FPGA-based low-complexity signal adaptive detection method and system

Abstract

The embodiment of the invention provides a low-complexity signal adaptive detection method and system based on an FPGA, and the method comprises the steps: obtaining a signal envelope which comprises a plurality of data samples; performing amplitude multi-point smoothing processing on the signal envelope to obtain a to-be-judged signal comprising a plurality of judgment samples; obtaining an adaptive threshold value based on the data sample; sequentially comparing the judgment samples with the self-adaptive threshold value; when M continuous judgment samples in the to-be-judged signal are greater than the adaptive threshold values corresponding to the judgment samples, outputting a high-level detection mark; and when N continuous judgment samples in the to-be-judged signal are smaller than the adaptive threshold values corresponding to the judgment samples, outputting a low-level detection mark. Amplitude multi-point smoothing processing is firstly carried out on a signal envelope, then rising edge pulse judgment and falling edge pulse judgment are respectively carried out by adopting multi-data-point accumulation (M-N method), and a relatively high correct judgment probability can also be achieved when the signal-to-noise ratio is relatively low.

Description

Technical Field

[0001] Multiple embodiments of this specification relate to the technical field of signal detection, and specifically to a low-complexity signal adaptive detection method and system based on FPGA. Background Art

[0002] Pulse signals are widely used in the field of communications. By receiving pulse signals and processing the received pulse signals, the processing results are finally combined with relevant technical means for analysis to obtain the transmitted information. In the process of processing pulse signals, the first step is to detect the pulse signal. Accurately and quickly detecting the rising and falling edges of the pulse signal is the premise for measuring signal parameters such as frequency, amplitude, phase, arrival time, and pulse width. In the receiver system, due to the thermal noise in the system itself, the received signal is generally a mixture of pulse signals and noise. The energy of the noise occupying the same frequency band as the pulse signal directly affects the system's ability to detect pulse signals. Therefore, the rationality of the detection threshold setting used to determine whether there is a pulse signal in the signal is closely related to the correctness of signal detection.

[0003] Since noise is always fluctuating, in order to improve the accuracy of signal detection, the detection threshold needs to be set adaptively according to the energy of the noise. Figure 1 In the existing signal adaptive detection technology, the pulse signal is received by the antenna, and the intermediate frequency analog signal is obtained after processing, which is converted to the digital domain by an analog-to-digital converter, and the real signal is converted into a complex signal through Hilbert transform for subsequent digital signal processing. Figure 1 The I and Q in the figure represent the real and imaginary data of the complex signal. The complex signal first enters the envelope detection module to obtain the amplitude characteristics of the signal (i.e., the signal envelope), and then enters the adaptive threshold setting module to calculate the adaptive threshold value. Finally, the pulse decision module uses a point-by-point decision method to compare the amplitude of the pulse signal with the calculated adaptive threshold value, and determines the rising and falling edges of the signal to determine whether there is a pulse signal in the received signal, thereby completing signal adaptive detection.

[0004] However, in the case of low signal-to-noise ratio, for signals with large signal envelope jitter, using a point-by-point judgment method to determine the rising and falling edges of the signal can easily split a large pulse signal into multiple small pulse signals, which will directly affect the signal detection results and make subsequent signal processing and analysis more difficult. Summary of the invention

[0005] The embodiments of this specification provide a low-complexity signal adaptive detection method and system based on FPGA, which first performs multi-point amplitude smoothing processing on the signal envelope, and then uses multiple data point accumulation (MN method) to perform rising edge and falling edge pulse judgment respectively. The idea of ​​the invention is to sacrifice the detection accuracy of the pulse signal edge, reduce the possibility of the pulse signal being segmented under low signal-to-noise ratio, improve the robustness of pulse signal detection, and achieve a higher probability of correct judgment even when the signal-to-noise ratio is low.

[0006] The technical solution is as follows:

[0007] In a first aspect, an embodiment of the present specification provides a low-complexity signal adaptive detection method based on FPGA, comprising:

[0008] acquiring a signal envelope comprising a plurality of data samples;

[0009] Performing multi-point smoothing processing on the signal envelope to obtain a signal to be decided including multiple decision samples;

[0010] Obtaining an adaptive threshold value based on the data sample;

[0011] Comparing the sizes of multiple decision samples and adaptive threshold values ​​in sequence;

[0012] When M consecutive decision samples in the signal to be decided are greater than the adaptive threshold value corresponding to the decision sample, a high-level detection flag is output;

[0013] When N consecutive decision samples in the signal to be decided are less than the adaptive threshold value corresponding to the decision sample, a low-level detection flag is output.

[0014] As a preferred solution, the step of acquiring a signal envelope including a plurality of data samples includes:

[0015] Acquire a correction factor A and a plurality of complex signals including real data I and imaginary data Q;

[0016] Respectively compare the sizes of the real data I and the imaginary data Q corresponding to the multiple complex signals to obtain the maxI,Q and minI,Q corresponding to the multiple complex signals; obtain the data samples corresponding to the multiple complex signals based on the maxI,Q, minI,Q and the correction factor A corresponding to the multiple complex signals to obtain the signal envelope including the multiple data samples;

[0017] The correction factor A is pre-calculated by the stepwise approximation method. To approach the target simulation, X and Y are random numbers.

[0018] As a preferred solution, the signal envelope is subjected to multi-point amplitude smoothing processing to obtain a signal to be decided including multiple decision samples, including:

[0019] Take any data sample as the target data sample:

[0020] Acquire a first preset number of data samples including the target data sample before and after the target data sample;

[0021] An average value of a first preset number of data samples is used as a decision sample corresponding to the target data sample;

[0022] Repeat the above steps to obtain the decision samples corresponding to each data sample, so as to obtain a signal to be decided including a plurality of decision samples.

