Time domain separation method and system for non-cooperative communication signal

Abstract

The embodiment of the invention provides a time domain separation method and system for a non-cooperative communication signal, and the method comprises the steps: firstly obtaining a mixed signal set which comprises the superposition of a plurality of different source signals in a non-cooperative communication scene, and enabling each superposed signal unit to carry a time domain feature mark of asynchronous transmission; performing dynamic characteristic association processing on the mixed signal set, and capturing an instantaneous frequency hopping characteristic and a phase continuity characteristic; constructing a time domain separation dynamic model based on the features, wherein the model comprises a feature association weight and a time domain boundary judgment rule; iterative boundary optimization processing is carried out on the mixed signal set through the model, the time domain independent distribution interval of each source signal is determined, and a single source signal unit is obtained; and finally, splicing and integrating the signal units in sequence, and outputting a time domain separation result set. According to the invention, the accuracy and effectiveness of non-cooperative communication signal separation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of non-cooperative communication, in particular to a time domain separation method and system for non-cooperative communication signals. BACKGROUND

[0002] In the field of non-cooperative communication, signal processing faces many complex and challenging problems. In the non-cooperative communication scenario, the receiving party often receives the superimposed mixed signals of non-cooperative communication signals from multiple different sources in the time domain, because the two parties of communication do not follow a unified cooperative protocol. The traditional signal separation method has obvious shortcomings in processing such mixed signals.

[0003] Some early methods are based only on the frequency domain features of the signals for separation, but the frequency domain characteristics of non-cooperative communication signals may have similarities, making it difficult to accurately distinguish signals from different sources. Although some other methods consider time domain features, they often only focus on the amplitude or simple time domain waveform of the signal, ignoring key features such as the instantaneous frequency jump and phase continuity of the signal in the time domain. These features are crucial for accurately separating non-cooperative communication signals from different sources, because the frequency mutation law and phase change trend of signals from different signal sources usually differ. In addition, most existing methods lack mechanisms for dynamic modeling and iterative optimization, and cannot adapt to the non-synchronous transmission characteristics of non-cooperative communication signals, making it difficult to accurately determine the time domain independent distribution interval of each source signal in complex mixed signals, thereby affecting the accuracy and effectiveness of signal separation. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a time domain separation method and system for non-cooperative communication signals.

[0005] In combination with the first aspect of the present application, a time domain separation method for non-cooperative communication signals is provided, which is applied to a time domain separation system for non-cooperative communication signals, and the method comprises:

[0006] Obtaining a mixed signal set in a non-cooperative communication scenario, the mixed signal set containing superimposed signal units of non-cooperative communication signals from multiple different sources in the time domain, and each superimposed signal unit carrying time domain feature markers of non-synchronous transmission;

[0007] Performing dynamic feature correlation processing on the mixed signal set to capture the instantaneous frequency jump feature and the phase continuity feature of each superimposed signal unit, the instantaneous frequency jump feature being used to describe the frequency mutation law of the signal in the time domain, and the phase continuity feature being used to reflect the coherent change trend of the signal phase over time;

[0008] construct a time-domain separation dynamic model based on the instantaneous frequency hopping feature and the phase continuity feature, the time-domain separation dynamic model containing feature correlation weights of different source signals and time-domain boundary determination rules;

[0009] perform iterative boundary optimization processing on the mixed signal set through the time-domain separation dynamic model to determine an independent distribution interval of each source signal in the time domain, and obtain a plurality of single-source non-cooperative communication signal units;

[0010] splice and integrate the plurality of single-source non-cooperative communication signal units according to the order of the time-domain distribution intervals, and output a time-domain separation result set of non-cooperative communication signals.

[0011] In combination with the second aspect of the present application, a time-domain separation system for non-cooperative communication signals is provided, which comprises a machine readable storage medium and a processor, the machine readable storage medium storing machine executable instructions, and the processor, when executing the machine executable instructions, implements the aforementioned time-domain separation method for non-cooperative communication signals.

[0012] In combination with the third aspect of the present application, a computer readable storage medium is provided, which stores computer executable instructions, and when the computer executable instructions are executed, the aforementioned time-domain separation method for non-cooperative communication signals is implemented.

[0013] In combination with any of the above aspects, by acquiring a mixed signal set containing a plurality of non-cooperative communication signals of different sources superimposed and performing dynamic feature correlation processing thereon, the instantaneous frequency hopping feature and the phase continuity feature of each superimposed signal unit can be accurately captured. The time-domain separation dynamic model constructed based on these features contains feature correlation weights of different source signals and time-domain boundary determination rules, which can dynamically adapt to the non-synchronous transmission characteristics of non-cooperative communication signals and more reasonably describe the complex relationship between signals. Through the iterative boundary optimization processing of the mixed signal set by the time-domain separation dynamic model, the independent distribution interval of each source signal in the time domain can be accurately determined, and false judgments and omissions in the signal separation process can be effectively avoided. Finally, the plurality of single-source non-cooperative communication signal units obtained by separation are spliced and integrated according to the order of the time-domain distribution intervals, and an accurate time-domain separation result set is output. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can be obtained without creative labor.

[0015] Figure 1 The flowchart of the time domain separation method for non-cooperative communication signals provided by the embodiments of the present application is shown. DETAILED DESCRIPTION

[0016] In order to make the person skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application.

[0017] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish different objects, not to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or end including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or end.

[0018] In this paper, the "embodiment" means that the specific features, structures or characteristics described in conjunction with the embodiment can be included in at least one embodiment of the present application. The phrase appears at various places in the specification does not necessarily refer to the same embodiment, nor is it independent or alternative to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0019] Figure 1 The flowchart of the time domain separation method for non-cooperative communication signals provided by the embodiments of the present application is shown. It should be understood that in other embodiments, the order of some steps of the time domain separation method for non-cooperative communication signals of the present embodiment can be shared with each other based on actual needs, or some steps can be omitted or maintained. The time domain separation method for non-cooperative communication signals includes:

[0020] Step S110: Obtain a mixed signal set in a non-cooperative communication scenario, the mixed signal set contains superimposed signal units of non-cooperative communication signals of different sources in the time domain, and each superimposed signal unit carries a time domain feature mark of non-synchronous transmission.

[0021] In this embodiment, a non-cooperative communication scenario of multi-device coexistence is selected, and there are multiple independently running communication devices in the scenario. These devices respectively send signals according to their own transmission protocols and time rhythms, and there is no cooperative communication mechanism between them, resulting in the inevitable superposition of their signals in the time domain during transmission. For example, there are device X, device Y and device Z, device X adopts frequency hopping communication mode, device Y adopts continuous wave communication mode, and device Z adopts burst communication mode. Their signals are received by the same receiving device after propagating in space, forming a mixed signal to be processed.

[0022] In order to obtain the mixed signal set, the receiving device needs to have a wideband receiving capability, which can cover the frequency range that these devices may use. The receiving device can continuously monitor the electromagnetic signals in the frequency band during operation, and when it detects that there is a signal, it will capture it. The captured signal is not a single source, but a composite signal of the signals of device X, device Y and device Z superimposed on each other.

[0023] These composite signals are divided into multiple superimposed signal units in chronological order, and each superimposed signal unit corresponds to a continuous time segment. Each superimposed signal unit carries the time domain feature mark of each source signal, which may include the starting time, duration, frequency change trend and other non-synchronous transmission traces of the signal. For example, the signal of device X may show a mark of rapid frequency hopping in a superimposed signal unit, the signal of device Y shows a mark of stable frequency, and the signal of device Z shows a mark of sudden appearance and disappearance.

[0024] Step S111: Capture a composite electrical signal in a non-cooperative communication scenario through a wideband receiving device, the composite electrical signal contains non-cooperative communication signal components of different frequency bands and different modulation modes.

