ADS-B signal detection method and system based on multi-dimensional feature joint judgment

Abstract

The invention relates to the technical field of wireless communication and signal processing, in particular to an ADS-B (Automatic Dependent Surveillance-Broadcast) signal detection method and system based on multi-dimensional feature joint judgment, and the method comprises the following steps: signal modeling and orthogonal demodulation: constructing an ADS-B baseband signal model according to an ICAO DO-260B protocol; sliding window energy accumulation and threshold judgment: performing energy accumulation on the demodulation signal based on a sliding window, and detecting an effective signal window in combination with dynamic threshold judgment logic; performing improved M-Rife frequency offset correction: performing FFT analysis on the detected signal in the effective signal window, judging whether dynamic spectrum shifting is needed or not according to spectrum characteristics, and realizing frequency offset compensation through secondary fine correction; and bipolar correlation and peak screening: performing bipolar transformation on the signal leading header after frequency offset correction, calculating a cross-correlation value with a standard sequence, and determining a leading pulse position through a multi-stage screening condition. According to the method, the detection precision, the robustness and the real-time processing efficiency of the ADS-B signal in a complex electromagnetic environment are remarkably improved.

Description

Technical Field

[0001] This invention relates to the fields of wireless communication and signal processing technology, and in particular to an ADS-B signal detection method and system based on multi-dimensional feature joint determination. Background Technology

[0002] ADS-B (Automatic Dependent Surveillance-Broadcast) is the core of a new generation of airborne surveillance technology, and its 1090MHz carrier signal uses PPM modulation. Existing detection technologies face three challenges: First, traditional energy detection methods have a high false alarm rate in low signal-to-noise ratio environments; second, Doppler frequency shifts caused by high-speed aircraft movement (up to ±1MHz) reduce signal demodulation reliability; and third, preamble detection is susceptible to pulse interference, leading to misjudgments.

[0003] Existing frequency offset correction methods, such as the classic Rife algorithm, show a significant drop in accuracy when the frequency offset exceeds the center region of the quantization frequency, resulting in residual phase errors in the corrected signal. Traditional preamble detection methods employ unipolar correlation schemes, with a false detection rate exceeding 40% when SNR < 10dB. Although improved envelope detection methods have been proposed, they have not yet solved the preamble identification problem under the combined effects of frequency offset and noise. Summary of the Invention

[0004] To address the problems of high false alarm rate due to low signal-to-noise ratio, severe frequency offset interference, and insufficient identification of the lead header in aviation ADS-B signal detection, this invention provides an ADS-B signal detection method and system based on multi-dimensional feature joint judgment. Through cascaded optimization of energy pre-screening, frequency offset robustness compensation, and polarity feature enhancement, the detection accuracy and robustness of ADS-B signals in complex electromagnetic environments are significantly improved.

[0005] To achieve the above objectives, the technical solution adopted is:

[0006] This invention provides an ADS-B signal detection method based on multi-dimensional feature joint determination, comprising the following steps:

[0007] Step 1, Signal Modeling and Quadrature Demodulation: Construct an ADS-B baseband signal model according to the ICAO DO-260B protocol. The receiver recovers the in-phase component I(t) and quadrature component Q(t) of the baseband signal through quadrature downconversion.

[0008] Step 2, Sliding window energy accumulation and threshold decision: The demodulated signal is accumulated based on the sliding window, and the valid signal window is detected by combining the dynamic threshold decision logic;

[0009] Step 3, Improve M-Rife frequency offset correction: Perform FFT analysis on the signal within the effective signal window detected in Step 2, determine whether dynamic spectrum shifting is required based on the spectral characteristics, and achieve frequency offset compensation through secondary fine calibration;

[0010] Step 4, Bipolar Correlation and Peak Filtering: Perform bipolar transformation on the preamble of the frequency offset corrected signal, calculate the cross-correlation value with the standard sequence, and determine the preamble pulse position through multi-level filtering conditions.

[0011] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, further, in step 1, the expression of the ADS-B baseband signal model is:

[0012] s(t)=A*cos(2πf c t+φ(t))*[p(t)+d(t)]+w(n)

[0013] Where A represents the signal amplitude, f c Let φ(t) represent the carrier frequency, φ(t) be the phase modulation term, p(t) be the preamble signal, d(t) be the data block signal, and w(n) be Gaussian white noise.

[0014] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, in step 2, the window length for sliding window energy accumulation is set to an integer multiple of the standard duration of the signal, and the energy statistics calculation formula is as follows:

[0015]

[0016] Wherein, the window length W = 120μs × n, n is a positive integer, ΔW is the step size, and I[k] and Q[k] are the discrete spectra of I(t) and Q(t) after Fourier transform; based on the Neyman-Pearson criterion, the dynamic threshold is set as follows:

[0017] λ = γ·max(E[n])

[0018] Where γ = 0.95, the detection and decision logic is as follows:

[0019]

[0020] If the energy in a certain window exceeds the threshold, a signal is considered to have been detected.

