Underwater acoustic signal channel coding blind identification method and system

Abstract

The invention relates to the technical field of signal analysis and processing, in particular to an underwater acoustic signal channel coding blind identification method and system.The method comprises the steps that firstly, preprocessed demodulation data are obtained, the data are in a bit stream form, and a threshold value is set; comparing the code word characteristics of the standard deleted convolutional code with a threshold value, and when the code word characteristics of the standard deleted convolutional code are greater than the threshold value, performing blind identification on the standard deleted convolutional code; otherwise, continuing subsequent judgment; comparing the code word characteristics of the RS code with a threshold value, and when the code word characteristics of the RS code are greater than the threshold value, performing RS code blind identification; otherwise, continuing subsequent judgment; comparing the code word characteristics of the BCH code with a threshold value, and performing BCH code blind identification when the code word characteristics of the BCH code are greater than the threshold value; otherwise, judging the coding format as an unknown coding format. According to the method, the BCH with the code length of 63 or below, the RS with the code length of 15 or below and the standard deletion convolutional codes of all code rates can be subjected to blind recognition, the requirement for the sample data size is small, and the good error code resistance performance is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of signal analysis and processing, and in particular to a method and system for blind recognition of underwater acoustic signal channel coding. Background Art

[0002] Due to the random movement of the sea surface, the random unevenness of the seabed, and the inhomogeneity of the water body, the underwater acoustic channel is unevenly distributed in space and is randomly time-varying. Actual signal transmission often uses short bursts. Considering the channel coding that matches this, it should mainly be coding with medium to short code lengths and strong error correction capabilities. Due to the limited number of codewords contained in the data under short burst conditions, how to complete the blind identification and blind decoding of the coding type under small sample conditions and noisy environments, and analyze its coding parameters, is the main problem and difficulty currently faced. Existing technologies are usually limited to using different code rates of the same code type, and can determine the code rate and complete decoding based on the information provided by the key field. Summary of the Invention

[0003] The present invention aims to solve the problem of how to complete blind identification of multiple coding types under small sample conditions and noisy environments, and proposes a method and system for blind identification of underwater acoustic signal channel coding, which can identify whether a group of underwater acoustic signals contains BCH codes, RS codes and standard deleted convolutional codes with medium and short code lengths. If so, their parameters are further determined to facilitate subsequent decoding.

[0004] In order to achieve the above purpose, the technical solutions adopted are:

[0005] The present invention provides a method for blind identification of underwater acoustic signal channel coding, comprising:

[0006] First, obtain the pre-processed demodulated data in the form of bit stream and set the threshold;

[0007] Compare the codeword feature of the standard erasure convolutional code with the threshold. If it is greater than the threshold, perform blind recognition of the standard erasure convolutional code; otherwise, continue with subsequent judgment.

[0008] Compare the codeword feature of the RS code with the threshold. If it is greater than the threshold, perform RS code blind recognition; otherwise, continue with subsequent judgment;

[0009] The codeword characteristics of the BCH code are compared with the threshold. If the codeword characteristics are greater than the threshold, the BCH code is blindly identified; otherwise, it is judged as an unknown coding format.

[0010] According to the blind identification method of underwater acoustic signal channel coding of the present invention, further, a check vector conformity method is used to perform blind identification on the standard erasure convolutional code, and the parameters for identifying the standard erasure convolutional code data include the code length n, two sub-generators g0, g1 and the data starting point q.

[0011] According to the blind identification method of underwater acoustic signal channel coding of the present invention, further, performing blind identification of standard erasure convolutional codes specifically includes:

[0012] S11, obtaining the demodulated data after preprocessing;

[0013] S12, setting a threshold value T;

[0014] S13. Set two sub-generators: g0 = 133, g1 = 171;

[0015] S14, initialize subcode length n=1;

[0016] S15. Update n=n+1. If n>8 and g0=133, set two sub-generators: g0=171 and g1=133, and return to step S14. If n>8 and g0=171, it means that the code is not a standard erasure convolutional code, and jump to step S112. If it is otherwise, continue.

[0017] S16. Select the (g0, g1) convolutional code as the parity check vector of the standard erasure convolutional code with a code rate of (n-1) / n of the mother code, denoted as h;

[0018] S17, initialization data starting point q=0;

[0019] S18. Update q=q+1. If q>n, return to step S15.

