Radio signal measuring method for OFDM (Orthogonal Frequency Division Multiplexing) detection

Abstract

The invention discloses a radio signal measurement method for OFDM (Orthogonal Frequency Division Multiplexing) detection, which relates to the field of signal processing and comprises the following steps: S1, acquiring received power data of a signal downlink and channel response data of a signal uplink measured by a target radio base station in an FDD (Frequency Division Duplexing) process, and establishing a channel mapping model; s2, screening out a first-order low-ebb frequency band with frequency fading, and allocating an MIMO multiplexing layer to the first-order low-ebb frequency band; s3, after an MIMO multiplexing layer is allocated, marking the MIMO multiplexing layer as a second-order off-peak frequency band, and performing fading repair on the second-order off-peak frequency band by using beam forming; and S4, comparing the re-measured receiving power data with the reference power range, and if the re-measured receiving power data do not accord with the reference power range, returning to the step S2. According to the method, the high-frequency-band deep fading area is enhanced through multi-stage repair such as channel mapping, MIMO optimization and beam forming, so that the signal repair process can be continuously adjusted, and the method has the advantages of remarkably reducing signal loss and improving power measurement stability.

Description

Technical Field

[0001] The present invention relates to the field of signal processing, and particularly to a radio signal measurement method for OFDM detection. Background Art

[0002] As one of the important indicators of the physical layer characteristics of the downlink signal of a radio base station, the reference signal received power, usually expressed as the RSRP value, directly reflects the intensity of the received signal at the receiving point. The measurement of signal intensity is an important link in radio supervision. Software-defined radio can, without changing the hardware platform, determine the functions implemented by the entire system through software programming. It transfers the signal processing work to a general-purpose computing device as much as possible, rather than designing a dedicated circuit board to implement signal processing. Currently, the common system architectures of software-defined radio mainly include analog-to-digital / digital-to-analog conversion for reference signals, DSP technology, and orthogonal frequency division multiplexing processing.

[0003] In the existing process of using orthogonal frequency division multiplexing to detect the received power of the uplink and downlink signals of a target radio base station, the common FDD frequency division duplex communication mode is used. Currently, in the FDD frequency division duplex structure, due to different uplink and downlink frequencies, the attenuation degree of the wireless channel for signals of different frequencies is different, resulting in signal non-uniformity in the frequency domain, and thus prone to selective fading of the detection frequency of the target detection signal. At the same time, due to different channel responses in the FDD system, the impact of frequency selective fading in the uplink and downlink is significantly different. The uplink frequency is lower and the multipath effect is weaker; while the downlink frequency is higher, the penetration ability is weaker, and multipath reflection and fading are more obvious, which more affects the calculation of the signal received power value, and further leads to the measurement result of the signal received power value not accurately reflecting the actual channel condition of the communication terminal, especially the channel frequency fading is more serious when detecting in the high-frequency band. Traditional signal received power measurement methods mainly consider the time domain and power mean, and lack consideration of the impact of frequency selective fading. Summary of the Invention

[0004] The present invention provides a radio signal measurement method for OFDM detection, which solves the problem that when detecting the signal received power during the FDD frequency division duplex process of frequency division multiplexing, the target signal is prone to frequency fading in the high-frequency downlink transmission link and the correction effect is not good.

[0005] The present invention is achieved by the following technical solutions: A radio signal measurement method for OFDM detection, the method comprising: Step S1: Obtain the received power data of the signal downlink and the channel response data of the signal uplink measured by the target radio base station during the FDD process, and establish a channel mapping model based on the received power data and the channel response data; Step S2: Use the channel mapping model to screen out the subcarrier frequency bands with frequency fading in the subcarrier frequency band of the signal downlink and label them as the first-order low valleys. Use the MIMO multiplexing optimization method to allocate MIMO multiplexing layers to the first-order low valleys; Step S3: Label the first-order low valleys with allocated MIMO multiplexing layers as the second-order low valleys. Use beamforming to repair the fading of the second-order low valleys and generate beamforming weights for each second-order low valley; Step S4: Apply the beamforming weights to the beamforming repair process of each second-order low valley, and set the reference power range. Retake the received power data of the repaired signal downlink, and compare the retaken received power data with the reference power range. If it does not meet the reference power range, return to Step S2.

