of the seventh decoding adder P ₆₁ ; a second input terminal of the seventh decoding basic unit C ₆₁ inputs an output signal of a sixth decoding adder P ₅₂ , and the output signal of the sixth decoding adder P ₅₂ is combined with the rotation factor The signals are multiplied in the seventh decoding multiplier _M61 and then input to the second input terminal of the seventh decoding adder _P61 . The seventh decoding adder _P61 is used to add the signals input from the first input terminal and the second input terminal and input the mapping symbol sent by the sending device through its output terminal. The symbol demapping module maps the symbol Demap into bit groups.

The present invention also provides an OFDM digital communication method based on reinforcement learning, which converts the bit information sequence into an OFDM symbol sequence through the above modulation module, and converts the OFDM symbol sequence into an OFDM symbol sequence through the above decoding module. Converted into the mapping symbol sequence {X(q)} sent by the transmitter.

The present invention also provides a computer program product, which encodes the modulation module and the decoding module into program codes that can be called by a processor and stored in a computer storage medium using a computer program language.

The present invention also provides a computer storage medium for storing the above program code, wherein the program code can be called and executed by a processor to implement the functions of a modulation module and a decoding module.

The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims (10)

1. An OFDM digital communication system based on reinforcement learning, comprising a sending device, characterized in that the sending device comprises a speech recognition module, an encoding module, a first storage module and a modulation module, the speech recognition module recognizes speech information as a text string, which comprises a self-attention mechanism model; the encoding module encodes the text string to generate a bit information sequence; the modulation module comprises a modulation segmentation module, a modulation symbol mapping module and an OFDM symbol generation module, the modulation segmentation module divides the bit information sequence into a group of K bits B={b ₀ … b _k … b _K-1 } to obtain a plurality of bit groups and provides them to the modulation symbol mapping module; the modulation symbol mapping module maps the bit information sequence into a mapping symbol sequence {X(n)}, n=0,1,…, The OFDM symbol generation module includes a modulation code bit reverse reading module, a modulation control counting module, a modulation rotation factor generation module and a modulation calculation module. The code bit reverse reading module reversely reads 2 ^M mapping symbols X(q) from the mapping symbol sequence {X(n)} in binary code bits and stores them in the first storage module. The first storage module outputs the 2 ^M mapping symbols X(q) input by the code bit reverse reading module in parallel. The modulation rotation factor generation module includes an M-level modulation rotation factor generation submodule. The m-th level modulation rotation factor generation submodule generates a rotation factor under the control of the modulation control counting module. , m=1,2,…,M, M is a positive integer greater than or equal to 2; p=0,1,2,…,2 ^M -1; q=0,1,2,…,2 ^M -1; The modulation calculation module includes M-level modulation calculation submodules, and the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, each of which includes two input terminals and one output terminal; the 2 ^Mm modulation basic units of the first-level modulation calculation submodule input 2 ^M mapping symbols X(q) in parallel, and the output of the modulation basic unit of the m-1-th level modulation calculation submodule is used as the input of the modulation basic unit of the m-th level modulation calculation submodule, and the rotation factors are used in turn. Perform calculations until the M-th level modulation calculation submodule sequentially outputs the OFDM symbols to be transmitted in series. . 2. According to the OFDM digital communication system based on reinforcement learning in claim 1, it is characterized in that each modulation basic unit is composed of a modulation multiplier and a modulation adder, wherein the signal inputted by the first input terminal of the modulation basic unit is provided to the first input terminal of the modulation adder, the signal inputted by the second input terminal of the modulation basic unit is multiplied by the rotation factor and then provided to the second input terminal of the modulation adder, and the modulation adder is used to perform addition operation on the signals inputted by the first input terminal and the second input terminal thereof, and provide the input signal or the OFDM symbol to be transmitted to the modulation basic unit of the modulation calculation submodule of the next level through its output terminal. . 3. The OFDM digital communication system based on reinforcement learning according to claim 2, characterized in that: M=3,{X(q)}={ _X1 (0), _X1 (4), _X1 (2), _X1 (6), _X1 (1), _X1 (5), _X1 (3), _X1 (7)} The modulation calculation module includes 3-level modulation calculation submodules, and the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, m = 1, 2, 3; q = 1, 2, ..., 7; The first-level modulation calculation submodule includes a first modulation basic unit, a second modulation basic unit, a third modulation basic unit and a fourth modulation basic unit. The first input end of the first modulation basic unit inputs X ₁ (0), and X ₁ (0) is input to the first input end of the first modulation adder; the second input end of the first modulation basic unit inputs X ₁ (4), and X ₁ (4) is combined with the rotation factor The signals are multiplied in the first modulation multiplier and then input