Signal detection circuit and method based on Brayster rank correlation coefficient

Abstract

The invention discloses a signal detection circuit and method based on a Bolester rank correlation coefficient. The signal detection circuit comprises three comparator arrays, three multiplier arrays, three adder trees, an accumulator, a divider and a register. Wherein the comparator array is an n * n matrix, and each element on the matrix is respectively subjected to comparison operation; the first multiplier array and the third multiplier array are both n3 2-input multipliers, and the second multiplier array is n4 3-input multipliers; the first adder tree and the third adder tree are both n3 input adders, and the second adder tree is an n4 input adder. By adopting the technical scheme of the invention, the interference of pulse noise can be well suppressed, so that the signal detection precision is greatly improved.

Description

Technical Field

[0001] The invention belongs to the technical field of signal detection, and in particular relates to a signal detection circuit and method based on the Breitbart rank correlation coefficient. Background Art

[0002] Signal detection in noise is a basic task in radar, sonar, communication and other systems. For example, in the time delay detection of echo signals in radar and sonar systems, in order to determine whether a specific signal has been received by the receiver, the signal needs to be detected. At present, the commonly used signal detection algorithm in the literature is the matched filter that maximizes the output signal-to-noise ratio. Due to its simple form, complete theory, and easy implementation, the matched filter has been widely used. In the additive Gaussian white noise environment, the matched filter has the best detection performance. However, the background noise in the real environment not only contains additive Gaussian noise, but also some outliers that obey the pulse distribution due to channel interference, cloud and rain reflection, etc. These pulse components will cause the signal-to-noise ratio of the received echo signal to decrease rapidly over a period of time, resulting in a serious deterioration in the performance of the matched filter. To determine whether there is a specific signal segment in the received signal with noise interference, it is possible to determine whether the received signal contains a signal segment with a high correlation with the specific signal, that is, to determine the degree of correlation between the received signal and the specific signal. The Pearson product-moment correlation coefficient (PPMCC) is the most classic indicator in the literature for measuring the degree of signal correlation, but it is very sensitive to the pulse component in the signal and thus loses its effect in the pulse noise environment. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a signal detection circuit and method based on the Breitbart rank correlation coefficient.

[0004] To achieve the above object, the present invention adopts the following technical solution:

[0005] A signal detection circuit based on the Breitbart rank correlation coefficient comprises three comparator arrays, three multiplier arrays, three adder trees, an accumulator, a divider and a register; the comparator arrays are all n×n matrices, and each element on the matrix is ​​subjected to a comparison operation; the first multiplier array and the third multiplier array are both n×n matrices. 3 2-input multipliers, the second multiplier array is n 4 3-input multipliers; the first adder tree and the third adder tree are both an n 3 The input adder, the second adder tree is an n 4 An adder for input; wherein received signal segments of the same length are sequentially selected from the received signal With specific signal It enters three comparator arrays, three multiplier arrays, three adder trees, accumulators, dividers and registers in turn for calculation and processing to obtain a series of Breitbart rank correlation coefficient values. Whether a specific signal is received is determined based on whether the time corresponding to the maximum value is close to the delay of the received signal.

[0006] Preferably, the signals x1,······,x n At the same time, send it to the first comparator array and the second comparator array, and convert x1,···,x i ,···,x n and x1,···,x k ,···,x n and x1,···,x j ,···,x n Conduct 2 Comparison operations are performed to obtain sgn(x i -x k ) and sgn(x i -x j ), where i, j, k = 1, 2, · · · , n; signal y1, · · · · · , y n At the same time, it is sent to the third comparator array to convert y1,···,y i ,···,y n with y1,···,y l ,···,y n Conduct 2 Comparison operation, get sgn(y i -y l ), where i,l=1,2,···,n, and n is the signal length.

[0007] As a preferred embodiment, the comparison results are sent to the corresponding multiplier arrays, and the comparison results sgn(x i -x k ) and sgn(x i -x j ) is simultaneously sent to the first multiplier array to complete the 2-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(x i -x k ); compare the result sgn(x i -x k )、sgn(x i -x j ) and sgn(y i -y l) is simultaneously sent to the second multiplier array to complete the 3-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(x i -x k )sgn(y i -y l ); compare the result sgn(x i -x j ) and sgn(y i -y l ) is simultaneously sent to the third multiplier array to complete the 2-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(y i -y l ).

