A method for blindly identifying a scrambling code generation polynomial of a direct spread scrambling signal

Abstract

The application discloses a method for blindly identifying a scrambling code generation polynomial of a direct spread scrambling signal, which is not limited by the polynomial order and can identify any high-order polynomial, and a reasonable K value setting can fully utilize the storage space. Moreover, the method for blindly identifying the scrambling code generation polynomial of the direct spread scrambling signal is robust to frequency offset, has no limit to the identifiable polynomial order, and requires a short length of the spread scrambling sequence, thus having strong practicability.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a blind identification method for scrambling code generator polynomials of direct-sequence spread spectrum scrambling signals. Background Technology

[0002] Direct Sequence Spread Spectrum (DSSS) technology is one of the commonly used communication technologies in modern communication systems (Yan Linbin. Research on Blind Estimation Technology of DS-CDMA Signal [D]. University of Electronic Science and Technology of China, 2017). Its basic principle is: at the signal transmitting end, the information code sequence is multiplied by a high-speed spreading code sequence to increase the transmission rate of the information code sequence, thereby expanding the signal spectrum and reducing the signal power spectral density. DSSS scrambling refers to the process of "scrambling" the spread sequence at the bit layer to change the statistical characteristics of the transmitted sequence and improve the confidentiality of information bits; that is, randomizing the scrambling code. At the receiving end, descrambling of the received signal must be completed before subsequent signal processing can proceed. The DSSS scrambling signal studied in this embodiment refers to a signal that uses a Walsh sequence or OVSF sequence to spread the information code sequence, and then scrambles it with a Gold sequence. For non-cooperative receivers, the spreading sequence number, Gold sequence generator polynomial, and register initial state of the transmitted signal are all unknown, making the analysis and processing of direct-sequence spread spectrum (DSSS) scrambling signals extremely difficult. Current research has addressed the identification of the spreading sequence number and Gold sequence register initial state, but the foundation for register initial state identification lies in the identification of the scrambling code generator polynomial. Therefore, in non-cooperative communication, achieving blind identification of the scrambling code generator polynomial for DSS scrambling signals is crucial for subsequent signal processing.

[0003] For the identification of generator polynomials, existing technologies (Yang Zhongli, Liu Yujun. Research on comprehensive algorithm of self-synchronizing scrambled sequence [J]. Information Technology, 2005(02):30-32) and existing technologies (Wu Wenjun, Huang Zhiping, Tang Guilin, Liu Chunwu. Fast recovery of scrambled sequence with error [J]. Acta Ordnance et al., 2009, 30(08):1134-1138) construct a system of scrambled equations using the iterative relationship of linear shift registers, and then use the Walsh-Hadamard transform to solve the system of scrambled equations to determine the generator polynomial. The advantage is that the identification accuracy is high and the principle is easy to understand. However, the disadvantage is that the higher the series, the more the computational load of the algorithm increases exponentially, and the series of generator polynomials of Gold sequences are all high, which is almost unbearable for hardware. Existing technology (Yuan Ye. Research on Blind Recognition of Linear Scrambling Codes [D]. University of Electronic Science and Technology of China, 2013) proposes a recognition algorithm based on probability distribution distance. This algorithm analyzes the mathematical model of the descrambling process and concludes that the values ​​of the bit groups corresponding to the scrambling code sequence and the generator polynomial are constrained by the source sequence and have an imbalance. Based on this, the generator polynomial of the scrambling code is calculated. The disadvantage of this algorithm is that it requires prior knowledge of the series and number of terms of the generator polynomial, which makes it impractical in engineering. In addition, the above algorithms all have high requirements for the source imbalance. For direct-sequence spread spectrum signals, the 0 and 1 bits of the spread sequence are perfectly balanced, which causes the above algorithms to fail completely.

