A Pulse Noise Suppression Method for Under-Ice Acoustic OFDM Communication in Polar Regions

Abstract

The object of the present invention is to provide a method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions. By using OFDM pilot subcarriers, channel estimation and equalization are performed on the received passband signal based on the least squares criterion, and then the background noise is estimated. At the same time, only this estimated value is used to obtain the position information of impulse noise, and the impulse noise is detected and marked using the 3σ principle based on the generalized signal-to-noise ratio. After suppressing the impulse noise using an adaptive median filter, the noise estimated value is added to the noise-free received signal to obtain the received passband signal after impulse noise suppression processing. Then, subsequent channel estimation, channel equalization, and decision-making processes are carried out, and finally the source information is decoded. The present invention effectively avoids the interference of OFDM signals on the impulse detection process; the 3σ impulse noise detection method based on the generalized signal-to-noise ratio realizes a detection threshold that can be adaptively adjusted according to impulse noise. The impulse noise suppression step reduces the computational complexity of the algorithm.

Description

Technical Field

[0001] The present invention relates to a method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions, belonging to the field of underwater acoustic communication in polar regions. Background Art

[0002] The Arctic Ocean is covered by ice all year round. Although the ice cover itself can reduce the generation of some noise, it is actually an important noise source. The processes of ice cracking and squeezing will generate a large amount of impulse noise, which causes the distribution characteristics of the background noise not to follow the Gaussian distribution, seriously reducing the performance of the underwater acoustic communication system. OFDM technology is a multi-carrier transmission technology with advantages such as high transmission rate, high spectrum utilization rate, and strong anti-multipath interference ability, and has been widely used in underwater acoustic communication. However, the OFDM signal has the problem of a high peak-to-average power ratio, which makes there are many peaks with relatively high amplitudes in the signal. When there is impulse noise in the underwater acoustic channel, the prominent OFDM signal peaks will affect the detection process of the impulse noise, and then lead to poor performance of the impulse noise suppression algorithm. Therefore, compared with single-carrier signals, it is more difficult to directly detect and eliminate impulse noise from OFDM received signals, which is also one of the reasons why the performance of many current impulse noise suppression algorithms fails at high signal-to-noise ratios. Therefore, there is an urgent need for a robust impulse noise suppression method that can avoid the influence of OFDM signal peaks. Summary of the Invention

[0003] The present invention proposes a method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions. The creativity of the invention lies in avoiding the influence of the OFDM signal itself on the impulse noise detection process by first estimating the background noise and then suppressing the noise. Based on the generalized signal-to-noise ratio defined by the SαS simulated impulse noise, the improved 3σ principle is used to detect impulse noise. During the impulse noise suppression process, only the detected impulse noise is processed, and an adaptively adjusted filtering window length is used, effectively suppressing impulse noise, improving the performance of OFDM communication, and reducing the computational complexity of the algorithm.

[0004] The purpose of the present invention is to improve the robustness of the underwater acoustic OFDM communication system under polar impulse interference, and provide an OFDM signal processing and impulse noise suppression method based on background noise estimation and adaptive median filtering under polar impulse noise interference. By first estimating the background noise and then detecting and suppressing the impulse noise, the influence of the OFDM signal peaks on the impulse noise detection process is avoided. The improved 3σ principle under the generalized signal-to-noise ratio is used to detect impulse noise. During the impulse noise suppression process, only the detected impulse noise is processed, and an adaptively adjusted filtering window length is used, effectively reducing the computational complexity of the algorithm.

[0005] The purpose of the present invention is achieved as follows:

[0006] The basic idea is to design a method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions. It mainly estimates the background noise and uses adaptive median filtering to effectively suppress impulse noise interference while avoiding the influence of OFDM signals, thereby improving the robustness of OFDM communication.

[0007] Step 1: Input the OFDM passband received signal, perform steps such as OFDM demodulation and downsampling, and use the known pilot information for channel estimation to obtain the channel estimation values in the time domain and frequency domain.

[0008] Step 2: Use the received signal and the channel estimation values in Step 1 for the channel equalization and decision-making process to obtain the transmitted signal symbols. After re-symbol mapping and OFDM modulation, convolve with the time-domain channel estimation values to reconstruct the noise-free OFDM received signal estimation value.

