A method for radio frequency interference optimization based on compliance constraints

Abstract

The present application provides a kind of based on compliance constraint radio frequency interference optimization method, method includes: obtaining wideband radio frequency signal and carrying out time-frequency analysis, by calculating the normalized abnormal score of each described frequency point in time window, abnormal frequency point is identified and extracted, and abnormal frequency point set is formed;Calculate trajectory connection cost, and the frequency hopping trajectory set constructed according to the trajectory connection cost;Based on frequency risk weight, overall risk assessment and local high-risk frequency band pruning are carried out to the frequency hopping trajectory set, and compliance trajectory set is obtained;According to the available resources of current system, the processing cost of each compliance trajectory of the compliance trajectory set is evaluated, and the executable trajectory in the current scheduling period is screened out as executable trajectory set;For the executable trajectory set, corresponding frequency index and abnormal score are generated interference parameters.The present application solves the systematic defects of prior art in real-time, generalization, compliance executability and energy economy.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of radio frequency interference, and particularly relates to a radio frequency interference optimization method based on compliance constraints. BACKGROUND

[0002] In recent years, low-cost small unmanned aerial vehicles have significantly increased activities in complex electromagnetic environments, and their control links often transmit in general frequency bands such as 2.4 GHz and 5.8 GHz in a low-power, frequency hopping, and bandwidth widening manner. Some models also use cellular networks for backhaul to enhance concealment.

[0003] Traditional countermeasures systems rely on rule templates, static thresholds, or radio frequency fingerprints for identification, and implement interference with fixed parameters in broadband suppression. In high-density communication scenarios such as cities, ports, and energy facilities, it is easy to cause spectrum overlap with wireless local area networks, Bluetooth, Internet of Things, and public safety communications, resulting in problems such as false alarms, missed alarms, and accompanying interference. Even if methods such as spectrum graph classification are introduced, there are still two common bottlenecks: one is the long processing chain, which is difficult to complete the "discovery-disposal" conversion in milliseconds under the real-time constraints of 100 MHz bandwidth and 100 million samples per second; the other is the insufficient extrapolation ability for unknown protocols and spectrum agile behavior, which is prone to unstable identification when facing frequency hopping and non-steady low-power signals.

[0004] More importantly, existing engineering solutions generally separate the implementation of "discovering abnormalities" and "executing interference", lack compliance clipping based on trajectories and execution access based on resource constraints, making it difficult to avoid interference near protected frequency bands in a timely manner, and difficult to ensure timely transmission when computational power is tight or concurrency is high. Ultimately, it shows long response delay, low energy utilization, and large disturbance to friendly communication.

[0005] In summary, there is an urgent need for an engineering method for complex electromagnetic environments that can quickly extract abnormal frequency points from wideband radio frequency data and reconstruct frequency hopping trajectories, perform fine-grained clipping on the trajectories within regulatory boundaries, and complete the immediate generation and delivery of interference parameters under limited computational power and strict time limits to achieve precise and low-side effect disposal of unknown, spectrum agile, and low-power control links. SUMMARY

[0006] The purpose of the present application is to design a radio frequency interference optimization method based on compliance constraints to solve the above problems.

[0007] In order to achieve the above purpose, the present application provides a radio frequency interference optimization method based on compliance constraints, which comprises:

[0008] S1, acquire a wideband radio frequency signal and perform time-frequency analysis to obtain a plurality of frequency points, identify and extract abnormal frequency points by calculating the normalized anomaly score of each frequency point in a time window, and form an abnormal frequency point set; wherein the abnormal frequency point set contains a plurality of abnormal frequency points, and the frequency point triplet of the abnormal frequency point includes time period number, frequency index and anomaly score;

[0009] S2, based on the abnormal frequency point set, calculate the trajectory connection cost by evaluating the time period number, frequency index and anomaly score between frequency points, and construct a frequency hopping trajectory set according to the trajectory connection cost, wherein each trajectory is a group of frequency point triplets arranged in ascending order of time;

[0010] S3, based on the pre-defined frequency risk weight, perform overall risk assessment and local high-risk frequency band clipping on the frequency hopping trajectory set to obtain a compliant trajectory set;

[0011] S4, according to the available resources of the current system, evaluate the processing cost of each compliant trajectory in the compliant trajectory set, and select the trajectories executable in the current scheduling period as the executable trajectory set;

[0012] S5, for the executable trajectory set, generate interference parameters combined with the corresponding frequency index and anomaly score; wherein the interference parameters include center frequency, dynamic adjustment bandwidth, modulation mode and duty cycle.

[0013] Further, the time-frequency analysis is specifically:

[0014] After windowing and segmenting the wideband radio frequency signal, perform short-time Fourier transform using the overlapping window method to generate frequency points;

[0015] Wherein, the normalized anomaly score is calculated based on the frequency point, and the median of the modulus and the median of the absolute deviation in the frequency dimension, which represents the normalized anomaly score of a frequency point in a time period.