[0023] As a preferred solution, the method of using the average value of the first preset number of data samples as the decision sample corresponding to the target data sample includes:

[0024] Obtaining an average value of a first preset number of data samples;

[0025] A truncation operation is performed on the average value of the first preset number of data samples, and the result of the truncation operation is used as the decision sample corresponding to the target data sample.

[0026] As a preferred solution, the step of obtaining an adaptive threshold value based on data samples includes:

[0027] Obtain the absolute value of noise corresponding to each of the multiple data samples;

[0028] Sequentially dividing a plurality of continuous data samples into a plurality of sample groups based on a second preset number;

[0029] For each sample group, the noise average value of the sample group is obtained based on the absolute value of the noise of each data sample in the sample group;

[0030] For each sample group, an adaptive threshold value of the sample group is obtained based on the noise average value of the sample group;

[0031] The step of sequentially comparing the multiple decision samples with the adaptive threshold value includes:

[0032] The decision samples corresponding to the multiple data samples in each sample group are compared with the adaptive threshold value of the sample group where the data samples are located.

[0033] As a preferred solution, the step of obtaining the absolute noise values ​​corresponding to each of the plurality of data samples includes:

[0034] Based on multiple data samples, a first-order backward difference method is used to obtain noise values ​​corresponding to each of the multiple data samples;

[0035] The noise absolute values ​​corresponding to the multiple data samples are obtained based on the noise values ​​corresponding to the multiple data samples.

[0036] In a second aspect, the embodiments of this specification provide a low-complexity signal adaptive detection system based on FPGA, including an envelope extraction and processing module, an adaptive dynamic threshold module and a pulse decision module;

[0037] The envelope extraction and processing module obtains a signal envelope including multiple data samples; performs amplitude multi-point smoothing processing on the signal envelope to obtain a signal to be decided including multiple decision samples;

[0038] The adaptive dynamic threshold module obtains an adaptive threshold value based on the data sample;

[0039] The pulse decision module compares the sizes of multiple decision samples and adaptive threshold values ​​in sequence; when M consecutive decision samples in the signal to be decided are greater than the adaptive threshold value corresponding to the decision samples, a high-level detection flag is output; when N consecutive decision samples in the signal to be decided are less than the adaptive threshold value corresponding to the decision samples, a low-level detection flag is output.

[0040] As a preferred solution, the adaptive dynamic threshold module includes a noise calculation unit, a multi-point mean value calculation unit and an adaptive threshold value calculation unit;

[0041] The noise calculation unit obtains the absolute value of noise corresponding to each of the plurality of data samples;

[0042] The multi-point mean value calculation unit divides the continuous plurality of data samples into a plurality of sample groups in sequence based on the second preset number; for each sample group, the noise average value of the sample group is obtained based on the absolute value of the noise of each data sample in the sample group;

[0043] The adaptive threshold value calculation unit obtains the adaptive threshold value of each sample group based on the noise average value of the sample group;

[0044] The pulse decision module sequentially compares the decision samples corresponding to the multiple data samples in each sample group with the adaptive threshold value of the sample group where the data samples are located.

[0045] As a preferred solution, the noise calculation unit includes a first-order backward difference subunit and an absolute value calculation subunit;

[0046] The first-order backward difference subunit is based on multiple data samples and uses a first-order backward difference method to obtain noise values ​​corresponding to each of the multiple data samples;

[0047] The absolute value calculation subunit obtains the noise absolute values ​​corresponding to the multiple data samples based on the noise values ​​corresponding to the multiple data samples.

[0048] In a third aspect, an embodiment of this specification provides a computer storage medium, wherein the computer storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing the steps described in the first aspect of the above embodiment.

[0049] The beneficial effects brought by the technical solutions provided by some embodiments of this specification include at least:

[0050] 1. In the traditional scheme, the adaptive threshold value of the signal envelope is directly calculated, and the amplitude of the pulse signal is compared with the size of the calculated adaptive threshold value by point-by-point judgment. If the amplitude of the pulse signal is greater than the adaptive threshold value, it is determined as a rising edge, and if the amplitude of the pulse signal is less than the adaptive threshold value, it is determined as a falling edge. In the case of low signal-to-noise ratio, for signals with large signal envelope jitter, it is easy to split a large pulse signal into multiple small pulse signals, affecting the signal detection result. The present invention first performs multi-point amplitude smoothing on the signal envelope, and then uses multi-data point accumulation (MN method) to perform rising edge and falling edge pulse judgment respectively, reducing the low possibility of pulse signal segmentation under low signal-to-noise ratio at the expense of the detection accuracy of the pulse signal edge, thereby improving the robustness of pulse signal detection, achieving a higher probability of correct judgment even when the signal-to-noise ratio is low, and solving the problem of low accuracy of signal adaptive detection technology under low signal-to-noise ratio in traditional schemes;

[0051] 2. In the traditional scheme, the amplitude of the signal (i.e., the signal envelope) is extracted by calculating the signal modulus value. The amplitude is Since the calculation accuracy of the amplitude size is not required to be high in actual detection, the present invention adopts the stepwise approximation method in advance and In order to obtain the correction factor A for approaching the target simulation, the data samples corresponding to the multiple complex signals are obtained based on the maxI,Q, minI,Q and correction factor A corresponding to the multiple complex signals. The original square root calculation is not required, which reduces the complexity of the algorithm and achieves the effect of greatly saving FPGA resource overhead with only a small part of the accuracy loss of the amplitude.