[0009] Step 3: Subtract the passband received signal in Step 1 from the received signal estimation value in Step 2 to obtain the background noise estimation value.

[0010] Step 4: Detect and classify the noise estimation value in Step 3. Use the improved 3σ principle to detect the positions of impulse noise and record them in the index vector, and mark the impulse noise through extreme value processing.

[0011] Step 5: Process the marked noise in Step 4, use an adaptive median filter to suppress impulse noise, and make the background noise approximately follow a Gaussian distribution. Then add it to the noise-free received signal estimation value in Step 2 to obtain the received signal after suppressing impulse noise.

[0012] Step 6: Re-perform processes such as OFDM demodulation, channel estimation, channel equalization, decision-making, and decoding on the received signal in Step 5, and finally restore and output the source information.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: A method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions is proposed. Aiming at the problem that OFDM signals have a great influence on the impulse noise detection process, the impulse noise in the background noise is converted into Gaussian noise by estimating the background noise and using an adaptive median filter, thereby realizing the suppression of impulse noise. Then, processes such as channel estimation, channel equalization, and decoding are re-performed on the processed signal.

[0014] (1) Maintain good performance when the noise intensity is low

[0015] The high peak of the OFDM signal will affect the pulse noise detection process. Especially when the noise intensity is low, it is usually very difficult for traditional algorithms to distinguish pulse noise, which will lead to a serious decline in communication performance. The present invention estimates and then suppresses pulse noise, thus avoiding the influence of the OFDM signal, enabling the algorithm to have good detection performance for pulse noise of different intensities, and effectively improving the reliability of the algorithm.

[0016] (2) Improved the calculation of the pulse noise detection threshold

[0017] Using the generalized signal-to-noise ratio under SαS simulated pulse noise to define the pulse noise variance, and then using the improved 3σ principle to detect the pulse noise position. This enables the detection threshold to be adjusted according to the characteristics of pulse noise and avoids the problem of difficult threshold determination in traditional pulse detection algorithms.

[0018] (3) Reduced the computational complexity

[0019] The algorithm only suppresses the detected pulse noise. During the suppression process, a filtering window length adaptively adjusted according to the number of pulse noises is used, effectively reducing the computational complexity of the system, thereby reducing the computational amount. Description of the Drawings

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for the description of the embodiments. The following described drawings are only some embodiments of the present invention. For other technical personnel, without creative work, other drawings can also be obtained according to these drawings, where:

[0021] Figure 1 is the flow chart of the pulse noise suppression method for polar ice-underwater acoustic OFDM communication;

[0022] Figure 2 is the flow chart of the pulse noise detection step;

[0023] Figure 3 is the flow chart of the pulse noise suppression step;

[0024] Figure 4 is the time-domain diagram of the pulse noise collected in the Arctic;

[0025] Figure 5 is the error code performance curve diagram of different algorithms in the Arctic ice-underwater pulse noise interference environment when the change range of the generalized signal-to-noise ratio is -10 to 20 dB. Detailed Embodiments

[0026] The following further describes the present invention in detail with reference to the drawings by way of examples.

[0027] A method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions provided by the present invention, the working process of the method is as follows Figure 1 , and the specific implementation manners are as follows:

[0028] For the preprocessing of the passband signal in Step 1, the received signal y needs to be subjected to basic OFDM demodulation and Fourier transform, and according to the known pilot information, the received signal Y at the pilot positions is extracted p . Since the background noise at this time follows a non-Gaussian distribution, the least squares algorithm is first used for channel pre-estimation, and the channel estimation results in the time-frequency domain are respectively and and are used in Steps 2 and 3.

[0029] For the restoration of the noise-free received signal in Step 2, channel equalization is performed according to the channel estimation result in Step 1, and OFDM modulation and inverse Fourier transform are performed again, and the estimated value of the transmitted signal in the passband can be obtained After convolution with the time-domain channel estimation result obtained in Step 1, the estimated value of the noise-free received signal

[0030] For the background noise estimation in Step 3, based on the passband received signal y input in Step 1 and the noise-free signal output in Step 2, the difference between the two is the estimated value of the background noise in the passband Due to the influence of the background noise itself and the errors brought by processes such as channel estimation, the here is difficult to be accurate, and it contains part of the information of the required signal. Therefore, it should not be directly eliminated as noise from the received signal, but is used to detect impulse noise in Step 4.