[0016] Further, the identification and extraction of abnormal frequency points to form an abnormal frequency point set are specifically:

[0017] If the normalized anomaly score exceeds a fixed threshold, it is determined that the point has abnormal energy behavior, and it is recorded as an abnormal frequency point, and combined with the anomaly score of the abnormal frequency point, an abnormal frequency point set is formed.

[0018] Further, the two evaluation frequency points in the trajectory connection cost need to meet the trajectory direction forward;

[0019] Wherein, the frequency hopping trajectory set constructed according to the trajectory connection cost specifically includes:

[0020] Selecting any two abnormal frequency points, arranging the abnormal frequency points in ascending order according to time period number, searching for a frequency point set with a time difference not exceeding a maximum time difference window after each abnormal frequency point, and calculating a trajectory connection cost;

[0021] If the trajectory connection cost of the first abnormal frequency point and the second abnormal frequency point is less than a preset threshold, the second abnormal frequency point is connected to the trajectory initiated by the first abnormal frequency point; wherein the direction from the first abnormal frequency point to the second abnormal frequency point satisfies the forward direction of the trajectory.

[0022] After the connection is completed, the system performs a simplification operation on all initial trajectories: removing trajectories with insufficient length , merging short trajectories with high time coincidence within adjacent frequency bands, and assigning a total score to each trajectory based on the abnormal score to represent the overall credibility;

[0023] If the total score is less than a preset threshold, the corresponding trajectory is directly removed.

[0024] Further, the calculation of the trajectory connection cost further includes a structure penalty term, which measures the deviation of the unit time frequency hopping rate between the two frequency points from the average frequency hopping rate; the average frequency hopping rate is the expected frequency hopping rate manually set according to the historical known target frequency spectrum behavior in the system.

[0025] Further, the frequency risk weight is a risk coefficient on each abnormal frequency point, with a value range .

[0026] Further, the S3 includes:

[0027] Obtaining the set of frequency hopping trajectories, wherein each frequency hopping trajectory is represented as a set of points arranged in ascending order of time;

[0028] Based on the set of frequency hopping trajectories, the total number of points in the frequency hopping trajectory, the frequency risk weight, and the corresponding abnormal score value are used to calculate the trajectory-level risk integral of each frequency hopping trajectory;

[0029] When the trajectory-level risk integral exceeds a set threshold, the frequency hopping trajectory is considered to fall in a sensitive frequency band as a whole and is removed; otherwise, it enters the local frequency band pruning;

[0030] Finally, output the compliant trajectory set.

[0031] Further, the local frequency band pruning is:

[0032] Based on the set of frequency hopping trajectories, for each abnormal frequency point in the trajectory, it is judged whether the frequency risk weight corresponding to the abnormal frequency point is higher than a threshold value, if yes, a connection penalty term is added to the frequency point, instead of directly deleting the point, but reducing its trajectory continuity weight to form a structure breaking point; if the trajectory continuity weight of the frequency point is reduced to zero, the frequency point is deleted When the high-risk segment of a point is found, the system performs a breaking operation to divide the track into two parts, and the tracks before and after the segmentation point are numbered and saved respectively.

[0033] Further, the S4 comprises:

[0034] The available resources of the current system are obtained, and an execution time estimation function is constructed to generate an execution time delay estimation value of the compliant track set as a processing cost.

[0035] The interference modulation complexity score corresponding to the track is scored by a system lookup function for the modulation mode of each abnormal frequency point, and then the maximum value is taken in the track.

[0036] The time interval from the current time to the end of the earliest task processing is recorded, and when the processing cost is less than or equal to the time interval of the compliant track in the compliant track set, the compliant track is put into the executable track set and sorted according to the total score to obtain the executable track set.

[0037] Further, the S5 comprises:

[0038] Based on the executable track set, a weighted calculation mechanism is introduced to weight the contribution of each frequency point in the executable track according to its abnormal score weight, so that the frequency point with higher confidence occupies a larger proportion in the interference target frequency selection, and the center frequency is obtained.

[0039] The dynamic adjustment bandwidth is calculated based on the standard deviation of the track abnormal score and the track frequency span.

[0040] The interference parameters are generated in combination with the center frequency, the dynamic adjustment bandwidth, and the modulation mode and duty cycle.

[0041] The beneficial technical effects of the present application are at least the following points:

[0042] The present application constructs a closed engineering link with trajectory as the core around "RF interference optimization under compliance constraints": first, a robust abnormal frequency point extraction and frequency hopping trajectory connection mechanism is proposed for a label-free background, and a time-frequency-confidence joint metric is used to shorten the decision path from wideband sampling to trajectory formation; secondly, a trajectory-level compliance pruning method is proposed, which combines the frequency band risk mapping generated before deployment with the internal continuous segment identification of the trajectory, and removes the high-risk segments without destroying the frequency hopping structure, and retains the trajectory sub-segments that can be disposed of legally; thirdly, the execution resource judgment is introduced, which uniformly estimates the trajectory length, frequency span, interference waveform complexity and modulation switching overhead with the current available time window of the platform, forms the "can it be completed on time" access constraint, and avoids the engineering mismatch that the strategy is feasible but the execution is overtime; finally, the interference strategy optimization generation is carried out on the executable trajectory, through the confidence weighted center frequency positioning and the bandwidth setting of adaptive compression with score dispersion, combined with the templated modulation mode and duty cycle mapping, the narrowband or multicarrier interference parameter group that can be directly issued is output, so that the target link is efficiently suppressed in the range allowed by the regulations with the minimum necessary coverage. The above invention points organically link abnormal extraction, trajectory reconstruction, compliance pruning, resource access and parameter generation in a unified process, and solve the systematic defects of the prior art in real-time, generalization, compliance executability and energy economy. BRIEF DESCRIPTION OF DRAWINGS

[0043] The present application is further illustrated by the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present application. For ordinary skilled in the art, other drawings can be obtained without creative labor according to the following drawings.

[0044] Figure 1 A flowchart of a radio frequency interference optimization method based on compliance constraints according to the present application. DETAILED DESCRIPTION

[0045] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference signs represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present application, and cannot be understood as a limitation on the present application.

[0046] In one or more embodiments, as shown in Figure 1 a radio frequency interference optimization method based on compliance constraints is disclosed, the method comprising the following:

[0047] S1, acquiring a wideband radio frequency signal and performing time-frequency analysis to obtain a plurality of frequency points, identifying and extracting abnormal frequency points by calculating the normalized anomaly score of each frequency point in a time window, and forming an abnormal frequency point set; wherein the abnormal frequency point set contains a plurality of abnormal frequency points, and the frequency point triplet of the abnormal frequency point includes a time period number, a frequency index, and an anomaly score.

[0048] Specifically, the purpose of this step is to identify frequency points with abnormal energy characteristics from wideband radio signals, which are used as initial inputs for subsequent frequency hopping trajectory identification and interference strategy analysis. The signal sampling device used by the system is a general software radio receiving platform, which has a working frequency range of about seventy megahertz to six thousand megahertz, and a sampling rate adjustable between ten megapoints per second to one hundred megapoints per second. The data input used in this step is a continuous complex sampling sequence, denoted as , where is the sampling point number in a single analysis time window. The sampling sequence is segmented by the window length and sliding step set by the system for spectral analysis.

[0049] Further, after windowing and segmenting the signal sequence , the short-time Fourier transform is performed using the overlapping window method to generate frequency points that vary with time, denoted as , where represents the time period number, represents the frequency index. The Fourier transform uses the data block processed by the weighting window function, typically set to two hundred fifty-six point discrete Fourier transform, with a window overlap ratio of fifty percent. Each frequency component represents the complex signal modulus at time , frequency , which is a representation of spectral energy.

[0050] Further, to suppress false positives caused by background energy fluctuations, the spectral components of each time period are normalized. Within each time period , the median and absolute deviation of the modulus in the frequency dimension are calculated, denoted as and , respectively. Then the normalized anomaly score of a frequency point in this time period is:

[0051] ;

[0052] where, represents the complex modulus in the th time period, frequency position , calculated by short-time Fourier transform; represents the median of the modulus of all frequency points in this time period, The absolute deviation median is represented. This normalization method has the ability to resist background drift, and is suitable for urban radio environment, near airports or large industrial sites, and other electromagnetic complex areas.

[0053] The system sets a fixed threshold for abnormal scores (e.g. three), if a frequency point meets the condition after normalization, it is determined that the point has abnormal energy behavior, and it is recorded as an abnormal frequency point , and its abnormal score is kept , forming an abnormal frequency point set:

[0054] ;

[0055] Among them, represents the time period number of the frequency point; is the frequency index of the frequency point, and its specific value is determined by the system set center frequency and transform resolution; is the abnormal score of the frequency point, and the value comes from the actual calculation result of . Each abnormal frequency point represents that the spectrum amplitude in a specific time period has a mutation beyond the background statistical range, and has the potential characteristics of control signal.

[0056] Taking an actual deployment example, if the system sets the bandwidth to one hundred megahertz, the total sampling point number is one million points, corresponding to about ten milliseconds of signal length, and after the above processing process, hundreds to thousands of frequency points with local energy mutation characteristics can be extracted. In the unmanned aerial vehicle spectrum frequency concentrated interval (such as two point four gigahertz and five point eight gigahertz), some frequency points continuously appear with high abnormal score in multiple time windows, which constitutes the basis for subsequent frequency hopping trajectory construction. While other discontinuous and isolated frequency points are automatically excluded in the next step.

[0057] The output of this step is the abnormal frequency point set , which is a three-tuple set , time period number , frequency index and abnormal score , all of which come from the short-time Fourier transform result of the input sampling sequence , without any training sample or manual template setting. The score value will be used as the credibility basis in the subsequent trajectory clustering, directly affecting the target identification and priority sorting results.