[0052] 3. The present invention performs truncation operation based on the average value of a first preset number of data samples. The signal amplitude stored in the sliding window register in the FPGA can be accumulated by an adder, and then divided by the first preset number to obtain the average value, and the truncation operation result of the average value is used as the decision sample corresponding to the target data sample, thereby achieving the effect of smoothing filtering. The structure for achieving this effect is simple, occupies few resources, and consumes only a small amount of FPGA resources;

[0053] 4. The traditional adaptive threshold detection scheme roughly extracts the background noise value after the high-pass filter module and the absolute value module, and then uses the method of multi-level smoothing filtering (generally two-level smoothing filtering) to obtain the fine extracted noise value of each point, thereby calculating the adaptive threshold value corresponding to each extracted signal amplitude, and using the point-by-point judgment method to compare one by one. In order to simplify the project implementation, from the perspective of project implementation complexity and effectiveness, the present invention uses a first-order backward difference operation to replace the high-pass filter, delays the signal by one beat, and then uses the method of finding the difference to extract the signal noise, eliminating the complex filter design steps, and after obtaining the absolute value of the noise, no multi-level smoothing filtering is performed. The continuous multiple data samples are divided into multiple sample groups in turn based on the second preset number. For each sample group, the adaptive threshold value of the sample group is obtained based on the noise average value of the sample group. When the pulse rising edge and falling edge judgment are performed, the judgment samples corresponding to each of the multiple data samples in each sample group are compared with the adaptive threshold value of the sample group where the data sample is located. This method is actually equivalent to a short smoothing filter, which makes the project implementation simpler and further saves the cost of FPGA resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] 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 some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 It is a schematic diagram of the architecture of the adaptive detection implementation process in the prior art;

[0056] Figure 2 It is a schematic diagram of the architecture of the adaptive detection implementation process in the present invention;

[0057] Figure 3 It is a flowchart of a low-complexity signal adaptive detection method based on FPGA provided in an embodiment of this specification;

[0058] Figure 4 It is the amplitude diagram calculated by using the square root method and the successive approximation method for the same series of sampling points, and the amplitude difference diagram of the two;

[0059] Figure 5 This is the signal envelope comparison diagram before and after multi-point smoothing when the signal-to-noise ratio is 7dB;

[0060] Figure 6 This is the signal envelope comparison diagram before and after multi-point smoothing when the signal-to-noise ratio is 15dB;

[0061] Figure 7 The signal envelope before and after multi-point smoothing when the signal-to-noise ratio is 9dB, as well as the comparison diagram between point-by-point pulse detection and MN method pulse detection;

[0062] Figure 8 It is a comparison diagram of the probability of correctly detecting a signal between the traditional signal adaptive detection scheme and the scheme of the present invention when the signal-to-noise ratio is 1-16dB;

[0063] Fig. 9 It is a structural schematic diagram of a low-complexity signal adaptive detection system based on FPGA provided in an embodiment of this specification. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of this specification will be described clearly and completely below in conjunction with the drawings in the embodiments of this specification.

[0065] The terms "first", "second", "third", etc. in the description and claims of this specification and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0066] The following description provides examples and does not limit the scope, applicability or examples set forth in the claims. Changes may be made to the functions and arrangements of the elements described without departing from the scope of the present specification. Various processes or components may be appropriately omitted, substituted or added to each example. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted or combined. In addition, features described with respect to some examples may be combined in other examples.

[0067] First, the principle of signal detection by setting adaptive threshold values ​​is explained:

[0068] In the signal receiver system, due to the existence of thermal noise in the system itself, the signal is generally mixed with the noise. The energy of the noise occupying the same frequency band as the signal directly affects the system's ability to detect small signals. Since noise always fluctuates, judging whether a signal exists or not becomes a statistical characteristic and must be judged according to certain statistical detection standards. The signal existence detection problem can be regarded as a binary hypothesis testing problem. According to the size of the selected statistic, the corresponding test criteria are designed to determine which state the received signal belongs to. Therefore, the signal detection problem can be modeled as follows:

[0069]

[0070] Among them, y(n) represents the received signal, x(n) represents the target signal, that is, the pulse signal, and w(n) represents Gaussian white noise. The two states of the judgment respectively represent whether the target signal exists in the received signal, H0 represents only noise, and H1 represents the presence of the target signal.

[0071] The signal presence detection method generally includes the following performance evaluation indicators:

[0072] Probability of Detection: denoted as P D =P{H1|H1}, that is, the probability of the detection method making a correct decision under the condition of the presence of the target signal;

[0073] Missing Detection Probability: denoted as P M =P{H0|H1}, that is, under the condition that the target signal exists, the detection method believes that there is only noise at this time and makes a wrong judgment probability;

[0074] False Alarm Probability: denoted as P FA =P{H1 |H0}, that is, under the condition that there is only noise and no target signal, the detection method believes that the target signal exists at this time and makes a wrong judgment probability.

[0075] The most commonly used criterion in signal detection is the Neyman-Pearson (NP) criterion. Its basic principle is to meet a certain false alarm probability P under a specific signal-to-noise ratio condition. FA At the same time, the detection probability P D Maximum. In the signal receiving system, a single pulse signal is channelized in the intermediate frequency part, followed by signal amplitude feature extraction, and then the envelope signal output by the detection is compared with a certain threshold value. If the output envelope value is greater than the threshold value, it is considered that the target signal exists, otherwise it is considered that there is no target signal. This is threshold detection. When the set threshold value is less than the noise, the pulse judgment will mistakenly treat the noise as a signal. This event is called a "false alarm". Therefore, "false alarm" should be an event that is avoided. Obviously, P FA It is determined by the noise power, the statistical characteristics of the noise and the set adaptive threshold value.