[0031] For the impulse noise detection in Step 4, mainly detect the position of the impulse noise existing in the background noise estimated value in Step 3. First, based on the SαS impulse noise model, parameter estimation is performed on to obtain its characteristic exponent α w and scale parameter γ w . Based on the generalized signal-to-noise ratio and the 3σ principle, the detection threshold of the impulse noise is defined as:

[0032]

[0033] where C g is the exponent of the Euler constant, approximately taking the value of 1.7811; c is a coefficient, usually taking the value of 1. For all the noise data points in , when the value of the data point is greater than the detection threshold th 3σWhen, record the position of the data point into the index vector E, and at mark it as "1". The specific process of Step 4 is shown in Figure 2 , define the processed noise as n div , and output it to Step 5.

[0034] For the impulse noise suppression described in Step 5, according to the index vector E in Step 4, process all the impulse noise points in n div . Determine a detection window with a size of 1 centered on the impulse noise. For the impulse noise point n div (j), its corresponding detection window is:

[0035] W 2r+1 (j) = [n div (j - r), …, n div (j), …, n div (j + r)]

[0036] Calculate the median of the data within the window. If the median is not "1", then replace the impulse noise with this median; if the median is "1", it means that the impulse noise dominates within the window. At this time, expand the size of the detection window and recalculate the median until the median replacement condition is met. The specific process of Step 5 is shown in Figure 3 . After processing all the impulse noises detected in Step 4 in Step 5, the estimated value of the background noise can approximately follow a Gaussian distribution. Define the processed noise as n trans , and add it to the estimated value of the noise - free received signal in Step 2 to obtain the received signal after suppressing the impulse noise

[0037] For the subsequent processing described in Step 6, perform normal OFDM demodulation, channel estimation, channel equalization, decision - making, decoding, etc. steps on , and obtain the final source information. Since the background noise in follows a Gaussian distribution at this time, the sparse Bayesian learning algorithm with better performance can be used in the channel estimation step here.

[0038] Next, the advantages of the present invention will be further explained in combination with the test data processing results. Obviously, the described test results are part of the embodiments of the present invention, rather than all embodiments, and are only used to illustrate and explain the present invention, and are not intended to limit the present invention. Based on the test data processing results in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0039] Test data verification:

[0040] Test conditions: Use the method proposed by the present invention to process test data and verify the performance of the method proposed by the invention. The communication data collected in the test is added with the impulse noise collected during the ninth Arctic scientific expedition as shown in Figure 4 . It can be seen from the figure that the occurrence time of the impulse noise is random and has a very high amplitude. Based on the generalized signal-to-noise ratio, the intensity of the noise is changed, and the bit error rate performance of several different algorithms under different signal-to-noise ratios is analyzed. The communication parameters in the test are as follows: The number of symbols of the OFDM signal is 20, each symbol contains 1024 subcarriers, and 24 null subcarriers are arranged on both sides of all subcarriers respectively. There are a total of 244 pilot subcarriers, which are evenly distributed at equal intervals among the remaining subcarriers. The center frequency of the carrier is 12 kHz, the bandwidth is 4 kHz, the passband sampling rate is 48 kHz, and the QPSK mapping method is used.

[0041] Analysis of test results: When there is interference from Arctic sub-ice impulse noise in the received signal, the test data processing results are as shown in Figure 5 . Figure 5 is the comparison of the bit error rates between the proposed method and the traditional method when the generalized signal-to-noise ratio ranges from -10 to 20 dB. The curve closer to the horizontal axis indicates that the bit error rate is smaller and the communication is more reliable under the same signal-to-noise ratio.