[0006] Due to the different channel responses in the FDD system, the impact of frequency-selective fading on the uplink and downlink is significantly different. The uplink frequency is relatively low and the multipath effect is weak; while the downlink frequency is relatively high, the penetration ability is weak, and multipath reflection and fading are more obvious, which more affects the calculation of the signal received power value, and further leads to the measurement result of the signal received power value not being able to accurately reflect the actual channel condition of the communication terminal, especially the channel frequency fading is more serious when detecting in the high-frequency band. The traditional signal received power measurement method mainly considers the time domain and power mean, and lacks consideration of the impact of frequency-selective fading. Based on this, the present invention provides a radio signal measurement method for OFDM detection to solve the problem that the target signal is prone to frequency fading and the correction effect is not good in the high-frequency band downlink transmission link during signal received power detection in the frequency-division duplexing process of FDD.

[0007] Further, the construction content of the channel mapping model includes: aligning the channel response data and the received power data in the form of interpolation timestamps, where each subcarrier frequency band contains at least one timestamp; mapping the channel response data to the signal downlink through linear frequency interpolation at the timestamp, labeling the channel response data located in the signal downlink after interpolation as the mapped response data, using the mapped response data to screen out the subcarrier frequency bands with frequency fading, and labeling the screened subcarrier frequency bands as the first-order low valleys.

[0008] Further, the acquisition process of the mapped response data includes: marking the uplink frequency points representing the signal uplink and the downlink frequency points representing the signal downlink at all timestamps; setting the uplink frequency point as FU, and setting the uplink frequency point ordinal number as i, and the frequency value of the i-th uplink frequency point is represented as FU(f i ); Set the downlink frequency point as FD, and set the downlink frequency point ordinal number as k, and the frequency value of the k-th downlink frequency point is represented as FD(fk );Let the mapped response data be denoted as HD and the interpolation weight be α. Then the calculation formula for the mapped response data of the k-th downlink frequency point is expressed as: , where f i' ‌ and f i'+1 represent the two uplink frequency points closest to the downlink frequency point f k .

[0009] Furthermore, the setting process of the interpolation weight is set as: The calculation formula for the interpolation weight α is expressed as: .

[0010] Furthermore, let the uplink frequency band range of the signal uplink be represented as a set of uplink frequency points; The form of the uplink frequency band range is: , where n represents the total number of frequency points in the uplink frequency band range; When the downlink frequency point exceeds the uplink frequency band range, that is, the frequency value of the downlink frequency point FD(f k ) < the frequency value of the uplink frequency point FU(f 1 ), or the frequency value of the downlink frequency point FD(f k ) > the frequency value of the uplink frequency point FD(f n ), the uplink frequency point closest to it is directly assigned to the target downlink frequency point.

[0011] Furthermore, the process of allocating MIMO multiplexing layers to the first-order low valley frequency band includes: allocating 2 redundant layers to the low valley frequency band and using maximum ratio transmission precoding for the low valley frequency band to focus signal energy.

[0012] Furthermore, the process of screening out the first-order low valley frequency band includes: Converting the complex channel response value in the mapped response data to a power value and setting it as the response power value, adding the received power data to all the response power values within each subcarrier and taking the average to calculate the power average; setting a subcarrier power threshold for the power average, and marking the subcarrier frequency band with a power average lower than the subcarrier power threshold as the first-order low valley frequency band.

[0013] Furthermore, using the gradient descent method to generate the optimal beamforming weight for each second-order low valley frequency band, the process includes: Constructing a mean square error loss function to quantify the beamforming weight deviation of the second-order low valley frequency band; Setting the learning rate value, which represents the step size of each update of the actual repair power of the second-order low valley frequency band; Set the preset number of updates, use beamforming to set the initial value of the beamforming weight for each second-order trough frequency band, use the mean square error loss function to update the beamforming weight, end the iteration after reaching the number of updates, and apply the finally updated beamforming weight to the beamforming for the second-order trough frequency band.

[0014] Further, set the expected reference power and denote it as P r , the expected reference power P r is within the reference power range; let the beamforming weight be denoted as w, and let the power value after fading repair for each second-order trough frequency band under the setting of the beamforming weight w be denoted as P t (w), and let the mean square error loss function be denoted as J(w), then the calculation formula of the mean square error loss function J(w) is expressed as: .