[0058] S2, based on the set of abnormal frequency points, calculate the trajectory connection cost by evaluating the time period number, frequency index and abnormal score between frequency points, and construct a set of frequency hopping trajectories according to the trajectory connection cost, wherein each trajectory is a group of frequency point triplets arranged in ascending order of time.

[0059] Specifically, this step is a key structure extraction link in the whole invention method, and the target is to convert the set of abnormal frequency points obtained in the previous step into an ordered trajectory set that can represent potential frequency hopping communication behavior . Since the target of the present invention is agile spectrum, frequency hopping control link type unmanned aerial vehicle signal, such signals often have the characteristics of rapid time hopping, discontinuous frequency, and power lower than background noise, so it is difficult to accurately reconstruct the communication trajectory using conventional time-frequency clustering methods (such as simple time proximity or frequency segment aggregation). This step proposes a frequency hopping trajectory construction method based on abnormal score driving, combining three types of constraints of time consistency, frequency change pattern and abnormality significance, and through a trajectory distance function with a regular control item, the stable reconstruction of the spectrum hopping target structure in a high clutter environment is realized.

[0060] The input of this step is , obtained from the output of step one. Each element in the set contains three components: time period number , frequency index and abnormal score . The time number is directly obtained from the short-time Fourier segment number, the frequency index is calculated from the system center frequency and the discrete transform resolution, and the score is the normalized score value of the frequency point deviating from the background spectrum energy in the corresponding time window, which has been normalized in the previous step.

[0061] It should be noted that the core idea of trajectory construction is to connect these abnormal frequency points in time and frequency dimensions. Traditional methods mostly use Euclidean distance or fixed frequency shift to judge trajectory connectivity, but in frequency hopping communication, the frequency shift is no longer fixed, and the path may present nonlinear transition. Therefore, this method introduces a trajectory connection cost function with directionality restriction, abnormality significance weighting and asymmetric time penalty term:

[0062] ;

[0063] Wherein: represents the path cost required to connect frequency point to frequency point ; is the time period number corresponding to the two points, and i.e. the direction of the trajectory is forward; is the frequency index; is the anomaly score of the two points, representing the respective credibility; is a structure penalty term, used to limit the occurrence of non-hopping feature connections; is an adjustable coefficient, the system default setting is , , , , which can be adjusted at deployment.

[0064] wherein the structure penalty term is defined as follows:

[0065] ;

[0066] This term measures the deviation of the hopping rate per unit time between two frequency points from the average hopping rate . is the expected hopping rate (in frequency points per time window) manually set according to the historical known target spectrum behavior in the system, for example, set to , which means an average of two frequency units per time window. This term serves as a regular term to penalize connection paths that do not conform to common hopping behavior, thereby reducing the risk of false trajectory connections.

[0067] Further, all abnormal frequency points are arranged in ascending order of , and the system searches for a set of frequency points with a time difference of no more than behind each frequency point, and calculates , if there is , , the frequency point is connected to the trajectory initiated by the frequency point . After the connection is completed, the system performs a simplification operation on all initial trajectories: removes trajectories with a length of less than , merges short trajectories with high time coincidence within adjacent frequency bands, and assigns each trajectory a total score , representing the overall credibility of the trajectory:

[0068] ;

[0069] wherein represents the number of frequency points contained in the trajectory , is the anomaly score value of each point in the trajectory. This score serves as an important basis for subsequent compliance screening and interference priority allocation.

[0070] wherein if the total score assigned to the trajectory is less than the preset threshold, the trajectory is directly removed.

[0071] For example, in a real-world collection, the spectrum range covers 2.3-2.5 GHz, the total sampling time is 10 ms, and the number of outliers is 1,500. After being constructed by the method, the system forms 12 trajectories, of which 7 are typical low-speed frequency hopping behaviors, and 5 show high-frequency multi-segment hopping, which can be preliminarily corresponded to the control channel and the image transmission link. The trajectory connection method proposed in this step realizes the reconstruction of the time-sparse and frequency-irregular hopping path without relying on label training, and is particularly suitable for high-robustness unmanned aerial vehicle countermeasure systems facing spectrum-uncertain targets.

[0072] The output is a set of frequency hopping trajectories Each trajectory is a set of frequency point triplets arranged in ascending order of time .

[0073] S3, based on the pre-defined frequency risk weight, overall risk assessment and local high-risk frequency band clipping are performed on the set of frequency hopping trajectories to obtain a set of compliant trajectories.