[0025] In the above scenario, when the wideband receiving device starts to work, it will first set its receiving frequency band to ensure that it can cover all frequency bands that device X, device Y and device Z may use. After the setting is completed, the receiving device starts the signal capture program, which receives the electromagnetic signals in space through the antenna and converts them into a composite electrical signal.

[0026] Since the device X, the device Y and the device Z use different frequency bands and modulation methods, the captured composite electrical signal contains these different signal components. For example, the device X works in the frequency band F1 and uses frequency keying modulation; the device Y works in the frequency band F2 and uses amplitude keying modulation; and the device Z works in the frequency band F3 and uses phase keying modulation. These different signal components together constitute the composite electrical signal.

[0027] During the capturing process, the wide-band receiving device can maintain a high sampling rate to ensure that the details of each component in the composite electrical signal can be accurately recorded. At the same time, in order to avoid signal loss, the receiving device continuously captures until a composite electrical signal with a sufficient time length is obtained for subsequent processing.

[0028] Step S112: performing time-frequency domain pre-decomposition processing on the composite electrical signal to decompose the composite electrical signal into a plurality of signal components with time-frequency focusing characteristics, each signal component corresponding to a specific time-frequency distribution range.

[0029] After obtaining the composite electrical signal, the next step is to perform time-frequency domain pre-decomposition processing. The purpose of this processing is to separate the signal components of different sources in the composite electrical signal and obtain a plurality of signal components with time-frequency focusing characteristics.

[0030] Firstly, a suitable time-frequency analysis method needs to be selected. Considering the complexity of the composite electrical signal, a suitable decomposition algorithm is used for processing in this embodiment. This algorithm can automatically decompose the composite electrical signal into a plurality of sub-signals according to the time-frequency characteristics of the signal, each sub-signal having a relatively concentrated distribution range on the time-frequency plane, i.e., having time-frequency focusing characteristics.

[0031] For example, after decomposition processing, the composite electrical signal originally containing the signal components of the device X, the device Y and the device Z can be decomposed into three signal components. One signal component mainly corresponds to the signal of the device X, and its distribution range on the time-frequency plane is concentrated in the frequency band F1 and the corresponding time interval. Another signal component corresponds to the signal of the device Y, and its distribution range is concentrated in the frequency band F2 and the corresponding time interval. The third signal component corresponds to the signal of the device Z, and its distribution range is concentrated in the frequency band F3 and the corresponding time interval.

[0032] Step S1121: performing decomposition processing on the composite electrical signal using a variational mode decomposition algorithm, setting a decomposition mode number parameter, and decomposing the composite electrical signal into a plurality of intrinsic mode functions.

[0033] In the time-frequency domain pre-decomposition processing of the composite signal, the variational mode decomposition algorithm is adopted in the embodiment. First, the decomposition mode number parameter needs to be determined, which needs to be determined according to the number of signal components that may be contained in the composite signal. In combination with the above scenario, since there are three signal sources of device X, device Y and device Z, the decomposition mode number parameter can be initially set to a suitable value to ensure that all possible signal components can be covered.

[0034] After setting the parameters, the variational mode decomposition algorithm starts to process the composite signal. The algorithm decomposes the composite signal into multiple intrinsic mode functions by constructing and solving a variational problem. Each intrinsic mode function has a specific frequency characteristic and a certain oscillation characteristic in the time domain. For example, after decomposition, multiple intrinsic mode functions IMF1, IMF2, IMF3… are obtained, which correspond to different frequency components of the composite signal.

[0035] Step S1122: Perform time-frequency analysis on each intrinsic mode function to generate a corresponding time-frequency distribution map through short-time Fourier transform, which reflects the energy distribution of the signal in the time and frequency dimensions.

[0036] After obtaining multiple intrinsic mode functions, time-frequency analysis needs to be performed on each intrinsic mode function. In the embodiment, short-time Fourier transform is used for this analysis. For each intrinsic mode function, a suitable window function and window length are first determined, and the selection of the window function will affect the time resolution and frequency resolution of the time-frequency analysis, which needs to be adjusted according to the characteristics of the intrinsic mode function.

[0037] Then, the intrinsic mode function is divided into multiple time segments according to the window length, and Fourier transform is performed on each time segment to obtain the frequency component and amplitude information corresponding to the time segment. Arranging these information in the dimensions of time and frequency generates a time-frequency distribution map corresponding to the intrinsic mode function.

[0038] In the time-frequency distribution map, the horizontal axis represents time, the vertical axis represents frequency, and the color or gray depth in the map represents the signal energy at that time and frequency point. For example, for the intrinsic mode function IMF1, its corresponding time-frequency distribution map has strong energy in some time intervals in the frequency band F1, which indicates that the intrinsic mode function may contain the signal component of device X; while the time-frequency distribution map of IMF2 has strong energy in the corresponding time interval of frequency band F2, which may correspond to the signal component of device Y.

[0039] Step S1123: Extract the energy aggregation area from the time-frequency distribution map, and determine the frequency range and time range corresponding to the energy aggregation area as the time-frequency focusing range of the intrinsic mode function.

[0040] After generating the time-frequency distribution map, it is necessary to extract the energy aggregation area from it. This process can be achieved by setting an energy threshold, when the energy value of a certain area exceeds the set threshold, the area is considered as an energy aggregation area.

[0041] For the time-frequency distribution map of each intrinsic mode function, the extraction of the energy aggregation area is carried out respectively. For example, in the time-frequency distribution map of IMF1, the energy aggregation area is extracted in the time interval T1 in the frequency band F1, then the frequency range F1 and the time range T1 corresponding to the energy aggregation area are determined as the time-frequency focusing range of IMF1; similarly, the frequency range F2 and the time range T2 corresponding to the energy aggregation area extracted in the time-frequency distribution map of IMF2 are determined as the time-frequency focusing range of IMF2.

[0042] Step S1124: screening the intrinsic mode functions according to the time-frequency focusing range, and retaining the intrinsic mode functions with clear time-frequency focusing range and no significant overlap as effective signal components.

[0043] After obtaining the time-frequency focusing range of each intrinsic mode function, it is necessary to screen the intrinsic mode functions according to these ranges. The main criteria for screening are whether the time-frequency focusing range is clear and whether there is significant overlap between the time-frequency focusing ranges of different intrinsic mode functions.

[0044] The clear time-frequency focusing range means that the boundary of the energy aggregation area is clear, and its frequency range and time range can be clearly determined; and the no significant overlap means that the overlapping part of the time-frequency focusing ranges of different intrinsic mode functions in frequency and time is small, which will not affect the distinction of different signal components.

[0045] For example, for intrinsic mode functions IMF1, IMF2 and IMF3, their time-frequency focusing ranges are (F1, T1), (F2, T2) and (F3, T3) respectively, and there is no significant overlap between these ranges, and each range is relatively clear, then these three intrinsic mode functions will be retained as effective signal components; while those intrinsic mode functions with ambiguous time-frequency focusing range or significant overlap with the ranges of other intrinsic mode functions will be excluded.

[0046] Step S1125: arranging the effective signal components in time sequence to form a plurality of signal component sequences with time-frequency focusing characteristics.

[0047] After obtaining the effective signal components through screening, it is necessary to arrange these effective signal components in time sequence. This is because in the subsequent processing, it is necessary to analyze and process the signal based on the time axis, and arranging in time sequence can ensure the coherence and accuracy of the processing.

[0048] For example, the time ranges corresponding to the effective signal components IMF1, IMF2 and IMF3 are T1, T2 and T3 respectively, and T1, T2 and T3 have a time sequence or are partially overlapped, and they are arranged in the time sequence to form a signal component sequence, each element in the sequence being an effective signal component with time-frequency focusing characteristics and arranged in time sequence.

