A similarity quantification method for cross-band wireless MIMO channels

Abstract

The present invention discloses a method for quantifying the similarity of cross-band wireless MIMO channels, covering the three dimensions of space, time and frequency of the channel. Specifically, it includes: 1) building a channel measurement system and planning measurement cases to measure the impulse response of the cross-band wireless MIMO channel; 2) calculating the frequency domain similarity index, that is, the delay power spectrum density correlation coefficient of each frequency band; 3) calculating the spatial domain similarity index, that is, the angle power spectrum density correlation coefficient of the channel; 4) calculating the time domain similarity index, that is, the small-scale fading correlation coefficient of the time-varying channel; 5) weighting and merging the indicators of the three domains to generate the final similarity measurement index of the cross-band channel. The present invention proposes a similarity measurement method for cross-band wireless MIMO channels for the first time based on the small-scale fading characteristics of the channel. It can be used to quantify the similarity of channels in different frequency bands, which has guiding significance for the deployment of cross-band communication systems; it can also be used as a performance indicator for the accuracy evaluation of cross-band channel models.

Description

Technical Field

[0001] The invention belongs to the technical field of wireless communication channel measurement and modeling, and relates to a similarity quantification method for cross-band wireless MIMO (Multiple-Input Multiple-Output) channels. Background Art

[0002] 2019 marks the entry of China into the commercial era of the fifth-generation (5G) mobile communication. Following the tradition of "commercializing one generation and planning the next generation", research on the sixth-generation (6G) mobile communication has begun. The 6G network will achieve seamless global coverage and meet the reliable connection of people, machines and objects anytime and anywhere. The realization of this vision is inevitably accompanied by a surge in data transmission. In order to meet the massive data transmission and considering the scarcity of spectrum resources, cross-band collaborative transmission has become an inevitable trend. Therefore, studying the channel similarity of different frequency bands has important practical value and can provide guidance for the deployment of cross-band communication systems. In addition, measuring the accuracy of the cross-band wireless MIMO channel model requires examining whether the channel model truly reflects the similarity of the measured channels of the two frequency bands. The existing patent results include terminal equipment and base station equipment for channel similarity acquisition, but they do not focus on describing the construction and calculation of channel similarity indicators, but focus on how to feedback channel similarity to reduce system overhead. Therefore, it is necessary to propose an indicator for comprehensively measuring MIMO channel similarity to fill the gap in current related research. Summary of the invention

[0003] The present invention aims to provide a similarity quantification method for cross-band wireless MIMO channels, so as to solve the problem of quantifying the similarity of cross-band wireless MIMO channels required for system deployment and model evaluation.

[0004] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0005] A similarity quantification method for cross-band wireless MIMO channels includes the following steps.

[0006] Step S1: Build a cross-band wireless MIMO channel measurement system, plan test cases, and calculate channel impulse response;

[0007] Step S2, calculating the delay power spectral density of each frequency band according to the channel impulse response;

[0008] Step S3, calculating the delay power spectrum correlation coefficient after preprocessing the delay power spectrum density as a measurement index of frequency domain similarity;

[0009] Step S4, calculating the Bartlett angle power spectra of the transmitting and receiving ends of the channels in different frequency bands, combining them and then quantizing them;

[0010] Step S5, calculating the angular power spectrum correlation coefficient of channels in different frequency bands as a measure of spatial domain similarity;

[0011] Step S6: After removing the large-scale fading, calculate the small-scale fading correlation coefficient of the mobile channels in different frequency bands as a measure of time domain similarity.

[0012] Step S7: assign different weights to the similarity indices in the space, time and frequency domains, and then merge them to generate a final cross-band channel similarity index.

[0013] The step S1 specifically includes the following steps:

[0014] Step S101: perform direct connection calibration on the channel detection system to obtain a direct connection calibration signal y th (t, τ);

[0015] Step S102: Use the antenna array to obtain the received signal y propagated through the wireless channel r (t, τ); Note that cross-band MIMO wireless channel measurement should meet the following requirements: 1. The bandwidth used by each frequency band is exactly the same; 2. Each frequency band should use an independent measurement system with the same dynamic range and different measurement systems using the same trigger signal; 3. Different frequency bands should use antennas of the same shape as much as possible; 4. The length of a single snapshot in each frequency band should be kept as similar as possible, which requires that the switching modes of the RF switch systems at both ends of the transmitter and receiver be kept as consistent as possible.