[0015] Further, let the learning rate value be denoted as μ, let the iteration ordinal number be denoted as k, and the gradient operator be denoted as ∇, then the update process of the beamforming weight is expressed as: .

[0016] Compared with the prior art, the present invention accurately and actively identifies subcarriers with severe fading through the channel mapping model, adopts MIMO multiplexing optimization and beamforming repair, and enhances the deep fading area of the high-frequency band through multi-stage repair processes such as channel mapping, MIMO optimization, and beamforming, enabling the signal repair process to be continuously adjusted, having the advantages of significantly reducing signal loss and improving the stability of power measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of this application, and do not limit the embodiments of the present invention. In the drawings: Figure 1 is a schematic structural diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in combination with embodiments and drawings. The illustrative embodiments and descriptions thereof of the present invention are only used to explain the present invention and do not limit the present invention.

[0019] Embodiment 1 As Figure 1 shown, this embodiment is a radio signal measurement method for OFDM detection, and the method includes: Step S1: Obtain the received power data of the signal downlink and the channel response data of the signal uplink measured by the target radio base station during the FDD process, and establish a channel mapping model based on the received power data and the channel response data; Step S2: Use the channel mapping model to screen out the subcarrier frequency bands with frequency fading in the subcarrier frequency band of the signal downlink and label them as the first-order low valley frequency bands, and allocate MIMO multiplexing layers to the first-order low valley frequency bands using the MIMO multiplexing optimization method; Step S3: Label the first-order low valley frequency bands allocated with MIMO multiplexing layers as the second-order low valley frequency bands, perform fading repair on the second-order low valley frequency bands using beamforming, and generate beamforming weights for each second-order low valley frequency band; Step S4: Apply the beamforming weights to the beamforming repair process of each second-order low valley frequency band, and set the reference power range. Re-measure the received power data of the repaired signal downlink, and compare the re-measured received power data with the reference power range. If it does not meet the reference power range, return to Step S2.

[0020] The FDD is a common frequency-division duplex mode in an OFDM wireless communication system. It realizes two-way communication by allocating uplink and downlink signals to different frequencies. In the FDD mode, the uplink and downlink use different frequencies and maintain a fixed frequency interval between them. The signal downlink refers to the signal transmission link from the target radio base station to the terminal. In specific applications, the terminal calculates the received power data by receiving the reference signal sent by the base station. The signal uplink refers to the signal transmission link from the terminal to the base station. In specific applications, the base station receives the pilot signal sent by the terminal and uses the pilot process to measure the channel response data. The channel response data mainly includes the channel frequency response, which represents the amplitude and phase characteristics of each subcarrier. The received power data represents the signal strength of each subcarrier in the current downlink. The channel response data represents the influence of the channel on different frequencies. In this embodiment, a channel mapping model is constructed to establish the mapping relationship between the uplink and downlink channels, so as to predict the signal fading situation of each subcarrier in the downlink. The channel mapping model is used to screen out the subcarrier frequency bands with frequency fading in the subcarrier frequency band of the signal downlink, and the screened subcarrier frequency bands with frequency fading are marked as the first-order low valley frequency bands, which is the first-stage screening process. The MIMO multiplexing layer represents the allocation level of different spatial streams in the MIMO environment, that is, how to use multiple transmit antennas and receive antennas to enhance the signal. In an OFDM system, each subcarrier may be subject to different degrees of fading. Generally, only one MIMO multiplexing layer may be used on a single subcarrier. For subcarriers that may have severe fading, multiple MIMO multiplexing layers can be additionally allocated, that is, the spatial multiplexing degree is increased to improve its anti-fading ability. In specific applications, to increase the MIMO multiplexing layer for the first-order low valley frequency band, multiplexing methods such as STBC / SFBC / V-BLAST can be used to enhance the signal.

