Ultra-wideband spectrum sensing radio measurement calibration method and system

Abstract

The present invention relates to the field of radio metrology technology, and discloses an ultra-wideband spectrum sensing radio metrology calibration method and system, the method comprising: performing time-frequency domain transformation and sliding interception on millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix; performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band to obtain an effective power density matrix; performing change rate calculation and band switching timing analysis on the power density data of each sub-band to obtain a sub-band scanning sequence and a minimum measurement time; performing adaptive optimization on system sampling parameters to obtain a target sampling parameter set; using the target sampling parameter set to re-measure the power density of each sub-band to obtain a measurement result, and performing error statistics on the measurement result to obtain a calibration compensation matrix, the calibration compensation matrix containing compensation coefficients of the frequency points to be calibrated. The present invention realizes the continuity and real-time performance of the measurement process and improves the calibration efficiency.

Description

Technical Field

[0001] The present invention relates to the field of radio metrology technology, and in particular to an ultra-wideband spectrum sensing radio metrology calibration method and system. Background Art

[0002] As wireless communication systems develop towards the millimeter wave frequency band, their ultra-wide spectrum characteristics have put forward higher requirements for radio metrology calibration. When facing millimeter wave ultra-wideband signals, the traditional calibration method has a significant increase in measurement delay due to the large bandwidth and large number of sampling points, making it difficult to meet real-time requirements. At the same time, millimeter wave signals are susceptible to external interference and multipath effects, which significantly reduces the accuracy of power density measurement and affects the calibration effect.

[0003] In order to obtain the true power characteristics of millimeter-wave ultra-wideband signals, the existing technology usually adopts a fixed sampling strategy and a single calibration scheme, but this method is difficult to cope with the dynamic change characteristics of the signal, and does not fully consider the signal correlation under the multi-channel receiving architecture, resulting in a waste of measurement resources and insufficient calibration accuracy. In addition, when performing spectrum sensing, due to the lack of real-time and continuity guarantees for power density measurement, the frequency band switching and parameter adjustment during the calibration process are relatively blind. In millimeter-wave communication systems, the accuracy of power density measurement directly affects the overall performance of the system. However, the existing calibration methods lack comprehensive consideration of factors such as noise influence and channel characteristics during the measurement process, and no effective calibration compensation mechanism has been established, resulting in large errors in the measurement results and unable to meet the requirements of high-precision calibration. Especially in the test of multiple frequency bands, due to the lack of a unified calibration strategy and optimization method, the consistency and reliability of the calibration results are difficult to guarantee. Summary of the invention

[0004] The present invention provides an ultra-wideband spectrum sensing radio measurement calibration method and system, which realizes the continuity and real-time performance of the measurement process and improves the calibration efficiency.

[0005] In a first aspect, the present invention provides an ultra-wideband spectrum sensing radio measurement and calibration method, the ultra-wideband spectrum sensing radio measurement and calibration method comprising:

[0006] Perform time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix;

[0007] Performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix;

[0008] Calculating the change rate and analyzing the frequency band switching timing of the power density data of each sub-band in the effective power density matrix to obtain a sub-band scanning sequence and a minimum measurement time;

[0009] Adaptively optimizing system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set;

[0010] The target sampling parameter set is used to remeasure the power density of each sub-frequency band to obtain a measurement result, and error statistics are performed on the measurement result to obtain a calibration compensation matrix, wherein the calibration compensation matrix includes compensation coefficients of the frequency points to be calibrated.

[0011] In a second aspect, the present invention provides an ultra-wideband spectrum sensing radio measurement and calibration system, the ultra-wideband spectrum sensing radio measurement and calibration system comprising:

[0012] A transformation module is used to perform time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix;

[0013] An analysis module, used for performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix;

[0014] A calculation module, used to calculate the change rate of the power density data of each sub-band in the effective power density matrix and analyze the frequency band switching timing to obtain a sub-band scanning sequence and a minimum measurement time;

[0015] An optimization module, configured to adaptively optimize system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set;

[0016] The error statistics module is used to remeasure the power density of each sub-band using the target sampling parameter set to obtain a measurement result, and to perform error statistics on the measurement result to obtain a calibration compensation matrix, wherein the calibration compensation matrix includes compensation coefficients of the frequency points to be calibrated.

[0017] In the technical solution provided by the present invention, power density characteristics are obtained through a multi-channel receiving architecture and a sliding interception method, and combined with time-frequency domain transformation and signal correlation analysis, the accuracy of power density measurement is significantly improved, and the influence of noise on the measurement results is effectively reduced; the optimal measurement period is determined by piecewise linear function approximation and dynamic programming optimization methods, and the measurement resources are reasonably allocated through an adaptive sampling strategy, which greatly reduces the measurement delay and ensures the measurement accuracy; a hidden Markov model is introduced for state tracking and error analysis, and a precise error compensation mechanism is established in combination with Bayesian parameter estimation and Viterbi algorithm, so that the average measurement error of the system is controlled within 0.1dB; sub-band priority sorting and band switching optimization strategy based on power stability indicators are implemented to achieve the continuity and real-time performance of the measurement process and improve the calibration efficiency; the optimal configuration of system resources is achieved by establishing a sampling parameter mapping relationship and iterative optimization and solution, which avoids resource waste and ensures the measurement accuracy requirements; a distributed calibration compensation scheme is adopted to integrate the calibration data of multiple bands, improve the consistency and reliability of the calibration results, and enable the system to maintain stable calibration performance in long-term operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 A schematic diagram of the steps of an ultra-wideband spectrum sensing radio measurement calibration method according to an embodiment of the present invention;

[0020] Figure 2 Schematic diagram of the structure of an ultra-wideband spectrum sensing radio measurement and calibration system in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Embodiments of the present invention provide an ultra-wideband spectrum sensing radio metrology calibration method and system. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 An embodiment of the ultra-wideband spectrum sensing radio measurement calibration method in the embodiment of the present invention includes:

[0023] Step S1, performing time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix;

[0024] It is understandable that the execution subject of the present invention may be an ultra-wideband spectrum sensing radio measurement and calibration system, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0025] Specifically, the sampling frequency and sampling duration are configured for multiple receiving channels, and the optimal sampling frequency value and sampling duration value are determined according to the actual application requirements and signal characteristics, and the configuration parameters of the receiving channels are generated. These configuration parameters determine the accuracy and efficiency of signal acquisition, ensuring that each receiving channel can synchronously acquire the millimeter wave ultra-wideband spectrum signal. Each receiving channel synchronously acquires the millimeter wave ultra-wideband spectrum signal according to the preset sampling frequency value and sampling duration value to generate an original signal data sequence. The original signal data sequence is processed by fast Fourier transform, and the signal originally in the time domain is converted into frequency domain distribution data to obtain a spectrum diagram. The frequency domain distribution data characterizes the relationship between the power of the millimeter wave signal and the frequency. The frequency domain distribution data is divided into sub-bands according to the sampling rate of the system to obtain a sub-band boundary sequence, where each sub-band represents a small range of the spectrum. The power of the frequency domain distribution data is calculated in each sub-band, so as to extract the power distribution characteristics of each sub-band and reflect the distribution of signal energy in each sub-band. Based on the power distribution characteristics of the sub-bands, frequency domain diffusion analysis is performed to study the energy leakage phenomenon of the signal between the sub-bands. By analyzing the distribution changes of the spectrum energy, the power leakage coefficient matrix is ​​calculated to reflect the degree of energy coupling between the sub-bands. Combining the power distribution characteristics and the power leakage coefficient matrix, a complete power density data structure is constructed, which contains the power density information of each sub-band and the corresponding correction coefficient. The power density data structure is matrixed to obtain the initial power density matrix. The power density data is organized in the form of a matrix, where the rows of the matrix correspond to different receiving channels and the columns represent the power density values ​​of different sub-bands.