[0074] Specifically, this step aims to perform compliance screening on the set of frequency hopping trajectories constructed in the previous step to ensure that the subsequent radio frequency interference strategy is only executed within the legal scope. This step is a key link in the "interference feasibility closed loop" of the entire invention system, which not only ensures that the interference behavior is legal and compliant, but also tries to preserve the complete structure of the frequency hopping communication trajectory, so that the system can implement precise strikes on unknown threat targets in complex spectrum environments while maintaining regulatory constraints. In actual deployment, spectrum resource use is subject to multiple regulations such as national telecommunications authorities, military wireless planning, and civil navigation and communication protection, especially in areas such as airports, power facilities, and government meeting places, where the compliance requirements for transmission behavior are extremely high. At the same time, modern unmanned aerial vehicles often use frequency hopping, spread spectrum, or multi-carrier communication methods, and their control link frequency distribution is not fixed, and some frequency points may cross protected communication bands. If a simple frequency whitelist is used for hard clipping, it is easy to cause the interruption of the trajectory structure, resulting in incomplete trajectory identification and continuous interference. Therefore, this step proposes a "trajectory-level compliant frequency band clipping mechanism" based on the combination of overall trajectory risk scoring and local segment weight weakening, which innovatively expands the compliance evaluation from frequency points to frequency hopping behavior structure.

[0075] The input of this step is a set of frequency hopping trajectories , where each trajectory is represented as a set of points arranged in ascending order of time:

[0076] ;

[0077] , where represents the time period of the th frequency point of the trajectory, is the frequency index of the frequency bin, is the anomaly score value, derived from the normalized energy deviation score in step one. The trajectory structure is aggregated by step two through the trajectory connection cost function with anomaly score, which has explicit frequency hopping characteristics and preliminary credibility guarantee.

[0078] Further, the first part of the compliance judgment is to quantify the risk level of each trajectory as a whole within the regulatory limited frequency band, and propose a trajectory-level "frequency legality risk score" indicator . This indicator considers the sensitivity of each trajectory's frequency position covered in the time-frequency space in the deployment area. The system uses the frequency risk weight function loaded before deployment, which defines the risk coefficient at each frequency point , with a value range , which is set as follows:

[0079] The weight of the frequency band explicitly prohibited by national regulations (such as the navigation signal frequency band) is set to ;

[0080] The military, police, and public safety communication frequency bands are set to ~ according to priority; ;

[0081] The active frequency band of the legal communication equipment in the environment (such as Wi-Fi, Bluetooth) is set to ~; ;

[0082] Other open frequency bands are set to .

[0083] The trajectory-level risk score is calculated as follows:

[0084] ;

[0085] In this formula: is the total number of points of the trajectory ; is the risk weight of the frequency bin in the frequency spectrum weight map; is the anomaly score value of the frequency bin; is a small constant to prevent the denominator from being zero, set to ; is the weight control factor of the anomaly score regularization term, used to reduce the compliance credibility of lower scoring points (such as ).

[0086] The innovation of the trajectory scoring item is to jointly include "compliance risk" and "abnormal credibility" into a unified scoring structure, preventing trajectories with low abnormal scores but in sensitive frequency bands from being misjudged as interferable targets due to scoring dilution. Unlike traditional methods that only mask the frequency whitelist, the present method has frequency sensitivity and abnormal intensity adjustment capability, and can better adapt to the judgment needs of uncertain target signals.

[0087] Further, when the trajectory-level risk integral exceeds a set threshold (such as ), it is considered that the trajectory as a whole falls in a sensitive frequency band and is rejected. Otherwise, it enters the local frequency band pruning process. In the local pruning process, the system does not delete the trajectory as a whole, but judges whether the risk weight of each frequency point in the trajectory on the spectrum weight map is higher than a threshold (such as ). If so, a connection penalty term is added to the frequency point, which does not directly delete the point, but reduces its trajectory continuity weight, forming a "structural breaking point". For a high-risk segment with more than points (such as ), the system performs a breaking operation to divide the trajectory into two parts, and the trajectories before and after the segmentation point are numbered and saved respectively.

[0088] For example, an original frequency hopping trajectory spans frequencies from 2.395GHz to 2.425GHz, and a certain segment at a frequency near 2.412GHz corresponds to (belonging to a local public safety communication channel) for 4 time window frequency points. Although the overall risk score of the trajectory does not exceed the full trajectory deletion threshold, because the number of continuous high-risk points exceeds the set threshold, the system triggers breaking and divides the trajectory into two sub-trajectories, which continue to participate in subsequent policy scoring, and the high-risk segment is no longer used for interference parameter generation.

[0089] The final output is a set of pruned compliance trajectories , each of which satisfies the following conditions: the overall risk score , and all high-risk frequency point segments have been deleted or segmented. Each trajectory still retains its time sequence structure and abnormal score value, providing a legal basis for subsequent system resource judgment and minimum interference strategy generation.

[0090] S4, according to the available resources of the current system, evaluate the processing cost of each compliant trajectory in the compliant trajectory set, and select the trajectories executable in the current scheduling period as the executable trajectory set.

[0091] Specifically, this step plays a key bridge role in connecting "target identification" and "interference execution" in the overall scheme of the application. The main goal is to evaluate the real-time performance and system load adaptability of each trajectory based on the compliance trajectory set , combined with the current system execution resource situation, to filter out the current "executable" trajectory set . The greatest innovation of this step is to introduce "trajectory structure complexity", "interference synthesis resource consumption" and "system load dynamic state" into a unified executability evaluation function, building a constraint judgment mechanism with hardware-task-signal coupling characteristics, ensuring the response closed loop of the system has controllability and stability.