[0021] Furthermore, the beamforming is a technology that enhances or focuses signals in a specific direction while suppressing interference in other directions by controlling the transmission weights of the antenna array, that is, the phase and amplitude of the signals transmitted by different antennas. In wireless communication, beamforming can enhance the strength of the received signal through the collaborative work of multiple antennas, especially for frequency bands with fading. After screening out the first-order low valley frequency bands through the channel mapping model and using MIMO multiplexing optimization for preliminary frequency fading correction, there may still be a large signal attenuation for each subcarrier. In this embodiment, beamforming is used to perform further frequency band correction, so that the signals in the second-order low valley frequency bands are strengthened, thereby compensating for the signal loss caused by fading. Fading may be caused by factors such as different obstacles, reflections, and multipath propagation encountered during signal propagation. Generating beamforming weights means calculating and allocating the corresponding antenna transmission weights for each second-order low valley frequency band. Beamforming repairs the fading of the subcarrier beams in the second-order low valley frequency bands by generating and applying beamforming weights, and tailors a phase and amplitude setting for transmitting signals for each frequency band, so that the signals in that frequency band are enhanced and the quality of signal reception is improved. The reference power range is used to evaluate whether the measured received power meets the expected standard, so as to determine whether the signal repair process is successful. Its specific setting process can be defined by historical data, empirical rules, or communication standards (such as 3GPP, IEEE, etc.) to ensure that the signal quality and coverage in different environments meet the requirements. For example, in the LTE system, the reference range of RSRP is generally set between -140 dBm and -44 dBm, and appropriate power values are set based on different network conditions, signal quality requirements, and the adaptability of user equipment.

[0022] Further, as a feasible implementation manner, the process of allocating MIMO multiplexing layers to the first-order low valley frequency bands includes: allocating 2 redundant layers to the low valley frequency bands and using maximum ratio transmission precoding for the low valley frequency bands to focus signal energy.

[0023] After detecting the first-order low-frequency band, two additional parallel transmission layers are added, and the same data is transmitted using two additional antennas to increase the spatial diversity gain and improve the anti-fading ability. It is used to enhance signal strength and reliability. These additional MIMO multiplexing layers do not transmit new data, but are used to improve the reception reliability of the original data, that is, to increase redundancy and reduce the impact of signal fading. The maximum ratio transmission precoding is used to adjust the phase and amplitude of the signals transmitted by different antennas, so that the signals of all transmitting antennas are phase coherent at the receiving end, maximizing the focusing of signal energy. The focused signal energy means that the signals of the transmitting antennas are adjusted through MRT precoding to ensure that when the signals are superimposed at the receiving end, a maximum signal gain can be formed. Even in the case of large fading in the low-frequency band, the received signal energy is still as strong as possible.

[0024] Further, as a feasible implementation, the construction content of the channel mapping model includes: aligning the channel response data and the received power data in the form of interpolation timestamps, where each subcarrier frequency band contains at least one timestamp; mapping the channel response data to the signal downlink through linear frequency interpolation at the timestamp, labeling the channel response data located in the signal downlink after interpolation as the mapped response data, using the mapped response data to screen out the subcarrier frequency bands with frequency fading, and labeling the screened subcarrier frequency bands as the first-order low-frequency band.

[0025] The channel mapping model is expressed as an uplink to downlink mapping relationship constructed based on a mathematical method of channel measurement. The purpose of the common mapping model is mainly based on the measured uplink channel response data, mapped to the position of the downlink channel through a mathematical relationship. The downlink and uplink frequency bands of the FDD system are different, and there are frequency selective fading and path loss differences, so it is necessary to establish a mapping relationship model; the channel relationship mapping between different frequency bands (such as 3.5GHz and 28GHz bands) supports cross-band scheduling optimization. After statistical analysis of a large amount of historical data, the signal data points before mapping construct a conversion relationship from the signal uplink to the signal downlink to form the original data set. The process may include data normalization, noise removal, and missing value filling and other data processing processes. In the prior art, the commonly used methods for constructing mapping models include polynomial regression or least squares method to fit the relationship between downlink power and uplink channel response. In this embodiment, the content of the channel mapping model is the statistical modeling mapping relationship between the received power data of the signal downlink and the channel response data of the signal uplink. Its working process is a method process for mapping the signal uplink to the signal downlink in a specific implementation method during the FDD measurement of the target radio base station; the base station coverage can be evaluated and the boundary power allocation of the target radio base station can be optimized through channel mapping. The timestamp refers to the assignment of a specific time mark to each data point, indicating the exact time when the data point is collected or generated. The channel response data and the received power data may use different time sampling rates, or there is a time delay, so one of the data sets needs to be interpolated to align the two data sets. The channel response data and the received power data are marked and aligned by timestamps to ensure that the data at each time point can be directly associated, and each subcarrier frequency band contains at least one timestamp to ensure that each subcarrier can be screened to avoid omissions. The channel response data corresponding to each timestamp is mapped to the spectrum of the signal downlink by a linear interpolation method, so as to obtain the value of the target frequency point by a linear calculation method based on the data between two adjacent frequency points. The linear frequency interpolation extrapolates the original discrete channel response data to generate data at more frequency points of the signal downlink. The purpose of the interpolation process is to enable the channel response data to be connected to the spectrum of the downlink to reflect the characteristics of the signal downlink. The mapped response data is used to screen out subcarrier frequency bands with frequency fading, mainly by analyzing the mapped channel response data to identify those frequency bands where the signal quality is significantly reduced due to factors such as frequency selective fading. In a specific implementation, the channel gain of each subcarrier frequency band can be calculated by mapping the response data, and the signal frequency fading can be intuitively judged by setting a threshold based on the channel gain or comparing adjacent frequency bands.