[0076] The following is a quantitative analysis of the relationship between them:

[0077] Usually, due to the thermal noise of the receiving antenna and the intermediate frequency filter (or intermediate frequency amplifier) ​​of the signal receiver, the noise in the system is generally broadband Gaussian noise, and its probability density function is shown as follows:

[0078]

[0079] where p(v)dv is the probability that the noise voltage is between v and p(v)dv; the noise has zero mean and a variance of σ 2 In the signal receiver, the signal processing process is generally to extract the signal through a narrowband intermediate frequency filter first, the intermediate frequency part performs channelization processing on the single pulse signal, and then enters the envelope detection module. In this process, the wideband noise is also processed in the same way. At this time, the envelope of the noise output from the envelope detection module obeys the Rayleigh distribution, and its probability density function is:

[0080]

[0081] In the formula, r represents the envelope value of Rayleigh noise. Assuming the adaptive threshold value is T, when the noise envelope value r exceeds the adaptive threshold value T, a false alarm will be triggered, so the false alarm probability P FA for:

[0082]

[0083] When the false alarm probability P FA When the threshold value T and variance σ are fixed, 2 The relationship is determined, and the relationship is:

[0084]

[0085] Since the noise follows the Rayleigh distribution, according to the relevant knowledge of probability theory, there is a fixed mathematical relationship between the mean of the noise and its standard deviation:

[0086]

[0087] Because it is much easier to calculate the noise mean than to directly calculate the noise variance, the noise standard deviation σ is replaced by the noise mean u, and equation (6) is substituted into equation (5), then:

[0088]

[0089] In the formula, is the threshold coefficient. If the false alarm probability remains unchanged, the adaptive detection threshold is only proportional to the mean value of the system noise. As long as the mean value u of the noise is calculated in time, the dynamically changing adaptive threshold T under constant false alarm can be obtained.

[0090] The above is that under a certain false alarm probability, by constantly estimating the value of the interference noise, the required threshold value is set according to the current noise amplitude value, so that the threshold value changes with the change of interference intensity. This is the so-called adaptive threshold.

[0091] However, in the case of low signal-to-noise ratio, for signals with large signal envelope jitter, traditional methods can easily split a large pulse signal into multiple small pulse signals, which will directly affect the signal detection results and make subsequent signal processing and analysis more difficult. Therefore, a low-complexity signal adaptive detection technology based on FPGA is proposed to meet high accuracy under low signal-to-noise ratio. Figure 2 , which is a schematic diagram of the architecture of the adaptive detection implementation process in the present invention.

[0092] The content of the present invention is described in detail below:

[0093] like Figure 3 As shown, Figure 3 It is a flowchart of a low-complexity signal adaptive detection method based on FPGA provided by one or more embodiments of this specification.

[0094] S202: Acquire a signal envelope including a plurality of data samples;

[0095] S204: performing amplitude multi-point smoothing processing on the signal envelope to obtain a signal to be decided including a plurality of decision samples;

[0096] S206: Acquire an adaptive threshold value based on the data sample;

[0097] S208: Compare the multiple decision samples and the adaptive threshold value in sequence;

[0098] S208A: When M consecutive decision samples in the to-be-decided signal are greater than the adaptive threshold value corresponding to the decision samples, a high-level detection flag is output;

[0099] S208B: When N consecutive decision samples in the to-be-decided signal are smaller than the adaptive threshold value corresponding to the decision samples, a low-level detection flag is output.

[0100] Illustratively, the pulse signal is received by the antenna, input into the processing module to obtain the intermediate frequency analog signal, which is converted to the digital domain by the analog-to-digital converter, and the received signal is Hilbert transformed to convert the real signal into a complex signal. The real and imaginary parts of the complex signal first enter the envelope detection module to obtain the amplitude characteristics of the signal (i.e., the signal envelope). In the traditional scheme, the amplitude of the signal is extracted by calculating the signal modulus value, and the amplitude is The data sample is the amplitude of the corresponding sampling point.

[0101] Illustratively, this module improves the system's ability to detect signals in low signal-to-noise ratio environments, which not only improves the receiver sensitivity but also controls the amount of calculation, making it very suitable for engineering implementation.

[0102] In order to adapt to low signal-to-noise ratio and accurately measure the arrival time and pulse width of pulse signals under low signal-to-noise ratio conditions, this solution does not directly compare the signal envelope with the adaptive threshold, but uses an amplitude multi-point smoothing technology to pre-process the signal envelope to avoid the problem of unclear spacing between high and low levels of the envelope signal due to interference caused by noise under low signal-to-noise ratio. This can be achieved using a filter. The judgment sample is the amplitude of the corresponding sampling point after smoothing.

[0103] Illustratively, the smoothed signal to be decided is delayed and waited, and after the adaptive threshold value corresponding to the current decision sample is calculated, the current decision sample and the corresponding adaptive threshold value are decided.