[0026] Step S2, performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix;

[0027] Specifically, the column vector of the initial power density matrix is ​​extracted to generate a sub-band power density sequence. Each sub-band power density sequence corresponds to a fixed sub-band and contains the power density values ​​of all receiving channels in the sub-band. The sub-band power density sequence is normalized to eliminate the power differences between different receiving channels caused by hardware characteristics or other factors, so that the data can be compared at the same level to obtain normalized power density data. The autocorrelation function of the normalized power density data is calculated to generate a power density correlation coefficient matrix, which reflects the signal correlation degree between different receiving channels and provides important information on the mutual correlation between spectrum signals. A singular value decomposition operation is performed based on the power density correlation coefficient matrix to separate the signal subspace and the noise subspace. Through singular value decomposition, the signal energy and the noise energy are separated, and the signal and the noise are distinguished according to the distribution of their eigenvalues ​​in the subspace. The ratio of the eigenvalues ​​is used for signal-to-noise ratio calculation to generate a sub-band signal-to-noise ratio sequence. The signal-to-noise ratio sequence reflects the quality of the signal of each sub-band and is an important indicator for judging the effectiveness of the signal. In order to distinguish valid signals from invalid signals, the sub-band signal-to-noise ratio sequence is subjected to threshold judgment processing. By setting a reasonable signal-to-noise ratio threshold, valid signals with a signal-to-noise ratio higher than the threshold are screened out, and the corresponding valid signal judgment result is generated. The valid signal judgment result is used to guide the data screening of the initial power density matrix, that is, to eliminate the noise data with a signal-to-noise ratio lower than the threshold, and only retain the power density data judged as valid to generate the initial screening power density data. Based on the initial screening power density data, matrix reconstruction is performed. The screened power density data is rearranged and combined in the format of the original matrix to ensure the integrity and consistency of the data structure. The final generated effective power density matrix only contains power density data that passes the signal-to-noise ratio judgment, eliminating the interference of low signal-to-noise ratio signals, and can more accurately reflect the true characteristics of the spectrum signal.

[0028] Step S3, calculating the change rate of the power density data of each sub-band in the effective power density matrix and analyzing the frequency band switching timing to obtain a sub-band scanning sequence and a minimum measurement time;

[0029] Specifically, the power density data of each sub-band is extracted from the effective power density matrix, and the data is segmented in units of time windows to form a time series of the power density of the sub-band. These time series reflect the dynamic characteristics of the power density in each sub-band over time, which is the basis for analyzing the change trend. The sub-band power density time series is subjected to differential operation, and the power difference between adjacent time points is calculated to generate a power change sequence, revealing the rate and amplitude of the power density of each sub-band over time, and providing necessary data support for stability analysis and change trend prediction. The power change rate matrix of each sub-band is calculated according to the power change sequence. The change rate matrix reflects the change rate and trend difference of each sub-band in the time dimension through standardization operation. Then, the data in the power change rate matrix is ​​statistically analyzed to generate a power stability index sequence. This sequence is an important parameter for quantifying the stability of the sub-band power change, and can effectively distinguish between sub-bands with large signal fluctuations and relatively stable sub-bands. Cluster analysis is performed based on the power stability index sequence. The sub-bands are divided into different stability levels, and the stability level division results are generated. The sub-band priority ranking is constructed based on the sub-band stability level division results. According to the stability level and power change rate of the sub-bands, the sub-bands with larger dynamic changes are processed first to construct the initial scanning sequence. The initial scanning sequence defines the basic order of scanning, but the switching overhead between bands needs to be considered. The frequency span between adjacent sub-bands is calculated according to the initial scanning sequence, and the band switching time sequence is generated to reflect the time cost required to switch from one sub-band to another. The initial scanning sequence is optimized using the dynamic programming method. Dynamic programming generates the sub-band scanning sequence by comprehensively considering the band switching time and the sub-band stability level. In this process, the goal is to minimize the total band switching time while ensuring that the scanning order has a higher priority for the sub-bands with larger changes, forming the final optimized scanning sequence. According to the optimized sub-band scanning sequence and the band switching time sequence, the time is accumulated to calculate the minimum measurement time required to complete the entire scanning process.

[0030] Step S4: adaptively optimize the system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set;

[0031] Specifically, the power change rate matrix of each sub-band is analyzed according to the sub-band scanning sequence, and the minimum sampling requirement of each sub-band in the time dimension is calculated according to the dynamic characteristics of the sub-band and its power change rate, so as to obtain the minimum sampling requirement matrix. The minimum sampling requirement matrix is ​​combined with the minimum measurement time, and the sampling point number sequence of each sub-band is generated by optimizing the allocation of limited sampling resources, that is, the number of sampling points required to meet the dynamic perception requirements of each sub-band. The sampling rate is calculated according to the sub-band sampling point number sequence and the bandwidth characteristics of each sub-band. The calculation of the sampling rate follows the basic principle of Nyquist sampling theorem, while taking into account the actual system constraints and the signal complexity of the sub-band, to ensure that the initial sampling rate sequence can meet the basic requirements of signal recovery. In order to improve the rationality of the sampling rate allocation, the initial sampling rate sequence is weighted and fused with the power stability index sequence. By assigning different weights to the power stability index, the optimization result of the sampling rate can be tilted on the sub-band with drastic dynamic changes, thereby improving the perception accuracy of the key band and generating an optimized sampling rate sequence. Based on the optimized sampling rate sequence and the sub-band sampling point sequence, the sampling parameter mapping relationship is constructed to reflect the coupling characteristics of the sampling rate and the number of sampling points between each sub-band. Through this relationship, a sampling parameter optimization model is established, which takes the maximization of sampling resource utilization and the optimization of scanning efficiency as the objective function, and takes the system's hardware resource limitation, time constraint and signal recovery accuracy as the constraint conditions. The sampling parameter optimization model is solved by an iterative optimization algorithm. Using the particle swarm optimization algorithm or genetic algorithm, a global optimal solution is found by gradually adjusting the sampling parameters of each sub-band to generate the sampling parameter optimization result. According to the sampling parameter optimization result, specific sampling configuration information is generated for each sub-band, including the final sampling rate, number of sampling points and sampling time window of each sub-band, forming a sub-band sampling scheme sequence. The sub-band sampling scheme sequence is parameterized, and the time loss of switching across sub-bands and the overall coordination of system resources are comprehensively considered to finally generate a target sampling parameter set. The target sampling parameter set can meet the dynamic perception needs of each sub-band, while maximizing the sampling resource utilization efficiency and system scanning performance.