to the second input terminal of the first modulation adder, and the first modulation adder is used to perform addition operation on the signals input from the first input terminal and the second input terminal; The first input terminal of the second modulation basic unit inputs X ₁ (2), and X ₁ (2) is input to the first input terminal of the second modulation adder; the second input terminal of the second modulation basic unit inputs X ₁ (6), and X ₁ (6) and the rotation factor Multiplied in the second modulation multiplier and then input to the second input terminal of the second modulation adder, the second modulation adder is used to add the signals input at its first input terminal and the second input terminal; The first input terminal of the third modulation basic unit inputs X ₁ (1), and X ₁ (1) is input to the first input terminal of the third modulation adder; the second input terminal of the third modulation basic unit inputs X ₁ (5), and X ₁ (5) and the rotation factor Multiplied in the third modulation multiplier and then input to the second input terminal of the third modulation adder, the third modulation adder for adding the signal input from the first input terminal and the second input terminal; The first input terminal of the fourth modulation basic unit inputs X ₁ (3), and X ₁ (3) is input to the first input terminal of the fourth modulation adder; the second input terminal of the fourth modulation basic unit inputs X ₁ (7), and X ₁ (7) and the rotation factor Multiplied in the fourth modulation multiplier and then input to the second input terminal of the fourth modulation adder, the fourth modulation adder for adding the signal input from the first input terminal and the second input terminal; The second-level modulation calculation submodule includes a fifth modulation basic unit and a sixth modulation basic unit. The first input end of the fifth modulation basic unit inputs the output signal of the first modulation adder, and the output signal of the first modulation adder is input to the first input end of the fifth modulation adder; the second input end of the fifth modulation basic unit inputs the output signal of the second modulation adder, and the output signal of the second modulation adder is input to the first input end of the fifth modulation adder. The signals are multiplied in the fifth modulation multiplier and then input to the second input terminal of the fifth modulation adder, and the fifth modulation adder is used to perform addition operation on the signals input from the first input terminal and the second input terminal; The output signal of the third modulation adder is input to the first input end of the sixth modulation adder; the output signal of the third modulation adder is input to the first input end of the sixth modulation adder; the output signal of the fourth modulation adder is input to the second input end of the sixth modulation adder, and the output signal of the fourth modulation adder is added to the rotation factor The signals are multiplied in the sixth modulation multiplier and then input to the second input terminal of the sixth modulation adder, and the sixth modulation adder is used to perform an addition operation on the signals input from the first input terminal and the second input terminal; The third-level modulation calculation submodule includes a seventh modulation basic unit, a first input end of the seventh modulation basic unit inputs the output signal of the fifth modulation adder, the output signal of the fifth modulation adder is input to the first input end of the seventh modulation adder; a second input end of the seventh modulation basic unit inputs the output signal of the sixth modulation adder, the output signal of the sixth modulation adder and the rotation factor The signals are multiplied in the seventh modulation multiplier and then input to the second input terminal of the seventh modulation adder. The seventh modulation adder is used to add the signals input from the first input terminal and the second input terminal to output the OFDM symbol to be transmitted. , p=1,2，…,7. 4. The OFDM digital communication system based on reinforcement learning according to claim 1 is characterized in that the modulation symbol mapping module includes 2 ^M mapping symbols X(q)}, each mapping symbol corresponds to a different combination of K bits of information, and the expression of the qth mapping symbol is: . 5. The OFDM digital communication system based on reinforcement learning according to claim 1, characterized in that it also includes a receiving device, the receiving device includes a decoding module and a second storage module, the decoding module includes a decoding segmentation module, a decoding code bit reverse reading module, a decoding control counting module, a decoding rotation factor generation module, a decoding calculation module and a symbol demapping module, the decoding segmentation module segments the received OFDM symbol sequence to obtain OFDM symbol groups, each group includes ^2M OFDM symbols Y(p); the decoding code bit reverse reading module reads ^2M OFDM symbols Y(p) from the OFDM symbol group in binary code bits and stores them in the second storage module, and the second storage module outputs the ^2M OFDM symbols Y(p) input by the decoding code bit reverse reading module in parallel; the decoding rotation factor generation module includes M decoding rotation factor generation submodules, and the m-th level decoding rotation factor generation submodule generates the rotation factor under the control of the decoding control counting module. ; The decoding calculation module includes M-level decoding calculation submodules, and the m-th level decoding calculation submodule includes 2 ^Mm decoding basic units, each decoding basic unit includes two input terminals and one output terminal; the 2 ^Mm decoding basic units of the first-level decoding calculation submodule input 2 ^M OFDM symbols in parallel, and the output of the decoding basic unit of the m-1-th level decoding calculation submodule is used as the input of the decoding basic unit of the m-th level decoding calculation submodule, and the rotation factors are used in turn. The operation is performed until the M-th level decoding calculation submodule sequentially outputs the mapping symbols sent by the sending device in series. ; The symbol demapping module will map the symbol /> Demap into bit groups. 