[0049] Step S113: performing time axis calibration processing on each signal component, superimposing and reorganizing the time axis calibrated signal components to form a signal sequence containing time domain superimposition information, each signal sequence unit corresponding to a signal superimposition state in a time segment.

[0050] After obtaining the signal component sequence, since different signal components can come from different devices, the clocks of these devices can have differences, resulting in that the start and end times of the signal components on the time axis are not synchronized, and therefore it is necessary to perform time axis calibration processing on each signal component.

[0051] After the time axis calibration processing is completed, the calibrated signal components are superimposed and reorganized according to the time axis. The superimposition process is to synthesize the amplitude values of the signal components at the same time point to obtain the superimposed signal value at the time point. In this way, a signal sequence containing time domain superimposition information is formed, and each signal sequence unit in the signal sequence corresponds to a time segment, and each unit reflects the superimposition state of the signal components in the time segment.

[0052] For example, the calibrated signal components IMF1, IMF2 and IMF3 have amplitude values A1, A2 and A3 in the time segment t1, and the amplitude value of the signal sequence unit corresponding to the time segment t1 in the signal sequence is the superimposed result of A1, A2 and A3, which reflects the superimposition state of the three signal components in the time segment t1.

[0053] Step S1131: adding an initial time mark to each signal component, the initial time mark being generated based on the signal receiving time.

[0054] When performing time axis calibration processing on the signal components, an initial time mark is first added to each signal component. The generation of the initial time mark is based on the signal receiving time. When the start time of the signal component is captured, the receiving device can record the time information of the time as the initial time mark of the signal component.

[0055] For example, the signal component IMF1 is captured at the receiving time t0, and the initial time mark added to IMF1 is t0; the signal component IMF2 is captured at the receiving time t1, and the initial time mark is t1, and so on.

[0056] Step S1132: Extracting feature points containing time reference information from each signal component, the feature points are peak points of signal amplitude or jump points of signal frequency.

[0057] After adding the initial time markers, feature points containing time reference information need to be extracted from each signal component. These feature points can reflect the key changes of the signal on the time axis, which is of great significance for time axis calibration.

[0058] The feature points can be peak points of signal amplitude, which are points when the signal amplitude reaches a local maximum value; or jump points of signal frequency, which are points when the signal frequency suddenly changes at a certain time point.

[0059] For example, in signal component IMF1, the signal amplitude reaches a peak value at time point t01, so t01 is a feature point; at time point t02, the signal frequency suddenly jumps from f0 to f1, so t02 is also a feature point. Similarly, similar feature points can also be extracted in IMF2 and IMF3.

[0060] Step S1133: Aligning the feature points of all signal components on the time axis, taking the feature point time of one signal component as the reference time, calculating the difference between the feature points of other signal components and the reference time as the time deviation.

[0061] After extracting the feature points of all signal components, these feature points are placed on the same time axis for alignment. First, select one of the signal components as the reference signal component, for example, select IMF1 as the reference, and take the time of a certain feature point of IMF1 as the reference time, assuming that t01 is selected as the reference time.

[0062] Then, the difference between the feature points of other signal components and the reference time is calculated respectively. For example, the time of a feature point of IMF2 is t11, so the difference between t11 and t01 is t11 minus t01, which is the time deviation of IMF2 relative to the reference signal component; similarly, the difference between the feature points of IMF3 and the reference time is calculated to obtain the time deviation of IMF3.

[0063] Step S1134: Correcting the initial time markers of the corresponding signal components according to the time deviation, and performing time interpolation processing on the corrected signal components to fill the signal gaps caused by the correction of the time markers.

[0064] After obtaining the time deviation amounts, the initial time marks of the corresponding signal components are corrected according to the time deviation amounts. For example, the initial time mark of IMF2 is t1, and the time deviation amount is Δt1, so the corrected initial time mark is t1 plus Δt1; the initial time mark of IMF3 is t2, and the time deviation amount is Δt2, so the corrected initial time mark is t2 plus Δt2.

[0065] Due to the correction of the time marks, the positions of the signal components on the time axis change, which may cause gaps between the signals. In order to fill the gaps, the corrected signal components need to be subjected to time interpolation processing. The interpolation processing can use linear interpolation or other suitable interpolation methods to calculate the amplitude value at the gap according to the amplitude values of the signal components at adjacent time points, so that the signal components remain continuous on the time axis.

[0066] For example, the corrected IMF2 has a signal gap in the time interval [t3, t4], and the amplitude value at t3 is A3 and the amplitude value at t4 is A4. Using linear interpolation, the amplitude value at any time point t between t3 and t4 is calculated as A3 plus (A4 minus A3) times (t minus t3) divided by (t4 minus t3), and the gap is filled in the above manner.

[0067] Step S114: determining the signal sequence as a mixed signal set in a non-cooperative communication scenario, and the superimposed signal units in the mixed signal set retaining the original time-frequency characteristics of the source signals.

[0068] The signal sequence obtained after the above processing contains superimposed information of multiple signal components in the time domain, each signal sequence unit corresponds to a superimposed state in a time segment, and retains the original time-frequency characteristics of the source signals. For example, in the signal sequence, the time-frequency characteristics (such as frequency hopping characteristics) of the signal of device X, the time-frequency characteristics (such as frequency stability characteristics) of the signal of device Y, and the time-frequency characteristics (such as burst characteristics) of the signal of device Z are all retained.

[0069] Therefore, the signal sequence is determined as a mixed signal set in a non-cooperative communication scenario, and the set will serve as input data for subsequent dynamic feature correlation processing.

[0070] Step S120: performing dynamic feature correlation processing on the mixed signal set, capturing the instantaneous frequency jump feature and the phase continuity feature of each superimposed signal unit, the instantaneous frequency jump feature being used to describe the frequency mutation law of the signal in the time domain, and the phase continuity feature being used to reflect the coherent change trend of the signal phase with time.

[0071] After obtaining the mixed signal set, dynamic feature correlation processing needs to be performed on it. The purpose of this processing is to analyze each superimposed signal unit in depth, and to capture the instantaneous frequency jump characteristics and phase continuity characteristics that can reflect the signal characteristics.

[0072] Instantaneous frequency jump characteristics focus on the pattern of sudden frequency changes in the time domain, such as the interval and amplitude of frequency jumps; while phase continuity characteristics focus on the degree of continuity of the signal phase change over time, and whether there are any discontinuities.

[0073] In the above scenario, the mixed signal set contains multiple superimposed signal units after the signals of device X, device Y and device Z are superimposed. Each unit is analyzed. For example, in a certain superimposed signal unit, the signal of device X may have multiple frequency jumps, the signal frequency of device Y is stable, and the phase of the signal of device Z may have discontinuities. Through dynamic feature correlation processing, these features are captured separately.

[0074] Step S121: Divide each superimposed signal unit in the mixed signal set into multiple time-domain analysis windows. Each time-domain analysis window has a variable time length, and the time length is adaptively adjusted according to the rate of change of the signal frequency.

[0075] When performing dynamic feature correlation processing on a mixed signal set, each superimposed signal unit first needs to be divided into multiple time-domain analysis windows. The division of time-domain analysis windows is to facilitate the analysis of signal features within different time segments.

[0076] The duration of each time-domain analysis window is not fixed but variable, and it adaptively adjusts according to the rate of change of the signal frequency. When the signal frequency changes rapidly, it indicates that the signal has undergone significant changes in a short period of time. In this case, the duration needs to be set shorter to improve time resolution and more accurately capture frequency jump characteristics. When the signal frequency changes slowly, the duration can be set longer to improve frequency resolution.

[0077] For example, for a certain superimposed signal unit, if the signal frequency changes rapidly in a certain segment, this segment is divided into multiple shorter time-domain analysis windows, such as windows W1, W2, etc., with each window having a short time length; while in the segment where the signal frequency changes slowly, it is divided into longer time-domain analysis windows, such as windows W3, W4, etc.