[0032] The sampling rate data of each sub-band in the optimized sampling rate sequence is subjected to correlation analysis. Through correlation analysis, the correlation degree of sampling rates between different sub-bands is quantified to generate a sampling rate correlation matrix. The calculation of the sampling rate correlation matrix is ​​based on statistical methods or correlation measurement technology, and its results reflect the relationship between the sampling rates of each sub-band in frequency distribution or time distribution. Based on this matrix, the original independent sub-band sampling point number sequence is reorganized to obtain a new sampling resource distribution vector. This vector is the result of the reorganization of sampling resources between different sub-bands, which can more effectively allocate limited resources to meet the needs of dynamic spectrum sensing. The sampling rate correlation matrix is ​​combined with the sampling resource distribution vector, and the sampling parameter initial mapping matrix is ​​constructed by the bilinear mapping method to describe the mapping relationship between the sampling rate and the number of sampling points in different sub-bands. Based on the sampling parameter initial mapping matrix, the sampling resource optimization objective function is constructed. The design of the optimization objective function aims to improve the overall measurement accuracy and resource utilization efficiency of the system, and balances the priority and resource allocation requirements between sub-bands by introducing weight parameters. The specific form of the objective function selects expressions such as weighted minimization error or maximization of resource utilization to meet the optimization needs of different application scenarios. After the objective function is constructed, constraints are added to the objective function according to the requirements of system measurement accuracy and hardware resource constraints. These constraints include factors such as the upper and lower limits of the sampling rate, the total resource allocation limit, and the sub-band switching time. The comprehensive objective function and the constraints constitute the sampling parameter optimization model. The sampling parameter optimization model is subjected to convex optimization decomposition to reduce its computational complexity, and the original problem is decomposed into a sequence of sub-problems that are easier to solve. Through convex optimization decomposition, the complex global optimization problem is transformed into multiple local optimization problems with good convergence properties. These sub-problems are solved by the alternating direction multiplier method. The alternating direction multiplier method is an efficient algorithm suitable for distributed optimization problems. By alternatingly updating the solutions of sub-problems, the global optimal solution is quickly approached. The results of each iteration constitute an iterative optimization sequence, and the sampling parameters are gradually optimized. In the iterative optimization process, convergence judgment and parameter update are performed on each iterative optimization sequence. By calculating the change between the current iteration result and the previous round result, it is judged whether the convergence condition is met. When the preset convergence standard is met, the iteration is stopped and the final optimization result is output as the sampling parameter optimization result. If the convergence condition is not met, the parameters are updated and the next iteration is entered. This process continuously adjusts the sampling rate and the number of sampling points, and finally generates sampling parameters that meet the optimization objectives.

[0033] Step S5: re-measure the power density of each sub-frequency band using the target sampling parameter set to obtain a measurement result, and perform error statistics on the measurement result to obtain a calibration compensation matrix, which includes compensation coefficients of the frequency points to be calibrated.

[0034] Specifically, sampling control is implemented on each sub-band according to the target sampling parameter set to ensure that the sampling rate and the number of sampling points meet the optimal configuration. The original re-measured data sequence generated by the sampling control is transformed into a re-measured power density sequence through time-frequency transformation. A hidden Markov observation sequence is constructed for the re-measured power density sequence. By treating the power density sequence as an observation value, the state transition matrix of the power density is established using the hidden Markov model to capture the change pattern of the power density over time. Based on the state transition matrix, the state probability distribution sequence is calculated using the forward algorithm to obtain the possible states and probabilities of the power density at each moment. The state probability distribution sequence reflects the statistical characteristics of the dynamic change of the power density. The re-measured power density sequence is state-decoded according to the state probability distribution sequence. The decoding result is traced back through the Viterbi algorithm to obtain the optimal state path. The optimal state path is the best explanation of the power density sequence under the hidden Markov model, indicating the state change law of the signal in time. Based on this path, the measurement error analysis of the re-measured power density sequence is performed to generate a power density error statistical matrix. The deviation between the re-measured power density and the true value is quantified. According to the error statistical matrix, the error distribution model is constructed through Bayesian parameter estimation. Bayesian estimation combines statistical data and prior knowledge to obtain a more reliable description of the error distribution. Based on the error distribution model, the measurement errors of each sub-band are classified and counted to generate a frequency band error distribution sequence. By detecting the threshold of this sequence, the frequency points with significant errors are screened out to form a set of frequency points to be calibrated. Based on the set of frequency points to be calibrated, the compensation coefficients of the error statistics matrix are calculated to generate an initial compensation coefficient matrix. The initial compensation coefficient matrix contains the coefficients required for power density correction for each frequency point to be calibrated. In order to ensure the applicability and stability of the coefficients, the initial compensation coefficient matrix is ​​regularized to obtain a normalized compensation matrix. The purpose of regularization is to prevent numerical instability caused by excessive or too small compensation coefficients and to improve the generalization ability of the matrix. The normalized compensation matrix is ​​combined with the optimal state path, and the compensation coefficients are adjusted through the optimization algorithm to obtain the optimal compensation coefficient sequence. The optimal compensation coefficient sequence is optimized based on the overall error model of the system and the specific measurement conditions, and has higher accuracy and stronger applicability. Through the frequency band mapping technology, the optimal compensation coefficient sequence is converted into a compensation parameter distribution vector so that it corresponds to a specific frequency band. The compensation parameter distribution vector is reconstructed to ensure the data structure integrity and calibration accuracy, and finally a calibration compensation matrix is ​​generated. Each element of the calibration compensation matrix corresponds to a compensation coefficient of a frequency point to be calibrated.

[0035] In the embodiment of the present invention, the power density characteristics are obtained by a multi-channel receiving architecture and a sliding interception method, and the time-frequency domain transformation and signal correlation analysis are combined to significantly improve the accuracy of power density measurement and effectively reduce the influence of noise on the measurement results; the optimal measurement period is determined by piecewise linear function approximation and dynamic programming optimization methods, and the measurement resources are reasonably allocated by an adaptive sampling strategy, which greatly reduces the measurement delay and ensures the measurement accuracy; the hidden Markov model is introduced for state tracking and error analysis, and an accurate error compensation mechanism is established by combining Bayesian parameter estimation and Viterbi algorithm, so that the average measurement error of the system is controlled within 0.1dB; the sub-band priority sorting and band switching optimization strategy based on the power stability index realize the continuity and real-time performance of the measurement process and improve the calibration efficiency; the optimal configuration of system resources is achieved by establishing the sampling parameter mapping relationship and iterative optimization solution, which avoids resource waste and ensures the measurement accuracy requirements; the distributed calibration compensation scheme is adopted to integrate the calibration data of multiple bands, improve the consistency and reliability of the calibration results, and enable the system to maintain stable calibration performance in long-term operation.