[0020] S19. Select data according to the starting point q, intercept the maximum length of the integer multiple code length n, and starting from the starting position, separate the vectors with the same length as h and overlapping with each other from the data; the starting bits of two adjacent vectors are n bits apart, and the total number of these vectors is recorded as L;

[0021] S110, calculate the inner product of L vectors with h in sequence, that is, verify whether the check relationship holds; if it is equal to 0, it holds, and if it is equal to 1, it does not hold. The number of valid relationships is recorded as L1;

[0022] S111. If L1 / L ≥ T, then output n, g0, g1, q and the program terminates; otherwise, return to step S18; where L1 / L is the codeword feature of the standard erasure convolutional code;

[0023] S112. The program terminates.

[0024] According to the blind recognition method of underwater acoustic signal channel coding of the present invention, further, the RS code is blindly recognized by using the code root statistics method, and the parameters for identifying the RS code data include the code length n, the finite field parameter m, the primitive polynomial p(x), the error correction capability t, the starting root power B, the shortened information bit s, the correct symbol starting point q and the correct code word starting point q m .

[0025] According to the blind recognition method for underwater acoustic signal channel coding of the present invention, further, performing blind recognition of RS code specifically includes:

[0026] S21, obtaining the pre-processed demodulated data;

[0027] S22, setting a threshold value T;

[0028] S23, initialize the finite field parameter m=5;

[0029] S24, update m=m-1. If m<3, it indicates that the code is not an RS code, and jump to step S218;

[0030] S25, initialization symbol starting point q=0;

[0031] S26, update the starting point q = q + 1, if q> m, return to step S24;

[0032] S27. Select data according to the starting point, intercept the maximum length of an integer multiple of m, and divide the data into non-overlapping m-tuples; in the order of the leftmost high bit, convert each m-tuple into an integer to obtain multi-encoded data;

[0033] S28. Find GF(2 m ) row, select a p(x) that has not been selected under the current starting condition, and the root is denoted as α;

[0034] S29, initialization n=2 m ;

[0035] S210, update n=n-1, if n=2 m-1 , then return to step S26;

[0036] S211, initialize q m =0;

[0037] S212, update starting point q m =q m +1, if q m >n, then return to step S210;

[0038] S213, according to the starting point q m Select multi-code data, truncate the maximum length of the integer multiple code length, and divide the data into L code words of length n that do not overlap each other;

[0039] S214, initialize B=0;

[0040] S215, update B=B+1;

[0041] S216, Verification α BIs it the root of L code words, with α B The number of code words with roots is recorded as L B , if B<2 m -1, then return to step S215; otherwise, continue;

[0042] S217, if in module 2 m On the integer ring of -1, there are no two consecutive Bs that satisfy L B / L≥T, then return to step S212; otherwise, continue; where L B / L is the code word feature of RS code;

[0043] S218, in module 2 m -1 integer ring, there are j consecutive positive integers B satisfying L B / L≥T, then t is the largest positive integer not exceeding j / 2, and the first positive integer is the power of the starting root;

[0044] S219, output parameters m,n,p(x),t,q,q m ,s=2 m-1 -1-n,B;

[0045] S220: The program terminates.

[0046] According to the blind identification method of underwater acoustic signal channel coding of the present invention, further, the code root statistics method is used to blindly identify the BCH code, and the parameters for identifying the BCH code data include the code length n, the primitive polynomial p(x), the error correction capability t, the shortened information bit number s, and the correct data starting point q.

[0047] According to the blind identification method of underwater acoustic signal channel coding of the present invention, further, performing BCH code blind identification specifically includes:

[0048] S31, obtaining the pre-processed demodulated data;

[0049] S32, setting a threshold T;

[0050] S33, initialization code length n=64;

[0051] S34. Update n=n-1. If n<5, it indicates that the code is not a BCH code. Go to step S316. Otherwise, continue.

[0052] S35, according to 2 m-1 ≤n≤2 m -1, calculate m;

[0053] S36, initialization data starting point q=0;

[0054] S37, update the starting point q = q + 1, if q> n, return to step S34;

[0055] S38. Select data according to the starting point q, intercept the maximum length of an integer multiple of the code length n, and divide the data into L non-overlapping codewords of length n; set t0 = 0, which is used to store the maximum error correction ability in the currently matched code.