[0021] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, further, in step 3, the specific steps of improving M-Rife frequency offset correction include:

[0022] Perform an N-point FFT on each valid signal segment to obtain the discrete spectrum S(k), and locate the index k of the maximum spectral line. max Used to calculate a coarse estimate of the frequency offset: Where N is the number of FFT points, f s f is the sampling frequency. c Indicates the carrier frequency;

[0023] when Dynamic spectrum shifting is performed in real time, and quantization frequency units are defined. Generate a transfer signal: Where, Δf shift It is the frequency shift amount;

[0024] Perform a second fine-tuning on the transfer signal s'(t) to obtain Δf fine ;

[0025] Finally, the frequency offset estimate Δf is obtained. final =Δf shift +Δf fine Using Δf final Output frequency offset compensation signal

[0026] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, the frequency shift calculation formula for dynamic spectrum shift is further as follows:

[0027]

[0028] in, This represents the floor function.

[0029] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, a second fine-tuning is further performed on the shift signal s'(t) to obtain Δf. fine include:

[0030] Perform a second FFT on the shifted signal s'(t) to obtain a new spectrum S'(k), and relocate the new maximum spectral line index k'. max Obtain the normalized difference:

[0031]

[0032] Calculate the fine-tuning value using the parabolic interpolation formula:

[0033]

[0034] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, further, in step 4, the bipolar transformation maps the binary sequence {0, 1} of the preamble header to {-1, 1}, and the standard preamble header sequence is:

[0035] s 0_w = [1,-1,1,-1,-1,-1,-1,1,-1,1,-1,-1,-1,-1,-1,-1,-1].

[0036] The cross-correlation value is obtained by converting the actual received signal x n The formula is obtained by multiplying and summing the standard preamble sequence s0 at different time delays m point by point.

[0037]

[0038] Where, x n This represents the actual received signal sequence. This represents the value of the standard preamble sequence at a delay of m, where m is the delay amount, representing the time offset of the standard preamble sequence relative to the received signal, and N is the length of the signal.

[0039] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, further, in step 4, peak selection specifically includes:

[0040] In the relevant results R xy Local maxima are identified in (m) to form a candidate main peak set;

[0041] Peak selection is performed using a method combining dynamic noise basis estimation and main lobe ratio.

[0042] σ noise = std(R[1:M])

[0043] Where σ noise Let M = 0.1·length(R) represent the dynamic noise basis estimation, where std represents the standard deviation, and (R[1:M]) represents the first M samples of cross-correlation values ​​R; then, a threshold T is set. h =max(3.5σ) noise ,α·max(R)), where α=0.7, retain those that satisfy R xy (m i )>Τ h The peak value.

[0044] According to the ADS-B signal detection method based on multi-dimensional feature joint determination of the present invention, the peak selection further includes: when a candidate main peak simultaneously meets the following conditions, it is determined to be a main peak, and the delay amount m corresponding to the main peak is... main The precise position of the preamble header in the received signal:

[0045] (a) The energy within the leading window centered on the main peak satisfies:

[0046]

[0047] Where, m main T is the index of the time delay sampling points for the candidate main peak. p η is the duration of the preamble pulse, r[k] is the cross-correlation value of the kth sampling point, η is the energy ratio threshold, and max(Energy) is the maximum energy window value of the entire signal.

[0048] (b) A valid PPM waveform must be detected within 112 μs after the main peak;

[0049] (c) The amplitude of the secondary peak does not exceed 50% of that of the primary peak, and the time interval is greater than 3.5 μs.

[0050] Furthermore, the present invention also provides an ADS-B signal detection system based on multi-dimensional feature joint determination, used to implement the above-mentioned ADS-B signal detection method based on multi-dimensional feature joint determination, comprising:

[0051] The signal modeling and quadrature demodulation module is used to construct an ADS-B baseband signal model according to the ICAO DO-260B protocol. The receiver recovers the in-phase and quadrature components of the baseband signal through quadrature downconversion.

[0052] The sliding window energy accumulation and threshold decision module is used to accumulate energy of the demodulated signal based on a sliding window and combine it with dynamic threshold decision logic to detect the valid signal window.

[0053] An improved M-Rife frequency offset correction module is used to perform FFT analysis on signals within the detected effective signal window, determine whether dynamic spectrum shifting is required based on spectral characteristics, and achieve frequency offset compensation through secondary fine calibration.

[0054] The bipolar correlation and peak filtering module is used to perform bipolar transformation on the preamble of the frequency offset corrected signal, calculate the cross-correlation value with the standard sequence, and determine the preamble pulse position through multi-level filtering conditions.

[0055] The beneficial effects achieved by adopting the above technical solution are:

[0056] 1. Improve signal acquisition capability in low signal-to-noise ratio environments

[0057] This invention constructs a joint feature space of time-frequency and statistical domains through orthogonal demodulation and adaptive sliding window energy accumulation. Combined with dynamic threshold decision logic based on the Neyman-Pearson criterion, it effectively suppresses noise interference, solves the problem of high false alarm rate in traditional single-dimensional energy detection under low signal-to-noise ratio environment, and significantly enhances the ability to capture weak signals.