[0092] The input of this step is , which is output by step three, representing the set of frequency hopping trajectories that are legally interferable after spectrum compliance pruning. Each trajectory is composed of several triplets, each containing a time index , a frequency index and an anomaly score , which come from the trajectory clustering algorithm in step two and the spectrum anomaly detection module in step one. Unlike the previous stage, this step no longer deals with the structure of the signal, but assesses whether it has the resources to complete processing and transmission within the time limit, i.e., whether it can "get there in time".

[0093] Further, considering that actual deployment systems usually use heterogeneous edge computing platforms, consisting of dedicated hardware (such as field programmable gate arrays), graphics processing modules (such as embedded graphics processing units) and radio frequency front ends (including modulators), the processing delay of the trajectory is subject to multiple factors, including but not limited to:

[0094] The length of the trajectory itself (i.e., the number of frequency points to be processed);

[0095] The frequency span of the trajectory (affecting the modulation range and frequency modulation overhead);

[0096] The complexity of the interference waveform corresponding to the frequency points involved in the trajectory (whether it is a multi-carrier or special modulation);

[0097] The current system state (number of parallel tasks, cache state, operation proportion);

[0098] This step builds an execution time estimation function that linearly combines multiple indicators to measure the processing cost of each trajectory :

[0099] ;

[0100] where: The execution time estimation of a track is calculated as the sum of the execution time estimations of all the tracks in the track set. The execution time estimation of a track is calculated as the sum of the execution time estimations of all the tracks in the track set. The length of a track is the number of frequency points in the track. The length of a track is the number of frequency points in the track. The frequency span of a track is the number of frequency points in the track. The frequency span of a track is the number of frequency points in the track. The interference modulation complexity score of a track is calculated by the system lookup table function for each frequency point's modulation mode, and then taking the maximum value in the track. The modulation switching overhead of a track is defined as the number of modulation mode changes of all adjacent frequency points in the track. The weight factor is set by the system to balance the influence of each index on the execution time.

[0101] The formula has the following creative design:

[0102] In traditional track processing time evaluation, only the length and frequency span of the track are considered, while the differences in interference synthesis are ignored. This method introduces two items, which respectively quantify "whether the track contains complex modulation frequency points" and "whether the track frequently switches between modulation modes", thus reflecting the modulation synthesis overhead. In complex scenarios, such as target unmanned aerial vehicle communication links based on frequency hopping spread spectrum superposition, different frequency points may correspond to different modulation structures (for example, part of the frequency offset disturbance and part of the phase disturbance). At this time, if the modulation overhead item is not introduced, the executability of the track will be incorrectly estimated, leading to strategy failure or system stall.

[0103] Further, the system updates the maximum time difference window for track connection search every second , which is calculated by the system scheduler according to the current task queue length, the occupancy rate of the graphics processing module, and the modulation module buffer condition. The system records the time interval from the current time to the end of the earliest task processing

[0104] , and makes the following determination:

[0105] ;

[0106] Tracks that meet the above conditions are considered "executable in this round of scheduling", and the remaining tracks will be moved to the waiting pool for reevaluation in the subsequent idle period.

[0107] Among the executable track set after resource checking (i.e. , the system scheduler maintains a priority queue. ​​Higher trajectory will be given higher execution priority, preferentially into the interference parameter generation link.

[0108] In a typical deployment scenario, for example, in the area where the UAV threat density is higher, the system processes about 30 trajectory tasks per second, the current GPU occupancy is 74%, the modulation module is synthesizing a multi-carrier three-carrier frequency jamming waveform, and the system calculates the current time window. A trajectory length of 10, a frequency span of 12 frequency units, a maximum modulation complexity of 1.5, and a modulation switching number of 3 are substituted to obtain , which is higher than the executable threshold, so it is not scheduled.

[0109] The output of this step is an executable trajectory set , the structure remains unchanged, each trajectory is still a time-ordered triple sequence, and carries an anomaly score , a frequency index and a trajectory number , which is used for the next step of interference parameter generation. All trajectories in this set have passed the resource constraint check, ensuring that subsequent execution will not fail due to timeout or module congestion.

[0110] S5, for the executable trajectory set, combined with the corresponding frequency index and anomaly score, generate interference parameters; wherein the interference parameters include center frequency, dynamic adjustment bandwidth, modulation mode and duty cycle.

[0111] Specifically, the executable trajectory set output by the previous stage is input, and for each trajectory , an actually executable interference parameter group is constructed, including center frequency, interference bandwidth, modulation mode and duty cycle, to finally form a control instruction for the interference module to call and execute. This step, as the last calculation link of the method chain of the present application, undertakes the conversion from "analysis and identification" to "executable control", and is the key output node in the structure closed loop of the invention. The scheme design emphasizes the principles of "trajectory feature driven", "resource state perception" and "minimum effective intervention within legal constraints", ensuring that the interference behavior is accurate, efficient and legal.

