阅读说明：本技术 多载波非相干水声通信的频响估计及信号传输方法、系统 (Frequency response estimation and signal transmission method and system for multi-carrier incoherent underwater acoustic communication ) 是由 武岩波 朱敏 于 2020-12-25 设计创作，主要内容包括：本发明公开了多载波非相干水声通信的频响估计及信号传输方法、系统。本发明利用对控制信息进行编码后的码字恒重的特性,对子载波的幅度频率响应进行初步的估计,通过深度神经网络对所有频点的幅频响应进行联合优化估计,之后进行根据各频点的幅频响应幅度对各个符号进行后验概率估计,获得比特对数似然序列,用于信道纠错码译码。本发明提高了非相干水声通信的传输速率、降低其对每比特能耗的要求。(The invention discloses a frequency response estimation and signal transmission method and system for multi-carrier incoherent underwater acoustic communication. The invention uses the character of code word constant weight after coding the control information to carry out preliminary estimation to the amplitude frequency response of the sub-carrier wave, carries out joint optimization estimation to the amplitude frequency response of all frequency points through a deep neural network, and then carries out posterior probability estimation to each symbol according to the amplitude frequency response amplitude of each frequency point to obtain a bit log likelihood sequence for decoding the channel error correcting code. The invention improves the transmission rate of the incoherent underwater acoustic communication and reduces the requirement on the energy consumption of each bit.)

1. A method for frequency response estimation and signal transmission in multi-carrier incoherent underwater acoustic communication, wherein each data packet transmitted comprises control information and data information, the method comprising the steps of:

the method comprises the following steps of transmitting end control information coding modulation:

carrying out dual K code and constant weight code two-stage cascade coding on a control information sequence representing control information, carrying out on-off keying (OOK) modulation on a code word coded by the constant weight code, and using a control information bit for multi-carrier transmission after coding modulation;

the method comprises the following steps of transmitting end data information coding modulation:

turbo code coding and interleaving are carried out on a data information sequence representing data information, multi-system conversion and Amplitude Shift Keying (ASK) modulation are carried out on a bit stream after interleaving, and data information bits are used for multi-carrier transmission after being coded and modulated;

and a transmitting end group packaging step:

the coded and modulated control information and data information are subjected to inverse Fourier transform (IFFT) to realize multi-carrier parallel transmission, wherein signals obtained by modulating the same constant-weight code word by on-off keying (OOK) are transmitted by the same subcarrier to ensure that the energy distributed by the subcarrier in the whole control information transmission is equal; the time domain waveform after inverse Fourier transform IFFT comprises a control information block and a data information block; in order to improve the anti-multipath capability, a cyclic prefix is inserted into a time domain waveform after inverse Fourier transform (IFFT), frames are formed, synchronous signals are inserted in front of and behind each frame, the synchronous signals adopt linear frequency modulation signals, an interval is reserved in front of and behind each synchronous signal, a plurality of frames form a data packet, and then the generation of transmitting signals is completed;

the receiving end synchronization, frequency point amplitude acquisition and control information decoding method comprises the following steps:

a receiving end receives a signal transmitted by an underwater acoustic channel, time synchronization and average Doppler compensation are completed through detection of a linear frequency modulation signal to obtain carrier amplitude, the carrier amplitude comprises a control information carrier amplitude and a data information carrier amplitude, then Fourier transform FFT and modulus calculation are carried out to obtain a control information carrier amplitude sequence and a data information carrier amplitude sequence, then constant-weight-code square-rate soft-decision detection is carried out on the obtained control information carrier amplitude sequence, multi-system Viterbi decoding of dual K codes is carried out to obtain a decoded control information sequence, and then dual K codes and constant-weight codes which are the same as a transmitting end are carried out on the decoded control information sequence to obtain a control information transmitting amplitude estimation sequence;

and (3) amplitude-frequency response estimation:

according to the received control information carrier amplitude sequence and the recovered control information transmission amplitude estimation sequence, firstly, carrying out energy averaging on all received control information subcarrier symbols with transmission amplitudes of zero to obtain noise variance estimation, then carrying out energy averaging on received control information subcarrier symbols with transmission amplitudes of non-zero of the same subcarrier, subtracting the noise variance estimation to obtain channel amplitude-frequency response estimation of each subcarrier, and obtaining an amplitude-frequency response vector according to the channel amplitude-frequency response estimation of all subcarriers;