[0004] For direct-sequence spread spectrum (DSSS) signals, the scrambling code used is a Gold sequence, which corresponds to a high generator polynomial series. If traditional identification algorithms are used to identify the scrambling code number, the computational complexity increases exponentially, the source imbalance causes the algorithm to fail, and even the frequency offset introduced during signal propagation makes traditional algorithms difficult to apply. In summary, blind identification of scrambling code parameters for intercepted DSSS scrambling signals is quite difficult in non-cooperative communication. This invention proposes a blind identification method for the generator polynomial of scrambling codes in DSSS scrambling signals in the context of non-cooperative communication. Summary of the Invention

[0005] To address the limitations and defects of existing technologies, this invention provides a blind identification method for scrambling codes generated by a multinomial in direct-sequence spread spectrum (DSSS) scrambling signals, comprising:

[0006] The direct-sequence spread spectrum (DSSS) scrambling signal transmitted by the transmitter is modeled as follows:

[0007] T(k)=S n (k)(β I X I (k)+jβ Q X Q (k)), k=1,2,3,...

[0008] Where, β I β is the gain of the I-channel signal. Q X is the gain of the Q-channel signal. I (k) is the sequence of I-way bits after being spread by a Walsh sequence or an OVSF sequence, X Q (k) is the sequence of Q-way bits after being spread by a Walsh sequence or an OVSF sequence, S n (k) is the Gold sequence used for scrambling, n is the index of the Gold sequence, and j is the imaginary unit;

[0009] The direct-sequence spread spectrum (DSSS) scrambling signal received at the receiver is modeled as follows:

[0010] T rec (k)=e jθ βS n (k)(β I X I (k)+jβ Q X Q (k))+n(k)

[0011] Where θ is the phase offset caused by the initial phase and frequency offset, n(k) is the noise component, and β is the channel gain. I β is the signal gain of path I. Q The signal gain of the Q-path;

[0012] Obtain sequence M rec (k), the sequence M rec The expression for (k) is as follows:

[0013]

[0014] Among them, w re (k) represents the noise component. This is a shifted sequence of the original Gold sequence;

[0015] The sequence M rec (k) The sequence of bits with a decision of 0 or 1 is defined as sequence R. rec (k);

[0016] The Gold sequence generator polynomial series is obtained to be 2L, and the sequence R is obtained. rec (k) satisfies a fault-containing system of (N-2L×2L+1)-dimensional equations, where N is the sequence R. rec The length of (k) is 2L, which is the generating polynomial series of the Gold sequence;

[0017] The solution to the system of equations containing errors is obtained as a vector [c 2L c 2L-1 ... c1 1]T , where c i ,0<i≤2L represents the coefficient factors of the generator polynomial, and T represents the transpose of the vector.

[0018] Optionally, the expression for the faulty system of equations is as follows:

[0019]

[0020] Optionally, the obtained sequence M rec (k) is preceded by:

[0021] The received signal T is processed according to a preset preprocessing method. rec (k) Perform preprocessing.

[0022] Optionally, the received signal T is processed according to a preset preprocessing method. rec (k) The preprocessing steps include:

[0023] For the received signal T rec (k) Perform conjugate difference and double downsampling processing.

[0024] Optional, also includes:

[0025] Selecting the real part of the processed signal yields the sequence M. rec (k).

[0026] Optional, also includes:

[0027] The coefficients of the generator polynomial are solved using the WH analysis method based on matrix update.

[0028] Optionally, the step of solving for the coefficients of the generating polynomial using the WH analysis method based on matrix update includes:

[0029] The N-2L row coefficients of the faulty equation system can be considered as N-2L groups of 2L+1 bits each, with each bit group having 2 2L +1 The possible values ​​are constructed with a dimension of 1×2. 2L+1 The vector S is used to count the number of times each bit group appears in the system of error-containing equations, accumulate the same bit group, and store the accumulated value in the decimal number position corresponding to the vector S.