[0056] S39. Find a row of GF(2 m ) in the primitive polynomial table, select a p(x) that has not been selected under the current starting point condition, and denote the root as α; if there is no unselected p(x), then return to step S37.

[0057] S310. Verify whether α is the root of the L codewords, and denote the number of codewords with α as the root as L1.

[0058] S311. If L1 / L < T, then return to step S39; otherwise, continue; where L1 / L is the codeword characteristic of the BCH code.

[0059] S312. Initialize i = 0.

[0060] S313. Update i = i + 1. If 2i + 1 > 2 m -1, then return to step S39.

[0061] S314. Verify whether α 2i+1 is the root of the L codewords, and denote the number of codewords with α as the root as L1. If L1 / L < T, then take t = i. If t > t0, then take t0 = t, and update n, p(x), t, q, s = 2 m-1 -1 - n; otherwise, return to step S313.

[0062] S315. Output the parameters n, p(x), t, q, s = 2 m-1 -1 - n.

[0063] S316. Terminate the program.

[0064] Furthermore, the present invention also provides an underwater acoustic signal channel coding blind recognition system, including:

[0065] A preprocessing module, which is used to obtain the preprocessed demodulated data. The data is in the form of a bit stream, and a threshold is set.

[0066] An encoding parameter recognition module, which is used to compare the codeword characteristics of the standard punctured convolutional code with the threshold. When it is greater than the threshold, perform blind recognition of the standard punctured convolutional code; otherwise, continue with subsequent judgments; compare the codeword characteristics of the RS code with the threshold. When it is greater than the threshold, perform blind recognition of the RS code; otherwise, continue with subsequent judgments; compare the codeword characteristics of the BCH code with the threshold. When it is greater than the threshold, perform blind recognition of the BCH code; otherwise, it is determined as an unknown coding format.

[0067] The beneficial effects achieved by adopting the above technical solution are:

[0068] Taking into account the difficulties faced by blind recognition of underwater acoustic signal coding, such as small samples, complex channel environment, and diverse coding parameters, the present invention adopts a blind recognition method for underwater acoustic signal channel coding that has strong anti-noise ability, is insensitive to the number of samples, and is easy to expand to parallel computing to improve recognition speed performance. The check vector conformity method is used for the analysis of standard erasure convolutional codes. This method calculates the check vector of the standard erasure convolutional code in advance, performs traversal matching verification on the samples, can identify information such as the starting point and code rate, and only requires a maximum of 100 subcodes to obtain a stable and reliable analysis conclusion, with strong anti-noise and anti-error capabilities. The code root statistics method is used for BCH code and RS code, and the matching analysis is performed by traversing parameters such as the starting point, primitive polynomial, and code length. Among them, the RS code needs to be converted from bits to symbols according to the setting of the primitive polynomial and then perform code root statistical analysis. Under the condition of a bit error rate of one thousandth, this method only requires a data volume of at most 20 code words to complete a stable and reliable analysis. The advantages of the present invention are summarized as follows:

[0069] (1) The present invention does not require a large amount of data and can adapt to recognition and analysis tasks under small sample conditions. Most types of encoding only require 1000 bits of data to complete stable analysis.

[0070] (2) The present invention has good noise and error resistance. The algorithm itself has good error resistance and can meet the recognition and analysis tasks under the condition of a bit error rate of one thousandth. At the same time, the error resistance performance can be further improved according to the threshold adjustment strategy. For example, combined with the judgment of the analyst, a low threshold is used for coarse screening, and then a high threshold is used for fine screening, thereby effectively reducing the error recognition rate.

[0071] (3) The present invention can also be further adapted for parallel computing. Since there are three types of codes—"standard deleted convolutional code," "BCH code," and "RS code"—the recognition engines can be run in parallel. Furthermore, each code contains several independent parameter combinations, allowing for parallel algorithms to simultaneously perform matching analysis, further improving analysis efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings of the embodiments of the present invention. The drawings are only used to illustrate some embodiments of the present invention, but not to limit all embodiments of the present invention thereto.

[0073] Figure 1 It is a flow chart of a blind identification method for underwater acoustic signal channel coding according to an embodiment of the present invention. DETAILED DESCRIPTION

[0074] The following will be combined with the accompanying drawings of specific embodiments of the present invention to clearly and completely describe the exemplary embodiments of the present invention. Unless otherwise defined, technical or scientific terms used in the present invention should be given the common meanings understood by people with ordinary skills in the relevant field.