[0008] As a preferred embodiment, the obtained multiplication results are simultaneously sent to the corresponding adder trees, wherein the multiplication results in the first multiplier array are sent to the first adder tree, and the multiplication results in the first multiplier array are synchronously and parallelly completed. 3 Input addition operation, get The multiplication results in the second multiplier array are fed into the second adder tree, completing n multiplications synchronously and in parallel. 4 Input addition operation, get The multiplication results in the third multiplier array are fed into the third adder tree to complete n 3 Input addition operation, get

[0009] As a preferred method, the obtained addition results are simultaneously sent to the accumulator to complete the linear accumulation and operation.

[0010]

[0011] As a preferred method, the accumulated sum is sent to a divider to complete the division operation.

[0012]

[0013] The present invention also provides a signal detection method based on the Brest rank correlation coefficient, comprising:

[0014] Step S1: Select received signal segments of the same length from the received signal in sequence With specific signal Enter three comparator arrays, three multiplier arrays, three adder trees, accumulators, dividers and registers in sequence for operation processing to obtain a series of values ​​of Breitbart rank correlation coefficients;

[0015] Step S2, judging whether a specific signal is received according to whether the time corresponding to the maximum value is close to the delay of the received signal;

[0016] The comparator arrays are all n×n matrices, and each element in the matrix is ​​compared. The first multiplier array and the third multiplier array are both n×n matrices. 3 2-input multipliers, the second multiplier array is n 4 3-input multipliers; the first adder tree and the third adder tree are both an n 3 The input adder, the second adder tree is an n 4 Input adder.

[0017] The present invention introduces the Breitbart rank correlation coefficient into the signal processing fields such as radar, sonar, and communication for the first time. By using the rank information of the data instead of the corresponding original data, the negative impact of impulse noise on signal correlation analysis can be greatly reduced, thereby improving the performance of signal detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0019] Figure 1 The structure diagram of the signal detection circuit based on the Brecht rank correlation coefficient of the present invention.

[0020] Figure 2 The schematic diagram of using the detection circuit of the present invention to detect specific signals is shown in FIG. 1 , wherein (a) is the specific signal transmitted by the transmitter, (b) is the ideal echo signal received by the receiver, (c) is the superposition of the pulse noise and the ideal echo signal, (d) is the specific signal scanning window, and (e) is the Braxton rank correlation coefficient value. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] Embodiment 1:

[0024] In the field of signal processing such as radar, sonar, and communication, signal detection is an indispensable task. However, in actual environments, the signal received by the receiver will introduce pulse components other than additive Gaussian noise, which brings great difficulties to signal detection. The matched filter has the best detection performance in the additive white Gaussian noise environment and is widely used, but its performance will be seriously deteriorated due to the interference of pulse noise on the received signal. The classic Pearson product-moment correlation coefficient is very sensitive to pulse noise, which brings estimation errors and also loses its detection ability in the pulse noise environment.

[0025] In order to accurately determine whether the received signal is interference noise or a specific signal and effectively estimate the delay of the echo signal, the embodiment of the present invention introduces the Breitbart rank correlation coefficient for signal detection in the fields of radar, sonar, and communication. The Breitbart rank correlation coefficient makes full use of the rank information of the signal, has good robustness in the pulse noise environment, and can solve the problems caused by pulse noise interference. represents a signal segment of length n in the noise-interfered received signal received by the receiver. represents a specific signal of length n sent by the transmitter without being interfered by noise, for elements x1,...,x i ,...x n In ascending order, x i Located at the kth position of the sorted data, k represents x i The rank of i Similarly, i The rank is denoted as Q i . Then the Breitbart rank correlation coefficient is defined as:

[0026]

[0027] From the above formula, it can be seen that, unlike the matched filter and the Pearson product-moment correlation coefficient, the Breitbart rank correlation coefficient only uses the rank information of the original data, which can well reduce the interference of impulse noise and has good robustness in the impulse noise environment. Therefore, the present invention introduces the Breitbart rank correlation coefficient for delay detection of signals in radar, sonar and communication systems, and uses it as the basis for FPGA circuit design through corresponding theoretical derivation.