[0058] 2. Expand the frequency offset compensation range and improve correction accuracy

[0059] The improved M-Rife algorithm uses a dynamic spectrum shifting strategy to force the main lobe of the frequency offset to the quantization center region. Combined with secondary Rife calibration, the estimation error is compressed to the Cramer-Rhodes lower limit. This breaks through the limitation of the classic Rife algorithm in terms of accuracy degradation when the frequency offset exceeds the quantization frequency center region. It can effectively compensate for the Doppler frequency offset generated by the high-speed movement of aircraft and ensure the reliability of signal demodulation.

[0060] 3. Enhance the reliability of leading header recognition.

[0061] The innovative bipolar correlation technology maps binary {0,1} to {-1,1}, enhancing the contrast of preamble pulse features. Combined with a multi-level peak screening mechanism of dynamic noise basis estimation, peak-to-peak ratio constraint and time-domain waveform matching, it effectively suppresses false peaks caused by noise and interference, solving the problem of false detection rate exceeding 40% in traditional unipolar correlation schemes when SNR < 10dB, and improving the positioning accuracy of the preamble header.

[0062] 4. Achieve efficient real-time processing

[0063] This invention employs a three-level collaborative processing architecture design involving energy detection, frequency offset correction, and bipolar correlation. This architecture is adaptable to parallel hardware architectures, ensuring detection performance while optimizing computational efficiency, thus meeting the real-time signal processing requirements of scenarios such as airborne electronic reconnaissance and airspace surveillance.

[0064] 5. Enhance robustness in complex electromagnetic environments

[0065] This invention comprehensively improves the detection accuracy and stability of ADS-B signals in complex electromagnetic environments such as noise, frequency offset, and pulse interference through cascaded optimization of time-frequency feature fusion, frequency offset robustness compensation, and polarity feature enhancement, overcoming the shortcomings of existing technologies in terms of detection performance degradation under the combined effect of multiple factors. Attached Figure Description

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. The drawings are merely illustrative of some embodiments of the present invention and are not intended to limit the scope of the present invention to all embodiments.

[0067] Figure 1 This is a flowchart illustrating the ADS-B signal detection method based on multi-dimensional feature joint determination according to an embodiment of the present invention.

[0068] Figure 2 This is a schematic diagram of the ADS-B signal pulse format according to an embodiment of the present invention;

[0069] Figure 3 This is a schematic diagram of the bipolar correlation of the ADS-B header pulse in an embodiment of the present invention;

[0070] Figure 4 This is a complete ADS-B signal modeling diagram (no noise and noisy) of an embodiment of the present invention;

[0071] Figure 5 This is a sliding window energy accumulation diagram according to an embodiment of the present invention;

[0072] Figure 6 This is a comparison chart of the detection performance under different frequency offsets in an embodiment of the present invention;

[0073] Figure 7 This is the time-frequency diagram of the short-time Fourier transform of the ADS-B corrected signal according to an embodiment of the present invention.

[0074] Figure 8 This is a comparison diagram of the correlation peaks before and after ADS-B correction in an embodiment of the present invention;

[0075] Figure 9 This is a diagram showing the unipolar correlation results of the leading header in an embodiment of the present invention;

[0076] Figure 10 This is a diagram showing the bipolar correlation results of the leading header in an embodiment of the present invention;

[0077] Figure 11 This is a graph showing the unipolar peak screening results of an embodiment of the present invention;

[0078] Figure 12 This is a graph showing the bipolar peak screening results of an embodiment of the present invention. Detailed Implementation

[0079] The exemplary solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art.

[0080] This invention discloses an ADS-B signal detection method based on multi-dimensional feature joint determination. The core technologies include: constructing a time-frequency-statistical domain joint detection framework; designing an adaptive sliding window energy accumulation algorithm; utilizing an improved M-Rife frequency offset correction algorithm and combining it with dynamic spectrum shifting to achieve subsampling rate accuracy correction; and innovating a bipolar correlation and dynamic threshold screening mechanism to improve the reliability of leader header detection. Figure 1 As shown, the specific implementation steps of this method are as follows:

[0081] Step S101, Signal Modeling and Quadrature Demodulation: Construct an ADS-B baseband signal model according to the ICAO DO-260B protocol. The receiver recovers the in-phase component I(t) and quadrature component Q(t) of the baseband signal through quadrature downconversion.