[0078] Step S122: Perform instantaneous frequency extraction processing on the signal within each time domain analysis window, calculate the instantaneous frequency value of the signal through Hilbert transform, and record the set of discrete points where the instantaneous frequency value changes over time.

[0079] After dividing the time-domain analysis window, instantaneous frequency extraction is performed on the signal within each window. In this embodiment, Hilbert transform is used to calculate the instantaneous frequency value.

[0080] For each time-domain analysis window, a Hilbert transform is first performed to obtain its analytic signal. The analytic signal contains the amplitude and phase information of the original signal. By differentiating the phase of the analytic signal and dividing by 2π, the instantaneous frequency value of the signal within that window can be obtained.

[0081] During the calculation, instantaneous frequency values ​​can be sampled at certain time intervals, and the instantaneous frequency value at each sampling moment can be recorded. These sampling points constitute a discrete set of instantaneous frequency values ​​that change over time. For example, within the time-domain analysis window W1, sampling is performed at smaller time intervals to obtain a series of instantaneous frequency values ​​f11, f12, f13, ..., with the corresponding time points t11, t12, t13, ..., thus forming a discrete set of points {(t11, f11), (t12, f12), (t13, f13) ...}. Similarly, a similar discrete set of points will be obtained within window W2.

[0082] Step S123: Perform abrupt change detection processing on the discrete point set, identify the time position where the instantaneous frequency value changes significantly, and determine the frequency change pattern between adjacent abrupt change points as instantaneous frequency jump features. The instantaneous frequency jump features include jump interval and frequency change amplitude.

[0083] After obtaining the set of discrete points, abrupt change detection processing is required to identify the time locations where the instantaneous frequency value changes significantly. These abrupt change points are key to analyzing the characteristics of instantaneous frequency jumps.

[0084] By comparing and analyzing the instantaneous frequency values ​​of adjacent time points in a discrete point set, it is determined whether there are significant changes. When the change reaches a certain level, the corresponding time position is identified as abrupt change point. Then, the frequency changes between adjacent abrupt change points are analyzed to determine the instantaneous frequency jump characteristics. The jump interval in this characteristic refers to the time difference between adjacent abrupt change points, and the frequency change amplitude is the difference in instantaneous frequency values ​​at adjacent abrupt change points.

[0085] For example, in the discrete point set {(t11, f11), (t12, f12), (t13, f13)...}, if a significant change in instantaneous frequency value is detected at time point t12 from f11 to f12, then t12 is a sudden change point. Then, another sudden change point is detected at t15. The time difference between t12 and t15 is the jump interval, and the difference between f15 and f12 is the frequency change amplitude within this interval. Together, they constitute the instantaneous frequency jump characteristic of this interval.

[0086] Step S1231: Perform differential calculation on the instantaneous frequency values ​​in the discrete point set to obtain the frequency change at adjacent time points.

[0087] When performing abrupt change detection on a set of discrete points, the first step is to perform difference calculation. For each instantaneous frequency value in the set of discrete points, the difference between it and the instantaneous frequency value at the previous time point is calculated. This difference represents the frequency change at adjacent time points.

[0088] For example, for instantaneous frequency values ​​f11, f12, f13, ... in a discrete set of points, subtracting f11 from f12 yields the first frequency change Δf1, subtracting f12 from f13 yields the second frequency change Δf2, and so on, resulting in a series of frequency changes Δf1, Δf2, Δf3, ... These frequency changes reflect the changes in instantaneous frequency between adjacent time points.

[0089] Step S1232: When the frequency change at any time point exceeds the frequency change threshold, mark that time point as a potential mutation point.

[0090] After obtaining the frequency change, a frequency change threshold needs to be set. This threshold is used to determine whether the frequency change is significant. When the frequency change at a certain time point exceeds this threshold, it indicates that the instantaneous frequency at that time point has changed significantly, and this time point is marked as a potential mutation point.

[0091] For example, if the frequency change threshold is set to Δf0, when the frequency change Δf2 exceeds Δf0, then the time point t13 corresponding to Δf2 is marked as a potential mutation point; similarly, other time points corresponding to frequency changes exceeding the threshold are also marked as potential mutation points.

[0092] Step S1233: Verify the potential mutation point by calculating the average frequency within a preset time window before and after the potential mutation point. When the difference between the average values ​​before and after the mutation point is consistent with the trend of the frequency change, the time point is confirmed as a mutation point.

[0093] After identifying potential mutation points, they need to be validated to rule out false positives. The validation method involves setting a preset time window before and after the potential mutation point and calculating the average instantaneous frequency within each window.

[0094] For example, for a potential mutation point t13, a time window [t13-Δt, t13] is set before it, and the mean instantaneous frequency μ within this window is calculated; a time window [t13, t13+Δt] is set after it, and the mean instantaneous frequency μ within this window is calculated. Then, the difference between μ before and μ after is compared. If the difference is consistent with the trend of the frequency change Δf2 at the potential mutation point (i.e., both are positive or both are negative), then the potential mutation point is considered a real mutation point and is confirmed; otherwise, it is judged as a false positive and the mark is removed.

[0095] Step S1234: Arrange all confirmed mutation points in chronological order to form a mutation point sequence. The time interval between two adjacent mutation points constitutes a frequency variation segment.

[0096] After verifying and confirming the mutation points, all confirmed mutation points are arranged in chronological order to form a mutation point sequence. For example, if the confirmed mutation points are t12, t15, t18, etc., they are arranged in chronological order as [t12, t15, t18, etc.], which is the mutation point sequence.

[0097] The time interval between two adjacent abrupt change points constitutes a frequency change segment. For example, the time interval between t12 and t15 is one frequency change segment, and the time interval between t15 and t18 is another frequency change segment. The instantaneous frequency changes within each frequency change segment follow a certain pattern.

[0098] Step S1235: Analyze the instantaneous frequency value change trend within each frequency change segment, calculate the average rate of change and direction of change of the frequency change segment, and combine them with the corresponding time interval to determine the instantaneous frequency jump characteristics.

[0099] For each frequency variation segment, it is necessary to analyze the trend of instantaneous frequency values. Calculate the total change in instantaneous frequency values ​​within that segment, and then divide it by the duration of that segment to obtain the average rate of change. The direction of change is determined by the sign of the total change: a positive total change indicates a positive direction, and a negative total change indicates a negative direction.

[0100] Combining the average rate of change, the direction of change, and the corresponding time interval constitutes the instantaneous frequency jump characteristic of that frequency change segment. For example, within the frequency change segment [t12, t15], the total change is Δf_total, the time length is Δt_total, the average rate of change is Δf_total divided by Δt_total, and the direction of change is positive. Therefore, the instantaneous frequency jump characteristic of this segment includes the average rate of change, the positive direction, and the time interval [t12, t15].

[0101] Step S124: Calculate the instantaneous phase value of the signal within each time domain analysis window, construct a continuous curve of phase change over time, perform smoothness analysis on the continuous curve, and calculate the rate of curvature change of the continuous curve.

[0102] When capturing phase continuity characteristics, the instantaneous phase value of the signal is first calculated within each time-domain analysis window. After obtaining the analytic signal through Hilbert transform, the phase of the analytic signal is the instantaneous phase value.

[0103] Based on the calculated instantaneous phase value, a continuous curve showing the phase change over time is constructed with time on the horizontal axis and the instantaneous phase value on the vertical axis. Then, a smoothness analysis is performed on this continuous curve, calculating its rate of change of curvature. The rate of change of curvature reflects the degree of curvature change of the curve; the smaller the rate of change of curvature, the smoother the curve; conversely, the larger the rate of change of curvature, the less smooth the curve.

[0104] For example, within the time-domain analysis window W3, multiple instantaneous phase values ​​φ31, φ32, φ33... are calculated, corresponding to time points t31, t32, t33... After constructing a continuous curve, the rate of curvature change of the curve at each point is calculated as k31, k32, k33...