[0042] Figure 5 The four curves in it are the bit error rate results obtained by processing the simulation test data without suppressing impulse noise, suppressing impulse noise using blanking algorithms, suppressing impulse noise using median filters, and suppressing impulse noise using the method proposed by the present invention. The results show that the traditional algorithm has certain performance advantages when the signal-to-noise ratio is low and the impulse noise is strong, but when the impulse noise is not obvious enough, its performance stability is poor, and the bit error rate is relatively high and it loses its performance advantages. The method proposed by the present invention, due to using the method of first estimating the background noise and then suppressing the impulse noise, can still maintain stable performance when the impulse noise is weak, making the bit error rate of OFDM communication under all signal-to-noise ratios significantly reduced, and having obvious performance advantages under different signal-to-noise ratio conditions.

[0043] From the above analysis of the test results, it can be obtained that the method proposed by the present invention has completely better simulation performance than the traditional method under polar impulse noise and can be applied to the underwater acoustic OFDM communication scenario under polar impulse noise interference.

Claims

1. A method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions, characterized in that, It includes the following steps: Step 1: After inputting the OFDM passband received signal, perform OFDM demodulation, downsampling, and Fourier transform steps; extract the received signal at the pilot positions, and combined with the known pilot transmitted signals, a preliminary rough channel estimation can be performed to obtain the channel estimation values in the time domain and frequency domain; since the background noise at this time follows a non-Gaussian distribution, the least squares algorithm is selected for the rough channel estimation here, reducing the influence of impulse noise on the accuracy of the channel estimation result; Step 2: Use the received signal and the channel estimation values in Step 1 for the channel equalization and decision-making process to obtain the transmitted signal symbols; after re-performing the symbol mapping and OFDM modulation steps, reconstruct the passband transmitted signal; convolve with the time-domain channel estimation value obtained in Step 1 to reconstruct the noise-free OFDM passband received signal estimation value; Step 3: Subtract the noisy received signal in Step 1 from the received signal estimation value in Step 2 to obtain the background noise estimation value; due to the influence of the background noise itself and the errors brought by the channel estimation process, the background noise estimation value here is difficult to be accurate and contains partial information of the desired signal, so it should not be directly eliminated as noise from the received signal, but is used to detect impulse noise in Step 4; Step 4: Perform noise detection and classification on the noise estimation value in Step 3; First, based on the noise model, perform parameter estimation on the noise estimation value, and calculate the pulse noise variance defined by the generalized signal-to-noise ratio; Furthermore, use the improved principle to perform pulse noise detection on the noise estimation value in Step 3, record the pulse noise positions in the index vector, and finally mark the pulse noise through extreme value processing; Step 5: Process the noise marked in Step 4 and use an adaptive median filter to suppress impulse noise; first, define a filtering window centered on each impulse noise point, and after removing all other impulse noise points from this window, it is considered that only Gaussian noise exists in the window at this time; use the median of all Gaussian noise points to replace the impulse noise point at the center of the current window to achieve the suppression of impulse noise; repeat this process until all the impulse noise detected in Step 4 has been processed; at this time, the background noise approximately follows a Gaussian distribution; then add the noise after suppressing impulse interference to the noise-free received signal estimation value in Step 2 to obtain the received signal after suppressing impulse noise; during the process of calculating the median, the size of the filtering window is adaptively adjusted according to the number of impulse noise points in the window; Step 6: Re-perform the OFDM demodulation, channel estimation, channel equalization, decision-making, and decoding processes on the received signal in Step 5, and finally restore and output the source information; at this time, the background noise of the received signal follows a Gaussian distribution, so the sparse Bayesian learning algorithm is used in the channel estimation step, thereby improving the accuracy of the final decoding.

2. A method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions according to claim 1, characterized in that: Through Step 1, Step 2, and Step 3, first estimate the background noise in the OFDM received signal, and use this estimation result to suppress the impulse noise component.

3. A method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions according to claim 1, characterized in that: In step 4, during the noise detection and classification of the background noise estimation value, the generalized signal-to-noise ratio based on impulse noise defines the variance of the model impulse noise, and an improvement is proposed principle for detecting the position of impulse noise.

4. A method for suppressing impulse noise in underwater acoustic OFDM communication in polar regions according to claim 1, characterized in that: In Step 5, use an adaptive median filter to process the estimation value of the background noise, convert the impulse noise in it into Gaussian noise, thereby achieving the suppression of impulse noise; and the processing process is only used for the marked impulse noise points, and an adaptive filter window length is also used.

Images