[0082] ADS-B signal pulse format as follows: Figure 2 As shown, the preamble p(t) of the ADS-B signal in S mode is fixed, consisting of four pulses with a width of 0.5μs and an interval of 0.5μs. The first pulse is in phase with the second pulse, and the third and fourth pulses are out of phase with the first two pulses. According to the PPM encoding rule, assuming a unit is 0.5μs, the preamble pulse is: p(t) = [1,0,1,0,0,0,0,1,0,1,0,0,0,0,0,0], a total of 16 bits. For the data block, since the data block length is 112μs, the data block can be represented as: d(t) = [a[1],a[2],a[3],......,a

[224] ]. Therefore, the expression of the ADS-B baseband signal model is:

[0083] s(t)=A*cos(2πf c t+φ(t))*[p(t)+d(t)]+w(n)

[0084] Where A represents the signal amplitude, f c Let φ(t) represent the carrier frequency (1090MHz in S-mode), p(t) be the phase modulation term, d(t) be the preamble signal, w(n) be the data block signal, and w(n) be Gaussian white noise. The receiver employs an orthogonal down-conversion structure, and the baseband signal recovery formula is:

[0085] I(t)=s(t)·cos(2πf c t)

[0086] Q(t)=s(t)·sin(2πf c t)

[0087] Step S102, Sliding window energy accumulation and threshold decision: After signal demodulation, energy accumulation of the demodulated signal is performed based on the sliding window, and the effective signal window is detected by combining dynamic threshold decision logic.

[0088] Based on the characteristics of ADS-B signals, energy detection is performed in noisy environments. ADS-B signals have a well-defined structure, strong periodicity, and a fixed carrier frequency of 1090MHz. Furthermore, the energy of ADS-B signals is mainly concentrated near the carrier frequency, with higher energy in the preamble and message portions and lower noise energy. Therefore, based on these characteristics, ADS-B signals exhibit distinct features in both the time and frequency domains, and can be effectively detected using methods such as energy detection, sliding window processing, and thresholding.

[0089] After demodulation, the received ADS-B signal can be used for energy accumulation detection using a sliding window algorithm. The window length for energy accumulation is set to an integer multiple of the standard signal duration, and the energy statistics are calculated using the following formula:

[0090]

[0091] Wherein, the window length W = 120μs × n, 120μs is the same as the duration of a standard ADS-B signal, n is a positive integer, ΔW is the step size, and I[k] and Q[k] are the discrete spectra of I(t) and Q(t) after Fourier transform. Finally, based on the Neyman-Pearson criterion, the dynamic threshold is set as follows:

[0092] λ = γ·max(E[n])

[0093] Where γ = 0.95, the detection and decision logic is as follows:

[0094]

[0095] If the energy in a certain window exceeds the threshold, a signal is considered to have been detected.

[0096] Step S103, Improved M-Rife frequency offset correction: Perform FFT analysis on the signal within the effective signal window detected in step S102, determine whether dynamic spectrum shift is required based on the spectral characteristics, and achieve frequency offset compensation through secondary fine calibration.

[0097] Frequency offset correction is performed on ADS-B signals. To address the frequency offset problem in ADS-B signals, this embodiment of the invention can employ the Rife algorithm, which boasts high accuracy with a root mean square error approaching CRLB. However, the accuracy of frequency estimation decreases when estimating the frequency quantization point range. Therefore, this embodiment of the invention uses an improved algorithm based on Rife: M-Rife. This algorithm first calculates the frequency offset using Rife and then continues calculation based on whether it falls within the high-precision region. If so, the Rife algorithm is used to calculate the frequency offset again; otherwise, the signal is spectrally shifted, and the calculation continues.

[0098] The time-domain ADS-B signal with Gaussian white noise contamination is represented as follows:

[0099]

[0100] Where A,f d , Let i represent amplitude, frequency offset, and initial phase, respectively. Let i be a positive integer. Let L represent the length of an ADS-B signal. Let w(n) represent Gaussian white noise. Let s[i] represent the original symbol sequence of the transmitter. Let x[n] represent the pulse shaping function. Let n represent the discrete-time index. Let N represent the total length of the signal.

[0101] Perform an N-point FFT on each valid segment s[n] to obtain the discrete spectrum S(k). Each valid signal segment contains a preamble header and a data block, and frequency offset correction and preamble detection need to be performed independently.

[0102]

[0103] Where n start n end This represents the start and end times of the effective signal segment, where N is the number of FFT points. Locating the maximum spectral line index: k max =arg max|S(k)|, to obtain a coarse estimate of the frequency offset: Where N is the number of FFT points, f s f is the sampling frequency. c Indicates the carrier frequency.

[0104] Define quantization frequency unit When the coarse estimate of frequency offset satisfies At that time, perform dynamic spectrum shifting: Where Δf shift Frequency shift:

[0105]

[0106] in, This represents the floor function. This operation ensures that the main lobe of the signal is located within [-Δf] after the shift. q / 2,+Δf q The interval is 1 / 2.

[0107] Perform a second FFT on the moved signal s'(t) to obtain the new spectrum S'(k), and relocate the new maximum spectral line index k'. max Obtain the normalized difference:

[0108]

[0109] Calculate the fine-tuning value using the parabolic interpolation formula:

[0110]

[0111] Finally, the frequency offset estimate Δf is obtained. final =Δf shift +Δf fine Using Δf final Output frequency offset compensation signal:

[0112] Considering the engineering implementation details, we can also calculate the frequency offset estimate from another perspective.