ASK detection and turbo decoding based on amplitude-frequency response estimation:

and for the amplitude sequence of the data information carrier, obtaining the log-likelihood ratio of each bit carried by the ASK symbol based on channel amplitude-frequency response estimation, de-interleaving each bit log-likelihood ratio of each ASK symbol to be used as the input of a turbo decoder, and recovering the data information sequence after iteration and hard decision to finish transmission.

2. The method of claim 1, wherein the Amplitude Shift Keying (ASK) modulation is mapped using Gray code and normalized for average energy, and a corresponding Amplitude Shift Keying (ASK) amplitude modulation mapping table is used according to ASK order.

3. The method of claim 1, wherein in the step of magnitude-frequency response estimation, after the magnitude-frequency response vector is obtained, the magnitude-frequency response vector is optimally adjusted by a Deep Neural Network (DNN).

4. The method of claim 3, wherein the deep neural network DNN comprises two stages of hidden layers, wherein the activation function of the hidden layers is a Relu function, and the mean square error is used as a performance loss function during optimization.

5. A multi-carrier incoherent underwater acoustic communication transmission apparatus, characterized by comprising:

the control information coding and modulating module is used for carrying out dual K code and constant weight code two-stage cascade coding on a control information sequence representing control information, carrying out on-off keying (OOK) modulation on a code word coded by the constant weight code, and using a control information bit for multi-carrier transmission after coding modulation;

a data information coding modulation module, which performs turbo code coding and interleaving on a data information sequence representing data information, performs multilevel system conversion and Amplitude Shift Keying (ASK) modulation on an interleaved bit stream, and uses the data information bits for multi-carrier transmission after coding modulation;

the packaging module is used for realizing multi-carrier parallel transmission of the control information and the data information after the coding modulation through inverse Fourier transform (IFFT), wherein signals obtained after the same constant-weight code word is subjected to on-off keying (OOK) modulation are transmitted by the same subcarrier so as to ensure that the energy distributed by the subcarrier in the whole control information transmission is equal; the time domain waveform after inverse Fourier transform IFFT comprises a control information block and a data information block; in order to improve the multipath resistance, a cyclic prefix is inserted into a time domain waveform after inverse Fourier transform (IFFT), frames are formed, synchronous signals are inserted in front of and behind each frame, the synchronous signals adopt linear frequency modulation signals, an interval is reserved in front of and behind each synchronous signal, a plurality of frames form a data packet, and then the generation of transmitting signals is completed.

6. The transmitting device of claim 5, wherein the Amplitude Shift Keying (ASK) modulation employs Gray code mapping and average energy normalization, and a corresponding ASK amplitude modulation mapping table is employed according to an ASK order.

7. A multi-carrier incoherent underwater acoustic communication receiving apparatus, characterized by comprising:

a synchronization, frequency point amplitude and control information decoding module, which receives signals transmitted through an underwater acoustic channel, completes time synchronization and average Doppler compensation by detecting a linear frequency modulation signal to obtain carrier amplitude, the carrier amplitude comprises two parts of control information carrier amplitude and data information carrier amplitude, then performs Fourier transform FFT and module calculation to obtain a control information carrier amplitude sequence and a data information carrier amplitude sequence, then performs constant repetition rate soft decision detection on the obtained control information carrier amplitude sequence, then performs multi-system Viterbi decoding on dual K codes to obtain a decoded control information sequence, and then performs dual K code and constant repetition code cascade coding on the decoded control information sequence with the same transmitting end to obtain a control information transmitting amplitude estimation sequence;