[0104] Explanatory, during signal detection and judgment, when M consecutive judgment samples in the signal to be judged exceed the corresponding adaptive threshold value, it is judged that the rising edge of the pulse signal is detected, and the rising edge of the detection flag is pulled high; when N consecutive judgment samples in the signal to be judged are lower than the corresponding adaptive threshold value, it is judged that the falling edge of the pulse signal is detected, and the falling edge of the detection flag is pulled low. Compared with the point-by-point detection in the traditional method, this processing can avoid detecting some interference of small pulse widths. At the same time, for signals with fluctuating envelopes, it can avoid the situation where the signal is divided into multiple small pulses when the signal envelope drops sharply. The robustness of the pulse detection algorithm is increased at the expense of the detection accuracy of the pulse signal edge in exchange for the low possibility of the pulse signal being divided under low signal-to-noise ratio.

[0105] Illustratively, the values ​​of M and N can be set according to actual conditions.

[0106] In one embodiment of the present specification, the acquiring a signal envelope including a plurality of data samples comprises:

[0107] Acquire a correction factor A and a plurality of complex signals including real data I and imaginary data Q;

[0108] Respectively compare the sizes of the real data I and the imaginary data Q corresponding to the multiple complex signals to obtain the maxI,Q and minI,Q corresponding to the multiple complex signals; obtain the data samples corresponding to the multiple complex signals based on the maxI,Q, minI,Q and the correction factor A corresponding to the multiple complex signals to obtain the signal envelope including the multiple data samples;

[0109] The correction factor A is pre-calculated by the stepwise approximation method. A*min(X, Y) is obtained by simulation approaching the target, and X and Y are random numbers.

[0110] Explanatory, because the actual detection does not require high calculation accuracy of the amplitude, the stepwise approximation method is used in advance and In order to obtain the correction factor A close to the target simulation, it is assumed that the contribution of the modulus value mainly comes from max(I, Q). To approximate the modulus value, it is only necessary to add the product of max(I, Q) and the correction factor A to the modulus value. Based on the max(I, Q), min(I, Q) and correction factor A corresponding to each of the multiple complex signals, the data samples corresponding to each of the multiple complex signals are obtained. There is no need to perform the original square root calculation. Only one comparison, one multiplication operation and one addition operation are required, which greatly reduces the complexity of the algorithm.

[0111] Specifically: The correction factor A obtained by simulation is 0.25, and the amplitude of the signal is:

[0112]

[0113] Exemplarily, the complex signal is Z=a+bi, when a>b, |Z|≈a+0.25b.

[0114] like Figure 4 As shown, Figure 4 The amplitude diagram and the amplitude difference diagram of the same series of sampling points are calculated using the square root method and the successive approximation method. From the corresponding values ​​on the vertical axis, it can be seen that the difference between the amplitudes of the same sampling point calculated by the two methods is very small relative to the amplitude of the sampling point. Only a small part of the amplitude accuracy loss is used to achieve the effect of greatly saving FPGA resource overhead.

[0115] In one embodiment of the present specification, performing multi-point smoothing of the signal envelope to obtain a signal to be decided including a plurality of decision samples includes:

[0116] Take any data sample as the target data sample:

[0117] Acquire a first preset number of data samples including the target data sample before and after the target data sample;

[0118] An average value of a first preset number of data samples is used as a decision sample corresponding to the target data sample;

[0119] Repeat the above steps to obtain the decision samples corresponding to each data sample, so as to obtain a signal to be decided including a plurality of decision samples.

[0120] Explanatory, in FPGA, a sliding window register is generally used to store the signal amplitude, and then an adder is used to accumulate the sampling points in the sliding window register, and the average value of the first preset number of data samples is used as the decision sample corresponding to the target data sample to obtain the effect of smooth filtering. Specifically, a pipeline FIR filter structure can be used, and the first preset number of data samples including the target data sample before and after the target data sample is obtained through the sliding window register, and the weighting coefficient of each data sample is set to 1 to achieve the sum, and then the decision sample of the target data sample can be obtained by dividing by the first preset number. Such a setting has a simple structure, occupies less resources, and can be done with a pipeline, which greatly improves the efficiency of calculation.

[0121] For example, Figure 5 , Figure 6 As shown, Figure 5 This is the signal envelope comparison diagram before and after multi-point smoothing when the signal-to-noise ratio is 7dB. Figure 6 The following is a comparison of the signal envelope before and after multi-point smoothing when the signal-to-noise ratio is 15dB. It can be seen that after the amplitude multi-point smoothing process, the signal amplitude is significantly enhanced, and the gap between the signal and the noise is widened, which not only reduces the difficulty of threshold setting, but also significantly improves the detection sensitivity.

[0122] In one embodiment of the present specification, the taking the average value of the first preset number of data samples as the decision sample corresponding to the target data sample includes:

[0123] Obtaining an average value of a first preset number of data samples;

[0124] A truncation operation is performed on the average value of the first preset number of data samples, and the result of the truncation operation is used as the decision sample corresponding to the target data sample.

[0125] Explanatory, since the average value may be a decimal, in order to reduce the consumption of FPGA resources and obtain an integer mean, the calculated mean value is processed using truncation operations. For example, truncation down: directly remove the decimal part and only keep the integer part. Rounding: if the decimal part is greater than or equal to 0.5, add 1 to the integer part; otherwise, only keep the integer part.

[0126] In one embodiment of the present specification, the acquiring of the adaptive threshold value based on the data sample includes:

[0127] Obtain the absolute value of noise corresponding to each of the multiple data samples;

[0128] Sequentially dividing a plurality of continuous data samples into a plurality of sample groups based on a second preset number;

[0129] For each sample group, the noise average value of the sample group is obtained based on the absolute value of the noise of each data sample in the sample group;

[0130] For each sample group, an adaptive threshold value of the sample group is obtained based on the noise average value of the sample group;

[0131] The step of sequentially comparing the multiple decision samples with the adaptive threshold value includes:

[0132] The decision samples corresponding to the multiple data samples in each sample group are compared with the adaptive threshold value of the sample group where the data samples are located.