[0030] Construction dimension is 2 K The Hadamard matrix H, where K < 2L+1, is used to represent the dimension of the matrix. This matrix H is used to iteratively generate a matrix of dimension 2. 2L+1 ×2 K The matrix, wherein the vector S is a matrix of dimension 2 2L+1 ×2 KThe product of the matrices is part of the Walsh spectrum, where the values ​​of K and L are used to ensure that the dimension is 2. 2L+1 ×2 K The matrix can be stored;

[0031] Traverse i=2 2L-K ,...,2 2L+1-K -1, convert the decimal number i being iterated over into a binary vector R of dimension 2L+1-K. i Read the vector R sequentially from the least significant bit to the most significant bit. i The bits, when the vector R i When the bit is 0, the matrix H is updated to H = [H; H], and when the vector R i When a bit is 1, update the matrix H to H = [H; -H], until the vector R is completely read. i After reading all bits of the vector R, i After all bits are processed, the matrix H is updated to a dimension 2. 2L+1 ×2 K matrix H i The vector S and the matrix H i Multiplying them together yields the vector W. i ;

[0032] All vectors W i Combining them in order yields a dimension of 1×2. 2L The vector W0, with dimensions 1×2 2L The vector 0 is combined with the vector W0 to obtain the final vector W, which is the Walsh spectrum. The binary vector corresponding to the maximum value index of the vector W is obtained, and the binary vector corresponding to the maximum value index of the vector W is the coefficient of the generator polynomial.

[0033] Optionally, the value of K is 13.

[0034] Optionally, the value of L is 7.

[0035] Optionally, the value of N can be 1200 or 2400.

[0036] The present invention has the following beneficial effects:

[0037] Compared to existing technologies, the blind identification method for scrambling generator polynomials of direct-sequence spread spectrum (DSSS) scrambling signals provided by this invention is not limited by the order of the generator polynomial and can identify generator polynomials of any high order. A reasonable K value setting can fully utilize storage space. Furthermore, the blind identification method for scrambling generator polynomials of DSS scrambling signals provided by this invention is robust to frequency offset, has no limitation on the order of the identifiable generator polynomial, and requires a relatively short spreading scrambling sequence, thus possessing strong practicality. Attached Figure Description

[0038] Figure 1 The flowchart of the WH algorithm based on matrix update provided in Embodiment 1 of the present invention is shown.

[0039] Figure 2(a) is a schematic diagram of the recognition accuracy of the generator polynomial when N=1200 provided in Embodiment 1 of the present invention.

[0040] Figure 2(b) is a schematic diagram of the recognition accuracy of the generator polynomial when N=2400 provided in Embodiment 1 of the present invention. Detailed Implementation

[0041] To enable those skilled in the art to better understand the technical solution of the present invention, the blind identification method of scrambling code generator polynomial for direct-sequence spread scrambling signals provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0042] Example 1

[0043] This embodiment proposes a blind identification method for scrambling generator polynomials of direct-spread scrambling signals. This method is designed for signals that are directly spread by Walsh or OVSF sequences and then scrambled by Gold sequences. It can perform blind identification of Gold sequence generator polynomials even with frequency offset.

[0044] Based on the relationship between the signal spreading gain and the period length of the spreading sequence, direct-sequence spread spectrum (DSSS) signals can be divided into short-code DSSS signals and long-code DSSS signals. A DSSS signal with a spreading gain equal to the period of the spreading sequence is called a short-code DSSS signal; a DSSS signal with a spreading gain less than the period of the spreading sequence is called a long-code DSSS signal. This embodiment mainly studies short-code DSSS signals.

[0045] The transmitted direct-sequence spread scrambling signal is modeled as follows:

[0046] T(k)=S n (k)(β I X I (k)+jβ Q X Q (k)), k=1,2,3,...

[0047] Where, β I and β Q X represents the gain of the I-channel and Q-channel signals, respectively. I (k) and X Q (k) represents the sequences of I-path and Q-path bits after being spread by Walsh or OVSF sequences, S n (k) represents the Gold sequence used for scrambling, and n is the index of the Gold sequence. The definition of the Gold sequence is existing technology and will not be repeated in this embodiment. The identification of its generator polynomial is the blind identification target of this embodiment.