[0112] The input trajectory comes from the output set after resource judgment, and its structure is a frequency point triple sequence arranged in ascending order of time . Among them represents the frequency index, which is the position of the th frequency point in the trajectory; is the normalized anomaly score extracted in step one, indicating the degree of deviation of the frequency point from the background noise; represents the corresponding time period number. All trajectories have been confirmed to have execution conditions within the scope of compliance and resource capacity.

[0113] Further, in order to improve the target alignment of the interference center frequency, the system does not use the traditional equal weight average method, but introduces a score weighted calculation mechanism. Specifically, the contribution of each frequency point in the trajectory is weighted according to its abnormal score weight, so that the frequency point with higher confidence occupies a larger proportion in the selection of the interference target frequency. The calculation of the center frequency is as follows:

[0114] ;

[0115] The design of this formula has two points of invention: first, through score weighting, it can suppress the deviation phenomenon caused by the sudden increase of energy of the occasional frequency point; second, when the target frequency control channel is weak, it can concentrate limited energy on the most credible area to improve the probability of actually breaking the control link.

[0116] In terms of bandwidth calculation, considering that the target trajectory may cover multiple discontinuous frequency segments, if the maximum and minimum frequency point span is simply used as the interference bandwidth, it is likely to result in excessive coverage, interference with legal communication, and even energy waste. This step uses a "confidence adjustment compression" method to scale and control the frequency span. The calculation method is as follows:

[0117] ;

[0118] wherein, is the dynamically adjusted bandwidth; represents the standard deviation of the trajectory abnormal score, reflecting the fluctuation degree of the score, is a control coefficient (such as between and ). When the scores in the trajectory are concentrated, the system considers that the target is more explicit and allows a wider bandwidth coverage; on the contrary, if the scores fluctuate greatly, it indicates that the trajectory may have "fragmentation" or "pseudo-target", the system automatically shrinks the bandwidth to avoid invalid interference. This mechanism is particularly effective in complex urban spectrum environment, and can dynamically adapt the interference energy coverage range to improve the intelligence of the system.

[0119] The selection of the modulation mode and the duty cycle is not calculated by formula, but is completed by mapping the parameter template library loaded during deployment. The system classifies the trajectory structure according to the trajectory frequency hopping rate (calculated by the difference between and ), the score distribution pattern, and the frequency band where the trajectory frequency is located, and matches the applicable interference strategy template. For example:

[0120] For a trajectory with a center frequency at 2.4 GHz and a uniform score distribution, a constant phase perturbation template is selected;

[0121] For the trajectory with wide span and frequent jump, the multi-carrier disturbance template with intermittent emission is selected;

[0122] For the trajectory with high score and short time window, the short pulse high power strategy is adopted to improve the attack efficiency.

[0123] Finally, the system outputs a set of parameters , wherein: is the weighted center frequency, used to modulate the transmitter to lock the main frequency; is the dynamically adjusted bandwidth, used to set the output range of the bandpass modulator; is the modulation mode number, which defines the corresponding interference waveform structure in the system template library; is the duty cycle, used for power amplifier transmission control.

[0124] The above parameters will be issued to the modulator, power control module and antenna control unit through the instruction interface of the interference system, used for final interference waveform generation and transmission. Combined with the priority management logic of the system scheduling module, the target trajectory with high score, stable structure and reasonable resource demand is executed in priority, forming a complete, compliant and deployable radio frequency closed-loop attack chain of the present application.

[0125] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The program can be stored in a computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiments. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM), etc.

[0126] The above is the preferred embodiment of the present application. It should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the present application. These improvements and refinements are also considered within the scope of protection of the present application.

Claims

1. A method for radio frequency interference optimization based on compliance constraints, comprising: The method comprises: S1, acquiring a wideband radio frequency signal and performing time-frequency analysis to obtain a plurality of frequency points, identifying and extracting abnormal frequency points by calculating the normalized anomaly score of each frequency point in a time window, and forming an abnormal frequency point set; wherein the abnormal frequency point set contains a plurality of abnormal frequency points, and the frequency point triplet of the abnormal frequency point includes time period number, frequency index and abnormal score; S2, based on the abnormal frequency point set, the time period number, frequency index and abnormal score between the frequency points are evaluated to calculate the trajectory connection cost, and a frequency hopping trajectory set is constructed according to the trajectory connection cost, wherein each trajectory in the frequency hopping trajectory set is a group of frequency point triplets arranged in ascending order of time; S3, based on the pre-defined frequency risk weight, the overall risk of the frequency hopping trajectory set is evaluated and the local high-risk frequency band is cut, and a compliant trajectory set is obtained; S4, according to the available resources of the current system, the processing cost of each compliant trajectory in the compliant trajectory set is evaluated, and the trajectories executable in the current scheduling period are screened out as an executable trajectory set; S5, for the executable trajectory set, the corresponding frequency index and abnormal score are combined to generate an interference parameter; wherein the interference parameter includes center frequency, dynamic adjustment bandwidth, modulation mode and duty cycle.