[0036] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0037] The sampling frequency and sampling duration are configured for multiple receiving channels to obtain receiving channel configuration parameters, where the receiving channel configuration parameters include sampling frequency values ​​and sampling duration values;

[0038] The millimeter-wave ultra-wideband spectrum signals of each receiving channel are synchronously collected according to the receiving channel configuration parameters to obtain an original signal data sequence, and the original signal data sequence is processed by fast Fourier transform to obtain frequency domain distribution data;

[0039] The frequency domain distribution data is divided into sub-bands according to the system sampling rate to obtain a sub-band boundary sequence, and the power of the frequency domain distribution data in each sub-band is calculated to obtain the sub-band power distribution characteristics;

[0040] Based on the sub-band power distribution characteristics, frequency domain diffusion analysis is performed to obtain a power leakage coefficient matrix, and a power density data structure is constructed according to the sub-band power distribution characteristics and the power leakage coefficient matrix to obtain initial power density data;

[0041] The initial power density data is matrixed to obtain an initial power density matrix, wherein the rows of the initial power density matrix represent different receiving channels, and the columns represent power density values ​​of different sub-frequency bands.

[0042] Specifically, according to the spectrum characteristics of the signal and system requirements, select the appropriate sampling frequency and sampling duration. The sampling frequency is usually expressed as It is defined as the number of samples per second in Hertz (Hz), and its size needs to satisfy the Nyquist sampling theorem, that is, ,in is the maximum bandwidth of the signal, in Hertz (Hz). It is defined as the total time of the acquired signal, and its size determines the frequency resolution , through the formula Calculated. Sampling frequency and sampling duration The combination of forms the receiving channel configuration parameters, using the set According to the configured sampling frequency and sampling duration, the millimeter wave ultra-wideband spectrum signal is collected for each receiving channel to generate the original signal data sequence. The original signal is represented by Indicates that is the index of the receiving channel, indicating the channels; These signals are sampled synchronously to ensure that the time points and spectrum ranges collected by each channel are aligned, providing a consistent basis for subsequent spectrum analysis. Perform fast Fourier transform to transform the signal from time domain to frequency domain and obtain frequency domain distribution data. Indicates that is the frequency in Hertz (Hz). The discrete form of the fast Fourier transform is:

[0043] ;

[0044] in, is the number of sampling points, is the frequency index, indicating the frequency points; is a discrete-time signal, indicating the Time sampling points; It is The complex amplitude corresponding to the frequency point is in square root of amplitude. Through fast Fourier transform, we get It is a complex representation of the signal spectrum, containing amplitude and phase information. , for frequency domain distribution data The sub-bands are divided. The sub-band boundaries are divided using the sequence Indicates that is the total number of sub-bands, Indicates For example, the sub-band range ,in and is the boundary frequency of adjacent frequency bands. In each sub-band, The power distribution characteristics are calculated by summing the square of the amplitude , which represents the power density distribution of the signal in different sub-bands, and its calculation formula is:

[0045] ;

[0046] in is the power density; Indicates the square of the spectrum amplitude. Based on the sub-band power distribution characteristics , frequency domain diffusion analysis is performed on each sub-band. Frequency domain diffusion analysis quantifies spectrum leakage by calculating the cross-correlation between sub-bands. The power leakage coefficient matrix The elements of are defined as:

[0047] ;

[0048] in It is the frequency band To frequency band The leakage coefficient; yes The numerator is the cross power of the signal between the two sub-bands, and the denominator is the sub-band Combined with the sub-band power distribution characteristics and the leakage coefficient matrix , construct the power density data structure, integrate the power density and diffusion effect of each sub-band, and generate the initial power density data. Matrix the initial power density data to generate the initial power density matrix ,in:

[0049] ;

[0050] matrix of rows Indicates different receiving channels, columns Represents different sub-bands, whose elements Indicates channel In sub-band This matrix represents the signal strength of multiple receiving channels in each sub-frequency band.

[0051] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0052] Extracting column vectors from the initial power density matrix to obtain a sub-band power density sequence, where the sub-band power density sequence includes power density values ​​of each receiving channel in the same sub-band;

[0053] Normalize the sub-band power density sequence to obtain normalized power density data, and calculate the autocorrelation function of the normalized power density data to obtain a power density correlation coefficient matrix, which reflects the signal correlation degree between each receiving channel;

[0054] Based on the power density correlation coefficient matrix, a singular value decomposition operation is performed to obtain the signal subspace and the noise subspace, and the signal-to-noise ratio is calculated according to the eigenvalue ratio of the signal subspace and the noise subspace to obtain the sub-band signal-to-noise ratio sequence;

[0055] Perform threshold judgment processing on the sub-band signal-to-noise ratio sequence to obtain a valid signal judgment result, and perform data screening on the initial power density matrix according to the valid signal judgment result to obtain preliminary screening power density data;

[0056] The primary screening power density data is reconstructed into a matrix to obtain an effective power density matrix, which only contains power density data judged by the signal-to-noise ratio.

[0057] Specifically, the initial power density matrix The rows represent the receiving channels, and the columns represent the power density values ​​of different sub-bands. Take out the column and get the sub-band power density sequence ,in Indicates the receiving channel In sub-band The power density value on is the total number of receiving channels. Through column vector extraction, the power density sequences of all sub-bands are obtained to describe the signal strength distribution of each sub-band on different receiving channels. Normalization is performed to eliminate the imbalance between different channels caused by hardware characteristics or signal strength differences, ensuring that subsequent analysis is performed on a unified scale. The normalization formula is:

[0058] ;

[0059] in is the normalized power density sequence, yes The normalized power density data Used to calculate the autocorrelation function to quantify the signal correlation between each receiving channel. Power density correlation coefficient matrix Elements Defined as:

[0060] ;

[0061] in Indicates the receiving channel and The correlation between is the total number of sub-bands. The correlation coefficient matrix Is a symmetric matrix, and its diagonal elements are always 1. Based on the power density correlation coefficient matrix , perform singular value decomposition on it, and the decomposition formula is:

[0062] ;

[0063] in is an orthogonal matrix containing the eigenvectors of the correlation coefficient matrix; It is a diagonal matrix, and its diagonal elements are the eigenvalues ​​of the correlation coefficient matrix. According to the results of singular value decomposition, the signal space is decomposed into signal subspace and noise subspace. The signal subspace corresponds to a larger eigenvalue, while the noise subspace corresponds to a smaller eigenvalue. By calculating the ratio of the eigenvalues ​​of the signal subspace and the noise subspace, the signal-to-noise ratio (SNR) is calculated, and the formula is:

[0064] ;

[0065] in yes It is a sub-band The signal-to-noise ratio of the sub-band is calculated. Perform threshold decision processing by setting the signal-to-noise ratio threshold , filter out the valid signal. , then the decision sub-band The signal is a valid signal, denoted by Otherwise it is considered as an invalid signal and recorded as . Valid signal judgment result It is used to guide the data screening of the initial power density matrix, retaining only the sub-band data judged as valid signals to generate the initial screening power density data. The initial screening power density data is reconstructed, and the sub-band power density after screening is rearranged into a matrix form to obtain the effective power density matrix .matrix The rows represent receiving channels, the columns represent sub-bands determined by the signal-to-noise ratio, and its elements only contain the power density data of the valid signal.

