阅读说明：本技术 一种脉冲噪声抑制水声信道估计方法 (Pulse noise suppression underwater acoustic channel estimation method ) 是由 韩笑 殷敬伟 高新博 孙宏放 朱瑞 李素文 于 2020-12-18 设计创作，主要内容包括：本发明提供一种脉冲噪声抑制水声信道估计方法,属于水声信号处理领域。涉及一种利用GAMP-SBL对脉冲噪声抑制并实现水声信道估计的方法。(1)输入基带接收信号、字典矩阵、迭代终止条件,以及相关参数初值；(2)利用GAMP-SBL估计脉冲噪声；(3)将脉冲噪声估计结果从基带接收信号中减去；(4)利用GAMP-SBL进行水声信道估计；本发明的优点在于：利用脉冲噪声在时域上的稀疏性,采用对脉冲噪声进行估计,并将其从基带信号中减去以抑制脉冲噪声,能够减少对信号结构的破坏,提高脉冲噪声抑制性能。(The invention provides an impulse noise suppression underwater sound channel estimation method, and belongs to the field of underwater sound signal processing. Relates to a method for suppressing impulse noise and realizing underwater acoustic channel estimation by utilizing GAMP-SBL. (1) Inputting a baseband receiving signal, a dictionary matrix, an iteration termination condition and a related parameter initial value; (2) estimating impulse noise using GAMP-SBL; (3) subtracting the impulse noise estimation result from the baseband received signal; (4) performing underwater acoustic channel estimation by using GAMP-SBL; the invention has the advantages that: by utilizing the sparsity of the impulse noise in the time domain, the impulse noise is estimated and subtracted from the baseband signal to suppress the impulse noise, so that the damage to a signal structure can be reduced, and the impulse noise suppression performance can be improved.)

1. An impulse noise suppression underwater acoustic channel estimation method is characterized in that: the method comprises the following steps:

the method comprises the following steps: the receiving end demodulates the received passband received signal into a baseband received signal and takes out the corresponding null sub-carrier Y in the OFDM signal_n；

Step two: dictionary matrix F is formed by rows corresponding to subcarriers in DFT matrix_nEstimating the impulse noise v by using GAMP-SBL;

step three: estimating the impulse noiseSubtracting from the time domain baseband signal to suppress impulse noise, and transforming the baseband signal from the time domain to the frequency domain to extract the pilot subcarrier Z_p；

Step four: according to dictionary matrix phi corresponding to pilot frequency subcarrier_pAnd estimating the underwater sound channel by utilizing GAMP-SBL.

2. The impulse noise suppression underwater acoustic channel estimation method according to claim 1, characterized in that: the second step of estimating the impulse noise v by using GAMP-SBL specifically comprises the following steps:

(1) inputting parameters: null subcarrier position received signal Y_nDictionary matrix F_nLet S be | F_n|²The gamma ray in gamma²Is carried out in units of elementsSquaring; assuming that impulse noise follows a gaussian independent homogeneous distribution with a mean value of 0 and a variance of a hyper-parameter γ ═ γ₁,γ₂,…,γ_N]^TN is the number of subcarriers; let Γ be diag (γ); to pairγ⁰The assignment, typically a vector greater than 0, where,can be understood as the variance of the impulse noise estimate; let initial noise variance (σ)²)⁰Is a constant greater than 0; s⁰,v⁰Is a 0 vector, where v is the time domain impulse noise; maximum number of SBL cycles K_maxMaximum number of cycles M of the GAMP algorithm_max(ii) a GAMP Algorithm stop condition ε_gampSBL stop condition ε_sbl(ii) a k is 1, m is 1, and k and m record the iteration times of SBL and GAMP respectively;

(2) channel estimation using GAMP:

order toμ^m＝1＝v^k-1，

If | | | mu^m+1-μ^m||²＜ε_gampOr M ═ M_maxStopping GAMP iteration;

(3) updating SBL parameters:

if it is notOr K ═ K_maxTerminating the iteration, otherwise returning to GAMP;

(4) outputting impulse noise estimation results

The specific form of the two scalar estimation functions and their derivatives is:

wherein: the multiplication between vectors or matrices is performed in units of elements_s,θ_h∈(0,1]For damping factors, for reducing the iteration speed and increasing the calculationAnd (4) convergence of the method.

Technical Field

The invention relates to an underwater acoustic signal processing method, in particular to an underwater acoustic channel estimation method for pulse noise suppression, which is a method for suppressing pulse noise and realizing underwater acoustic channel estimation by utilizing generalized approximate message transfer-sparse Bayesian learning (GAMP-SBL).