[0133] In one embodiment of the present specification, obtaining the absolute value of noise corresponding to each of the plurality of data samples includes:

[0134] Based on multiple data samples, a first-order backward difference method is used to obtain noise values ​​corresponding to each of the multiple data samples;

[0135] The noise absolute values ​​corresponding to the multiple data samples are obtained based on the noise values ​​corresponding to the multiple data samples.

[0136] In the receiver, pulse signals are usually processed. The pulse transmitter sends a pulse every fixed time T. τ (Repetition frequency) sends a pulse signal with a certain width τ (pulse width) carrying basic information. The envelope of the signal a(t) can generally be regarded as noise n(t) and a series of pulse signals g with a width of τ τ (t), where a(t) is:

[0137] a(t)=n(t)+g τ (t-NT τ ) (8)

[0138] Where n(t) is the noise signal that follows Rayleigh distribution, T(t-NT τ ) is the repetition frequency T τ The pulse signal g τ The expression of (t) in a single cycle is:

[0139]

[0140] g τ The Fourier transform of (t) is:

[0141]

[0142] From the above formula, we can see that the pulse energy is basically concentrated in the frequency band of ω=±2π / τ. Therefore, only one cutoff angular frequency ω is needed. sA high-pass filter greater than 2π / τ can extract the noise signal. The cutoff frequency f of the high-pass filter is s Need to meet:

[0143]

[0144] In this way, the noise signal can be completely extracted from the envelope signal. At this time, equation (8) becomes:

[0145] a(t)=n(t) (12)

[0146] When the threshold coefficient K is fixed, a high-pass filter can be used to extract the high-frequency noise from the amplitude sequence, and then the mean value u of the noise can be calculated. The purpose of the high-pass filter is to extract the noise in the envelope and filter out the low-frequency part of the envelope. Assume that the pulse width requirement of the pulse signal in the system is 1μs to 10μs. From the Fourier transform theory, it can be known that the frequency of the first zero point of a pulse with a pulse width of τ in the frequency domain is the inverse of the pulse width, that is, f=1 / τ, so its frequency range is 10KHz to 1MHz. In other words, as long as the cutoff frequency of the high-pass filter is greater than 1MHz, the pulse envelope can be completely filtered out, thereby extracting the noise signal.

[0147] Explanatory, the present invention uses a method of delaying the signal envelope by one beat and then using a differential method to extract signal noise. The first-order backward difference method is a commonly used signal processing method, which can be used to perform differential operations on time series data to obtain a first-order derivative. The process of extracting noise by the differential method is equivalent to a low-order high-pass filtering operation. The advantage of using this method is that the implementation complexity is low, there is no need to design and use a high-pass filter, and the engineering implementation is relatively simple.

[0148] Illustratively, the absolute value of the extracted noise is calculated. In the traditional method, after obtaining the absolute value of the noise, a multi-stage smoothing filter (generally a two-stage smoothing filter) method is used to obtain the finely extracted noise value of each point, and the adaptive dynamic threshold value corresponding to each point is obtained by formula (7). In order to simplify the engineering implementation, starting from the complexity and effectiveness of the project implementation, the present invention no longer calculates the adaptive threshold value for each decision sample in the system, but after extracting the absolute value of the noise by the first-order backward difference and absolute value calculation, the absolute value of the noise corresponding to the second preset number of decision samples is calculated to obtain a common adaptive dynamic threshold value for multiple decision samples in the sample group. This method is actually equivalent to a short smoothing filter. It further reduces the complexity of the engineering implementation and can also achieve a good detection effect.

[0149] For example, in conjunction with Figure 7, assuming that the modulation type of the received signal is QPSK signal, the pulse width is 2μs, the duty cycle is 50%, the sampling frequency is set to 250MHz, and the signal-to-noise ratio is 9dB. In this simulation, the first preset number of envelope smoothing points and the second preset number of threshold mean accumulation points are both set to 32, and the false alarm rate is taken as 5*10 -6 , M and N are both 48. Figure 7 The following are the signal envelopes before and after multi-point smoothing when the signal-to-noise ratio is 9dB, as well as the comparison between point-by-point pulse detection and MN method pulse detection. The first line is the original signal, in which the interval between the high level and the low level is not obvious; the second line is the corresponding signal after signal smoothing and filtering, and the signal obtained after amplitude multi-point smoothing is significantly improved; the third line is the adaptive threshold, corresponding to the calculated adaptive threshold value; the fourth line is the pulse result obtained by point-by-point pulse detection in the traditional method; the fourth line is the pulse result obtained by MN method pulse detection in the present invention. It can be clearly seen that the present invention reduces the system calculation complexity and saves FPGA resources, and at a lower signal-to-noise ratio, the envelope jitter of the signal after smoothing the signal envelope is significantly reduced, and the rising and falling edges of the signal can be correctly detected after the MN method is used for judgment; in contrast, the traditional method has a high probability of error in detecting the signal, and the pulse signal will be detected as multiple small pulses.