[0066] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0067] The power density data of each sub-band in the effective power density matrix is ​​processed by time window segmentation to obtain the sub-band power density time series, and the sub-band power density time series is subjected to difference operation to obtain the power variation series;

[0068] The power change rate matrix of each sub-frequency band is calculated according to the power change sequence to obtain the power stability index sequence, and the power stability index sequence is clustered and analyzed to obtain the sub-frequency band stability grade division result;

[0069] Based on the result of the sub-band stability level division, a sub-band priority ranking is constructed to obtain an initial scanning sequence, and the frequency span between adjacent sub-bands is calculated according to the initial scanning sequence to obtain a frequency band switching time sequence;

[0070] The frequency band switching time-consuming sequence is dynamically optimized to obtain a sub-band scanning sequence, and the time is accumulated according to the sub-band scanning sequence and the frequency band switching time-consuming sequence to obtain the minimum measurement time.

[0071] Specifically, the power density data of each sub-band is extracted from the effective power density matrix and segmented according to the time window. Assume that the effective power density matrix is , whose elements Indicates the receiving channel In sub-band and time The time window segmentation process divides the power density data of each sub-band into fixed time lengths. Divide into multiple time slices to form a sub-band power density time series, using Indicates. Among them, It is a sub-band The average power density in each time window is defined as:

[0072] ;

[0073] in It is The start time of the time window, is the total number of receiving channels. Perform differential operation to obtain the power change sequence , describes the change in power density between adjacent time windows. The difference formula is:

[0074] ;

[0075] in, It is a sub-band In time According to the power change sequence , calculate the power change rate matrix of each sub-band . Power change rate Indicates sub-band In time The power change rate is defined as:

[0076] ;

[0077] Power Change Rate Matrix Elements Indicates sub-band In time The rate of change of power. By statistically analyzing the power change rate matrix The time mean and variance of , generating the power stability index sequence The power stability index is used to quantify the sub-band The dynamic characteristics of , the calculation formula is:

[0078] ;

[0079] in and Sub-bands The standard deviation and mean of the power change rate. Indicates that the sub-band power changes dramatically, with smaller Indicates stable change. Based on the power stability index sequence , the sub-bands are divided into different stability levels through cluster analysis. For example, using - Mean value clustering algorithm, divides the sub-bands into The sub-band stability level division results are obtained. ,in Indicates that it belongs to The sub-bands are divided into a set of sub-bands with different levels. The sub-band priority ranking is constructed based on the stability level division results. The priority ranking is sorted from the sub-band with the largest dynamic change to the sub-band with the smallest change according to the stability level to generate the initial scanning sequence ,in Indicates According to the initial scanning sequence, the frequency span between adjacent sub-bands is calculated. , defined as:

[0080] ;

[0081] in It is a sub-band The frequency span determines the time required to switch from one sub-band to another, generating a frequency band switching time sequence ,in Indicates that the sub-band Switch to sub-band The scanning sequence is optimized by dynamic programming, with the objective function of minimizing the total switching time. Suppose the optimized scanning sequence is , the goal of dynamic programming is to minimize:

[0082] ;

[0083] After the optimization is completed, according to the optimized scanning sequence Band switching time sequence Accumulate time and calculate the minimum measurement time :

[0084] ;

[0085] in It is a sub-band The measurement time.

[0086] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0087] The power change rate matrix of each sub-band is analyzed according to the sub-band scanning sequence to obtain the minimum sampling requirement matrix, and the sampling resource allocation calculation is performed on the minimum sampling requirement matrix and the minimum measurement time to obtain the sub-band sampling point number sequence;

[0088] The sampling rate is calculated according to the sub-band sampling point number sequence and the bandwidth characteristics of each sub-band to obtain an initial sampling rate sequence, and the initial sampling rate sequence and the power stability index sequence are weightedly fused to obtain an optimized sampling rate sequence;

[0089] A sampling parameter mapping relationship is constructed based on the optimized sampling rate sequence and the sub-band sampling point number sequence to obtain a sampling parameter optimization model, and the sampling parameter optimization model is iteratively optimized and solved to obtain a sampling parameter optimization result;

[0090] The sampling configuration information of each sub-band is generated according to the sampling parameter optimization result, a sub-band sampling scheme sequence is obtained, and the sub-band sampling scheme sequence is parameter integrated to obtain a target sampling parameter set.

[0091] Specifically, the power change rate matrix of each sub-band is analyzed according to the sub-band scanning sequence. Elements Indicates sub-band In time Based on these change rates, the minimum sampling requirement matrix of the sub-band is defined. , whose elements Indicates sub-band The minimum sampling requirement is calculated by the following formula:

[0092] ;

[0093] in is the total measurement time of the system, Indicates time Find the maximum value of the operation, in bits per second. Minimum sampling requirement matrix In order to accurately capture the power variation within the sub-band, the minimum sampling resource that needs to be allocated is represented by and the minimum measurement time of the system , by optimizing the allocation of sampling resources, calculating the sub-band sampling point sequence The calculation formula of the sampling point sequence is:

[0094] ;

[0095] in It is a sub-band bandwidth. Indicates sub-band The number of sampling points required. This formula shows that the number of sampling points in a sub-band depends on the power change rate, measurement time, and frequency band bandwidth. After obtaining the sub-band sampling point sequence, the initial sampling rate sequence is calculated based on the bandwidth characteristics of the sub-band. The calculation of the sampling rate follows the Nyquist sampling theorem, and the formula is:

[0096] ;

[0097] in It is a sub-band The initial sampling rate must meet the requirements of bandwidth and number of sampling points. and power stability index sequence Perform weighted fusion to obtain the optimized sampling rate sequence The weighted fusion formula is:

[0098] ;

[0099] in and is a weight factor that satisfies , used to balance the influence of sampling rate requirements and stability indicators. The selection of weight factors depends on the priority requirements in specific application scenarios. and the sub-band sampling point sequence , construct the sampling parameter mapping relationship. The sampling parameter mapping relationship is represented by the following model:

[0100] ;

[0101] in It is a sub-band The sampling parameter set includes the optimized sampling rate and number of sampling points. In order to optimize the resource allocation efficiency, a sampling parameter optimization model is constructed for the above mapping relationship. The objective function of the optimization model is defined as minimizing the total resource consumption of the system:

[0102] ;

[0103] in is the total number of sub-bands. Also add constraints, including hardware sampling capabilities and total resource limits ,in is the maximum sampling rate of the system, is the maximum number of sampling points of the system. The optimization model is iteratively optimized and solved, using gradient descent or Lagrange multiplier method to gradually adjust and To approach the optimal solution. The optimization results generate the sampling configuration information of each sub-band , including sampling rate and number of sampling points. Integrate the sampling configuration information of all sub-bands to form a target sampling parameter set .