[0113] Let k0 be the maximum spectral line, and take the maximum spectral line value |S(k0)| to calculate the frequency at that point using the following formula:

[0114]

[0115] In the formula, T = NΔt, and r represents the direction parameter, which takes the value of +1 or -1. When |S(k0+1)| < |S(k0-1)|, r = -1; when |S(k0+1)| ≥ |S(k0-1)|, r = +1.

[0116] Whether a signal needs correction can be determined using the following formula. If it satisfies:

[0117]

[0118] In the formula, Δf is the DFT quantization frequency interval, and it is assumed that... If the signal is located in the center region of the quantization frequency, correction is required; otherwise, correction is necessary. In this case, to make the frequency of the estimated signal as close as possible to the midpoint of the quantization frequency, the signal S(n) is shifted left or right by R. k Quantize frequency units and calculate frequency shift R k The formula is:

[0119]

[0120] The time-domain signal after the first Rife algorithm calculation is:

[0121]

[0122] The frequency domain signal is:

[0123]

[0124] when When r = 1, the spectral line shifts to the right; conversely, when r = -1, the spectral line shifts to the left.

[0125] Following the principle of the M-Rife algorithm, taking r=1 as an example, the Rife algorithm is used again to estimate the frequency offset. The calculated frequency offset is:

[0126]

[0127] Finally, the frequency offset corrected ADS-B signal s0(n) is obtained:

[0128]

[0129] Step S104, Bipolar Correlation and Peak Filtering: Perform bipolar transformation on the preamble of the frequency offset corrected signal, calculate the cross-correlation value with the standard sequence, and determine the preamble pulse position through multi-level filtering conditions.

[0130] The generated standard ADS-B signal is compared with the actual ADS-B signal for correlation detection. A higher cross-correlation result indicates a better correlation. The formula for calculating the cross-correlation of the signals is:

[0131]

[0132] In the formula, f(t) represents the actual ADS-B signal, f0(t) represents the standard ADS-B signal, and the autocorrelation function of the bipolar preamble satisfies:

[0133]

[0134] In the formula T s =0.5μs is the pulse interval, which makes the main lobe peak significantly higher than the noise floor.

[0135] A bipolar transformation is performed on the standard preamble of the ADS-B signal. This transformation maps binary {0, 1} to {-1, 1}, satisfying the power normalization constraint. The resulting bipolar-transformed standard preamble sequence is: s 0_w = [1,-1,1,-1,-1,-1,-1,1,-1,1,-1,-1,-1,-1,-1,-1], such as Figure 3 As shown; since the preamble of the ADS-B signal is constant, the formula for correlating the simulated standard preamble signal with the actual preamble signal is as follows:

[0136]

[0137] Where, x n This represents the actual received signal sequence. This represents the value of the standard preamble sequence at a delay of m, where m is the delay amount, representing the time offset of the standard preamble sequence relative to the received signal; N is the signal length. The maximum value is obtained when the actual preamble pulse perfectly matches the simulated standard preamble pulse. However, when the actual preamble pulse contains parts that differ from the standard preamble pulse, the value is adjusted according to the calculated R. xy The value will decrease, which is reflected in the amplitude spectrum as a decrease in peak value. Then, a method combining dynamic noise basis estimation and main lobe ratio can be used to perform threshold screening, and finally obtain the accurate position of the ADS-B signal preamble.

[0138] The correlation of the preamble header after bipolar transformation is calculated using the cross-correlation formula. When the actual preamble pulse is completely consistent with the simulated standard preamble pulse, the obtained R... xy Represented as:

[0139] R xy1max(m)=4M·A

[0140] In the formula, M represents the maximum cross-correlation value under ideal conditions, and A represents the calculated cross-correlation result between the actual received signal and the standard preamble pulse. However, when the actual preamble pulse has a portion that is inconsistent with the standard preamble pulse, R calculated according to the cross-correlation formula... xy The value of R will decrease, at which point R will... xy The calculation formula is:

[0141] R xy1 (m)=4M·A-nM·A

[0142] The value of n in the formula represents the number of error pulses.

[0143] The following section performs peak filtering on the cross-correlation results. After bipolar correlation, the location of the peak is clearly found, determining the starting point of the ADS-B signal. In the correlation result R... xy To identify local maxima in (m), the mathematical condition is: R xy (m)>R xy (m-1) and R xy (m)>R xy (m+1), all points {m} that satisfy this condition i} constitutes the candidate main peak set {(m i ,R xy (m i ))}.

[0144] Then, a method combining dynamic noise basis estimation and main lobe ratio is used for peak selection:

[0145] σ noise = std(R[1:M])

[0146] Where σ noise The dynamic noise basis estimation is represented by M = 0.1·length(R), std represents the standard deviation, and (R[1:M]) represents the first M samples of the cross-correlation value R. The basis level of noise can be obtained by calculating the standard deviation of the first M samples of the correlation value sequence R.