[0150] For example, in conjunction with Figure 8 , assuming that the received signal modulation type is QPSK signal, the pulse width is 2μs, the duty cycle is 50%, and the sampling frequency is set to 250MHz. In this simulation, the envelope smoothing points and the threshold mean accumulation points are both set to 32, and the false alarm rate is taken as 5*10 -6 When the signal-to-noise ratio is 1 to 16 dB, the simulation is performed 1000 times under each set of parameters. Figure 8 The figure is a comparison of the probability of correctly detecting signals between the traditional signal adaptive detection scheme and the scheme in the present invention when the signal-to-noise ratio is 1-16dB. It can be clearly seen that when the signal-to-noise ratio is 1-16dB, the probability of correctly detecting signals obtained by the method of the present invention is significantly improved. In addition, when the signal-to-noise ratio is greater than or equal to 8dB, the scheme can achieve a 99% probability of correctly detecting signals. To achieve a 99% probability of correctly detecting signals, the traditional scheme requires a signal-to-noise ratio greater than or equal to 11dB.

[0151] It should be noted that in the prior art, the signal envelope also needs to be delayed when calculating the adaptive threshold value. In the present invention, the smoothed signal to be determined will be processed by a series of processes such as first-order backward difference, absolute value, mean value, and multiplication by the threshold coefficient K. It takes multiple clock cycles to obtain the adaptive threshold value corresponding to the current decision sample. Therefore, it is necessary to delay the signal to be determined by a corresponding clock cycle so that the current decision sample corresponds to the corresponding adaptive threshold value one by one. In the present invention, since the delay clock cycle involved is too long, the FIFOIP is used to check the signal modulus value for delay.

[0152] It should be noted that the false alarm rate, the first preset number of smoothing points and the second preset number of threshold mean points in the signal adaptive detection technology need to be designed according to specific system indicators and are not discussed in this patent.

[0153] Implementation principle: Combined with the Figure 2 , firstly, the input signal is envelope extracted, and then it enters the upper and lower parts of amplitude multi-point smoothing and adaptive dynamic threshold estimation. The upper part, i.e. the first part, performs amplitude multi-point smoothing on the signal envelope to further reduce the system's requirements for the signal-to-noise ratio, and sends the signal amplitude after multi-point smoothing to the delay unit to wait for the corresponding adaptive threshold value calculation to be completed; the lower part, i.e. the second part, is the adaptive threshold estimation module, which uses the first-order backward difference, absolute value, multi-point mean calculation and adaptive threshold value operation on the signal envelope to output the adaptive threshold value, and finally performs pulse judgment on the signal amplitude after multi-point smoothing and the adaptive threshold value to obtain the detection pulse.

[0154] See next Fig. 9 , Fig. 9 It is a structural schematic diagram of a low-complexity signal adaptive detection system based on FPGA provided in an embodiment of this specification, including an envelope extraction and processing module, an adaptive dynamic threshold module and a pulse decision module;

[0155] The envelope extraction and processing module obtains a signal envelope including multiple data samples; performs amplitude multi-point smoothing processing on the signal envelope to obtain a signal to be decided including multiple decision samples;

[0156] The adaptive dynamic threshold module obtains an adaptive threshold value based on the data sample;

[0157] The pulse decision module compares the sizes of multiple decision samples and adaptive threshold values ​​in sequence; when M consecutive decision samples in the signal to be decided are greater than the adaptive threshold value corresponding to the decision samples, a high-level detection flag is output; when N consecutive decision samples in the signal to be decided are less than the adaptive threshold value corresponding to the decision samples, a low-level detection flag is output.

[0158] As a preferred solution, the adaptive dynamic threshold module includes a noise calculation unit, a multi-point mean value calculation unit and an adaptive threshold value calculation unit;

[0159] The noise calculation unit obtains the absolute value of noise corresponding to each of the plurality of data samples;

[0160] The multi-point mean value calculation unit divides the continuous plurality of data samples into a plurality of sample groups in sequence based on the second preset number; for each sample group, the noise average value of the sample group is obtained based on the absolute value of the noise of each data sample in the sample group;

[0161] The adaptive threshold value calculation unit obtains the adaptive threshold value of each sample group based on the noise average value of the sample group;

[0162] The pulse decision module sequentially compares the decision samples corresponding to the multiple data samples in each sample group with the adaptive threshold value of the sample group where the data samples are located.

[0163] As a preferred solution, the noise calculation unit includes a first-order backward difference subunit and an absolute value calculation subunit;

[0164] The first-order backward difference subunit is based on multiple data samples and uses a first-order backward difference method to obtain noise values ​​corresponding to each of the multiple data samples;

[0165] The absolute value calculation subunit obtains the noise absolute values ​​corresponding to the multiple data samples based on the noise values ​​corresponding to the multiple data samples.

[0166] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the adaptive detection system embodiment, since it is basically similar to the adaptive detection method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the adaptive detection method embodiment.

[0167] The embodiments of this specification also provide a computer-readable storage medium, which stores instructions, and when the instructions are executed on a computer or a processor, the computer or the processor executes one or more steps in the above-mentioned adaptive detection method embodiment. If the components of the above-mentioned adaptive detection system are implemented in the form of software functional units and sold or used as independent products, they can be stored in the computer-readable storage medium.

[0168] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of this specification is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, data center, etc. that contains one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0169] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes. In the absence of conflict, the technical features in this embodiment and the implementation scheme can be combined arbitrarily.

[0170] The embodiments described above are merely preferred embodiments of this specification and are not intended to limit the scope of this specification. Without departing from the design spirit of this specification, various modifications and improvements made to the technical solutions of this specification by ordinary technicians in this field should fall within the scope of protection determined by the claims of this specification.