Background

Generally, the underwater acoustic channel estimation algorithm assumes that the background noise follows a gaussian distribution, however, in some more complex underwater acoustic environments, the background noise may contain some impulse noise. In this case, the algorithm assuming gaussian noise will suffer from impulse noise and performance will be degraded. The conventional impulse noise suppression method is difficult to set a proper filtering threshold, is easy to damage the structure of a signal and has poor suppression performance on impulse noise. Considering that impulse noise has sparseness in the time domain, impulse noise can be estimated using the null sub-carriers of the OFDM signal in combination with the gam-SBL and then subtracted from the baseband signal to suppress the impulse noise. After the impulse noise suppression, the GAMP-SBL can be used again for underwater acoustic channel estimation.

Disclosure of Invention

The invention aims to provide an impulse noise suppression underwater sound channel estimation method.

The purpose of the invention is realized as follows: the method comprises the following steps: the receiving end demodulates the received passband received signal into a baseband received signal and takes out the corresponding null sub-carrier Y in the OFDM signal_n；

Step two: dictionary matrix F is formed by rows corresponding to subcarriers in DFT matrix_nEstimating the impulse noise v by using GAMP-SBL;

step three: estimating the impulse noiseSubtracting from the time domain baseband signal to suppress impulse noise, and transforming the baseband signal from the time domain to the frequency domain to extract the pilot subcarrier Z_p；

Step four: according to dictionary matrix phi corresponding to pilot frequency subcarrier_pAnd estimating the underwater sound channel by utilizing GAMP-SBL.

The invention also includes such structural features:

1. the second step of estimating the impulse noise v by using GAMP-SBL specifically comprises the following steps:

(1) inputting parameters: null subcarrier position received signal Y_nDictionary matrix F_nLet S be | F_n|²The gamma ray in gamma²The square is carried out by taking an element as a unit; assuming that impulse noise follows a gaussian independent homogeneous distribution with a mean value of 0 and a variance of a hyper-parameter γ ═ γ₁,γ₂,…,γ_N]^TN is the number of subcarriers; let Γ be diag (γ); to pairγ⁰The assignment, typically a vector greater than 0, where,can be understood as the variance of the impulse noise estimate; let initial noise variance (σ)²)⁰Is a constant greater than 0; s⁰,v⁰Is a 0 vector, where v is the time domain impulse noise; maximum number of SBL cycles K_maxMaximum number of cycles M of the GAMP algorithm_max(ii) a GAMP Algorithm stop condition ε_gampSBL stop condition ε_sbl(ii) a k is 1, m is 1, and k and m record the iteration times of SBL and GAMP respectively;

(2) channel estimation using GAMP:

order toμ^m＝1＝v^k-1，

If | | | mu^m+1-μ^m||²＜ε_gampOr M ═ M_maxStopping GAMP iteration;

(3) updating SBL parameters:

if it is notOr K ═ K_maxTerminating the iteration, otherwise returning to GAMP;

(4) outputting impulse noise estimation results

The specific form of the two scalar estimation functions and their derivatives is:

wherein: the multiplication between vectors or matrices is performed in units of elements_s,θ_h∈(0,1]The method is a damping factor and is used for reducing the iteration speed and improving the algorithm convergence.

Compared with the prior art, the invention has the beneficial effects that: (1) the impulse noise is estimated by utilizing the time domain sparsity of the impulse noise and combining GAMP-SBL, and is subtracted from the baseband signal to suppress the impulse noise. The damage to a signal structure is reduced, and the suppression effect on pulse noise is enhanced; (2) the underwater acoustic channel is estimated by combining the GAMP-SBL, so that the calculation complexity of the SBL is reduced under the condition of no large performance loss; specifically, the method comprises the following steps: the invention provides an impulse noise suppression underwater acoustic channel estimation method based on GAMP-SBL, aiming at the problem that the performance of an algorithm which takes Gaussian noise as an assumption in an impulse noise environment is reduced due to the influence of impulse noise. The method estimates and inhibits the impulse noise by using GAMP-SBL by utilizing the time domain sparsity of the impulse noise, and then realizes the underwater acoustic channel estimation by combining the GAMP-SBL. Compared with the traditional pulse noise suppression method, the method can reduce the damage to the signal structure and improve the pulse noise suppression performance.

Drawings

FIG. 1 is a schematic diagram of a GAMP-SBL based impulse noise suppression underwater acoustic channel estimation method;

FIG. 2 is a flow chart of the GAMP-SBL algorithm;

FIG. 3 is a time domain graph of arctic under-ice noise;

FIG. 4 is a comparison of normalized mean square error and bit error rate performance based on arctic under-ice noise.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

The invention discloses an impulse noise suppression underwater acoustic channel estimation method based on generalized approximate message transmission-sparse Bayesian learning, which is mainly characterized in that the GAMP-SBL method is used for estimating and suppressing impulse noise; utilizing GAMP-SBL to realize underwater acoustic channel estimation; the impulse noise has sparseness in a time domain, the impulse noise is estimated by combining a null subcarrier of the OFDM signal with GAMP-SBL, and a noise estimation result is subtracted from a baseband receiving signal to suppress the impulse noise. After the impulse noise is suppressed, the underwater acoustic channel is estimated using GAMP-SBL. The method can reduce the calculation complexity of the SBL without great performance loss.