[0147] The threshold is set as follows:

[0148] T h =max(3.5σ) noise ,α·max(R))

[0149] Where α = 0.7, and at the same time retain the condition that R satisfies xy (m i )>Τ h The peak value.

[0150] The verification conditions for the main peak also include:

[0151] (a) Energy verification: The energy within the leading window centered on the main peak should meet the following requirements.

[0152]

[0153] Where, m main T is the index of the time delay sampling points for the candidate main peak. p It is the duration of the preamble pulse, taken as T. p =8μs, r[k] is the cross-correlation value of the kth sampling point (from R). xy (m)), where η is the energy proportion threshold, taken as η = 0.8, and max(Energy) is the maximum energy window value of the entire signal. This constraint ensures that the preamble energy accounts for more than 80% of the total signal energy.

[0154] (b) Uniqueness verification: Select the global maximum peak value m main =argmax(R), which is the main peak. For the other secondary peaks, the following must be satisfied:

[0155] |m i -m main |>Δt min

[0156] Where Δt min =3.5μs, while the amplitude of the secondary peak does not exceed 50% of that of the main peak.

[0157] (c) Waveform verification: T after the main peak d =A valid pulse position modulation (PPM) waveform must be detected within 112μs, and its normalized correlation coefficient must satisfy:

[0158]

[0159] Where d[k] is the standard data segment template generated by simulation, r represents the normalized cross-correlation coefficient between the signal waveform and the standard template, and d represents the minimum distance threshold between the secondary peak and the main peak.

[0160] This method achieves efficient signal detection through a three-level collaborative processing architecture: First, coarse signal acquisition is completed based on multi-dimensional feature joint judgment. A joint time-frequency domain feature space is constructed through orthogonal demodulation and adaptive sliding window energy accumulation, and low signal-to-noise ratio interference is eliminated using dynamic threshold decision. Second, a modified M-Rife algorithm is used for frequency offset correction. A dynamic spectrum shifting strategy forces the frequency offset main lobe to the quantization center region, and combined with secondary Rife fine correction, the estimation error is compressed to the Cramer-Rhodes lower limit. Finally, bipolar correlation is used to enhance the contrast of the leader pulse features, and a multi-level screening mechanism (dynamic noise floor estimation, peak-to-peak ratio constraint, and time-domain waveform matching) is designed to achieve accurate header localization. This solves the problems of small signal amplitude, low signal-to-noise ratio, and low signal detection probability encountered in traditional ADS-B signal detection.

[0161] Corresponding to the above method, embodiments of the present invention also disclose an ADS-B signal detection system based on multi-dimensional feature joint determination, comprising:

[0162] The signal modeling and quadrature demodulation module is used to construct an ADS-B baseband signal model according to the ICAO DO-260B protocol. The receiver recovers the in-phase and quadrature components of the baseband signal through quadrature downconversion.

[0163] The sliding window energy accumulation and threshold decision module is used to accumulate energy of the demodulated signal based on a sliding window and combine it with dynamic threshold decision logic to detect the valid signal window.

[0164] An improved M-Rife frequency offset correction module is used to perform FFT analysis on signals within the detected effective signal window, determine whether dynamic spectrum shifting is required based on spectral characteristics, and achieve frequency offset compensation through secondary fine calibration.

[0165] The bipolar correlation and peak filtering module is used to perform bipolar transformation on the preamble of the frequency offset corrected signal, calculate the cross-correlation value with the standard sequence, and determine the preamble pulse position through multi-level filtering conditions.

[0166] To verify the effectiveness of this scheme, the following explanation is further illustrated with simulation data:

[0167] To simulate the random distribution of multiple ADS-B signals in a real-world environment, Figure 4 A complete signal containing multiple ADS-B signals was constructed. The preamble and message signals were combined, and zero padding was generated before and after the signal. Considering that ADS-B signals do not appear densely in reality, the interval was set to 1 / 3 of the length of the ADS-B signal. Then, Gaussian white noise with a signal-to-noise ratio of 10dB was added to the signal and modulated onto a carrier frequency of 1090MHz.

[0168] To verify the advantages of this solution in ADS-B signal detection capabilities in noisy environments. Figure 5 The results show that after applying sliding window energy accumulation to the analog signal, the results demonstrate that in a low signal-to-noise ratio environment, the energy accumulation algorithm can accurately detect the ADS-B signal and maintain the same number of generated ADS-B signals without causing signal loss.