[0104] In a specific embodiment, the execution step constructs a sampling parameter mapping relationship based on the optimized sampling rate sequence and the sub-band sampling point number sequence to obtain a sampling parameter optimization model, and iteratively optimizes and solves the sampling parameter optimization model. The process of obtaining the sampling parameter optimization result may specifically include the following steps:

[0105] Performing correlation analysis on the sampling rate data of each sub-band in the optimized sampling rate sequence to obtain a sampling rate correlation matrix, and reorganizing the sub-band sampling point number sequence according to the sampling rate correlation matrix to obtain a sampling resource distribution vector;

[0106] Perform bilinear mapping on the sampling rate correlation matrix and the sampling resource distribution vector to obtain the initial mapping matrix of the sampling parameters;

[0107] An optimization function is constructed based on the initial mapping matrix of sampling parameters to obtain a sampling resource optimization objective function, and constraints are added to the sampling resource optimization objective function according to the system measurement accuracy requirements and hardware resource limitations to obtain a sampling parameter optimization model;

[0108] The sampling parameter optimization model is decomposed by convex optimization to obtain a sub-problem sequence, and the sub-problem sequence is solved by the alternating direction multiplier method to obtain an iterative optimization sequence;

[0109] Convergence judgment and parameter update are performed on the iterative optimization sequence to obtain the sampling parameter optimization results.

[0110] Specifically, according to the optimized sampling rate sequence Starting from, the sampling rates of each sub-band are compared pairwise to quantify the correlation between them. Indicates sub-band The sampling rate correlation matrix is ​​constructed by calculating the Pearson correlation coefficient between the sampling rates. The elements of the matrix Defined as:

[0111] ;

[0112] in, It is a sub-band and The covariance of the sampling rates, and Sub-bands and The standard deviation of the sampling rate. Through this formula, The value range is [-1,1], where positive values ​​indicate positive correlation and negative values ​​indicate negative correlation. The closer the value is to 1, the stronger the correlation is. , the sampling point sequence of the sub-band Reorganize the data to obtain the sampling resource distribution vector . Sampling point sequence Indicates allocation to sub-band The purpose of the reorganization is to readjust the distribution of the number of sampling points based on the sampling rate correlation between sub-bands, so that sub-bands with strong correlation share more sampling resources. The calculation formula of the sampling resource distribution vector is:

[0113] ;

[0114] in, It is a sub-band The reorganized sampling resources are expressed in points. Through this formula, the sampling resources are adjusted according to the correlation and are preferentially allocated to the sub-bands with stronger correlation. and the sampling resource distribution vector Perform bilinear mapping to generate the initial mapping matrix of sampling parameters The mapping formula is:

[0115] ;

[0116] in, is a diagonal matrix consisting of sampled resource distribution vectors, is the transposed matrix of the correlation matrix. Elements Indicates sub-band and The initial allocation relationship of sampling parameters is used for further optimization. Based on the initial mapping matrix of sampling parameters , construct the sampling resource optimization objective function. The objective function aims to minimize the total resource consumption of the system while satisfying the sampling accuracy and hardware constraints. The optimization objective function is defined as:

[0117] ;

[0118] in, is the total resource consumption in bits per second (bps). Also add the following constraints: , that is, the sampling rate of each sub-band cannot exceed the maximum sampling rate of the system; , that is, the total number of sampling points cannot exceed the hardware capability of the system. The optimization model is subjected to convex optimization decomposition processing to decompose the complex global optimization problem into a sequence of sub-problems that are easy to solve. The decomposition method is to divide the objective function and constraints into multiple parts and optimize them separately. Each sub-problem represents an optimization task for some variables. The alternating direction multiplier method (ADMM) is used to solve the sequence of sub-problems. ADMM introduces multipliers to transform the constraints in the optimization problem into penalty terms, and gradually approaches the optimal solution in iterations. Specifically, assume and There are two sets of variables, and the goal is to minimize And meet , then each iteration includes the following steps: fix ,optimization ;fixed ,optimization ; Update the Lagrange multiplier. In each iteration, calculate the iterative optimization sequence , until the convergence condition is met ,in is the convergence threshold. After the iteration is completed, the final sampling parameter optimization result is generated , including the optimized sampling rate and number of sampling points for each sub-band.

[0119] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0120] The sampling of each sub-frequency band is controlled according to the target sampling parameter set to obtain the original re-measurement data sequence, and the original re-measurement data sequence is transformed in time-frequency to obtain the re-measurement power density sequence;

[0121] A hidden Markov observation sequence is constructed for the re-measured power density sequence to obtain a power density state transfer matrix, and a forward algorithm is performed based on the power density state transfer matrix to obtain a state probability distribution sequence;

[0122] The re-measured power density sequence is state-decoded according to the state probability distribution sequence to obtain a power density measurement state sequence, and the power density measurement state sequence is backtracked by a Viterbi path to obtain an optimal state path;

[0123] The measurement error analysis of the optimal state path is performed to obtain the power density error statistical matrix, and the Bayesian parameter estimation is performed based on the power density error statistical matrix to obtain the error distribution model;

[0124] The sub-band measurement errors are classified and counted according to the error distribution model to obtain a frequency band error distribution sequence, and the frequency band error distribution sequence is threshold-detected to obtain a set of frequency points to be calibrated;

[0125] The compensation coefficients of the power density error statistical matrix are calculated based on the frequency point set to be calibrated to obtain an initial compensation coefficient matrix, and the initial compensation coefficient matrix is ​​regularized to obtain a normalized compensation matrix;

[0126] Compensation parameters are optimized according to the normalized compensation matrix and the optimal state path to obtain an optimal compensation coefficient sequence, and the optimal compensation coefficient sequence is frequency-band mapped to obtain a compensation parameter distribution vector;

[0127] The compensation parameter distribution vector is reconstructed into a matrix to obtain a calibration compensation matrix, wherein each element of the calibration compensation matrix corresponds to a compensation coefficient of a frequency point to be calibrated.

[0128] Specifically, according to the target sampling parameter set Sampling control for each sub-band. Set the sampling rate and number of sampling points for each sub-band, control the receiving channel to accurately sample the millimeter wave signal, and generate the original re-measurement data sequence ,in It is a sub-band The discrete-time signal, Represents the sampling point index. Perform time-frequency transformation and use short-time Fourier transform to capture the time-frequency characteristics of the signal and obtain the re-measured power density sequence The formula for short-time Fourier transform is:

[0129] ;

[0130] in It is a sub-band The time-frequency representation of is a window function that controls the signal distribution within a local time window. is the power density. Based on the retest power density sequence Construct a hidden Markov observation sequence to capture the state changes of signal power density over time. Suppose the state set of power density is , each state Indicates a power density range. The distribution of ,in Indicates The observation state at the moment. According to the observation sequence and state collection , construct the state transfer matrix , whose elements Defined as:

[0131] ;

[0132] in Indicates from the state Transfer to state The probability of. Based on the state transfer matrix and the observation sequence , use the forward algorithm to calculate the state probability distribution sequence. Assume Indicates at time Observed sequence and in state The probability of , the recursive formula of the forward algorithm is:

[0133] ;

[0134] in is the observed probability distribution. According to the state probability distribution sequence, the re-measured power density sequence is decoded and the optimal state path is solved by the Viterbi algorithm. The recursive formula of the Viterbi algorithm is:

[0135] ;

[0136] in It's time state The maximum probability path. By backtracking, the optimal state path is obtained . Perform measurement error analysis on the optimal state path and calculate the power density error statistical matrix The elements of the matrix Indicates sub-band At the moment The power density error is defined as:

[0137] ;

[0138] in is the true power density. Based on the error statistics matrix , perform Bayesian parameter estimation and obtain the error distribution model. Assume that the error distribution is Gaussian distribution, and its parameters are calculated by maximum a posteriori estimation:

[0139] ;

[0140] in and Sub-bands According to the error distribution model, the sub-band measurement errors are classified and counted to obtain the frequency band error distribution sequence By setting the threshold , filter out the frequency points whose errors exceed the threshold, and form a set of frequency points to be calibrated Based on the frequency set to be calibrated, the error statistics matrix Calculate the compensation coefficients and generate the initial compensation coefficient matrix The compensation coefficient is defined as:

[0141] ;

[0142] Through regularization processing, the normalized compensation coefficient matrix :

[0143] ;

[0144] According to the normalized compensation matrix and the optimal state path , optimize the compensation parameters and obtain the optimal compensation coefficient sequence . Perform frequency band mapping on the optimal compensation coefficient sequence to generate the compensation parameter distribution vector . For the compensation parameter distribution vector Perform matrix reconstruction to generate calibration compensation matrix , each element of which corresponds to the compensation coefficient of the frequency point to be calibrated, thereby completing the calibration process of power density measurement.