[0028] Define the step function H(t), when t>0, H(t)=1, when t<0, H(t)=0, when t=0, According to the definition of rank, we can get the following expression:

[0029]

[0030] Where i,j,l=1,2,···,n, where n is the signal length.

[0031] Define the sign function sgn(t), when t>0, sgn(t)=1; when t<0, sgn(t)=-1; when t=0, sgn(t)=0. Based on the relationship between the step function and the sign function, that is:

[0032] 2H(t)=sgn(t)+1

[0033] The rank of the data can be expressed as:

[0034]

[0035] Where i,j,l=1,2,···,n, where n is the signal length.

[0036] Expanding the Breitbart rank correlation coefficient, we can get:

[0037]

[0038] Define the summation term of the right half as U, that is:

[0039]

[0040] Decomposing U yields:

[0041]

[0042] Since sgn(x i -x j )+sgn(x j -x i )=0, and sgn(0)=0, we can get:

[0043]

[0044] Similarly, we can get:

[0045]

[0046] Therefore, the expression of the Breitbart rank correlation coefficient can be obtained as follows:

[0047]

[0048] The present invention provides a signal detection circuit based on the Braxton rank correlation coefficient. represents a signal segment of length n in the noise-interfered received signal received by the receiver. Represents a specific signal of length n sent by the transmitter without being interfered by noise, using Figure 1 The circuit structure shown in the figure processes the fragment, including three comparator arrays, three multiplier arrays, three adder trees, an accumulator, a divider, and a register. The comparator arrays are all n×n matrices, and each element in the matrix is ​​compared separately; the first multiplier array and the third multiplier array are both n×n. 3 2-input multipliers, the second multiplier array is n 4 3-input multipliers; the first adder tree and the third adder tree are both an n 3 The input adder, the second adder tree is an n 4 The adder performs the input; the accumulator performs the linear addition and operation; the divider performs the 2-input division operation; the register stores the final operation result.

[0049] The operation steps of the signal after it is sent into the circuit are as follows:

[0050] 1)Signal x1,······,x n At the same time, send it to the first comparator array and the second comparator array, and convert x1,···,x i ,···,x n and x1,···,x k ,···,x n and x1,···,x j ,···,x n Conduct 2 Comparison operations are performed to obtain sgn(x i -x k ) and sgn(x i -x j ), where i, j, k = 1, 2, · · · , n; signal y1, · · · · · , y n At the same time, it is sent to the third comparator array to convert y1,···,y i ,···,y n with y1,···,y l ,···,y n Conduct 2 Comparison operation, get sgn(y i -y l ), where i,l=1,2,···,n, n is the signal length;

[0051] 2) The comparison results are sent to the corresponding multiplier arrays, where the comparison result sgn(x i -x k ) and sgn(x i -x j) is simultaneously sent to the first multiplier array to complete the 2-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(x i -x k ); compare the result sgn(x i -x k )、sgn(x i -x j ) and sgn(y i -y l ) is simultaneously sent to the second multiplier array to complete the 3-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(x i -x k )sgn(y i -y l ); compare the result sgn(x i -x j ) and sgn(y i -y l ) is simultaneously sent to the third multiplier array to complete the 2-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(y i -y l );

[0052] 3) The obtained multiplication results are simultaneously sent to the corresponding adder trees, where the multiplication results in the first multiplier array are sent to the first adder tree, and n multiplications are completed synchronously and in parallel. 3 Input addition operation, get The multiplication results in the second multiplier array are fed into the second adder tree, completing n multiplications synchronously and in parallel. 4 Input addition operation, get The multiplication results in the third multiplier array are fed into the third adder tree to complete n 3 Input addition operation, get

[0053] 4) The addition results are sent to the accumulator at the same time to complete the linear accumulation and operation:

[0054]

[0055] 5) The accumulated sum is sent to the divider to complete the division operation:

[0056]

[0057] 6) The division result is sent to a register for storage, and the value of the Brecht rank correlation coefficient is obtained, and the entire circuit operation is completed.

[0058] Select received signal segments of the same length (length is n) from the received signal in sequence With specific signal According to steps 1) to 6), correlation analysis is performed to obtain a series of values ​​of the Braxton rank correlation coefficient. Whether a specific signal is received is determined based on whether the time corresponding to the maximum value is close to the delay of the received signal.