[0048] The spread spectrum scrambling signal received at the receiver is modeled as follows:

[0049] T rec (k)=e jθ βS n (k)(β I X I (k)+jβ Q X Q (k))+n(k)

[0050] Where θ is the phase offset caused by the initial phase and frequency offset, n(k) is the noise component, and β is the channel gain. I β is the signal gain of path I. Q This is the signal gain of the Q-path.

[0051] To eliminate the effects of the spreading code, the initial phase of the received signal, and frequency offset on the phase of the received signal, it is first necessary to process the received data T. rec (k) Preprocessing is performed to obtain the shifted sequence of the original Gold sequence. This embodiment employs a preset preprocessing method, performing conjugate difference and double downsampling on the received data, and then taking the real part of the processed signal to obtain M. rec (k):

[0052]

[0053] Among them, w re (k) represents the noise component. This is a shifted sequence of the original Gold sequence. For ease of subsequent analysis, sequence M is... rec (k) The bit sequence with a decision of 0 or 1 is defined as R. rec (k).

[0054] Before blindly identifying the scrambling generator polynomial, this embodiment first calculates that the series of the Gold sequence generator polynomial is 2L. Further, the sequence R... rec (k) satisfies the following system of faulty equations:

[0055]

[0056] The above equation is a (N-2L×2L+1) dimensional system of faulty equations. The vector [c] that satisfies the above equation as many times as possible is [c]. 2L c 2L-1 ... c1 1] T This is called the solution to the system of equations containing errors, and the solution is the coefficient of the generating polynomial.

[0057] Based on the above equation, in order to obtain the coefficients of the generator polynomial, the traditional WH analysis method needs to first statistically analyze the bit states of the faulty equation system and store them in a 1×2... 2L+1 In a 2-dimensional vector S, generate a vector of dimension 2. 2L+1 The Hadamard matrix is ​​then used. Subsequently, the Walsh spectral coefficients obtained by multiplying the vector S by the Hadamard matrix are used to calculate the polynomial coefficients, but this is impossible for generator polynomials with high series. To address the problem of the Hadamard matrix being too large to store and compute, this embodiment uses 2... 2L+1 The Hadamard matrix is ​​segmented into several smaller matrices of lower dimension to facilitate storage and computation. Walsh spectral coefficients are then calculated piecewise based on these smaller matrices, and finally combined to form the final Walsh spectral coefficients. For the above purpose, this embodiment proposes a matrix-updated WH analysis method to solve for the generator polynomial coefficients.

[0058] Figure 1 The flowchart of the WH algorithm based on matrix update provided in Embodiment 1 of the present invention is as follows:

[0059] a. Based on sequence R rec (k) Construct a system of equations containing errors.

[0060] b. Consider the N-2L row coefficients of the faulty system of equations as N-2L groups of 2L+1 bits each, where each bit group may have 2... 2L +1 Consider the possible values. Construct a 1×2... 2L+1 A dimensional vector S is used to count the number of times each bit group appears in the system of equations. The cumulative value of the same bit group is stored in the corresponding decimal position of S.

[0061] c. Construct a 2 K A 2-dimensional Hadamard matrix H, where K < 2L+1, is used to iteratively generate a 2-dimensional Hadamard matrix. 2L+1 ×2 K A small matrix of dimension S, the product of which is a vector S and this matrix constitutes a portion of the final desired Walsh spectrum. The values ​​of K and L should guarantee 2... 2L+1 ×2 K A matrix of a certain dimension can be stored.

[0062] d. Traverse i=2 2L-K ,...,2 2L+1-K -1, for a decimal number i in a certain traversal, convert it into a binary vector R of dimension 2L+1-K. i Read R sequentially from least significant bit to most significant bit. iThe initial matrix H is updated to H = [H; H] when a certain bit is 0, and to H = [H; -H] when it is 1, until the vector R is completely read. i All bits are read. After reading all bits, the H matrix is ​​updated to a matrix with dimension 2. 2L+1 ×2 K matrix H i Finally, using vector S and matrix H i Multiplying them together yields the vector W. i .