The invention mainly comprises the following steps with reference to fig. 1:

(1) the receiving end demodulates the received passband received signal into a baseband received signal and takes out the corresponding null sub-carrier Y in the OFDM signal_n；

(2) Dictionary matrix F is formed by rows corresponding to subcarriers in DFT matrix_nEstimating and suppressing the impulse noise v by using GAMP-SBL;

(3) estimating the impulse noiseSubtracting from the time domain baseband signal to suppress impulse noise, and transforming the baseband signal from the time domain to the frequency domain to extract the pilot subcarrier Z_p；

(4) According to dictionary matrix phi corresponding to pilot frequency subcarrier_pAnd estimating the underwater sound channel by utilizing GAMP-SBL.

The implementation flow of the impulse noise estimation based on GAMP-SBL in the step (2) is described with reference to FIG. 2.

(1) Inputting parameters: null subcarrier position received signal Y_nDictionary matrix F_nLet S be | F_n|²The gamma ray in gamma²The square is carried out by taking an element as a unit; assuming that impulse noise follows a gaussian independent homogeneous distribution with a mean value of 0 and a variance of a hyper-parameter γ ═ γ₁,γ₂,…,γ_N]^TN is the number of subcarriers; let Γ be diag (γ); to pairγ⁰The assignment, typically a vector greater than 0, where,as a result of impulse noise estimationVariance; let initial noise variance (σ)²)⁰Is a constant greater than 0; s⁰,v⁰Is a 0 vector, where v is the time domain impulse noise; maximum number of SBL cycles K_maxMaximum number of cycles M of the GAMP algorithm_max(ii) a GAMP Algorithm stop condition ε_gampSBL stop condition ε_sbl(ii) a k is 1, m is 1, and k and m record the iteration times of SBL and GAMP respectively;

(2) channel estimation using GAMP

Order toμ^m＝1＝v^k-1，

If | | | mu^m+1-μ^m||²＜ε_gampOr M ═ M_maxStopping GAMP iteration;

(3) SBL parameter update

If it is notOr K ═ K_maxTerminating the iteration, otherwise returning to GAMP;

(4) outputting impulse noise estimation results

The specific form of the two scalar estimation functions and their derivatives is:

in the above algorithm flow, the multiplication between vectors or matrices is performed by θ in units of elements_s,θ_h∈(0,1]The method is a damping factor and is used for reducing the iteration speed and improving the algorithm convergence.

The simulation study of the invention is as follows:

the pulse noise suppression underwater acoustic channel estimation method based on GAMP-SBL provided by the invention is verified by using the ice noise data acquired by the ninth north pole scientific investigation, and the two sections of intercepted noise data are shown in figure 3. The communication signal is an OFDM signal, the center frequency is 12kHz, the frequency band range is 9 kHz-15 kHz, 1024 sub-carriers, 256 pilot sub-carriers and 384 null sub-carriers are totally, 8 symbols are totally, and data are modulated by QPSK.

Comparing SBL, GAMP-SBL, Clipping + GAMP-SBL with the method provided by the invention, the normalized mean square error and error rate performance are shown in figure 4. Wherein (a) and (c) are normalized mean square error and bit error rate corresponding to the noise (a) in fig. 3; (b) and (d) is the normalized mean square error and bit error rate for noise (b) in fig. 3. As can be seen, the proposed algorithm achieves a bit error rate of 10 at 22dB and 18dB SNR, respectively^-2The error rate is lower than that of SBL and GAMP-SBL without pulse noise suppression, so the proposed algorithm has certain performance advantages for SBL and GAMP-SBL; with the increase of the signal-to-noise ratio, the normalization mean square error and the error rate curve turning of the method are increased and the performance is reduced because the Clipping method destroys the structure of the signal. Therefore, the proposed method is more robust with respect to the Clipping method.

In summary, the present invention provides an impulse noise suppression underwater acoustic channel estimation method based on generalized approximate messaging-sparse bayesian learning (GAMP-SBL). The invention belongs to the field of underwater acoustic signal processing. Relates to a method for suppressing impulse noise and realizing underwater acoustic channel estimation by utilizing GAMP-SBL. (1) Inputting a baseband receiving signal, a dictionary matrix, an iteration termination condition and a related parameter initial value; (2) estimating impulse noise using GAMP-SBL; (3) subtracting the impulse noise estimation result from the baseband received signal; (4) performing underwater acoustic channel estimation by using GAMP-SBL; the invention has the advantages that: by utilizing the sparsity of the impulse noise in the time domain, the impulse noise is estimated and subtracted from the baseband signal to suppress the impulse noise, so that the damage to a signal structure can be reduced, and the impulse noise suppression performance can be improved.