[0145] In this embodiment, the step of state decoding the remeasured power density sequence according to the state probability distribution sequence to obtain the power density measurement state sequence, and performing Viterbi path backtracing on the power density measurement state sequence to obtain the optimal state path also includes: constructing a two-layer belief network structure for the power density measurement state sequence to obtain an initial belief node matrix, wherein the initial belief node matrix includes a measurement state node and a calibration state node; performing probability decomposition on the power density state transfer matrix according to the initial belief node matrix to obtain a state transfer probability tree, and constructing a belief propagation rule based on the state transfer probability tree to obtain a belief update model; cross-validation training is performed on the belief update model to obtain a measurement state belief layer, and a calibration state belief layer is constructed based on the measurement state belief layer to obtain a two-layer belief propagation network; belief reasoning is performed on the optimal state path based on the two-layer belief propagation network to obtain a state credibility sequence , and weighted fusion of the power density measurement results is performed according to the state credibility sequence to obtain a credible measurement result; a measurement uncertainty evaluation model is constructed according to the credible measurement result to obtain a measurement uncertainty matrix, and belief network parameters are optimized based on the measurement uncertainty matrix to obtain an optimized belief parameter set; back-propagation training is performed on the optimized belief parameter set to obtain a calibration compensation belief mapping function, and calibration compensation is performed on the credible measurement results according to the calibration compensation belief mapping function to obtain an optimized calibration result; calibration reliability evaluation is performed based on the optimized calibration result and the measurement uncertainty matrix to obtain a calibration reliability index sequence, and the optimal state path is dynamically corrected according to the calibration reliability index sequence to obtain a corrected state path; the corrected state path is input into the power density measurement state sequence for state update to obtain an updated power density measurement state sequence for subsequent error analysis and calibration compensation.

[0146] The ultra-wideband spectrum sensing radio measurement calibration method according to the embodiment of the present invention is described above. The ultra-wideband spectrum sensing radio measurement calibration system according to the embodiment of the present invention is described below. Figure 2 In one embodiment of the present invention, an ultra-wideband spectrum sensing radio measurement and calibration system includes:

[0147] A transformation module is used to perform time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix;

[0148] An analysis module is used to perform correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix;

[0149] A calculation module is used to calculate the change rate of the power density data of each sub-band in the effective power density matrix and analyze the frequency band switching timing to obtain the sub-band scanning sequence and the minimum measurement time;

[0150] An optimization module is used to adaptively optimize the system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set;

[0151] The error statistics module is used to remeasure the power density of each sub-band using the target sampling parameter set to obtain the measurement results, and to perform error statistics on the measurement results to obtain a calibration compensation matrix, which includes compensation coefficients of the frequency points to be calibrated.

[0152] Through the cooperation of the above components, the power density characteristics are obtained through a multi-channel receiving architecture and a sliding interception method. Combined with time-frequency domain transformation and signal correlation analysis, the accuracy of power density measurement is significantly improved, and the influence of noise on the measurement results is effectively reduced. The optimal measurement period is determined by piecewise linear function approximation and dynamic programming optimization methods, and the measurement resources are reasonably allocated through an adaptive sampling strategy, which greatly reduces the measurement delay and ensures the measurement accuracy. The hidden Markov model is introduced for state tracking and error analysis. Combined with Bayesian parameter estimation and Viterbi algorithm, an accurate error compensation mechanism is established to control the average measurement error of the system within 0.1dB. The sub-band priority sorting and band switching optimization strategy based on the power stability index realize the continuity and real-time performance of the measurement process and improve the calibration efficiency. The optimal configuration of system resources is achieved by establishing the sampling parameter mapping relationship and iterative optimization solution, avoiding resource waste and ensuring the measurement accuracy requirements. The distributed calibration compensation scheme is adopted to integrate the calibration data of multiple bands, improve the consistency and reliability of the calibration results, and enable the system to maintain stable calibration performance in long-term operation.

[0153] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0154] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0155] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An ultra-wideband spectrum sensing radio measurement calibration method, characterized in that: The method comprises: Perform time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix; Performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix; Calculating the change rate and analyzing the frequency band switching timing of the power density data of each sub-band in the effective power density matrix to obtain a sub-band scanning sequence and a minimum measurement time; Adaptively optimizing system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set; The target sampling parameter set is used to remeasure the power density of each sub-frequency band to obtain a measurement result, and the measurement result is subjected to error statistics to obtain a calibration compensation matrix, wherein the calibration compensation matrix contains compensation coefficients of the frequency points to be calibrated; specifically, the following steps are used: sampling and controlling each sub-frequency band according to the target sampling parameter set to obtain an original remeasurement data sequence, and performing time-frequency transformation on the original remeasurement data sequence to obtain a remeasurement power density sequence; constructing a hidden Markov observation sequence for the remeasurement power density sequence to obtain a power density state transfer matrix, and performing forward algorithm calculation based on the power density state transfer matrix to obtain a state probability distribution sequence; state decoding is performed on the remeasurement power density sequence according to the state probability distribution sequence to obtain a power density measurement state sequence, and Viterbi path backtracing is performed on the power density measurement state sequence to obtain an optimal state path; and measurement error analysis is performed on the optimal state path. , obtain a power density error statistical matrix, and perform Bayesian parameter estimation based on the power density error statistical matrix to obtain an error distribution model; classify and count the sub-band measurement errors according to the error distribution model to obtain a frequency band error distribution sequence, and perform threshold detection on the frequency band error distribution sequence to obtain a set of frequency points to be calibrated; calculate the compensation coefficients of the power density error statistical matrix based on the set of frequency points to be calibrated to obtain an initial compensation coefficient matrix, and perform regularization on the initial compensation coefficient matrix to obtain a normalized compensation matrix; optimize the compensation parameters according to the normalized compensation matrix and the optimal state path to obtain an optimal compensation coefficient sequence, and perform frequency band mapping on the optimal compensation coefficient sequence to obtain a compensation parameter distribution vector; perform matrix reconstruction on the compensation parameter distribution vector to obtain a calibration compensation matrix, wherein each element of the calibration compensation matrix corresponds to a compensation coefficient of a frequency point to be calibrated.