[0105] Step S125: Determine the phase continuity characteristics of the signal based on the curvature change rate. When the curvature change rate is lower than a preset threshold, it is determined that the signal phase has continuity within the corresponding time period; otherwise, it is determined that there is a phase breakpoint.

[0106] After obtaining the rate of change of curvature, a preset threshold is set to determine whether the phase is continuous. When the rate of change of curvature within a certain period is lower than the preset threshold, it indicates that the phase change curve over time is relatively smooth and the phase change is continuous, thus determining that the signal phase is continuous within the corresponding period. When the rate of change of curvature is higher than the preset threshold, it indicates that the curvature of the curve changes significantly and the phase change is discontinuous, thus determining that there is a phase discontinuity.

[0107] For example, with a preset threshold of k0, if the rate of curvature change k32 is lower than k0 within the time period [t32, t33] of the time domain analysis window W3, then the signal phase is determined to be continuous within this time period; while in the time period [t33, t34], the rate of curvature change k33 is higher than k0, then a phase discontinuity is determined to exist within this time period. These determinations together constitute the phase continuity characteristic of the signal.

[0108] Step S130: Construct a time-domain separation dynamic model based on the instantaneous frequency jump feature and the phase continuity feature. The time-domain separation dynamic model includes feature association weights of signals from different sources and time-domain boundary determination rules.

[0109] After capturing the instantaneous frequency jump characteristics and phase continuity characteristics, a time-domain separation dynamic model is constructed based on these characteristics. The role of this model is to formulate a separation strategy based on the characteristics of the signal, thereby achieving time-domain separation of signals from different sources in a mixed signal.

[0110] The model includes feature association weights for signals from different sources, used to measure the importance of instantaneous frequency jump features and phase continuity features in the separation process; the time-domain boundary determination rule is used to determine the boundary points between different source signals in the time domain. The combination of these two parts enables the model to dynamically adapt to changes in different signal characteristics, improving the accuracy of separation.

[0111] In the above scenario, based on the instantaneous frequency jump characteristics and phase continuity characteristics of the signals of device X, device Y and device Z, a corresponding time-domain separation dynamic model is constructed. This model can identify the boundaries of these three signals in the time domain.

[0112] Step S131: Perform cluster analysis on the instantaneous frequency jump features, and group features with similar jump intervals and frequency change amplitudes into the same frequency feature cluster. Each frequency feature cluster corresponds to the frequency characteristics of a type of signal source.

[0113] When constructing the time-domain separated dynamic model, the instantaneous frequency jump features are first subjected to cluster analysis. The cluster analysis is based on the jump interval and the frequency change amplitude, and instantaneous frequency jump features with similar jump intervals and frequency change amplitudes are grouped into the same frequency feature cluster.

[0114] Each frequency feature cluster represents the frequency characteristics of a class of signal sources, because signals from the same device often have similar instantaneous frequency jump characteristics. For example, the signal from device X has a specific jump interval and frequency change amplitude, and its related instantaneous frequency jump characteristics are classified into frequency feature cluster C1; the signal from device Y has another set of similar jump intervals and frequency change amplitudes, and is classified into frequency feature cluster C2; and the signal from device Z is classified into frequency feature cluster C3.

[0115] Step S132: Calculate the correlation degree of the phase continuity feature, analyze the similarity of phase curves in different time domain analysis windows, and generate a phase correlation matrix. The elements in the phase correlation matrix represent the degree of phase continuity between two windows.

[0116] The correlation degree of phase continuity characteristics is calculated to analyze the similarity of phase curves within different time-domain analysis windows. The correlation degree can be calculated by comparing factors such as the shape and trend of two phase curves. The higher the correlation degree, the better the similarity between the two curves and the higher the degree of phase continuity.

[0117] The correlation results between all pairwise time-domain analysis windows are arranged in matrix form to generate a phase correlation matrix. The rows and columns of the matrix correspond to different time-domain analysis windows, and the element Pij represents the phase continuity between the i-th and j-th windows. For example, the element P12 in the phase correlation matrix represents the phase continuity between windows W1 and W2; a larger value indicates a higher similarity and better continuity in their phase curves.

[0118] Step S133: Determine the feature association weights of signals from different sources based on the frequency feature clusters and the phase correlation matrix. The feature association weights are used to measure the degree of influence of frequency features and phase features in the separation process.

[0119] By combining frequency feature clusters and phase correlation matrices, feature correlation weights for signals from different sources are determined. The magnitude of the feature correlation weights reflects the degree of influence of frequency features (i.e., instantaneous frequency jump features) and phase features (i.e., phase continuity features) on the separation results during the separation process.

[0120] For a given frequency feature cluster, if the signal corresponding to it exhibits a high degree of phase continuity in the phase correlation matrix, the weight of the frequency feature represented by that cluster in the feature correlation weights may be adjusted accordingly to reflect its importance in separation. For example, if the signal corresponding to frequency feature cluster C1 has a high degree of phase continuity with multiple windows in the phase correlation matrix, then the feature correlation weight w1 assigned to C1 will be relatively large, indicating that the frequency feature has a significant impact on separating the source signal.

[0121] Step S1331: Assign an initial weight value to each frequency feature cluster, the initial weight value being determined based on the proportion of signal features in that frequency feature cluster.

[0122] When determining the feature association weights, an initial weight value is first assigned to each frequency feature cluster. The initial weight value is determined based on the proportion of the number of signal features in that frequency feature cluster to the total number of signal features in all frequency feature clusters.

[0123] For example, if the total number of signal features in all frequency feature clusters is N, and frequency feature cluster C1 contains n1 signal features, then the initial weight value w1initial of C1 is n1 / N; the number of signal features in frequency feature cluster C2 is n2, and its initial weight value w2initial is n2 / N, and so on. This method allows the initial weight values ​​to initially reflect the proportion of signal features in each frequency feature cluster.

[0124] Step S1332: Calculate the correspondence between each frequency feature cluster and the high correlation region in the phase correlation matrix. When any frequency feature cluster has a high degree of overlap with the high correlation region, increase the weight value of the frequency feature cluster. The high correlation region is the region with a correlation degree greater than the set correlation degree.

[0125] A correlation degree value is set as a boundary, and regions in the phase correlation matrix with a correlation degree greater than this value are identified as high correlation regions. Then, the correspondence between each frequency feature cluster and these high correlation regions is calculated, that is, the degree of temporal overlap between the signal features represented by the frequency feature cluster and the high correlation regions is analyzed.

[0126] When a frequency feature cluster has a high degree of overlap with a highly correlated region, it indicates that the signal corresponding to that frequency feature cluster has good phase continuity, and its frequency characteristics may be more representative during the separation process. Therefore, it is necessary to increase the weight value of that frequency feature cluster. For example, if the frequency feature cluster C1 has an 80% overlap with the highly correlated region, while other clusters have a low overlap, then the weight value of C1 should be increased by a certain percentage, such as 20%, based on the initial weight value.

[0127] Step S1333: Analyze the phase continuity characteristics of signals within frequency feature clusters. For frequency feature clusters whose phase continuity meets the set conditions, increase their weight ratio in feature association.

[0128] Define conditions for phase continuity, such as the proportion of phase-continuous periods of a signal within a certain frequency feature cluster exceeding a certain value. Analyze whether the phase continuity characteristics of the signal within each frequency feature cluster satisfy these conditions. For frequency feature clusters that satisfy the conditions, increase their weight in feature correlation.

[0129] For example, if the set condition is that the proportion of continuous phase time exceeds 70% and the proportion of continuous phase time of the signal in frequency feature cluster C2 is 75%, and the set condition is met, then the weight of C2 in feature association is increased, such as by 10% on the basis of the previously adjusted weight.