[0075] This embodiment discloses a method for blind recognition of underwater acoustic signal channel coding. Figure 1 As shown, including:

[0076] First, obtain the pre-processed demodulated data in the form of a bit stream and set the threshold (the default value is 0.8).

[0077] Compare the codeword feature of the standard erasure convolutional code with the threshold. If it is greater than the threshold, perform blind recognition of the standard erasure convolutional code; otherwise, continue with subsequent judgment.

[0078] Compare the codeword feature of the RS code with the threshold. If it is greater than the threshold, perform RS code blind recognition; otherwise, continue with subsequent judgment.

[0079] The codeword characteristics of the BCH code are compared with the threshold. If the codeword characteristics are greater than the threshold, the BCH code is blindly identified; otherwise, it is judged as an unknown coding format.

[0080] The following introduces the recognition and analysis methods of three codes respectively. The standard erasure convolutional code uses the check vector coincidence method for blind recognition, and the RS code and BCH code use the code root statistics method for blind recognition.

[0081] (1) Identification and analysis of standard erasure convolutional codes

[0082] The mother code of a standard erasure convolutional code uses a (2, 1, 6) convolutional code with taps 133 and 171 as its two sub-generators, or swapping the order to use (171, 131). Following the erasure template, high-rate codes can be obtained, such as 2 / 3, 3 / 4, ..., 7 / 8.

[0083] Table 1: Erasure patterns of (133, 171) convolutional codes

[0084]

[0085] Table 2: Erasure patterns of (171, 133) convolutional codes

[0086]

[0087] This embodiment adopts a parity check vector conformance analysis method to obtain the parity check vectors of the erasure codes of the (133, 171) and (171, 133) convolutional codes in advance.

[0088] Table 3. Check vectors of erasure codes for (133,171) convolutional codes

[0089]

[0090] Table 4. Check vectors of erasure codes for (171,133) convolutional codes

[0091]

[0092] The steps of the standard erasure convolutional code identification and analysis method are as follows:

[0093] S11. Obtain pre-processed demodulated data.

[0094] S12. Set the threshold T (the default value is 08, which can be modified).

[0095] S13. Set two sub-generators: g0=133, g1=171.

[0096] S14. Initialize the subcode length n=1.

[0097] S15. Update n=n+1. If n>8 and g0=133, set two sub-generators: g0=171, g1=133, and return to step S14. If n>8 and g0=171, it means that the code is not a standard erasure convolutional code, and jump to step S112. If it is other cases, continue.

[0098] S16. Select the (g0, g1) convolutional code as the parity check vector of the standard erasure convolutional code with a code rate of (n-1) / n of the mother code, denoted as h.

[0099] S17. Initialize the data starting point q=0.

[0100] S18. Update q=q+1. If q>n, return to step S15.

[0101] S19. Select data according to the starting point q, truncate the maximum length of the integer multiple code length n, and separate from the data vectors with the same length h and overlapping with each other starting from the starting position; the starting bits of two adjacent vectors are n bits apart, and the total number of these vectors is recorded as L.

[0102] S110. Calculate the inner product of the L vectors with h in sequence, i.e., verify whether the check relationship holds. If it is equal to 0, it holds, and if it is equal to 1, it does not hold. The number of valid relationships is recorded as L1.

[0103] S111. If L1 / L ≥ T, then output n, g0, g1, q and the program terminates; otherwise, return to step S18; wherein L1 / L is the codeword feature of the standard erasure convolutional code.

[0104] S112. The program terminates.

[0105] (2) Recognition and analysis of RS codes

[0106] The list of RS code parameters that need to be identified is as follows:

[0107] Table 5 RS code parameters

[0108]

[0109] The parameters for identifying RS code data include: code length n, finite field parameter m, primitive polynomial p(x), error correction capability t, starting root power B, shortened information bit number s, correct symbol starting point q (demodulated data is a complete symbol starting from the qth bit) and correct codeword starting point q m (The complete symbol sequence of the demodulated data starts from the qth m symbols to start with (a complete codeword).