[0026] Example 2 The process of obtaining the mapping response data includes: marking the uplink frequency points representing the signal uplink and the downlink frequency points representing the signal downlink at all time stamps; setting the uplink frequency point as FU, and setting the uplink frequency point ordinal number as i, and the frequency value of the i-th uplink frequency point is represented as FU(f i ); setting the downlink frequency point as FD, and setting the downlink frequency point ordinal number as k, and the frequency value of the k-th downlink frequency point is represented as FD(f k ); setting the mapping response data as HD, and the interpolation weight as α, then the calculation formula of the mapping response data of the k-th downlink frequency point is expressed as: , where f i' ‌ and f i'+1 represent the two uplink frequency points closest to the downlink frequency point f k .

[0027] The interpolation weight α is used to represent the relationship between the two uplink frequency points and the downlink frequency point; the f i represents the i-th frequency point. The f i' ‌ and f i'+1 are represented as the two uplink frequency points closest to the downlink frequency point f k , that is, FU(f i' ‌) and FU(f i'+1 ) represent the frequency values of the two uplink frequency points closest to the downlink frequency point FD(f k ). The meaning of i' is the same as that of i, both representing the uplink frequency point ordinal number, but i' also represents the closest distance relationship to the downlink frequency point f k . (1 - α)∙FU(f i' ‌) represents the contribution part of the uplink frequency point, (1 - α) represents the relative position relationship between the downlink frequency point f k and the uplink frequency point f i' , which is the channel response data of the uplink frequency point, usually the signal quality of a specific frequency point in the uplink, specifically, it can be signal strength, gain or signal-to-noise ratio, etc.; α∙FU(f i'+1 ‌) represents another contribution part of the uplink frequency point, representing the channel response data of the (i + 1)-th uplink frequency point, usually the signal quality of another frequency point in the uplink, corresponding to FU(f i' ‌). By performing interpolation calculation on the distance relationship between the uplink and downlink frequency points, the response data of the downlink frequency point is obtained. This method can make full use of the information of the uplink to calculate the channel state when the measurement accuracy of the downlink is poor, so as to correct the power value of the signal downlink.

[0028] Furthermore, as a feasible implementation method, the setting process of the interpolation weight is set as: The calculation formula of the interpolation weight α is expressed as: .

[0029] The k' represents the downlink frequency point ordinal number in the process of target calculation, and the FD(f k' ) represents the frequency value of the kth downlink frequency point in the process of target calculation. The interpolation weight α is a numerical value between 0 and 1, which is a linear interpolation coefficient, reflecting the position of the downlink frequency point relative to the uplink frequency point and used to balance the interpolation contributions of two uplink frequency points. The FD(f k' ) - FU(f i' ‌) represents the frequency difference between the target downlink frequency point f k' and the uplink frequency point f i' of one of the frequency bands with the closest adjacent distance currently being calculated, indicating the relative position of the target downlink frequency point f k' in this uplink frequency point interval. The larger this difference is, the greater the gap between f k' and f i' , so the interpolation weight α is larger, indicating that the downlink frequency point depends more on f i'+1 . The FU(f i'+1 ‌) - FU(f i' ‌) represents the frequency difference between two adjacent uplink frequency points, which is used to normalize the relative position of the target downlink frequency point f k' in this interval, so that the interpolation calculation varies between [0, 1]. This difference reflects the span of the uplink frequency points and determines the relative position of the downlink frequency point within this range.