the module firstly carries out energy averaging on all received control information subcarrier symbols with zero transmission amplitude to obtain noise variance estimation, then carries out energy averaging on the received control information subcarrier symbols with non-zero transmission amplitude of the same subcarrier, subtracts the noise variance estimation to obtain channel amplitude-frequency response estimation of each subcarrier, and obtains an amplitude-frequency response vector according to the channel amplitude-frequency response estimation of all subcarriers;

the ASK detection and turbo decoding module based on amplitude-frequency response estimation obtains the log-likelihood ratio of each bit carried by an ASK symbol for a data information carrier amplitude sequence based on channel amplitude-frequency response estimation, the log-likelihood ratio of each bit of each ASK symbol is used as the input of a turbo decoder after being deinterleaved, and after iteration and hard decision, the data information sequence is recovered to finish transmission.

8. The receiving device of claim 7, further comprising a magnitude-frequency response optimization module that performs optimization adjustment on the magnitude-frequency response vector obtained by the magnitude-frequency response estimation module through a Deep Neural Network (DNN).

9. The receiver apparatus of claim 8, wherein the deep neural network DNN comprises two stages of hidden layers, wherein an activation function of the hidden layers is a Relu function, and a mean square error is used as a performance loss function during optimization.

10. A frequency response estimation and signal transmission system for multi-carrier incoherent underwater acoustic communication, characterized by comprising a transmitting device and a receiving device, wherein the transmitting device adopts the transmitting device as claimed in one of claims 5 to 6, and the receiving device adopts the receiving device as claimed in one of claims 7 to 9.

Technical Field

The invention belongs to the field of underwater acoustic communication, and particularly relates to a frequency response estimation and signal transmission method and system for multi-carrier incoherent underwater acoustic communication, aiming at frequency point allocation, amplitude-frequency response estimation and high-order modulation of the multi-carrier incoherent underwater acoustic communication.

Background

In the submarine observation wireless network, data and instruction transmission among the fixed observation submerged buoy, the mobile observation submersible, the wireless relay submerged buoy and the wireless gateway is realized in an underwater acoustic communication mode. The incoherent communication adopts a simple energy detection mode, does not need to track a rapidly changing phase, and has the advantages of no pilot frequency overhead, small influence of Doppler effect and the like. Due to the adoption of a multi-carrier parallel transmission mode, the multi-path-resistant multi-carrier parallel transmission system has stronger multi-path resistance. The incoherent communication is the most common transmission system in the wireless expansion system and is responsible for network control instruction transmission and return of a large amount of sensor data. Technical indexes such as error rate, transmission rate and energy consumption per bit of incoherent communication directly determine the working performance and the life cycle of the whole wireless expanded network. Because of the severe multipath of the underwater acoustic channel, which causes the channel to have severe frequency selectivity, the existing algorithm usually avoids the estimation of the amplitude response of each frequency point, but modulates the information sequence onto a plurality of subcarriers by adopting constant weight code words and On-off keying (OOK) at the transmitting end, and the receiving end adopts an energy combination detection mode.

In the existing incoherent underwater acoustic communication scheme, each subcarrier can only carry 1 bit after being coded at most, and meanwhile, the use of constant-weight codes further reduces the communication rate, and the decoding error rate is high when the signal-to-noise ratio is low because the frequency response difference of different carriers is not distinguished.

Disclosure of Invention

The invention aims to improve the transmission rate of incoherent underwater acoustic communication and reduce the requirement on energy consumption of each bit.