[0110] Table 6 Primitive polynomials used by RS

[0111]

[0112] When the code length n satisfies 2 m-1 ≤n≤2 m -1, take the finite field GF(2 m ) and the corresponding primitive polynomials to construct RS codes. Each finite field can be constructed with multiple primitive polynomials. It is necessary to identify which primitive polynomial is used for an RS code.

[0113] This example uses the code root statistics method combined with the idea of traversal search to complete the recognition and analysis of RS codes. The steps are as follows:

[0114] S21. Obtain pre-processed demodulated data.

[0115] S22. Set the threshold T (the default value is 0.8, which can be modified).

[0116] S23. Initialize the finite field parameter m=5.

[0117] S24. Update m=m-1. If m<3, it indicates that the code is not an RS code, and jump to step S218.

[0118] S25. Initialize the symbol starting point q=0.

[0119] S26. Update the starting point q=q+1. If q>m, return to step S24.

[0120] S27. Select data according to the starting point, cut off the maximum length of an integer multiple of m, and divide the data into non-overlapping m-tuples; in the order of high bits on the left, convert each m-tuple into an integer to obtain multi-encoded data.

[0121] S28. Find GF(2 m ) in a row, select a p(x) that has not been selected under the current starting condition, and the root is denoted as α.

[0122] S29, initialization n=2 m .

[0123] S210, update n=n-1, if n=2 m-1 , then return to step S26.

[0124] S211, initialize q m =0.

[0125] S212, update starting point q m =q m +1, if q m >n, then return to step S210.

[0126] S213, according to the starting point q m Select multi-code data, truncate the maximum length of the integer multiple code length, and divide the data into L code words of length n that do not overlap.

[0127] S214. Initialize B=0.

[0128] S215. Update B=B+1.

[0129] S216, Verification α B Is it the root of L code words, with α B The number of code words with roots is recorded as L B , if B<2 m -1, return to step S215; otherwise, continue.

[0130] S217, if in module 2 m On the integer ring of -1, there are no two consecutive Bs that satisfy L B / L≥T, then return to step S212; otherwise, continue; where L B / L is the code word feature of RS code.

[0131] S218, in module 2 m -1 integer ring, there are j consecutive positive integers B satisfying L B / L≥T, then t is the largest positive integer not exceeding j / 2, and the first positive integer is the power of the starting root.

[0132] S219, output parameters m,n,p(x),t,q,q m ,s=2 m-1 -1-n,B.

[0133] S220: The program terminates.

[0134] (3) Identification and analysis of BCH codes

[0135] The list of BCH code parameters that need to be identified is as follows:

[0136] Table 7 BCH code parameters

[0137]

[0138]

[0139] The parameters for identifying BCH code data include: code length n, primitive polynomial p(x), error correction capability t, shortened information bit number s, and correct data starting point q (the demodulated data is a complete codeword starting from the qth bit).

[0140] Table 8 Primitive polynomial table

[0141]

[0142]

[0143] When the code length n satisfies 2 m-1 ≤n≤2 m -1, take the finite field GF(2 m ) and the corresponding primitive polynomials to construct BCH codes. Each finite field can be constructed with multiple primitive polynomials. It is necessary to identify which primitive polynomial is used for a BCH code.

[0144] This example uses the code root statistics method combined with the idea of traversal search to complete the recognition and analysis of BCH codes. The steps are as follows:

[0145] S31. Obtain pre-processed demodulated data.

[0146] S32. Set the threshold T (the default value is 0.8, which can be modified).

[0147] S33. Initialization code length n=64.

[0148] S34. Update n=n-1. If n<5, it indicates that the code is not a BCH code. Go to step S316. Otherwise, continue.

[0149] S35, according to 2 m-1 ≤n≤2 m -1, calculate m.

[0150] S36. Initialize the data starting point q=0.

[0151] S37. Update the starting point q = q + 1. If q > n, then return to step S34.

[0152] S38. Select data according to the starting point q, intercept the maximum length that is an integer multiple of the code length n, and divide the data into L non - overlapping codewords of length n; set t0 = 0, which is used to store the maximum error - correcting ability in the currently matched code.

[0153] S39. Find a row of the primitive polynomial table for GF(2 m ). Select a p(x) that has not been selected under the current starting - point condition, and denote its root as α; if there is no unselected p(x), then return to step S37.