[0130] Step S1334: Construct a feature association weight adjustment function, which takes the stability and phase correlation of the frequency feature cluster as input parameters and outputs dynamically adjusted weight values.

[0131] A feature correlation weight adjustment function is constructed, whose input parameters include the stability of the frequency feature cluster and the phase correlation degree. The stability of the frequency feature cluster can be measured by the dispersion of the signal features within the cluster; the smaller the dispersion, the higher the stability. The phase correlation degree is taken from the correlation degree value related to the frequency feature cluster in the phase correlation matrix.

[0132] The specific form of the function can be set according to the actual situation. For example, the function can be an expression that comprehensively considers stability and phase correlation. When the stability is high and the phase correlation is large, the output weight value is relatively large; otherwise, it is smaller. By dynamically adjusting the weight value through this function, the adjusted weight value can better reflect the actual impact of frequency feature clusters in the separation process.

[0133] Step S1335: Assign the adjusted weight values ​​to the corresponding frequency feature clusters to form a feature association weight set for signals from different sources. This feature association weight set is updated in real time as the signal features change.

[0134] The adjusted weight values ​​obtained through the feature association weight adjustment function are assigned to the corresponding frequency feature clusters. For example, the adjusted weight values ​​w1 tune are assigned to C1, w2 tune to C2, etc. These weight values ​​together constitute the feature association weight set {w1 tune, w2 tune, w3 tune, ...} of signals from different sources.

[0135] As the signal characteristics in the mixed signal set change, such as the addition of new superimposed signal units or changes in the original signal characteristics, the feature association weight set will also be updated in real time to ensure that the weight values ​​can always accurately reflect the influence of each frequency feature cluster.

[0136] Step S134: Construct a time-domain boundary determination rule by combining feature correlation weights. The time-domain boundary determination rule stipulates that when the frequency jump feature changes across clusters and the phase correlation is lower than the critical value, it is determined to be the time-domain boundary point of signals from different sources.

[0137] After determining the feature association weights, a time-domain boundary determination rule is constructed by combining these weights. The core of this rule is to determine the boundary points in the time domain for signals from different sources.

[0138] When a frequency jump characteristic at a certain time point changes across clusters (i.e., a characteristic from one frequency feature cluster changes to a characteristic from another), and the phase correlation at that time point is below a set threshold, the time point is determined to be the time-domain boundary between signals from different sources according to the rules. For example, at time point t, the frequency jump characteristic changes from the characteristic of C1 to the characteristic of C2, and the phase correlation P at t is below the threshold P0, then t is determined to be the time-domain boundary between the signals of device X and device Y.

[0139] Step S135: Integrate the frequency feature clusters, phase correlation matrix, feature correlation weights, and time domain boundary determination rules to construct a time domain separation dynamic model for dynamically adjusting the separation strategy.

[0140] The frequency feature cluster, phase correlation matrix, feature correlation weight, and time domain boundary determination rules are integrated into an organic whole, namely the time domain separation dynamic model.

[0141] This model can dynamically adjust the separation strategy based on changes in the characteristics of the input mixed signal. For example, when new signal features are added, the model will re-perform cluster analysis, correlation calculation, and weight adjustment, and update the time-domain boundary determination rules to adapt to the new signal separation requirements, ensuring the accuracy and reliability of the separation results.

[0142] Step S140: Perform iterative boundary optimization processing on the mixed signal set through the time-domain separation dynamic model to determine the independent distribution range of each source signal in the time domain, and obtain multiple single-source non-cooperative communication signal units.

[0143] Using the established temporal separation dynamic model, iterative boundary optimization is performed on the mixed signal set. This process continuously detects and adjusts the temporal boundary points to make the determined boundary points more accurate, thereby enabling the determination of the independent distribution range of each source signal in the temporal domain, ultimately resulting in multiple single-source non-cooperative communication signal units.

[0144] In the above scenario, the model analyzes the signal superposition areas of device X, device Y and device Z in the mixed signal set. Through multiple iterations to optimize the boundary, it gradually clarifies the independent distribution range of each device signal and separates them from the mixed signal.

[0145] Step S141: Call the initial boundary detection module of the time-domain separation dynamic model, and preliminarily identify potential time-domain boundary points in the mixed signal set based on the time-domain boundary determination rules to obtain the initial boundary point set.

[0146] The initial boundary detection module of the time-domain separation dynamic model is started. This module will perform a comprehensive scan of the mixed signal set according to the time-domain boundary determination rules already built in the model, and initially identify the positions of the time-domain boundary points that may be signals from different sources. These positions constitute the initial boundary point set.

[0147] For example, in a mixed signal set, the frequency jump characteristics at time points t21, t23, and t25 are found to have cross-cluster changes and the phase correlation is lower than the critical value. According to the time domain boundary determination rules, these time points are initially identified as potential time domain boundary points, forming an initial boundary point set {t21, t23, t25}.

[0148] Step S142: Divide the mixed signal set into multiple preliminary separation intervals according to the initial boundary point set, with each preliminary separation interval corresponding to a candidate signal unit.

[0149] Using the time points in the initial set of boundary points as boundaries, the mixed signal set is divided into multiple consecutive time intervals, i.e., preliminary separation intervals. The signal within each preliminary separation interval is considered a candidate signal unit. This unit may contain signals from a single source, or there may still be signal superposition, which requires further verification.

[0150] For example, based on the initial set of boundary points {t21, t23, t25}, the mixed signal set is divided into preliminary separation intervals [t_start, t21), [t21, t23), [t23, t25), and [t25, t_end]. Each interval corresponds to a candidate signal unit, which are denoted as S1, S2, S3, and S4, respectively.

[0151] Step S143: Extract the instantaneous frequency jump features and phase continuity features of each candidate signal unit, input them into the feature verification module of the time-domain separation dynamic model, and calculate the matching degree between the instantaneous frequency jump features and phase continuity features and the corresponding frequency feature clusters.

[0152] For each candidate signal unit, its instantaneous frequency jump features and phase continuity features are re-extracted, using the same extraction method as described in step S120. These extracted features are then input into the feature verification module of the time-domain separation dynamic model.

[0153] The feature verification module compares the features of candidate signal units with the frequency feature clusters in the model and calculates the matching degree. The matching degree calculation can comprehensively consider factors such as transition interval, frequency change amplitude, and phase continuity. The higher the matching degree, the better the consistency between the features of the candidate signal unit and the corresponding frequency feature cluster.

[0154] For example, after extracting the instantaneous frequency jump features and phase continuity features of candidate signal unit S1 and inputting them into the feature verification module, it is calculated that its matching degree with frequency feature cluster C1 is 85%, its matching degree with C2 is 30%, and its matching degree with C3 is 20%.

[0155] Step S144: When the matching degree is lower than the preset standard, start the model iterative optimization mechanism, adjust the feature association weights and re-detect the time domain boundary point to form a new set of boundary points.

[0156] Set a preset standard for matching degree, such as 70%. When the highest matching degree between the features of a candidate signal unit and the corresponding frequency feature cluster is lower than the preset standard, it indicates that the candidate signal unit may contain signals from multiple sources, or the position of the initial boundary point is inaccurate, and the model iterative optimization mechanism needs to be activated.

[0157] The iterative optimization mechanism first adjusts the feature association weights, for example, by increasing or decreasing the weight values ​​of certain frequency feature clusters, to change the model's attention to different features. Then, based on the adjusted weight values, it re-detects temporal boundary points and identifies new potential boundary points by reapplying the temporal boundary determination rules, forming a new set of boundary points.

[0158] For example, the highest matching degree of candidate signal unit S2 is 65%, which is lower than the preset standard of 70%, so the iterative optimization mechanism is initiated. After adjusting the feature association weights, a new set of boundary points {t21, t22, t23, t25} is obtained by re-detection.