2. The method of claim 1, wherein, The time-frequency analysis is specifically: After windowing and segmenting the wideband radio frequency signal, the short-time Fourier transform is performed using the overlapping window method to generate frequency points; Wherein, the normalized anomaly score is calculated based on the frequency point, the median of the module length and the median of the absolute deviation in the frequency dimension, and represents the normalized anomaly score of a frequency point in a time period.

3. The method of claim 2, wherein, The identification and extraction of abnormal frequency points to form an abnormal frequency point set are specifically: If the normalized anomaly score exceeds a fixed threshold, it is determined that the frequency point has abnormal energy behavior, and it is recorded as an abnormal frequency point, and an abnormal frequency point set is formed in combination with the abnormal score of the abnormal frequency point.

4. The method of claim 1, wherein, The two evaluation frequency points in the trajectory connection cost must satisfy the forward direction of the trajectory; Wherein, the construction of the frequency hopping trajectory set according to the trajectory connection cost specifically includes: Selecting any two abnormal frequency points, arranging the abnormal frequency points in ascending order of time period number, searching for a frequency point set with a time difference not exceeding a maximum time difference window behind each abnormal frequency point, and calculating the trajectory connection cost; If the trajectory connection cost of the first abnormal frequency point and the second abnormal frequency point is less than a preset threshold, the second abnormal frequency point is connected to the trajectory initiated by the first abnormal frequency point; wherein the direction from the first abnormal frequency point to the second abnormal frequency point satisfies the forward direction of the trajectory. After the connection is completed, the system performs a simplification operation on all initial trajectories: removing trajectories with insufficient length , merging short trajectories with high time coincidence within adjacent frequency bands, and assigning a total score to each trajectory based on the anomaly score to represent the overall credibility; If the total score is less than a preset threshold, the corresponding trajectory is directly excluded.

5. The method of claim 1, wherein, The calculation of the trajectory connection cost also includes a structure penalty term, which measures the deviation of the unit time frequency hopping rate between the two frequency points from the average frequency hopping rate; the average frequency hopping rate is the expected frequency hopping rate manually set according to the historical known target spectrum behavior in the system.

6. The method of claim 1, wherein, The frequency risk weight is a risk coefficient on each abnormal frequency point, and the value range is .

7. The method of claim 1, wherein, The S3 includes: Obtain the frequency hopping trajectory set, wherein each frequency hopping trajectory is represented as a point set arranged in ascending order of time; Based on the set of frequency hopping trajectories, the total number of points of the frequency hopping trajectory, the frequency risk weight, and the corresponding abnormal score value are used to calculate the trajectory-level risk integral of each frequency hopping trajectory; When the trajectory-level risk integral exceeds a set threshold, the frequency hopping trajectory is considered to fall in a sensitive frequency band and is removed; otherwise, it enters the local frequency band pruning; Finally, the compliant trajectory set is output.

8. The method of claim 7, wherein, The local frequency band pruning is: Based on the set of frequency hopping trajectories, for each abnormal frequency point in the trajectory, it is judged whether the frequency risk weight of the abnormal frequency point should be higher than a threshold value, if yes, a connection penalty term is added to the frequency point, not directly deleting the point, but reducing its trajectory continuous weight, forming a structure breaking point; for a high-risk segment with more than one point, the system performs a breaking operation to divide the trajectory into two parts, and the trajectories before and after the segmentation point are numbered and saved respectively.

9. The method of claim 1, wherein, The S4 includes: The compliant trajectory set and the available resources of the current system are obtained, an execution time estimation function is constructed, an execution delay estimation value of the compliant trajectory set is generated as a processing cost; wherein the available resources of the current system include the trajectory length, the trajectory frequency span, the interference modulation complexity score corresponding to the trajectory, and the modulation switching overhead of the compliant trajectory set; The interference modulation complexity score corresponding to the trajectory is scored by a system lookup table function for each abnormal frequency point, and then the maximum value is taken in the trajectory; the modulation switching overhead of the trajectory is the number of changes of the modulation mode of all adjacent frequency points in the trajectory; The time interval from the current time to the end of the earliest task processing is recorded, and when the processing cost is less than or equal to the time interval, the compliant trajectory in the compliant trajectory set is put into the executable trajectory set, and is sorted according to the total score to obtain the executable trajectory set.

10. The method of claim 1, wherein, The S5 includes: Based on the executable trajectory set, a weighted calculation mechanism is introduced, the contribution of each frequency point in the executable trajectory is weighted according to the abnormal score weight, so that the frequency point with higher confidence occupies a larger proportion in the interference target frequency selection, and the center frequency is obtained; The standard deviation of the trajectory abnormal score and the trajectory frequency span are used to calculate the dynamic adjustment bandwidth; The center frequency, the dynamic adjustment bandwidth, and the modulation mode and duty cycle are combined to generate the interference parameters.

Images