[0154] S310. Verify whether α is a root of the L codewords, and denote the number of codewords with α as the root as L1.

[0155] S311. If L1 / L < T, then return to step S39; otherwise, continue; where L1 / L is the code - word characteristic of the BCH code.

[0156] S312. Initialize i = 0.

[0157] S313. Update i = i + 1. If 2i + 1>2 m - 1, then return to step S39.

[0158] S314. Verify whether α 2i+1 is a root of the L codewords, and denote the number of codewords with α as the root as L1. If L1 / L < T, then take t = i. If t > t0, then take t = t0, and update n, p(x), t, q, s = 2 m-1 - 1 - n; otherwise, return to step S313.

[0159] S315. Output the parameters n, p(x), t, q, s = 2 m-1 - 1 - n.

[0160] S316. Terminate the program.

[0161] Corresponding to the above method, this embodiment also discloses an underwater acoustic signal channel - coding blind - recognition system, including:

[0162] A pre - processing module, which is used to obtain the pre - processed demodulated data. The data is in the form of a bit stream, and a threshold is set;

[0163] The coding parameter identification module is used to compare the codeword characteristics of the standard erasure convolutional code with a threshold value. When the codeword characteristics are greater than the threshold value, the standard erasure convolutional code is blindly identified; otherwise, subsequent judgment is continued; the codeword characteristics of the RS code are compared with a threshold value. When the codeword characteristics are greater than the threshold value, the RS code is blindly identified; otherwise, subsequent judgment is continued; the codeword characteristics of the BCH code are compared with a threshold value. When the codeword characteristics are greater than the threshold value, the BCH code is blindly identified; otherwise, it is judged as an unknown coding format.

[0164] This invention employs a traversal search approach, uses a check vector consistency method, and employs code root statistics. In the design of each algorithm module, the order in which each parameter is set is carefully considered to improve algorithm efficiency. This technique has a wide range of applications and can perform blind recognition for BCH codes with a code length of 63 or less, RS codes with a code length of 15 or less, and standard erasure convolutional codes of all code rates.

[0165] The effectiveness and superiority of the present invention are further verified below in combination with experimental results.

[0166] ① 14 signals are generated sequentially using BCH codes. These signals are transmitted over a channel. When received, they are affected by noise, resulting in some bit errors. The bit error rate represents the degree of noise. Using the aforementioned blind recognition method, the code parameters are correctly determined, and the correct information can be recovered through decoding.

[0167] Table 9 Coding parameters of 14 groups of BCH codes

[0168]

[0169]

[0170] Effect: Under the conditions that the total length of the code sequence data does not exceed 1500 and the bit error rate is greater than 1‰, all code words can be correctly identified and the decoding error rate is 0, that is, all information can be correctly recovered.

[0171] ② Generate 11 signals encoded using RS codes. These signals are transmitted through a channel. When received, they are affected by noise, resulting in some bit errors. The bit error rate represents the degree of noise. Using the blind recognition method described above, the code parameters are correctly obtained, and the correct information can be recovered through decoding.

[0172] Table 10 Coding parameters of 14 groups of RS codes

[0173] Group number n M T p(x) B Bit Error Rate Number of codewords Data length Decoding error rate 1 15 4 5 19 10 0.0017 10 600 0 2 14 4 4 19 7 0.0036 10 560 0 3 13 4 3 25 11 0.0019 10 520 0 4 12 4 4 25 5 0.0021 10 480 0 5 11 4 2 19 0 0.0023 10 440 0 6 10 4 2 25 11 0.0075 10 400 0 7 9 4 1 19 14 0.0056 10 360 0 8 8 4 2 19 6 0.0031 10 320 0 9 7 3 1 11 0 0.0030 16 336 0 10 6 3 1 13 3 0.0069 16 288 0 11 5 3 1 13 3 0.0056 30 450 0

[0174] Effect: Under the conditions that the total length of the code sequence data does not exceed 600 and the bit error rate is greater than 1‰, all code words can be correctly identified and the decoding error rate is 0, that is, all information can be correctly recovered.

[0175] ③. 14 sets of signals are generated sequentially using standard erasure convolutional codes. When receiving these signals, they are affected by noise, resulting in some bit errors. The bit error rate represents the degree of noise impact. Using the aforementioned blind recognition method, the code parameters are correctly obtained, and the correct information is recovered through decoding.