2. The ultra-wideband spectrum sensing radio measurement calibration method according to claim 1, characterized in that: The method of performing time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix includes: Performing sampling frequency and sampling duration configuration on multiple receiving channels to obtain receiving channel configuration parameters, wherein the receiving channel configuration parameters include sampling frequency values ​​and sampling duration values; Synchronously collecting the millimeter-wave ultra-wideband spectrum signals of each receiving channel according to the receiving channel configuration parameters to obtain an original signal data sequence, and performing fast Fourier transform processing on the original signal data sequence to obtain frequency domain distribution data; Dividing the frequency domain distribution data into sub-bands according to the system sampling rate to obtain a sub-band boundary sequence, and performing power calculation on the frequency domain distribution data in each sub-band to obtain a sub-band power distribution feature; Perform frequency domain diffusion analysis based on the sub-band power distribution characteristics to obtain a power leakage coefficient matrix, and construct a power density data structure based on the sub-band power distribution characteristics and the power leakage coefficient matrix to obtain initial power density data; The initial power density data is matrixed to obtain an initial power density matrix, wherein rows of the initial power density matrix represent different receiving channels, and columns represent power density values ​​of different sub-frequency bands.

3. The ultra-wideband spectrum sensing radio measurement calibration method according to claim 2, characterized in that: The performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix includes: Extracting a column vector from the initial power density matrix to obtain a sub-band power density sequence, wherein the sub-band power density sequence includes power density values ​​of each receiving channel in the same sub-band; Normalizing the sub-band power density sequence to obtain normalized power density data, and performing autocorrelation function calculation on the normalized power density data to obtain a power density correlation coefficient matrix, wherein the power density correlation coefficient matrix reflects the signal correlation degree between each receiving channel; Performing a singular value decomposition operation based on the power density correlation coefficient matrix to obtain a signal subspace and a noise subspace, and performing a signal-to-noise ratio calculation based on an eigenvalue ratio of the signal subspace to the noise subspace to obtain a sub-band signal-to-noise ratio sequence; Performing threshold judgment processing on the sub-band signal-to-noise ratio sequence to obtain a valid signal judgment result, and performing data screening on the initial power density matrix according to the valid signal judgment result to obtain preliminary screening power density data; The primary screening power density data is subjected to matrix reconstruction to obtain an effective power density matrix, wherein the effective power density matrix only includes power density data determined by the signal-to-noise ratio.

4. The ultra-wideband spectrum sensing radio measurement calibration method according to claim 3, characterized in that: The step of calculating the rate of change of the power density data of each sub-band in the effective power density matrix and analyzing the frequency band switching timing to obtain a sub-band scanning sequence and a minimum measurement time includes: Performing time window segmentation processing on the power density data of each sub-band in the effective power density matrix to obtain a sub-band power density time series, and performing a difference operation on the sub-band power density time series to obtain a power variation sequence; Calculating the power change rate matrix of each sub-frequency band according to the power change amount sequence to obtain a power stability index sequence, and performing cluster analysis on the power stability index sequence to obtain a sub-frequency band stability grade classification result; Constructing a sub-frequency band priority ranking based on the sub-frequency band stability level division result to obtain an initial scanning sequence, and calculating the frequency span between adjacent sub-frequency bands according to the initial scanning sequence to obtain a frequency band switching time sequence; Dynamically optimize the frequency band switching time-consuming sequence to obtain a sub-frequency band scanning sequence, and perform time accumulation according to the sub-frequency band scanning sequence and the frequency band switching time-consuming sequence to obtain a minimum measurement time.

5. The ultra-wideband spectrum sensing radio measurement calibration method according to claim 4, characterized in that: The step of adaptively optimizing the system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set includes: Analyzing the power change rate matrix of each sub-frequency band according to the sub-frequency band scanning sequence to obtain a minimum sampling requirement matrix, and performing sampling resource allocation calculation on the minimum sampling requirement matrix and the minimum measurement time to obtain a sub-frequency band sampling point number sequence; Calculating the sampling rate according to the sub-band sampling point number sequence and the bandwidth characteristics of each sub-band to obtain an initial sampling rate sequence, and weighted fusion of the initial sampling rate sequence and the power stability index sequence to obtain an optimized sampling rate sequence; A sampling parameter mapping relationship is constructed based on the optimized sampling rate sequence and the sub-band sampling point number sequence to obtain a sampling parameter optimization model, and the sampling parameter optimization model is iteratively optimized and solved to obtain a sampling parameter optimization result; The sampling configuration information of each sub-band is generated according to the sampling parameter optimization result to obtain a sub-band sampling scheme sequence, and the sub-band sampling scheme sequence is integrated with parameters to obtain a target sampling parameter set.

6. The ultra-wideband spectrum sensing radio measurement calibration method according to claim 5, characterized in that: The step of constructing a sampling parameter mapping relationship based on the optimized sampling rate sequence and the sub-band sampling point number sequence to obtain a sampling parameter optimization model, and iteratively optimizing and solving the sampling parameter optimization model to obtain a sampling parameter optimization result includes: Performing correlation analysis on the sampling rate data of each sub-frequency band in the optimized sampling rate sequence to obtain a sampling rate correlation matrix, and reorganizing the data of the sub-frequency band sampling point number sequence according to the sampling rate correlation matrix to obtain a sampling resource distribution vector; Performing bilinear mapping on the sampling rate correlation matrix and the sampling resource distribution vector to obtain an initial mapping matrix of sampling parameters; An optimization function is constructed based on the sampling parameter initial mapping matrix to obtain a sampling resource optimization objective function, and constraints are added to the sampling resource optimization objective function according to system measurement accuracy requirements and hardware resource limitations to obtain a sampling parameter optimization model; Performing convex optimization decomposition processing on the sampling parameter optimization model to obtain a sub-problem sequence, and solving the sub-problem sequence by an alternating direction multiplier method to obtain an iterative optimization sequence; Convergence judgment and parameter update are performed on the iterative optimization sequence to obtain sampling parameter optimization results.

7. An ultra-wideband spectrum sensing radio measurement and calibration system, characterized in that: Used to perform the ultra-wideband spectrum sensing radio metrology calibration method according to any one of claims 1 to 6, the ultra-wideband spectrum sensing radio metrology calibration system comprising: A transformation module is used to perform time-frequency domain transformation and sliding interception on the millimeter-wave ultra-wideband spectrum signals collected by multiple receiving channels to obtain an initial power density matrix; An analysis module, used for performing correlation analysis and signal-to-noise ratio calculation on the signals of each sub-band in the initial power density matrix to obtain an effective power density matrix; A calculation module, used to calculate the change rate of the power density data of each sub-band in the effective power density matrix and analyze the frequency band switching timing to obtain a sub-band scanning sequence and a minimum measurement time; An optimization module, configured to adaptively optimize system sampling parameters according to the sub-band scanning sequence and the minimum measurement time to obtain a target sampling parameter set; The error statistics module is used to remeasure the power density of each sub-band using the target sampling parameter set to obtain a measurement result, and to perform error statistics on the measurement result to obtain a calibration compensation matrix, wherein the calibration compensation matrix includes compensation coefficients of the frequency points to be calibrated.

Images