In order to achieve the above object, the present invention provides a method for estimating frequency response and transmitting signals of multi-carrier incoherent underwater acoustic communication, wherein each data packet to be transmitted comprises control information and data information, and the method comprises the following steps:

the method comprises the following steps of transmitting end control information coding modulation:

carrying out dual K code and constant weight code two-stage cascade coding on a control information sequence representing control information, carrying out on-off keying (OOK) modulation on a code word coded by the constant weight code, and using a control information bit for multi-carrier transmission after coding modulation;

the method comprises the following steps of transmitting end data information coding modulation:

turbo code coding and interleaving are carried out on a data information sequence representing data information, multi-system conversion and Amplitude Shift Keying (ASK) modulation are carried out on a bit stream after interleaving, and data information bits are used for multi-carrier transmission after being coded and modulated;

and a transmitting end group packaging step:

the coded and modulated control information and data information are subjected to inverse Fourier transform (IFFT) to realize multi-carrier parallel transmission, wherein signals obtained by modulating the same constant-weight code word by on-off keying (OOK) are transmitted by the same subcarrier to ensure that the energy distributed by the subcarrier in the whole control information transmission is equal; the time domain waveform after inverse Fourier transform IFFT comprises a control information block and a data information block; in order to improve the anti-multipath capability, a cyclic prefix is inserted into a time domain waveform after inverse Fourier transform (IFFT), frames are formed, synchronous signals are inserted in front of and behind each frame, the synchronous signals adopt linear frequency modulation signals, an interval is reserved in front of and behind each synchronous signal, a plurality of frames form a data packet, and then the generation of transmitting signals is completed;

the receiving end synchronization, frequency point amplitude acquisition and control information decoding method comprises the following steps:

a receiving end receives a signal transmitted by an underwater acoustic channel, time synchronization and average Doppler compensation are completed through detection of a linear frequency modulation signal to obtain carrier amplitude, the carrier amplitude comprises a control information carrier amplitude and a data information carrier amplitude, then Fourier transform FFT and modulus calculation are carried out to obtain a control information carrier amplitude sequence and a data information carrier amplitude sequence, then constant-weight-code square-rate soft-decision detection is carried out on the obtained control information carrier amplitude sequence, multi-system Viterbi decoding of dual K codes is carried out to obtain a decoded control information sequence, and then dual K codes and constant-weight codes which are the same as a transmitting end are carried out on the decoded control information sequence to obtain a control information transmitting amplitude estimation sequence;

and (3) amplitude-frequency response estimation:

according to the received control information carrier amplitude sequence and the recovered control information transmission amplitude estimation sequence, firstly, carrying out energy averaging on all received control information subcarrier symbols with transmission amplitudes of zero to obtain noise variance estimation, then carrying out energy averaging on received control information subcarrier symbols with transmission amplitudes of non-zero of the same subcarrier, subtracting the noise variance estimation to obtain channel amplitude-frequency response estimation of each subcarrier, and obtaining an amplitude-frequency response vector according to the channel amplitude-frequency response estimation of all subcarriers;

ASK detection and turbo decoding based on amplitude-frequency response estimation:

and for the amplitude sequence of the data information carrier, obtaining the log-likelihood ratio of each bit carried by the ASK symbol based on channel amplitude-frequency response estimation, de-interleaving each bit log-likelihood ratio of each ASK symbol to be used as the input of a turbo decoder, and recovering the data information sequence after iteration and hard decision to finish transmission.

Further, the amplitude shift keying ASK modulation adopts gray code mapping and average energy normalization, and a corresponding amplitude shift keying ASK amplitude modulation mapping table is adopted according to ASK order.

Further, in the step of amplitude-frequency response estimation, after the amplitude-frequency response vector is obtained, the amplitude-frequency response vector is optimally adjusted through a deep neural network DNN.

Furthermore, the deep neural network DNN comprises two stages of hidden layers, the Relu function is selected as an activation function of the hidden layers, and the mean square error is used as a performance loss function during optimization.