[0176] Table 11 Coding parameters of 14 groups of standard erasure convolutional codes

[0177] Group number Generator Bitrate Bit Error Rate Number of subcodes Data length Decoding error rate 1 (133,171) 1 / 2 0.0125 80 160 0 2 (133,171) 2 / 3 0.0083 40 120 0 3 (133,171) 3 / 4 0.0063 80 320 0 4 (133,171) 4 / 5 0.002 100 500 0 5 (133,171) 5 / 6 0.0021 80 480 0 6 (133,171) 6 / 7 0.0071 40 280 0 7 (133,171) 7 / 8 0.0031 80 640 0 8 (171,133) 1 / 2 0.0050 300 600 0 9 (171,133) 2 / 3 0.0017 200 600 0 10 (171,133) 3 / 4 0.0050 100 400 0 11 (171,133) 4 / 5 0.0040 100 500 0 12 (171,133) 5 / 6 0.0017 100 600 0 13 (171,133) 6 / 7 0.0018 80 560 0 14 (171,133) 7 / 8 0.0016 80 640 0

[0178] Effect: Under the conditions that the total length of the code sequence data does not exceed 700 and the bit error rate is greater than 1‰, all code words can be correctly identified and the decoding error rate is 0, that is, all information can be correctly recovered.

[0179] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A blind identification method for underwater acoustic signal channel coding, characterized in that: include: First, obtain the pre-processed demodulated data in the form of bit stream and set the threshold; Compare the codeword feature of the standard erasure convolutional code with the threshold. If it is greater than the threshold, perform blind recognition of the standard erasure convolutional code; otherwise, continue with subsequent judgment. Compare the codeword feature of the RS code with the threshold. If it is greater than the threshold, perform RS code blind recognition; otherwise, continue with subsequent judgment; The codeword characteristics of the BCH code are compared with the threshold. If the codeword characteristics are greater than the threshold, the BCH code is blindly identified; otherwise, it is judged as an unknown coding format.

2. The method for blind identification of underwater acoustic signal channel coding according to claim 1, characterized in that: The standard erasure convolutional code is blindly identified using a check vector coincidence method. The parameters for identifying the standard erasure convolutional code data include the code length n, two sub-generators g0 and g1, and the data starting point q.

3. The method for blind identification of underwater acoustic signal channel coding according to claim 2, characterized in that: The blind recognition of standard erasure convolutional codes specifically includes: S11, obtaining the demodulated data after preprocessing; S12, setting a threshold value T; S13. Set two sub-generators: g0 = 133, g1 = 171; S14, initialize subcode length n=1; S15. Update n=n+1. If n>8 and g0=133, set two sub-generators: g0=171 and g1=133, and return to step S14. If n>8 and g0=171, it means that the code is not a standard erasure convolutional code, and jump to step S112. If it is otherwise, continue. S16. Select the (g0, g1) convolutional code as the parity check vector of the standard erasure convolutional code with a code rate of (n-1) / n of the mother code, denoted as h; S17, initialization data starting point q=0; S18. Update q=q+1. If q>n, return to step S15. S19. Select data according to the starting point q, intercept the maximum length of the integer multiple code length n, and starting from the starting position, separate the vectors with the same length as h and overlapping with each other from the data; the starting bits of two adjacent vectors are n bits apart, and the total number of these vectors is recorded as L; S110, calculate the inner product of L vectors with h in sequence, that is, verify whether the check relationship holds; if it is equal to 0, it holds, and if it is equal to 1, it does not hold. The number of valid relationships is recorded as L1; S111. If L1 / L ≥ T, then output n, g0, g1, q and the program terminates; otherwise, return to step S18; where L1 / L is the codeword feature of the standard erasure convolutional code; S112. The program terminates.

4. The method for blind identification of underwater acoustic signal channel coding according to claim 1, characterized in that: For RS code, the code root statistics method is used for blind recognition. The parameters for identifying RS code data include code length n, finite field parameter m, primitive polynomial p(x), error correction capability t, starting root power B, shortened information bit s, correct symbol starting point q and correct code word starting point q. m .