[0030] When substituting the calculation formula of the interpolation weight α into the calculation formula of the mapping response data of the kth downlink frequency point: When α = 0, the target downlink frequency point f k' is exactly equal to f i' , that is, at this time, the target downlink frequency point f k' completely uses the uplink response data located at f i' .

[0031] When α = 1, the target downlink frequency point f k' is exactly equal to f i'+1 , that is, at this time, the target downlink frequency point f k' completely uses the uplink response data at f i'+1 .

[0032] When 0 < α < 1, the target downlink frequency point f k' is between f i' and f i'+1 , and at this time, its corresponding mapping response data is calculated by linear interpolation.

[0033] In this way, this embodiment can perform weighted interpolation on the downlink frequency point between two uplink frequency points.

[0034] Further, as a feasible implementation manner, let the uplink frequency band range of the signal uplink be represented as a set of uplink frequency points; The form of the uplink frequency band range is: , where n represents the total number of frequency points in the uplink frequency band range; When the downlink frequency point exceeds the uplink frequency band range, that is, the frequency value FD(f k ) < the frequency value FU(f 1 ) of the uplink frequency point, or the frequency value FD(f k ) > the frequency value FD(f n ) of the uplink frequency point, directly assign the nearest uplink frequency point to the target downlink frequency point.

[0035] When FD(f k ) is less than the minimum frequency point FU(f i' ) of the uplink frequency band or greater than the maximum frequency point FU(f n ‌), match the frequency value at the frequency point f k with the frequency value of the uplink frequency point closest to f k . When the downlink frequency point is less than the minimum frequency point of the uplink frequency band, directly assign FU(f i ) to the downlink frequency point FD(f k ), indicating that the signal of the downlink frequency point can directly use the response data of the starting frequency point of the uplink frequency band; when the downlink frequency point is greater than the maximum frequency point of the uplink frequency band, directly assign FU(f n ‌) to the downlink frequency point FD(f k ), indicating that the signal of the downlink frequency point can directly use the response data of the ending frequency point of the uplink frequency band. The purpose of this implementation manner is to "map" the frequency value of the downlink frequency point to the nearest uplink frequency point to ensure that even if the downlink frequency point exceeds the uplink frequency band range, the data of the uplink link signal can still be used for estimation. This processing method can avoid data loss or calculation errors when the frequency point exceeds the range, ensuring that the system can still operate normally even when the frequency band exceeds the range and can reasonably process the downlink signal.

[0036] Further, as a feasible implementation manner, the process of screening out the first-order low-frequency valley band includes: Convert the complex channel response values in the mapped response data into power values and set them as the response power values. Add all the response power values within each subcarrier to the received power data and then take the average to calculate the power average. Set a subcarrier power threshold for the power average, and label the subcarrier frequency bands with power averages lower than the subcarrier power threshold as first-order low valley frequency bands.

[0037] The complex channel response value refers to the complex data form in the channel response data. To calculate the signal strength, we need to convert it into the actual power value because power is a key indicator for measuring the signal strength. In specific implementation, for each complex channel response value, its power can be calculated by the square of its amplitude modulus to obtain the response power value. By adding the response power value to the received power data and taking the average, the signal quality of each subcarrier frequency band can be determined more accurately, especially in an environment with frequency-selective fading. This helps to screen out the low valley frequency bands, that is, the frequency bands with poor signal quality, providing effective data support for subsequent optimization processing. The subcarrier power threshold is a set threshold used to distinguish the high and low signal strengths. If the power average of a subcarrier is lower than the subcarrier power threshold, it means that the signal of this subcarrier frequency band is weak and there may be frequency fading. The subcarrier power threshold can be set according to engineering experience and network deployment requirements.

[0038] Embodiment 3 Use the gradient descent method to generate the optimal beamforming weights for each second-order low valley frequency band, and the process includes: Construct a mean square error loss function to quantify the beamforming weight deviation of the second-order low valley frequency band; Set the learning rate value, which represents the step size of each update of the actual repair power of the second-order low valley frequency band; Preset the number of updates. Use beamforming to set the initial value of the beamforming weights for each second-order low valley frequency band, use the mean square error loss function to update the beamforming weights, end the iteration after reaching the number of updates, and apply the finally updated beamforming weights to the beamforming for the second-order low valley frequency band.