[0159] Step S145: Repeat the boundary point detection, interval division and feature verification steps until the feature matching degree of all candidate signal units meets the preset standard, and determine the corresponding candidate signal unit as a single-source non-cooperative communication signal unit.

[0160] The initial separation intervals are redefined based on the new set of boundary points to obtain new candidate signal units, and features are extracted again for matching degree calculation. This process is repeated continuously, namely boundary point detection, interval division, and feature verification, until the matching degree between the features of all candidate signal units and their corresponding frequency feature clusters reaches the preset standard.

[0161] When the matching degree of all candidate signal units meets the requirements, it means that these candidate signal units can represent signals from a single source, and they are identified as single-source non-cooperative communication signal units. For example, after multiple iterations, the matching degree of all candidate signal units exceeds 70%, where S1 corresponds to the signal of device X, S2 corresponds to the signal of device Y, and S3 corresponds to the signal of device Z. These units are identified as single-source non-cooperative communication signal units.

[0162] Step S150: The multiple single-source non-cooperative communication signal units are spliced ​​and integrated according to the order of their time-domain distribution intervals, and the time-domain separation result set of the non-cooperative communication signals is output.

[0163] After obtaining multiple single-source non-cooperative communication signal units, they need to be spliced ​​and integrated according to their chronological order of distribution in the time domain. The splicing and integration process involves connecting each signal unit sequentially along the time axis to form a complete signal sequence. This sequence contains all the separated single-source signals and maintains their temporal order.

[0164] For example, the time-domain distribution intervals corresponding to the non-cooperative communication signal units from a single source are [t_start, t21), [t21, t23), and [t23, t_end], which correspond to the signal units of device X, device Y, and device Z, respectively. They are spliced ​​together in chronological order to form a time-domain separation result set. This set can clearly show the distribution of each device signal in the time domain. Finally, this result set is output to complete the time-domain separation process of the non-cooperative communication signal.

[0165] Throughout the process, signal acquisition and processing are involved, which may involve some communication signal-related information. However, this method only analyzes and processes the characteristics of the signal, without parsing the signal content. Furthermore, conventional signal reception techniques are used during signal acquisition, and no privacy-sensitive data is collected; therefore, there is no issue of privacy leakage. In addition, all processing steps of this method comply with standard practices in the field of communication technology and do not violate laws, social ethics, or harm public interests.

[0166] In the above embodiments, the time-domain separation system for non-cooperative communication signals used to perform the above method embodiments has at least one processor, a control module (chipset) coupled to at least one of the processors, a memory coupled to the control module, a non-volatile memory (NVM) / storage device coupled to the control module, at least one load to / output device coupled to the control module, and a network interface coupled to the control module.

[0167] The processor may include at least one single-core or multi-core processor, and may include any combination of general-purpose processors or special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). For some alternative implementations, a time-domain separation system for non-cooperative communication signals can serve as the gateway or other electronic device described in the embodiments of this application.

[0168] In some alternative implementations, a time-domain separation system for non-cooperative communication signals may include at least one computer-readable medium (e.g., a memory or NVM / storage device) having instructions and at least one processor fused with the at least one computer-readable medium and configured to execute the instructions to implement the module thereby performing the actions described in this disclosure.

[0169] In one embodiment, the control module may include any suitable interface controller to provide any suitable interface to at least one of the processors and / or any suitable device or component communicating with the control module.

[0170] The control module may include a memory controller module to provide an interface to the memory. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0171] The memory can be used, for example, to load and store data and / or instructions for a time-domain separation system for non-cooperative communication signals. In one embodiment, the memory may include any suitable volatile memory, such as suitable DRAM.

[0172] In one embodiment, the control module may include at least one load-to-output controller to provide an interface to the NVM / storage device and (at least one) load-to-output device.

[0173] For example, an NVM / storage device can be used to store data and / or instructions. An NVM / storage device may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (at least one) non-volatile storage device (e.g., at least one hard disk drive (HDD), at least one optical disc (CD) drive, and / or at least one digital universal optical disc (DVD) drive).

[0174] NVM / storage devices may include storage resources that are physically part of a device mounted on which a time-domain separation system for non-cooperative communication signals is installed, or that can be accessed by the device without needing to be part of the device. For example, an NVM / storage device may be accessed over a network via at least one load-to-output device.

[0175] At least one loading / output device may provide an interface for the time-domain separation system for non-cooperative communication signals to communicate with any other suitable device. The loading / output device may include communication components, phonetic components, sensor components, etc. A network interface may provide an interface for the time-domain separation system for non-cooperative communication signals to communicate based on at least one network. The time-domain separation system for non-cooperative communication signals may wirelessly communicate with at least one component of a wireless network based on at least one wireless network prior and / or protocol, such as accessing a communication prior-based wireless network.

[0176] In one embodiment, at least one of the processors may be integrated with the logic of at least one controller of the control module (e.g., a memory controller module). In one embodiment, at least one of the processors may be integrated with the logic of at least one controller of the control module to form a system-level integration. In one embodiment, at least one of the processors may be fused with the logic of at least one controller of the control module on the same die. In one embodiment, at least one of the processors may be fused with the logic of at least one controller of the control module on the same die to form a system-on-a-chip (SoC).

[0177] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

[0178] This invention discloses a computer read storage medium that stores a computer program for electronic data interchange, wherein the computer program causes a computer to execute the steps in the time-domain separation method for non-cooperative communication signals described in the foregoing embodiments.

[0179] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the time-domain separation method for non-cooperative communication signals described in the foregoing embodiments.

[0180] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0181] Finally, it should be noted that the above-disclosed embodiments are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A time-domain separation method for non-cooperative communication signals, characterized in that, The method includes: Obtain a mixed signal set in a non-cooperative communication scenario. The mixed signal set contains superimposed signal units of non-cooperative communication signals from multiple different sources in the time domain, and each superimposed signal unit carries a time domain feature marker of asynchronous transmission. Dynamic feature correlation processing is performed on the mixed signal set to capture the instantaneous frequency jump feature and phase continuity feature of each superimposed signal unit. The instantaneous frequency jump feature is used to describe the frequency change law of the signal in the time domain, and the phase continuity feature is used to reflect the coherent change trend of the signal phase over time. A time-domain separation dynamic model is constructed based on the instantaneous frequency jump characteristics and the phase continuity characteristics. The time-domain separation dynamic model includes feature association weights of signals from different sources and time-domain boundary determination rules. The mixed signal set is iteratively optimized using the time-domain separation dynamic model to determine the independent distribution range of each source signal in the time domain, thereby obtaining multiple single-source non-cooperative communication signal units. The multiple single-source non-cooperative communication signal units are spliced ​​and integrated according to the order of their time-domain distribution intervals to output a set of time-domain separation results of non-cooperative communication signals.

2. The time-domain separation method for non-cooperative communication signals according to claim 1, characterized in that, The acquisition of the mixed signal set in the non-cooperative communication scenario includes: The composite electrical signal in a non-cooperative communication scenario is captured by a broadband receiving device. The composite electrical signal contains non-cooperative communication signal components of different frequency bands and different modulation methods. The composite electrical signal is subjected to time-frequency domain pre-decomposition processing to decompose the composite electrical signal into multiple signal components with time-frequency focusing characteristics, each signal component corresponding to a specific time-frequency distribution range; Each signal component is time-axis calibrated, and the time-axis calibrated signal components are superimposed and recombined to form a signal sequence containing time-domain superposition information. Each signal sequence unit corresponds to the signal superposition state within a time segment. The signal sequence is determined as a set of mixed signals in a non-cooperative communication scenario, and the superimposed signal units in the mixed signal set retain the original time-frequency characteristics of each source signal.