6. The OFDM digital communication system based on reinforcement learning according to claim 5 is characterized in that each decoding basic unit is composed of a decoding multiplier and a decoding adder, wherein the signal input at the first input end of the decoding basic unit is provided to the first input end of the decoding adder, and the signal input at the second input end of the decoding basic unit is combined with the rotation factor The multiplication is then provided to the second input terminal of the decoding adder, and the decoding adder is used to perform an addition operation on the signals inputted from the first input terminal and the second input terminal and provide the input signal or the received mapping symbol to the decoding basic unit of the next level decoding calculation submodule through its output terminal. . 7. The OFDM digital communication system based on reinforcement learning according to claim 6, characterized in that: M=3,{Y(p)}={Y ₁ (0), Y ₁ (4), Y ₁ (2), Y ₁ (6), Y ₁ (1), Y ₁ (5), Y ₁ (3), Y ₁ (7)} The decoding calculation module includes 3 levels of decoding calculation submodules, and the m-th level decoding calculation submodule includes 2 ^Mm decoding basic units, m = 1, 2, 3; p = 1, 2, ..., 7; The first level decoding calculation submodule includes a first decoding basic unit, a second decoding basic unit, a third decoding basic unit and a fourth decoding basic unit. The first input end of the first decoding basic unit inputs Y ₁ (0), and Y ₁ (0) is input to the first input end of the first decoding adder; the second input end of the first decoding basic unit inputs Y ₁ (4), and Y ₁ (4) is combined with the rotation factor The signals are multiplied in the first decoding multiplier and then input to the second input terminal of the first decoding adder, and the first decoding adder is used to perform addition operation on the signals input from the first input terminal and the second input terminal; The first input terminal of the second decoding basic unit inputs Y ₁ (2), and Y ₁ (2) is input to the first input terminal of the second decoding adder; the second input terminal of the second decoding basic unit inputs X ₁ (6), and X ₁ (6) and the rotation factor Multiplied in the second decoding multiplier and then input to the second input terminal of the second decoding adder, the second decoding adder is used to add the signal input to its first input terminal and the second input terminal; The first input terminal of the third decoding basic unit inputs X ₁ (1), and X ₁ (1) is input to the first input terminal of the third decoding adder; the second input terminal of the third decoding basic unit inputs X ₁ (5), and X ₁ (5) and the rotation factor Multiplied in the third decoding multiplier and then input to the second input terminal of the third decoding adder, the third decoding adder is used to add the signal input from the first input terminal and the second input terminal; The first input terminal of the fourth decoding basic unit inputs X ₁ (3), and X ₁ (3) is input to the first input terminal of the fourth decoding adder; the second input terminal of the fourth decoding basic unit inputs X ₁ (7), and X ₁ (7) and the rotation factor Multiplied in the fourth decoding multiplier and then input to the second input terminal of the fourth decoding adder, the fourth decoding adder for adding the signal input from the first input terminal and the second input terminal thereof; The second-level calculation submodule includes a fifth decoding basic unit and a sixth decoding basic unit, wherein the first input end of the fifth decoding basic unit inputs the output signal of the first decoding adder, and the output signal of the first decoding adder is input to the first input end of the fifth decoding adder; the second input end of the fifth decoding basic unit inputs the output signal of the second decoding adder, and the output signal of the second decoding adder is added to the rotation factor The signals are multiplied in the fifth decoding multiplier and then input to the second input terminal of the fifth decoding adder, and the fifth decoding adder is used to perform addition operation on the signals input from the first input terminal and the second input terminal; The output signal of the third decoding adder is input to the first input end of the sixth decoding basic unit, and the output signal of the third decoding adder is input to the first input end of the sixth decoding adder; the output signal of the fourth decoding adder is input to the second input end of the sixth decoding basic unit, and the output signal of the fourth decoding adder is input to the rotation factor The signals are multiplied in the sixth decoding multiplier and then input to the second input terminal of the sixth decoding adder, and the sixth decoding adder is used to perform addition operation on the signals input from the first input terminal and the second input terminal; The third-level calculation submodule includes a decoding basic unit, the first input end of the seventh decoding basic unit inputs the output signal of the fifth decoding adder, the output signal of the fifth decoding adder is input to the first input end of the seventh decoding adder; the second input end of the seventh decoding basic unit inputs the output signal of the sixth decoding adder, the output signal of the sixth decoding adder and the rotation factor The signals are multiplied in the seventh decoding multiplier and then input to the second input terminal of the seventh decoding adder. The seventh decoding adder is used to add the signals input from the first input terminal and the second input terminal and input the mapping symbol sent by the sending device through its output terminal. , q=1,2，…,7. 