[0039] The mean square error loss function is used to quantify the gap between the predicted beamforming weights and the actual patching power, and its goal is to make the received signal power after beamforming as close as possible to the target power. The learning rate value determines the step size for each update and controls the magnitude of the weight update during each gradient descent iteration. The number of updates specifies how many iterations the gradient descent method will perform; during each iteration, the beamforming weights will be adjusted according to the gradient of the loss function to minimize the error. Through iterative updates, the received power in each second-order low-frequency band is made close to the normal power value, reducing frequency fading. The process of using the gradient descent method to generate beamforming weights for the second-order low-frequency band is essentially to continuously optimize the weights to reduce the beamforming error and make the received power of the frequency band close to the target value.

[0040] Further, as a feasible implementation, a desired reference power is set and denoted as P r , and the desired reference power P r is within the reference power range; let the beamforming weights be denoted as w, and let the power value after fading patching for each second-order low-frequency band under the setting of the beamforming weights w be denoted as P t (w), and let the mean square error loss function be denoted as J(w). Then the calculation formula of the mean square error loss function J(w) is expressed as: .

[0041] The desired reference power P r is the preset target power, within the reference power range, and is a fixed value used as the standard or reference for optimization to ensure that the final fading patching can reach the expected power level. The power value P t (w) after beamforming patching is used to compare with the desired reference power P r to evaluate the effect of beamforming patching. The mean square error loss function J(w) is a common way to measure the gap between the desired reference power P r and the actual patching power P t (w), and quantifies the error between the two by squaring the differences and summing them. The smaller the value of the mean square error loss function J(w), the closer the power value P t (w) after beamforming patching is to the desired reference power P r . The goal is to minimize J(w) by adjusting w to achieve the best beamforming effect and ensure that the signal can be restored to the desired received power in the fading area. By minimizing the mean square error loss function, on each second-order low-frequency band, the power value P t (w) after beamforming patching is made as close as possible to the desired reference power P r , thereby achieving effective patching of frequency fading and improving signal quality.

[0042] Further, as a feasible implementation manner, let the learning rate value be denoted as μ, let the iteration ordinal number be denoted as k, and the gradient operator be denoted as ∇. Then the update process of the beamforming weight is expressed as: .

[0043] The w k represents the current value of the beamforming weight in the k-th iteration, and the weight value will be updated in each iteration process to optimize the fading repair effect. The w k+1 represents the beamforming weight after the (k + 1)-th iteration, that is, the weight updated by gradient descent. Its goal is to gradually reduce the value of the loss function J(w), so that the power value after beamforming repair is closer to the desired reference power. The learning rate value μ is a constant that controls the update amplitude during each gradient descent. The gradient operator ∇ represents the gradient of the loss function J(w) with respect to the beamforming weight w, and ∇∙J(w) represents the partial derivative vector of the loss function J(w) with respect to the beamforming weight w, indicating the change rate of the loss function at the current weight value. The gradient operator ∇ provides the direction of the steepest descent of the loss function. The core idea of the gradient descent algorithm is to update the parameters along the opposite direction of the gradient because this can reduce the value of the loss function. According to the gradient and the learning rate, update the current beamforming weight. Specifically, the new beamforming weight w k+1 is obtained by subtracting the product of the gradient and the learning rate μ from the current weight w k . The updated weight will be adjusted in the direction of reducing the loss function J(w), gradually reducing the error. Through this iterative process, the beamforming weight w is continuously updated, so that the fading repair effect of each second-order low-frequency band is continuously optimized, and finally the repaired power value P t (w) is close to the desired reference power P r , improving the reception quality of the signal.

[0044] The above specific implementation manners further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above is only the specific implementation manners of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A radio signal measurement method for OFDM detection, characterized in that: The method includes: Step S1: acquiring downlink signal reception power data and uplink signal channel response data measured by the target radio base station in the FDD process, and establishing a channel mapping model based on the reception power data and the channel response data; Step S2: using the channel mapping model to screen out subcarrier frequency bands with frequency fading in the subcarrier frequency bands of the signal downlink and marking them as first-order trough frequency bands, and using the MIMO multiplexing optimization method to allocate MIMO multiplexing layers to the first-order trough frequency bands; Step S3: marking the first-order trough frequency band to which the MIMO multiplexing layer is allocated as a second-order trough frequency band, using beamforming to perform fading repair on the second-order trough frequency band, and generating a beamforming weight for each second-order trough frequency band; Step S4: Apply the beamforming weight to the beamforming repair process of each second-order trough frequency band, set the reference power range, re-measure the receiving power data of the signal downlink after the repair, and compare the re-measured receiving power data with the reference power range. If it does not meet the reference power range, return to step S2.