3. The time-domain separation method for non-cooperative communication signals according to claim 2, characterized in that, The step of performing time-frequency domain pre-decomposition processing on the composite electrical signal, which decomposes the composite electrical signal into multiple signal components with time-frequency focusing characteristics, includes: The composite electrical signal is decomposed using a variational mode decomposition algorithm. The number of decomposition modes is set to decompose the composite electrical signal into multiple intrinsic mode functions. Time-frequency analysis is performed on each intrinsic mode function, and a corresponding time-frequency distribution map is generated by short-time Fourier transform. The time-frequency distribution map reflects the energy distribution of the signal in the time and frequency dimensions. Extract the energy accumulation region from the time-frequency distribution map, and determine the frequency range and time range corresponding to the energy accumulation region as the time-frequency focusing range of the intrinsic mode function; The intrinsic mode functions are screened based on the time-frequency focusing range, and intrinsic mode functions with clear time-frequency focusing ranges and no significant overlap are retained as effective signal components; The effective signal components are arranged in chronological order to form multiple signal component sequences with time-frequency focusing characteristics.

4. The time-domain separation method for non-cooperative communication signals according to claim 2, characterized in that, The time-axis calibration process for each signal component includes: An initial time stamp is added to each signal component, the initial time stamp being generated based on the signal reception time; Extract feature points containing time reference information from each signal component; the feature points are peak points of signal amplitude or frequency jump points. Align the feature points of all signal components on the time axis, take the time of the feature point of one signal component as the reference time, and calculate the difference between the feature points of other signal components and the reference time as the time deviation. The initial time marker of the corresponding signal component is corrected according to the time deviation, and the corrected signal component is subjected to time interpolation to fill the signal gap caused by the time marker correction.

5. The time-domain separation method for non-cooperative communication signals according to claim 1, characterized in that, The dynamic feature correlation processing performed on the mixed signal set to capture the instantaneous frequency jump characteristics and phase continuity characteristics of each superimposed signal unit includes: Each superimposed signal unit in the mixed signal set is divided into multiple time-domain analysis windows. Each time-domain analysis window has a variable time length, and the time length is adaptively adjusted according to the rate of change of the signal frequency. For each time-domain analysis window, instantaneous frequency extraction is performed on the signal. The instantaneous frequency value of the signal is calculated through Hilbert transform, and the set of discrete points where the instantaneous frequency value changes over time is recorded. The discrete point set is subjected to abrupt change detection processing to identify the time position where the instantaneous frequency value changes significantly, and the frequency change pattern between adjacent abrupt change points is determined as the instantaneous frequency jump feature. The instantaneous frequency jump feature includes the jump interval and the frequency change amplitude. The instantaneous phase value of the signal is calculated within each time domain analysis window, a continuous curve of phase change over time is constructed, smoothness analysis is performed on the continuous curve, and the rate of change of curvature of the continuous curve is calculated. The phase continuity characteristics of the signal are determined based on the rate of curvature change. When the rate of curvature change is lower than a preset threshold, it is determined that the signal phase is continuous within the corresponding time period; otherwise, it is determined that there is a phase breakpoint.

6. The time-domain separation method for non-cooperative communication signals according to claim 5, characterized in that, The step of performing abrupt change point detection processing on the discrete point set, identifying the time locations where significant changes in instantaneous frequency values ​​occur, and determining the frequency change pattern between adjacent abrupt change points as instantaneous frequency jump features includes: The frequency change at adjacent time points is obtained by performing differential calculations on the instantaneous frequency values ​​in the discrete point set. When the frequency change at any point in time exceeds the frequency change threshold, that point in time is marked as a potential mutation point. The potential mutation point is verified by calculating the average frequency within a preset time window before and after the potential mutation point. When the difference between the average values ​​before and after the mutation point is consistent with the trend of the frequency change, the time point is confirmed as a mutation point. All confirmed mutation points are arranged in chronological order to form a mutation point sequence. The time interval between two adjacent mutation points constitutes a frequency variation segment. Analyze the instantaneous frequency value change trend within each frequency change segment, calculate the average rate of change and direction of change of the frequency change segment, and combine them with the corresponding time interval to determine the instantaneous frequency jump characteristics.

7. The time-domain separation method for non-cooperative communication signals according to claim 1, characterized in that, The construction of the time-domain separated dynamic model based on the instantaneous frequency jump characteristics and the phase continuity characteristics includes: Cluster analysis is performed on the instantaneous frequency jump features to group features with similar jump intervals and frequency change amplitudes into the same frequency feature cluster, and each frequency feature cluster corresponds to the frequency characteristics of a type of signal source. The correlation degree of the phase continuity feature is calculated, the similarity of phase curves in different time domain analysis windows is analyzed, and a phase correlation matrix is ​​generated. The elements in the phase correlation matrix represent the degree of phase continuity between two windows. Based on the frequency feature cluster and the phase correlation matrix, the feature correlation weights of signals from different sources are determined. The feature correlation weights are used to measure the degree of influence of frequency features and phase features in the separation process. A time-domain boundary determination rule is constructed by combining feature correlation weights. The time-domain boundary determination rule stipulates that when the frequency jump feature changes across clusters and the phase correlation degree is lower than the critical value, it is determined to be the time-domain boundary point of signals from different sources. By integrating the frequency feature clusters, phase correlation matrix, feature correlation weights, and time domain boundary determination rules, a time domain separation dynamic model is constructed for dynamically adjusting the separation strategy.

8. The time-domain separation method for non-cooperative communication signals according to claim 7, characterized in that, The step of determining the feature correlation weights of signals from different sources based on the frequency feature clusters and the phase correlation matrix includes: An initial weight value is assigned to each frequency feature cluster, and the initial weight value is determined based on the proportion of signal features in that frequency feature cluster. Calculate the correspondence between each frequency feature cluster and the high correlation region in the phase correlation matrix. When any frequency feature cluster has a high degree of overlap with the high correlation region, increase the weight value of that frequency feature cluster. The high correlation region is the region with a correlation degree greater than a set correlation degree. Analyze the phase continuity characteristics of signals within frequency feature clusters, and increase the weight ratio of frequency feature clusters whose phase continuity meets the set conditions in feature correlation. A feature association weight adjustment function is constructed, which takes the stability and phase correlation of frequency feature clusters as input parameters and outputs dynamically adjusted weight values. The adjusted weight values ​​are assigned to the corresponding frequency feature clusters to form a feature association weight set for signals from different sources. This feature association weight set is updated in real time as the signal features change.

9. The time-domain separation method for non-cooperative communication signals according to claim 1, characterized in that, The iterative boundary optimization process performed on the mixed signal set using the time-domain separation dynamic model determines the independent distribution interval of each source signal in the time domain, resulting in multiple single-source non-cooperative communication signal units, including: The initial boundary detection module of the time-domain separation dynamic model is invoked to initially identify potential time-domain boundary points in the mixed signal set based on the time-domain boundary determination rules, thereby obtaining an initial boundary point set; The mixed signal set is divided into multiple preliminary separation intervals based on the initial set of boundary points, and each preliminary separation interval corresponds to a candidate signal unit. Extract the instantaneous frequency jump features and phase continuity features of each candidate signal unit, input them into the feature verification module of the time-domain separation dynamic model, and calculate the matching degree between the instantaneous frequency jump features and phase continuity features and the corresponding frequency feature clusters; When the matching degree is lower than the preset standard, the model iterative optimization mechanism is activated to adjust the feature association weights and re-detect the time domain boundary point to form a new set of boundary points. Repeat the boundary point detection, interval division, and feature verification steps until the feature matching degree of all candidate signal units meets the preset standard, and then determine the corresponding candidate signal unit as a single-source non-cooperative communication signal unit.

10. A time-domain separation system for non-cooperative communication signals, characterized in that, The method includes a processor and a computer-readable storage medium storing machine-executable instructions that, when executed by a computer, implement the time-domain separation method for non-cooperative communication signals as described in any one of claims 1-9.

Images