The invention also provides a multi-carrier incoherent underwater acoustic communication sending device, which comprises:

the control information coding and modulating module is used for carrying out dual K code and constant weight code two-stage cascade coding on a control information sequence representing control information, carrying out on-off keying (OOK) modulation on a code word coded by the constant weight code, and using a control information bit for multi-carrier transmission after coding modulation;

a data information coding modulation module, which performs turbo code coding and interleaving on a data information sequence representing data information, performs multilevel system conversion and Amplitude Shift Keying (ASK) modulation on an interleaved bit stream, and uses the data information bits for multi-carrier transmission after coding modulation;

the packaging module is used for realizing multi-carrier parallel transmission of the control information and the data information after the coding modulation through inverse Fourier transform (IFFT), wherein signals obtained after the same constant-weight code word is subjected to on-off keying (OOK) modulation are transmitted by the same subcarrier so as to ensure that the energy distributed by the subcarrier in the whole control information transmission is equal; the time domain waveform after inverse Fourier transform IFFT comprises a control information block and a data information block; in order to improve the multipath resistance, a cyclic prefix is inserted into a time domain waveform after inverse Fourier transform (IFFT), frames are formed, synchronous signals are inserted in front of and behind each frame, the synchronous signals adopt linear frequency modulation signals, an interval is reserved in front of and behind each synchronous signal, a plurality of frames form a data packet, and then the generation of transmitting signals is completed.

Further, the amplitude shift keying ASK modulation adopts gray code mapping and average energy normalization, and a corresponding ASK amplitude modulation mapping table is adopted according to the ASK order.

The invention also provides a multi-carrier incoherent underwater acoustic communication receiving device, which comprises:

a synchronization, frequency point amplitude and control information decoding module, which receives signals transmitted through an underwater acoustic channel, completes time synchronization and average Doppler compensation by detecting a linear frequency modulation signal to obtain carrier amplitude, the carrier amplitude comprises two parts of control information carrier amplitude and data information carrier amplitude, then performs Fourier transform FFT and module calculation to obtain a control information carrier amplitude sequence and a data information carrier amplitude sequence, then performs constant repetition rate soft decision detection on the obtained control information carrier amplitude sequence, then performs multi-system Viterbi decoding on dual K codes to obtain a decoded control information sequence, and then performs dual K code and constant repetition code cascade coding on the decoded control information sequence with the same transmitting end to obtain a control information transmitting amplitude estimation sequence;

the module firstly carries out energy averaging on all received control information subcarrier symbols with zero transmission amplitude to obtain noise variance estimation, then carries out energy averaging on the received control information subcarrier symbols with non-zero transmission amplitude of the same subcarrier, subtracts the noise variance estimation to obtain channel amplitude-frequency response estimation of each subcarrier, and obtains an amplitude-frequency response vector according to the channel amplitude-frequency response estimation of all subcarriers;

the ASK detection and turbo decoding module based on amplitude-frequency response estimation obtains the log-likelihood ratio of each bit carried by an ASK symbol for a data information carrier amplitude sequence based on channel amplitude-frequency response estimation, the log-likelihood ratio of each bit of each ASK symbol is used as the input of a turbo decoder after being deinterleaved, and after iteration and hard decision, the data information sequence is recovered to finish transmission.

Further, the receiving device further comprises an amplitude-frequency response optimization module, and the module optimizes and adjusts the amplitude-frequency response vector obtained by the amplitude-frequency response estimation module through a deep neural network DNN.

Furthermore, the deep neural network DNN comprises two stages of hidden layers, the Relu function is selected as an activation function of the hidden layers, and the mean square error is used as a performance loss function during optimization.

The invention also provides a frequency response estimation and signal transmission system of multi-carrier incoherent underwater acoustic communication, which is characterized by comprising a transmitting device and a receiving device, wherein the transmitting device adopts the transmitting device as claimed in the claim, and the receiving device adopts the receiving device.