2. A radio signal measurement method for OFDM detection according to claim 1, characterized in that: The construction of the channel mapping model includes: aligning the channel response data and the received power data in the form of interpolation timestamps, wherein each subcarrier frequency band contains at least one timestamp; mapping the channel response data to the signal downlink through linear frequency interpolation at the timestamp, marking the channel response data located in the signal downlink after the interpolation as mapping response data, using the mapping response data to screen out the subcarrier frequency bands with frequency fading, and marking the screened subcarrier frequency bands as first-order trough frequency bands.

3. A radio signal measurement method for OFDM detection according to claim 2, characterized in that: The acquisition process of the mapping response data includes: marking the uplink frequency representing the uplink signal and the downlink frequency representing the downlink signal on all time stamps; assuming that the uplink frequency is denoted as FU, and the uplink frequency sequence is set to i, and the frequency value of the i-th uplink frequency is denoted as FU(f i ); Let the downlink frequency be denoted as FD, and the downlink frequency number be k, and the frequency value of the kth downlink frequency be denoted as FD(f k ); Let the mapped response data be HD, the interpolation weight be α, Then the mapping response data calculation formula of the kth downlink frequency point is expressed as: ， where f i' ‌ and f i'+1 Indicates the distance to the downlink frequency point f k The two nearest uplink frequencies.

4. A radio signal measurement method for OFDM detection according to claim 3, characterized in that: The interpolation weight setting process is set as: The calculation formula of the interpolation weight α is expressed as: .

5. A radio signal measurement method for OFDM detection according to any one of claims 3 to 4, characterized in that: Assume that the uplink frequency band range of the signal uplink is represented as a set of uplink frequency points; The uplink frequency band range is in the form of: , Where n represents the total number of frequencies in the uplink frequency band; When the downlink frequency exceeds the uplink frequency range, the frequency value of the downlink frequency FD (f k )<the frequency value of the uplink frequency point FU(f1), or the frequency value of the downlink frequency point FD(f k )>Frequency value of uplink frequency point FD(f n ), the nearest uplink frequency point is directly assigned to the target downlink frequency point.

6. A radio signal measurement method for OFDM detection according to claim 1, characterized in that: The process of allocating MIMO multiplexing layers to the first-order trough frequency band includes: allocating 2 redundant layers to the trough frequency band, and using maximum ratio transmission precoding to focus signal energy on the trough frequency band.

7. A radio signal measurement method for OFDM detection according to claim 2, characterized in that: The process of screening out the first-order trough frequency band includes: The complex channel response value in the mapped response data is converted into a power value and set as the response power value. All response power values ​​in each subcarrier are added to the received power data and the average is taken to calculate the power average. The subcarrier power threshold is set according to the power average, and the subcarrier frequency band whose power average is lower than the subcarrier power threshold is marked as a first-order trough frequency band.

8. The radio signal measurement method for OFDM detection according to claim 1, characterized in that: The gradient descent method is used to generate the optimal beamforming weights for each second-order trough frequency band. The process includes: Construct a mean square error loss function to quantify the beamforming weight deviation in the second-order trough frequency band; Set the learning rate value, which represents the update step size of the actual patch power in the second-order trough frequency band; The number of updates is preset, and beamforming is used to set the initial value of the beamforming weight for each second-order trough frequency band. The beamforming weight is updated using the mean square error loss function. After the number of updates is reached, the iteration ends, and the final updated beamforming weight is applied to the beamforming for the second-order trough frequency band.

9. A radio signal measurement method for OFDM detection according to claim 8, characterized in that: Set the desired reference power and denote it as P r , the expected reference power P r is within the reference power range; let the beamforming weight be w, let the power value after fading repair for each second-order trough frequency band under the beamforming weight w be P t (w), and let the mean square error loss function be J(w), Then the mean square error loss function J(w) is calculated as: .

10. A radio signal measurement method for OFDM detection according to claim 9, characterized in that: Let the learning rate value be μ, let the iteration number be k, and let the gradient operator be ∇. Then the updating process of the beamforming weights is expressed as: .

Images