[0063] e. Put all vectors W i Combining them in sequence yields a 1×2 dimension. 2L The vector W0. A vector with dimension 1×2... 2L Combining the zero vector with the W0 vector yields the final vector W, which is the desired Walsh spectrum. The binary vector corresponding to the maximum value index of vector W is the coefficient of the generator polynomial.

[0064] It is worth noting that in step d, the reason why parameter i changes from 2... 2L-K Start iterating up to 2 2L+1-K The reason for using -1 instead of starting the iteration from 0 is that the highest bit of the generator polynomial coefficients is 1, and the decimal number corresponding to the generator polynomial coefficients must be located in the latter half of the vector W index. Therefore, the first half of the vector W can be skipped. Additionally, to fully utilize the simulation storage space and simultaneously speed up execution, the value of K is set to 13 during simulation.

[0065] Compared to the traditional WH algorithm, the matrix-updated WH analysis method proposed in this embodiment is no longer limited by the number of generator polynomial series. It can identify generator polynomials of any high order, and a reasonable K value setting can also ensure full utilization of storage space, making it highly practical for engineering applications.

[0066] This embodiment uses a Gold sequence of order L=7 for simulation, and the generator polynomial of the preferred pair of m sequences is x. 7 +x 3 +1 and x 7 +x 3 +x 2 +x+1. Generate a spread spectrum signal of length N, spread using either Walsh or OVSF codes, with the spreading code chosen randomly. Then, scramble the spread signal by multiplying it with an equal-length Gold sequence, setting the symbol rate to 1 Msps. Pass the scrambled spread signal through an AWGN channel, introducing a frequency offset. Set the signal-to-noise ratio to -2 to 4 dB, and the sequence length N to 1200 and 2400, respectively. Add frequency offsets of 0 kHz, 1 kHz, and 2 kHz for each signal-to-noise ratio. Perform 500 Monte Carlo simulations.

[0067] Figure 2(a) shows the accuracy of generator polynomial recognition when N=1200 according to Embodiment 1 of the present invention. Figure 2(b) shows the accuracy of generator polynomial recognition when N=2400 according to Embodiment 1 of the present invention. The received signal is preprocessed according to the above process, and then the generator polynomial is recognized according to the above process. Based on the blind recognition scheme designed in this embodiment, the simulation results of the recognition accuracy of Gold code generator polynomials are shown in Figure 2(a) and Figure 2(b).

[0068] As shown in the figure above, the recognition accuracy of the direct-sequence spread spectrum (DSSS) scrambling signal generator polynomial blind recognition algorithm gradually increases with the increase of the signal-to-noise ratio (SNR). With a spread spectrum sequence length of 1200 and an SNR of 3dB, the recognition accuracy can reach over 95%; with a spread spectrum sequence length of 2400 and an SNR of 3dB, the recognition accuracy can reach over 100%. This demonstrates that increasing the sequence length can improve the algorithm's recognition accuracy to some extent. Furthermore, comparing the results under different frequency offsets shows that the frequency offset has almost no impact on the recognition accuracy of this algorithm.

[0069] In summary, the blind identification method for scrambling generator polynomials of direct-spread scrambling signals provided in this embodiment is robust to frequency offset, has no limitation on the order of the identifiable generator polynomial, and requires a short spread spectrum scrambling sequence, thus possessing strong practicality.