5. The method for blind identification of underwater acoustic signal channel coding according to claim 4, characterized in that: The RS code blind recognition specifically includes: S21, obtaining the pre-processed demodulated data; S22, setting a threshold value T; S23, initialize the finite field parameter m=5; S24, update m=m-1. If m<3, it indicates that the code is not an RS code, and jump to step S218; S25, initialization symbol starting point q=0; S26, update the starting point q = q + 1, if q> m, return to step S24; S27. Select data according to the starting point, intercept the maximum length of an integer multiple of m, and divide the data into non-overlapping m-tuples; in the order of the leftmost high bit, convert each m-tuple into an integer to obtain multi-encoded data; S28. Find GF(2 m ) row, select a p(x) that has not been selected under the current starting condition, and the root is denoted as α; S29, initialization n=2 m ; S210, update n=n-1, if n=2 m-1 , then return to step S26; S211, initialize q m =0; S212, update starting point q m =q m +1, if q m >n, then return to step S210; S213, according to the starting point q m Select multi-code data, truncate the maximum length of the integer multiple code length, and divide the data into L code words of length n that do not overlap each other; S214, initialize B=0; S215, update B=B+1; S216, Verification α B Is it the root of L code words, with α B The number of code words with roots is recorded as L B , if B<2 m -1, then return to step S215; otherwise, continue; S217, if in module 2 m On the integer ring of -1, there are no two consecutive Bs that satisfy L B / L≥T, then return to step S212; otherwise, continue; where L B / L is the code word feature of RS code; S218, in module 2 m -1 integer ring, there are j consecutive positive integers B satisfying L B / L≥T, then t is the largest positive integer not exceeding j / 2, and the first positive integer is the power of the starting root; S219, output parameters m,n,p(x),t,q,q m ,s=2 m-1 -1-n,B; S220: The program terminates.

6. The method for blind identification of underwater acoustic signal channel coding according to claim 1, characterized in that: For BCH codes, the code root statistical method is used for blind recognition. The parameters for identifying BCH code data include the code length n, the primitive polynomial p(x), the error correction capability t, the shortened information bit number s, and the correct data starting point q.

7. The method for blind identification of underwater acoustic signal channel coding according to claim 6, characterized in that: The specific steps for blind recognition of BCH codes include: S31. Obtain the preprocessed demodulation data; S32. Set the threshold T; S33. Initialize the code length n = 64; S34. Update n = n - 1. If n < 5, it indicates that this code is not a BCH code, and go to step S316; otherwise, continue; S35, according to 2 m-1 ≤n≤2 m -1, calculate m; S36. Initialize the data starting point q = 0; S37. Update the starting point q = q + 1. If q > n, return to step S34; S38. Select data according to the starting point q, intercept the maximum length that is an integer multiple of the code length n, and divide the data into L non-overlapping codewords of length n; set t0 = 0, which is used to store the maximum error correction capability in the currently matched coding; S39, find GF(2 m ) row, select a p(x) that has not been selected under the current starting condition, and record the root as α; if there is no unselected p(x), return to step S37; S310. Verify whether α is a root of the L codewords, and record the number of codewords with α as the root as L1; S311. If L1 / L < T, return to step S39; otherwise, continue; where L1 / L is the codeword feature of the BCH code; S312. Initialize i = 0; S313, update i=i+1, if 2i+1>2 m -1, then return to step S39; S314. Verify α 2i+1 Whether it is the root of L codewords. The number of codewords with α as the root is denoted as L1. If L1 / L < T, then take t = i. If t > t0, then take t0 = t, and update n, p(x), t, q, s = 2 m-1 -1 - n; otherwise, return to step S313; S315, output parameters n, p(x), t, q, s = 2 m-1 -1-n; S316. Terminate the program.

8. A blind recognition system for underwater acoustic signal channel coding, characterized in that: It includes: A preprocessing module, which is used to obtain the preprocessed demodulation data. The data is in the form of a bit stream and set the threshold; An encoding parameter recognition module, which is used to compare the codeword features of the standard punctured convolutional code with the threshold. When it is greater than the threshold, perform blind recognition of the standard punctured convolutional code; otherwise, continue with subsequent judgments; compare the codeword features of the RS code with the threshold. When it is greater than the threshold, perform blind recognition of the RS code; otherwise, continue with subsequent judgments; compare the codeword features of the BCH code with the threshold. When it is greater than the threshold, perform blind recognition of the BCH code; otherwise, it is judged as an unknown coding format.

Images