Advantageous effects

The invention improves the transmission rate of the incoherent underwater acoustic communication and reduces the requirement of the incoherent underwater acoustic communication on the energy consumption of each bit, and particularly comprises the following steps:

(1) the invention adopts the waveform of the control frame to estimate in the estimation of the amplitude-frequency response, does not need to set a special training pilot frequency, and can avoid the pilot frequency overhead.

(2) Because of the nonlinear influence of module value calculation, the traditional linear minimum mean square error channel estimation algorithm is not applicable to incoherent underwater acoustic communication, the invention firstly utilizes a method of average energy in the estimation of amplitude-frequency response, combines the amplitude-frequency estimation of a plurality of symbol times with low calculation complexity, adopts a deep neural network mode for further optimization, utilizes the nonlinear correlation among the amplitude-frequency responses of all frequency points, improves the precision of frequency response estimation, and does not depend on a channel model at the same time.

(3) After the amplitude-frequency response of each frequency point is obtained, the modulation information detection of each frequency point is more accurate, the channel utilization rate can be improved by adopting a modulation mode of high-order Amplitude Shift Keying (ASK), and pilot frequency overhead is not needed.

(4) The detection form of the invention for the modulation information on each carrier is based on the log-likelihood ratio of the posterior probability, and is more suitable for the requirement of a turbo decoder for the bit information form compared with the traditional energy detection.

Drawings

Fig. 1 is a block diagram of a frequency response estimation and signal transmission method of multi-carrier incoherent underwater acoustic communication according to the present invention;

FIG. 2 is a schematic diagram of the structure of the transmitting packet in the present invention;

FIG. 3 is a simulation comparison graph of the frequency response estimation effect and the true value in the present invention;

fig. 4 is a comparison graph of the relationship between the number of decoding iterations and the bit error rate under different frequency response estimation and detection modes.

Detailed Description

The following describes in detail embodiments of the present invention with reference to the drawings.

The invention provides a frequency response estimation and signal transmission method and system for multi-carrier incoherent underwater acoustic communication, and transmitting equipment and receiving equipment thereof based on a transmitting waveform generation scheme capable of being used for subcarrier amplitude detection, accurate amplitude response estimation of a receiving end to each subcarrier and posterior probability detection of information on each carrier.

The invention uses the character of code word constant weight after coding the control information to carry out preliminary estimation to the amplitude frequency response of the sub-carrier wave, carries out joint optimization estimation to the amplitude frequency response of all frequency points through a deep neural network, and then carries out posterior probability estimation to each symbol according to the amplitude frequency response amplitude of each frequency point to obtain a bit log likelihood sequence for decoding the channel error correcting code.

The invention discloses a frequency response estimation and signal transmission method for multi-carrier incoherent underwater acoustic communication, which comprises the following steps:

the method comprises the following steps of transmitting end control information coding modulation:

the transmit signal flow is shown in figure 1. Each transmission packet needs to transmit two parts of control information and data information. Control information bit number of N_CtrlAll control information bits are noted asAfter the dual K code coding, the coded multilevel sequence is obtained asWhereinR_DualCode rate, N, for dual K codes_CWIThe number of input bits per packet is encoded for the subsequent stage constant weight code. N is a radical of_CWOThe number of output bits of the packet is encoded for a constant weight code. Constant weight code mapping matrix is notedThat is, when the multilevel value of the constant weight code mapping input is i, the jth bit result of the binary sequence of the mapping output isThe matrix is generated by adopting a Hadamard matrix, and the probability of 1 appearing in the constant weight code isThe code rate of the constant-weight code isThe control information bit is cascade coded by dual K and constant weight code, after OOK modulation, it adopts multi-carrier transmission mode, N_CarrierThe number of parallel multi-carriers. The same constant weight code word is transmitted by the same subcarrier, so as to ensure that the energy distributed by the subcarrier in the whole control information transmission is equal. Ith of the ith subcarrierThe amplitude of the j multicarrier symbols is:

wherein i is more than or equal to 0 and less than N_Carrier，0≤j＜N_CtrlSymb，For the rounding-down function, the number of multicarrier symbols used for the control information transmission is

The method comprises the following steps of transmitting end data information coding modulation:

the data information bits to be transmitted are marked asWherein N is_DataIs the number of data information bits. After turbo code encoding and interleaving, the bit vector obtained isWhereinOutputting the number of bits, R, for turbo coding_TurboIs the code rate of turbo coding.