[0059] In an impulse noise environment, the performance of matched filters and Pearson product-moment correlation coefficient is greatly reduced, while the Breitbart rank correlation coefficient can reduce the harmful effects of impulse noise on signal detection and has good robustness, including a mathematical expectation that is very close to the true value and a small standard deviation.

[0060] Related experiments:

[0061] In order to compare the performance of matched filter, Pearson product-moment correlation coefficient, and Brest rank correlation coefficient under impulse noise, we will verify it through Monte Carlo experiment. Figure 2 It is a schematic diagram of signal detection. Among them, (a) is the specific signal emitted by the transmitter, (b) is the ideal echo signal received by the receiver, which has the same waveform as the emitted signal but has a certain delay T0, (c) is the superposition of pulse noise and ideal echo signal, (d) is the specific signal as the scanning window, each time a segment of the same length is selected from the received signal for correlation analysis with the specific signal, so as to obtain the Breitbart rank correlation coefficient value, (e) is the Breitbart rank correlation coefficient value, where the time corresponding to the maximum value in (e) is the delay T0 of the received signal.

[0062] The experimental parameters are set as follows: the specific signal template is set to a sine signal

[0063]

[0064] Impulse noise is generated by a Gaussian mixture model:

[0065] Z~(1-ε)Ν(0,σ 2 )+εΝ(0,δ 2 )

[0066] Where ε = 0.03 represents the probability of the pulse component occurring in the entire pulse noise environment, and δ = 50 >> σ represents the standard deviation of the pulse component. Then the signal-to-noise ratio of the received signal can be defined as:

[0067]

[0068] Through Monte Carlo experiments, the performance of matched filter, Pearson product-moment correlation coefficient, and Breitbart rank correlation coefficient under different signal-to-noise ratios can be compared, which verifies that Breitbart rank correlation coefficient is robust in impulse noise environments. The number of experiments is 500, and the first received signal position is T0=1000. The experimental schematic diagram is shown in the figure. Figure 2 The experimental results are shown in Table 1.

[0069] Table 1

[0070] SNR Matched filter Pearson product-moment correlation coefficient Brest rank correlation coefficient 0 971.1±222.5768 966.7±209.4104 1000±1.8402 -1 956.7±211.3405 962.1±199.9424 999.8±2.0301 -2 962.1±202.1801 974.1±192.3078 1000.1±2.1184 -3 965.8±232.5636 976.4±193.3436 1000.0±2.3807 -4 957.2±216.0070 966.9±188.6113 999.9±2.5321 -5 961.6±227.9085 960.8±218.5415 999.9±2.8266 -6 950.8±223.3328 948.1±221.3363 999.8±3.1605 -7 936.0±250.4841 953.0±220.9771 1000.1±3.4448 -8 950.9±225.7330 955.2±225.8030 999.9±3.9445 -9 936.3±231.7735 958.5±216.1614 1000.0±4.5674

[0071] The data in Table 1 are the "average ± standard deviation" of the echo signal positions detected multiple times. From the experimental results in Table 1, it can be seen that when the signal is interfered by impulse noise, the matched filter and Pearson product-moment correlation coefficient basically fail, while the Breitbart rank correlation coefficient shows good robustness in the impulse noise environment, including a mathematical expectation close to the true value and a small standard deviation, which shows that the Breitbart rank correlation coefficient can be used as a powerful tool for signal detection in the impulse noise environment.

[0072] Embodiment 2:

[0073] The embodiment of the present invention also provides a signal detection method based on the Braxton rank correlation coefficient, comprising the steps of:

[0074] Step S1: Select received signal segments of the same length from the received signal in sequence With specific signal Enter three comparator arrays, three multiplier arrays, three adder trees, accumulators, dividers and registers in sequence for operation processing to obtain a series of values ​​of Breitbart rank correlation coefficients;

[0075] Step S2, judging whether a specific signal is received according to whether the time corresponding to the maximum value is close to the delay of the received signal;

[0076] The comparator arrays are all n×n matrices, and each element in the matrix is ​​compared. The first multiplier array and the third multiplier array are both n×n matrices. 3 2-input multipliers, the second multiplier array is n 4 3-input multipliers; the first adder tree and the third adder tree are both an n 3 The input adder, the second adder tree is an n 4 Input adder.