[0169] To verify the effectiveness of this algorithm in frequency offset correction. Figure 6 The comparison of ADS-B signal detection accuracy under different frequency offset conditions is shown. The results indicate that the larger the frequency offset, the lower the detection accuracy under low signal-to-noise ratio conditions. Figure 7 The time-frequency plot of the signal after frequency offset correction is shown. Short-time Fourier transform analysis of the frequency-offset corrected ADS-B signal reveals that the main lobe energy of the corrected signal is significantly concentrated at the baseband center frequency, and the peak amplitude is significantly increased compared to the uncorrected state. The frequency offset correction process reconstructs the phase consistency of the signal, effectively improving the timing alignment characteristics of Pulse Position Modulation (PPM) symbols, while suppressing out-of-band noise interference caused by phase transitions. Figure 8 The comparison of correlation peaks before and after frequency offset correction is shown. After correction, the main lobe of the correlation peak is narrower, the side lobes are suppressed, and the peak value is higher. This indicates that the frequency offset is effectively eliminated, the matching degree between the signal and the template is improved, and the anti-interference ability of the signal is enhanced, especially in low signal-to-noise ratio environments.

[0170] To verify the advantages of bipolar correlation and peak screening in detecting ADS-B signal leader headers, Figure 9 and Figure 10 The amplitude spectra of the ADS-B signal preamble are shown after unipolar and bipolar correlation. As can be seen from the figure, the amplitude of the preamble after bipolar correlation is not only significantly higher than that after unipolar correlation, but also higher than that of the data block portion. This can provide a basis for subsequent peak selection. Figure 11 and Figure 12 The results of peak filtering after unipolar and bipolar correlation are shown. The cross-correlation value using bipolar transformation has a significant peak near the leader header, which can be used to identify the location of the ADS-B leader header. However, the unipolar correlation method has more peaks and it is difficult to identify the location of the ADS-B leader header.

[0171] The above experimental results further demonstrate the superiority of this scheme in joint detection of ADS-B signals, which can maintain a high detection accuracy while simplifying the algorithm process.

[0172] It should be noted that the terms "sliding window energy accumulation," "dynamic spectrum shifting," and "bipolar correlation" used in this article are only used to describe the functional characteristics of different technical modules and do not limit their implementation order or structural hierarchy. For example, "sliding window energy accumulation" refers to the energy focusing method based on a time window, which is an independent module in the process of the frequency correction method "dynamic spectrum shifting." Its naming is only for the sake of clarity of technical description and is not a mandatory constraint on the execution order of the algorithms. Those skilled in the art should understand that such terms can be combined in different ways according to the needs of the actual application scenario (such as placing frequency offset correction before or after energy detection) to achieve better system performance adaptation.

[0173] In this specification, descriptions of "exemplary embodiments" and "implementation methods" are intended to illustrate the core innovations of the present invention through specific technical scenario examples, rather than limiting its application scope. For example, the "dynamic threshold decision" described in the embodiments can be adjusted to a segmented threshold model based on noise statistical characteristics; and the polarity mapping rules in "bipolar correlation" (such as mapping the standard preamble header to a [-1,+1] or [+1,-1] sequence) can also be redefined based on hardware characteristics. The technical features mentioned in different embodiments (such as the sliding window step size optimization strategy) can be flexibly combined through parameterized configuration to adapt to diverse application scenarios such as spaceborne platforms and ground base stations.

[0174] Furthermore, the formula references in the claims of this invention strictly correspond to those in the specification and drawings to ensure the rigor of the legal documents. The experimental data in the specification are merely examples to verify the technical effects; in practical applications, they can be adjusted according to the signal type and carrier frequency parameters. All derivative solutions based on the technical concept of this invention, including but not limited to extending the bipolar correlation mechanism to preamble detection of pulse modulation signals such as Mode-S and UAT, and variant methods that dynamically optimize the sliding window length (W) based on the real-time signal quantity and channel conditions, fall within the potential protection scope of this invention.

Claims

1. An ADS-B signal detection method based on multi-dimensional feature joint determination, characterized in that, Includes the following steps: Step 1, Signal Modeling and Quadrature Demodulation: Construct an ADS-B baseband signal model according to the ICAO DO-260B protocol. The receiver recovers the in-phase component I(t) and quadrature component Q(t) of the baseband signal through quadrature downconversion. Step 2, Sliding window energy accumulation and threshold decision: The demodulated signal is accumulated based on the sliding window, and the valid signal window is detected by combining the dynamic threshold decision logic; Step 3, Improve M-Rife frequency offset correction: Perform FFT analysis on the signal within the effective signal window detected in Step 2, determine whether dynamic spectrum shifting is required based on the spectral characteristics, and achieve frequency offset compensation through secondary fine calibration; Step 4, Bipolar Correlation and Peak Filtering: Perform bipolar transformation on the preamble of the frequency offset corrected signal, calculate the cross-correlation value with the standard sequence, and determine the preamble pulse position through multi-level filtering conditions.

2. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 1, characterized in that, In step 1, the expression for the ADS-B baseband signal model is: s(t)=A*cos(2πf c t+φ(t))*[p(t)+d(t)]+w(n) Where A represents the signal amplitude, f c Let φ(t) represent the carrier frequency, φ(t) be the phase modulation term, p(t) be the preamble signal, d(t) be the data block signal, and w(n) be Gaussian white noise.

3. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 1, characterized in that, In step 2, the window length for sliding window energy accumulation is set to an integer multiple of the standard signal duration, and the energy statistics are calculated using the following formula: Wherein, the window length W = 120μs × n, n is a positive integer, ΔW is the step size, and I[k] and Q[k] are the discrete spectra of I(t) and Q(t) after Fourier transform; based on the Neyman-Pearson criterion, the dynamic threshold is set as follows: λ = γ·max(E[n]) Where γ = 0.95, the detection and decision logic is as follows: If the energy in a certain window exceeds the threshold, a signal is considered to have been detected.

4. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 1, characterized in that, In step 3, the specific steps for improving the M-Rife frequency offset correction include: Perform an N-point FFT on each valid signal segment to obtain the discrete spectrum S(k), and locate the index k of the maximum spectral line. max Used to calculate a coarse estimate of the frequency offset: Where N is the number of FFT points, f s f is the sampling frequency. c Indicates the carrier frequency; when Dynamic spectrum shifting is performed in real time, and quantization frequency units are defined. Generate a transfer signal: Where, Δf shift It is the frequency shift amount; Perform a second fine-tuning on the transfer signal s'(t) to obtain Δf fine ; Finally, the frequency offset estimate Δf is obtained. final =Δf shift +Δf fine Using Δf final Output frequency offset compensation signal 5. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 4, characterized in that, The formula for calculating the frequency shift during dynamic spectrum shifting is: in, This represents the floor function.

6. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 4, characterized in that, Perform a second fine-tuning on the transfer signal s'(t) to obtain Δf fine include: Perform a second FFT on the shifted signal s'(t) to obtain a new spectrum S'(k), and relocate the new maximum spectral line index k'. max Obtain the normalized difference: Calculate the fine-tuning value using the parabolic interpolation formula:

7. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 1, characterized in that, In step 4, the bipolar transformation maps the binary sequence {0, 1} of the preamble header to {-1, 1}, and the standard preamble header sequence is: s 0_w ＝[1,-1,1,-1,-1,-1,-1,1,-1,1,-1,-1,-1,-1,-1,-1]。 The cross-correlation value is obtained by converting the actual received signal x n The formula is obtained by multiplying and summing the standard preamble sequence s0 at different time delays m point by point. Where, x n This represents the actual received signal sequence. This represents the value of the standard preamble sequence at a delay of m, where m is the delay amount, representing the time offset of the standard preamble sequence relative to the received signal, and N is the length of the signal.

8. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 7, characterized in that, In step 4, peak filtering specifically includes: In the relevant results R xy Local maxima are identified in (m) to form a candidate main peak set; Peak selection is performed using a method combining dynamic noise basis estimation and main lobe ratio. s noise =std(R[1:M]) Where σ noise Let M = 0.1·length(R) represent the dynamic noise basis estimation, where std represents the standard deviation, and (R[1:M]) represents the first M samples of cross-correlation values ​​R; then, a threshold T is set. h =max(3.5σ) noise ,α·max(R)), where α=0.7, retain those that satisfy R xy (m i )>Τ h The peak value.

9. The ADS-B signal detection method based on multi-dimensional feature joint determination according to claim 8, characterized in that, Peak selection also includes: when a candidate peak simultaneously meets the following conditions, it is determined to be a main peak, and the corresponding delay m is... main The precise position of the preamble header in the received signal: (a) The energy within the leading window centered on the main peak satisfies: Where, m main T is the index of the time delay sampling points for the candidate main peak. p η is the duration of the preamble pulse, r[k] is the cross-correlation value of the kth sampling point, η is the energy ratio threshold, and max(Energy) is the maximum energy window value of the entire signal. (b) A valid PPM waveform must be detected within 112 μs after the main peak; (c) The amplitude of the secondary peak does not exceed 50% of that of the primary peak, and the time interval is greater than 3.5 μs.

10. An ADS-B signal detection system based on multi-dimensional feature joint determination, characterized in that, A method for implementing the ADS-B signal detection method based on multi-dimensional feature joint determination as described in any one of claims 1-9 includes: The signal modeling and quadrature demodulation module is used to construct an ADS-B baseband signal model according to the ICAO DO-260B protocol. The receiver recovers the in-phase and quadrature components of the baseband signal through quadrature downconversion. The sliding window energy accumulation and threshold decision module is used to accumulate energy of the demodulated signal based on a sliding window and combine it with dynamic threshold decision logic to detect the valid signal window. An improved M-Rife frequency offset correction module is used to perform FFT analysis on signals within the detected effective signal window, determine whether dynamic spectrum shifting is required based on spectral characteristics, and achieve frequency offset compensation through secondary fine calibration. The bipolar correlation and peak filtering module is used to perform bipolar transformation on the preamble of the frequency offset corrected signal, calculate the cross-correlation value with the standard sequence, and determine the preamble pulse position through multi-level filtering conditions.

Images