Carrying out multi-system sequence conversion on the coded and interleaved bit stream to obtain

Wherein N is_ASKFor the number of bits carried by each ASK symbol, for different ASK orders, gray code mapping is adopted and average energy normalization is performed to obtain an ASK amplitude mapping (modulation) table, wherein the mapping table of 1 to 3 orders is as follows:

whereinRepresenting ASK amplitude of the output when input.

The amplitude of the jth multicarrier symbol of the ith subcarrier carrying data information code is:

wherein i is more than or equal to 0 and less than N_Carrier，0≤j-N_CtrlSymb＜N_DataSymbAnd the number of the multi-carrier symbols used for data information transmission is

The step of group packing of the transmitting end:

the component structure of the transmission packet is shown in fig. 2, and after control information and data information are coded and mapped, multicarrier mode parallel transmission is realized through Inverse Fast Fourier Transform (IFFT). The lowest and highest subcarrier frequencies are f_LAnd f_HSubcarrier spacing of f_δ. The effective length of the sub-carrier being T_δ. The multicarrier symbols of the preceding control information block portion carry control information and the multicarrier symbols of the following data information block carry data information. The same multicarrier symbol contains multiple subcarriers that carry the same type of information. To improve the ability to overcome multipath, a cyclic prefix of length T is inserted in the time domain waveform_g. One frame includes N_SymbPerFrmAnd the multi-carrier symbols are inserted with synchronous signals before and after each frame, so that the generation of the transmitting signals is completed. The synchronization signal is in the form of a linear frequency modulation, the lowest and highest frequencies of which are f₀And f₁Having a duration of T_Sync. The blank length left before and after each synchronous signal is T_Gap. Each frame having a length of T_Frm. According to the number of frames in the sending packet and the frame length, the transmission length of the packet is T_packet。

The receiving end synchronization, frequency point amplitude acquisition and control information decoding method comprises the following steps:

the signal flow at the receiving end is shown in fig. 1. Time synchronization and average Doppler synchronization are completed through detection of linear frequency modulation signals, Fourier Transform (FFT) calculation is carried out, the amplitude of the jth symbol of the ith subcarrier in the whole packet is obtained and is y (i, j), wherein i is more than or equal to 0 and less than N_Carrier，0≤j＜N_CtrlSymb+N_DataSymb. Its simulation generation model is expressed as

y(i，j)＝|h(i)x(i，j)+w(i，j)|

Where h (i) is the amplitude-frequency response of the ith subcarrier, which is the amount to be estimated, and w (i, j) is the additive noise at the corresponding subcarrier. According to the received carrier amplitude carrying control information, firstly, the traditional square rate soft decision detection is carried out on the constant weight code, then the multilevel Viterbi decoding of the dual K code is carried out, and the decoded control information sequence is obtainedThe control information sequence obtained by decoding is subjected to dual K code and constant repetition code cascade coding which are the same at the sending end, and the emission amplitude estimation of the amplitude of the jth symbol of the ith subcarrier in the control information is obtained, namely the estimation is carried outWherein i is more than or equal to 0 and less than N_Carrier，0≤j＜N_CtrlSymb。

Amplitude-frequency response estimation step and improved estimation based on deep neural network

According to the receiving amplitude sequence of the control information and the control information transmitting amplitude estimation recovered by the receiving end, firstly, the energy average is carried out on the received subcarrier symbol with the transmitting amplitude of 0, and the noise variance estimation is obtained