[0077] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A signal detection circuit based on the Braxton rank correlation coefficient, characterized in that: It includes three comparator arrays, three multiplier arrays, three adder trees, an accumulator, a divider, and a register; the comparator arrays are all n×n matrices, and each element on the matrix is ​​compared separately; the first multiplier array and the third multiplier array are both n 3 2-input multipliers, the second multiplier array is n 4 3-input multipliers; the first adder tree and the third adder tree are both an n 3 The input adder, the second adder tree is an n 4 An adder for input; wherein received signal segments of the same length are sequentially selected from the received signal With specific signal It enters three comparator arrays, three multiplier arrays, three adder trees, accumulators, dividers and registers in turn for calculation and processing to obtain a series of Breitbart rank correlation coefficient values. Whether a specific signal is received is determined based on whether the time corresponding to the maximum value is close to the delay of the received signal.

2. The signal detection circuit based on the Braxton rank correlation coefficient as claimed in claim 1, characterized in that: Signal x1,······,x n At the same time, send it to the first comparator array and the second comparator array, and convert x1,···,x i ,···,x n and x1,···,x k ,···,x n and x1,···,x j ,···,x n Conduct 2 Comparison operations are performed to obtain sgn(x i -x k ) and sgn(x i -x j ), where i, j, k = 1, 2, · · · , n; signal y1, · · · · · , y n At the same time, it is sent to the third comparator array to convert y1,···,y i ,···,y n with y1,···,y l ,···,y n Conduct 2 Comparison operation, get sgn(y i -y l ), where i,l=1,2,···,n, and n is the signal length.

3. The signal detection circuit based on the Blaisdt rank correlation coefficient as claimed in claim 2, characterized in that: The comparison results are sent to the corresponding multiplier arrays, and the comparison results sgn(x i -x k ) and sgn(x i -x j ) is simultaneously sent to the first multiplier array to complete the 2-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(x i -x k ); compare the result sgn(x i -x k )、sgn(x i -x j ) and sgn(y i -y l ) is simultaneously sent to the second multiplier array to complete the 3-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(x i -x k )sgn(y i -y l ); compare the result sgn(x i -x j ) and sgn(y i -y l ) is simultaneously sent to the third multiplier array to complete the 2-input multiplication operation synchronously and in parallel, and sgn(x i -x j )sgn(y i -y l ).

4. The signal detection circuit based on the Blaisdt rank correlation coefficient as claimed in claim 3, characterized in that: The multiplication results are sent to the corresponding adder trees at the same time, where the multiplication results in the first multiplier array are sent to the first adder tree, and n multiplications are completed synchronously and in parallel. 3 Input addition operation, get The multiplication results in the second multiplier array are fed into the second adder tree, completing n multiplications synchronously and in parallel. 4 Input addition operation, get The multiplication results in the third multiplier array are fed into the third adder tree to complete n 3 Input addition operation, get 5. The signal detection circuit based on the Braxton rank correlation coefficient as claimed in claim 4, characterized in that: The addition results are sent to the accumulator at the same time to complete the linear accumulation and operation.

6. The signal detection circuit based on the Braxton rank correlation coefficient as claimed in claim 5, characterized in that: The accumulated sum is sent to the divider to complete the division operation.

7. A signal detection method based on the Braxton rank correlation coefficient, characterized in that: include Step S1: Select received signal segments of the same length from the received signal in sequence With specific signal Enter three comparator arrays, three multiplier arrays, three adder trees, accumulators, dividers and registers in sequence for operation processing to obtain a series of values ​​of Breitbart rank correlation coefficients; Step S2, judging whether a specific signal is received according to whether the time corresponding to the maximum value is close to the delay of the received signal; The comparator arrays are all n×n matrices, and each element in the matrix is ​​compared. The first multiplier array and the third multiplier array are both n×n matrices. 3 2-input multipliers, the second multiplier array is n 4 3-input multipliers; the first adder tree and the third adder tree are both an n 3 The input adder, the second adder tree is an n 4 Input adder.

Images