Claims

1. A low-complexity signal adaptive detection method based on FPGA, characterized in that: include: acquiring a signal envelope comprising a plurality of data samples; Performing multi-point smoothing processing on the signal envelope to obtain a signal to be decided including multiple decision samples; Obtaining an adaptive threshold value based on the data sample; Comparing the sizes of multiple decision samples and adaptive threshold values ​​in sequence; When M consecutive decision samples in the signal to be decided are greater than the adaptive threshold value corresponding to the decision sample, a high-level detection flag is output; When N consecutive decision samples in the signal to be decided are less than the adaptive threshold value corresponding to the decision sample, a low-level detection flag is output.

2. The low-complexity signal adaptive detection method based on FPGA according to claim 1, characterized in that: The obtaining of a signal envelope including a plurality of data samples comprises: Acquire a correction factor A and a plurality of complex signals including real data I and imaginary data Q; Compare the sizes of the real data I and the imaginary data Q corresponding to each of the multiple complex signals respectively to obtain the max(I,Q) and min(I,Q) corresponding to each of the multiple complex signals; obtain the data samples corresponding to each of the multiple complex signals based on the max(I,Q), min(I,Q) and correction factor A corresponding to each of the multiple complex signals to obtain a signal envelope including the multiple data samples; The correction factor A is pre-calculated by the stepwise approximation method. To approach the target simulation, X and Y are random numbers.

3. The low-complexity signal adaptive detection method based on FPGA according to claim 1, characterized in that: The signal envelope is subjected to multi-point amplitude smoothing processing to obtain a signal to be decided including a plurality of decision samples, including: Take any data sample as the target data sample: Acquire a first preset number of data samples including the target data sample before and after the target data sample; An average value of a first preset number of data samples is used as a decision sample corresponding to the target data sample; Repeat the above steps to obtain the decision samples corresponding to each data sample, so as to obtain a signal to be decided including a plurality of decision samples.

4. The low-complexity signal adaptive detection method based on FPGA according to claim 3, characterized in that: The method of taking the average value of the first preset number of data samples as the decision sample corresponding to the target data sample includes: Obtaining an average value of a first preset number of data samples; A truncation operation is performed on the average value of the first preset number of data samples, and the result of the truncation operation is used as the decision sample corresponding to the target data sample.

5. The low-complexity signal adaptive detection method based on FPGA according to claim 3, characterized in that: The step of obtaining an adaptive threshold value based on the data sample includes: Obtain the absolute value of noise corresponding to each of the multiple data samples; Sequentially dividing a plurality of continuous data samples into a plurality of sample groups based on a second preset number; For each sample group, the noise average value of the sample group is obtained based on the absolute value of the noise of each data sample in the sample group; For each sample group, an adaptive threshold value of the sample group is obtained based on the noise average value of the sample group; The step of sequentially comparing the multiple decision samples with the adaptive threshold value includes: The decision samples corresponding to the multiple data samples in each sample group are compared with the adaptive threshold value of the sample group where the data samples are located.

6. The low-complexity signal adaptive detection method based on FPGA according to claim 5, characterized in that: The obtaining of the absolute noise values ​​corresponding to the plurality of data samples comprises: Based on multiple data samples, a first-order backward difference method is used to obtain noise values ​​corresponding to each of the multiple data samples; The noise absolute values ​​corresponding to the multiple data samples are obtained based on the noise values ​​corresponding to the multiple data samples.

7. A low-complexity signal adaptive detection system based on FPGA, characterized in that: It includes envelope extraction and processing module, adaptive dynamic threshold module and pulse decision module; The envelope extraction and processing module obtains a signal envelope including multiple data samples; performs amplitude multi-point smoothing processing on the signal envelope to obtain a signal to be decided including multiple decision samples; The adaptive dynamic threshold module obtains an adaptive threshold value based on the data sample; The pulse decision module sequentially compares the sizes of multiple decision samples and the adaptive threshold value; When M consecutive decision samples in the signal to be decided are greater than the adaptive threshold value corresponding to the decision sample, a high-level detection flag is output; When N consecutive decision samples in the signal to be decided are less than the adaptive threshold value corresponding to the decision sample, a low-level detection flag is output.

8. The low-complexity signal adaptive detection system based on FPGA according to claim 7, characterized in that: The adaptive dynamic threshold module includes a noise calculation unit, a multi-point mean value calculation unit and an adaptive threshold value calculation unit; The noise calculation unit obtains the absolute value of noise corresponding to each of the plurality of data samples; The multi-point mean value calculation unit divides the continuous plurality of data samples into a plurality of sample groups in sequence based on the second preset number; for each sample group, the noise average value of the sample group is obtained based on the absolute value of the noise of each data sample in the sample group; The adaptive threshold value calculation unit obtains the adaptive threshold value of each sample group based on the noise average value of the sample group; The pulse decision module sequentially compares the decision samples corresponding to the multiple data samples in each sample group with the adaptive threshold value of the sample group where the data samples are located.

9. The low-complexity signal adaptive detection system based on FPGA according to claim 8, characterized in that: The noise calculation unit includes a first-order backward difference subunit and an absolute value calculation subunit; The first-order backward difference subunit is based on multiple data samples and uses a first-order backward difference method to obtain noise values ​​corresponding to each of the multiple data samples; The absolute value calculation subunit obtains the noise absolute values ​​corresponding to the multiple data samples based on the noise values ​​corresponding to the multiple data samples.

10. A computer storage medium storing a plurality of instructions, wherein the instructions are loaded by a processor and execute the steps of the FPGA-based low-complexity signal adaptive detection method as described in any one of claims 1 to 6.

Images