Then, for each subcarrier, averaging the energy of the received subcarrier symbols with the transmission amplitude different from 0, and subtracting the noise variance to obtain the channel amplitude frequency response of the ith subcarrier, wherein the channel amplitude frequency response is estimated as

Adjusting the amplitude-frequency response vector through a Deep Neural Network (DNN), and recording the input amplitude-frequency response vector and the output amplitude-frequency response vector as

The DNN internally comprises two stages of hidden layers, and the activation function of the hidden layers is selected from a ReLu function and is marked as f_ReLu(. cndot.). The output vector of the DNN can thus be expressed as:

wherein { W₀，b₀，W₁，b₁，W₂，b₂And (4) training and optimizing to obtain the parameters by an Adam tool, wherein the mean square error is used as a performance loss function during optimization.

ASK detection and turbo decoding based on amplitude-frequency response estimation:

for the jth symbol of the ith subcarrier carrying data information (i is more than or equal to 0 and less than N)_Carrier，N_CtrlSymb≤j≤N_CtrlSymb+N_DataSymb) With an ASK amplitude ofThe posterior probability of (a) is proportional to the non-centering parameter ofDimension parameterHas a probability density of the Leiss distribution, i.e.

Wherein the probability density function of the Leise distribution is

And I₀(. cndot.) is a modified Bessel function of order 0. The log-likelihood ratio of the nth bit carried by the ASK symbol is

WhereinFor solving the nth bit of the integer m corresponding to the ASK symbol, the solving process is Is a rounded down function. Deinterleaving the bit log-likelihood ratios of ASK symbols as the input of turbo decoder, and performing multiple operationsAfter sub-iteration and hard decision, the data information bit stream is recovered, i.e.The transfer process is completed.

Typical parameters and values used in the present invention are as follows:

according to the parameters, performance simulation under the multipath channel is carried out. The channel multipath employs independent uncorrelated rayleigh distribution with a multipath spreading length of 0.625ms and is used to generate frequency response samples for DNN training. Fig. 3 shows the frequency response comparison obtained by different methods. The method comprises the steps of real frequency response, amplitude-frequency response estimation obtained by carrying out average energy estimation by using frequency points of control information, and amplitude-frequency estimation after DNN adjustment, and can be seen to be closer to a real value after DNN adjustment.

The encoder of the data information uses a turbo code of rate 1/3, the component codes of which are recursive systematic convolutional codes of polynomial [37,21 ]. Fig. 4 shows a relationship curve between different decoding iteration times and Bit Error Rate (BER) under different detection modes of ASK of 4 th order. The signal-to-noise ratio used was 10 dB. In the rayleigh distribution ASK detection mode of frequency-response-free estimation in the conventional method, the iteration times are increased, the BER cannot be improved, and the reliable transmission of high-order modulation cannot be performed by using the conventional incoherent technology. After the frequency response estimation based on the control information constant-weight codes, which is provided by the invention, the high-order ASK detection based on the Laisi distribution hypothesis becomes credible. For two different frequency response estimation modes, BER can be improved through multiple iterations. After adjusting the frequency response estimation by using DNN, the BER of the system is lower.

Based on the same inventive concept, the specific embodiment of the invention also provides a frequency response estimation and signal transmission system for multi-carrier incoherent underwater acoustic communication, and transmitting equipment and receiving equipment in the system. The sending device comprises a control information coding and modulating module, a data information coding and modulating module and a packaging module. The receiving equipment comprises a synchronization, frequency point amplitude and control information decoding module, an amplitude-frequency response estimation module, an amplitude-frequency response optimization module and an ASK detection and turbo decoding module based on amplitude-frequency response estimation. The modules and the corresponding equipment of the modules complete the corresponding functions of the frequency response estimation and signal transmission method of the multi-carrier incoherent underwater acoustic communication.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.