8. According to any one of claims 1-7, the OFDM digital communication system based on reinforcement learning is characterized in that the speech recognition module includes a control component, a control module, a display module, a speech acquisition module and a language model, wherein the control component is configured to be operated by a user so as to input the user's intention into the control module, and the control module selects a text from the first storage module according to the instruction formed by the control component and outputs it to the display module for display; the user reads the displayed text according to the display of the display module, so that the speech acquisition module acquires the user's voice and inputs the acquired voice into the language model; the language model is trained using the user's voice acquired by the speech acquisition module and the recognized text is displayed on the display module, and the read text is compared with the text recognized by the language model. If the self-similarity is greater than or equal to the set threshold, the language model training is stopped and the trained language model is applied to the OFDM digital communication system. If the self-similarity is less than the set threshold, the language model continues to be trained using the voice of the read displayed text. 9. An OFDM digital communication method based on reinforcement learning, characterized in that a bit information sequence is converted into an OFDM symbol sequence by a modulation module, the modulation module includes a modulation segmentation module, a modulation symbol mapping module and an OFDM symbol generation module, the modulation segmentation module divides the bit information sequence into a group of K bits B = {b ₀ ...b _k ...b _K-1 } to obtain multiple groups of bit groups and provides them to the modulation symbol mapping module; the modulation symbol mapping module maps the bit information sequence into a mapping symbol sequence {X(n)}, n = 0, 1, ..., The OFDM symbol generation module includes a modulation code bit reverse reading module, a modulation control counting module, a modulation rotation factor generation module and a modulation calculation module. The code bit reverse reading module reversely reads 2 ^M mapping symbols X(q) from the mapping symbol sequence {X(n)} in binary code and stores them in the first storage module. The first storage module outputs the 2 ^M mapping symbols X(q) input by the code bit reverse reading module in parallel. The modulation rotation factor generation module includes M modulation rotation factor generation submodules. The m-th modulation rotation factor generation submodule generates rotation factors respectively under the control of the control counting module/> , m=1,2,…,M, M is a positive integer greater than or equal to 2; p=0,1,2,…,2 ^M -1; q=0,1,2,…,2 ^M -1; The modulation calculation module includes M-level modulation calculation submodules, and the m-th level modulation calculation submodule includes 2 ^Mm modulation basic units, each of which includes two input terminals and one output terminal; the 2 ^Mm modulation basic units of the first-level modulation calculation submodule input 2 ^M mapping symbols in parallel, and the output of the modulation basic unit of the m-1-th level modulation calculation submodule is used as the input of the modulation basic unit of the m-th level modulation calculation submodule, and the rotation factors are used in turn. Perform calculations until the M-th level modulation calculation submodule sequentially outputs the OFDM symbols to be transmitted in series. . 10. The OFDM digital communication method based on reinforcement learning according to claim 9 is characterized in that the OFDM symbol sequence is converted into a mapping symbol sequence sent by the transmitter through a decoding module, the decoding module includes a decoding segmentation module, a decoding code bit reverse reading module, a decoding control counting module, a decoding rotation factor generation module, a decoding calculation module and a symbol demapping module, the decoding segmentation module segments the received OFDM symbol sequence to obtain OFDM symbol groups, each group includes ^2M OFDM symbols Y(p); the decoding code bit reverse reading module reads ^2M OFDM symbols Y(p) from the OFDM symbol group in binary code bits and stores them in a second storage module, and the second storage module outputs the ^2M OFDM symbols Y(p) input by the decoding code bit reverse reading module in parallel; the decoding rotation factor generation module includes M decoding rotation factor generation submodules, which respectively generate rotation factors under the control of the decoding control counting module. ; The decoding calculation module includes M-level decoding calculation submodules, and the m-th level decoding calculation submodule includes 2 ^Mm decoding basic units, each decoding basic unit includes two input terminals and one output terminal; the 2 ^Mm decoding basic units of the first-level decoding calculation submodule input 2 ^M OFDM symbols in parallel, and the output of the decoding basic unit of the m-1-th level decoding calculation submodule is used as the input of the decoding basic unit of the m-th level decoding calculation submodule, and the rotation factors are used in turn. The operation is performed until the M-th level decoding calculation submodule sequentially outputs the mapping symbols sent by the sending device in series. ; The symbol demapping module receives the mapping symbol /> Demap into bit groups.