[0070] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A blind identification method using a scrambling code generator polynomial for direct-sequence spread spectrum scrambling signals, characterized in that, include: The direct-sequence spread spectrum (DSSS) scrambling signal transmitted by the transmitter is modeled as follows: ， in, For the gain of the I-channel signal, For the gain of the Q-channel signal, This is a sequence of I-way bits spread using a Walsh or OVSF sequence. This is a sequence of Q-way bits spread using a Walsh or OVSF sequence. The Gold sequence used for scrambling, where n is the index of the Gold sequence. j The imaginary unit; The direct-sequence spread spectrum (DSSS) scrambling signal received at the receiver is modeled as follows: ， in, The phase shift is caused by the initial phase and frequency offset. For noise components, For channel gain, for I The signal gain of the path, for Q The signal gain of the path; Obtain sequence The sequence The expression is as follows: ， in, For noise components, This is a shifted sequence of the original Gold sequence; The sequence A sequence of bits with a decision value of 0 or 1 is defined as a sequence. ; The Gold sequence generator polynomial series is obtained to be 2. L To obtain the sequence Satisfied A system of fault-containing equations of dimension, wherein, N For the sequence Length, 2 L The generating polynomial series of the Gold sequence; The solution to the system of equations containing errors is obtained as a vector. ,in, The coefficient factors of the generating polynomial are represented. T Represents the transpose of a vector; The expression for the faulty system of equations is as follows: ； Also includes: The coefficients of the generator polynomial are solved using the WH analysis method based on matrix update. The steps for solving the coefficients of the generator polynomial using the WH analysis method based on matrix update include: The faulty system of equations The row coefficient is regarded as indivual Bit groups of 2 bits each 2L+1 The possible values ​​are constructed with a dimension of 1×2. 2L+1 The vector S is used to count the number of times each bit group appears in the system of error-containing equations, accumulate the same bit group, and store the accumulated value in the decimal number position corresponding to the vector S. Construction dimension is 2 K The Hadamard matrix H, where, K <2 L +1, used to represent the dimension of the matrix, where matrix H is used to iteratively generate a matrix of dimension 2. 2L+1 ×2 K The matrix, wherein the vector S is a matrix of dimension 2 2L+1 ×2 K The product of the matrices is part of the Walsh spectrum, where K and L The value is used to ensure that the dimension is 2. 2L+1 ×2 K The matrix can be stored; Traversal The decimal numbers that will be traversed i Convert to dimension binary vector R i Read the vector R sequentially from the least significant bit to the most significant bit. i The bits, when the vector R i When a bit is 0, the matrix H is updated to... When the vector R i When a bit is 1, the matrix H is updated to... until the vector R is completely read i After reading all bits of the vector R, i After all bits are processed, the matrix H is updated to a dimension 2. 2L+1 ×2 K matrix H i The vector S and the matrix H i Multiplying them together yields the vector W. i ; All vectors W i Combining them in order yields a dimension of 1×2. 2L The vector W0, with dimensions 1×2 2L The vector 0 is combined with the vector W0 to obtain the final vector W, which is the Walsh spectrum. The binary vector corresponding to the maximum value index of the vector W is obtained, and the binary vector corresponding to the maximum value index of the vector W is the coefficient of the generator polynomial.

2. The blind identification method for scrambling code generator polynomials of direct-sequence spread scrambling signals according to claim 1, characterized in that, The obtained sequence The steps beforehand include: The received signal is processed according to a preset preprocessing method. Preprocessing is performed.

3. The blind identification method for scrambling codes generated by the generator polynomial of a direct-sequence spread scrambling signal according to claim 2, characterized in that, The received signal is processed according to a preset preprocessing method. The preprocessing steps include: For the received signal Perform conjugate difference and double downsampling processing.

4. The blind identification method for scrambling code generator polynomial of direct-sequence spread scrambling signal according to claim 3, characterized in that, Also includes: The sequence is obtained by selecting the real part of the processed signal. .

5. The blind identification method for scrambling code generator polynomials of direct-sequence spread scrambling signals according to claim 4, characterized in that, K The value is 13.

6. The blind identification method for scrambling code generator polynomials of direct-sequence spread scrambling signals according to claim 5, characterized in that, L The value is 7.

7. The blind identification method for scrambling code generator polynomials of direct-sequence spread scrambling signals according to claim 6, characterized in that, N The value is 1200 or 2400.

Images