[0016] Step S103: Calculate the measurement channel impulse response h(t, τ) based on the inverse Fourier transform IFFT(·), expressed as

[0017] h(t, τ)=IFFT{H(t, f)}=IFFT{Y r (t,f) / Y th (t, f)}

[0018] where Y r (t, f) and Y th (t, f) are y r (t, τ) and y th The frequency domain transfer function of (t, τ). The dimension of the channel impulse response in the expression is M t ×M r ×M f ×M s , where M t is the number of transmitting antennas, M r is the number of receiving antennas, Mf is the number of subcarriers, M s is the number of channel snapshots, and H(t, f) is the channel transmission matrix.

[0019] The step S2 specifically includes the following steps:

[0020] Calculate the frequency band f based on the channel impulse response c The delayed power spectral density is expressed as:

[0021]

[0022] In the formula The dimension is M f ×1 vector, |·| means taking the absolute value of each element in the matrix, and m and n represent the serial numbers of the receiving and transmitting antennas respectively.

[0023] The step S3 specifically includes the following steps, S301-S303 are performed separately for each compared frequency band:

[0024] Step S301: extract all delay points within the effective range of the delay power spectrum density.

[0025] Step S302: Determine all effective paths of the delay power spectrum density by searching for the local maximum value.

[0026] Step S303: Select the delay power spectrum density of the frequency band with fewer effective paths as a reference, and match the effective paths of the delay power spectrum densities of all frequency bands.

[0027] Step S304: Calculate the delay power spectrum density correlation coefficient, which is expressed as:

[0028]

[0029] in{·} T Represents a transpose operation.

[0030] Step S305: Calculate ρ for all m and n τ,m,n , and obtain ρ τ,m,n Statistics.

[0031] The step S4 specifically includes the following steps, which are performed separately for each compared frequency band:

[0032] Step S401: Calculate the angle power spectrum density of the channel to reach the angle power spectrum density And the departure angle power spectrum Here we use the Bartlett method, and its expressions are:

[0033]

[0034]

[0035] in,{·} H represents the Hermit transpose, and c(θ,φ) represents the steering vector when the pitch angle is θ and the azimuth angle is φ. Since the present invention is mainly carried out for a stationary channel, the angular power spectrum density of different antenna pairs is close, so when calculating the angular power spectrum density of the transmitting and receiving ends, the antenna pair with the highest signal-to-noise ratio can be selected for calculation.

[0036] Step S402: merge the departure and arrival angle power spectrum density matrices and quantize them, that is:

[0037]

[0038] where vec{·} represents the operation of vectorizing a matrix.

[0039] The step S5 specifically comprises the following steps:

[0040] Step S501, calculate the angular power spectrum density correlation coefficient of two frequency bands, the expression is as follows:

[0041]

[0042] Step S502: Calculate ρ for all f B,f , and obtain ρ B,f Statistics.

[0043] The step S6 specifically includes the following steps, S601-S603 are performed separately for each compared frequency band:

[0044] Step S601: Calculate the channel gain after removing small-scale fading:

[0045]

[0046] Where Δτ and Δt represent the delay resolution of the channel matrix and the duration of a single snapshot, respectively. W represents the window length, which is generally 40 wavelengths.

[0047] Step S602: Path loss is obtained by fitting the antenna gain and receiving and sending power of the transmitting and receiving ends:

[0048]

[0049] where d = τ LOS c, τ LOS Obtained by the time-delay power spectral density; Indicates the least squares fitting method and d k expression.

[0050] Step S603: remove the path loss from the channel gain to obtain a small-scale fading value:

[0051]

[0052] Step S604: Calculate the Pearson correlation coefficient of small-scale fading in two frequency bands:

[0053]

[0054] in ρ S When it is a negative number, it means that the small-scale fading of the two frequency bands is negatively correlated, a positive number indicates a positive correlation, and 0 indicates linear independence.

[0055] Step S605: Calculate ρ for all mobility measurement routes S , and obtain ρ S Statistics.

[0056] Step S701: merge the three indicators in the following manner:

[0057]

[0058] where α 1 , α 2 and α 3 Respectively represent the weights of the similarity index in the frequency domain, spatial domain, and time domain, satisfying α 1 +α 2 +α 3 =1.

[0059] The similarity calculation method of the cross-band wireless MIMO channel of the present invention has the following advantages:

[0060] Based on the collected channel impulse responses, the present invention calculates similarity indicators for the frequency domain, spatial domain and time domain respectively, and can comprehensively compare the similarities between various frequency bands. The indicators all have clear physical meanings and value ranges, and the quantification of similarities is achieved. It can support the study of the channel similarity of different frequency bands and provide a basis for measuring the accuracy of cross-band wireless MIMO channel models. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a flow chart of a similarity calculation method of a cross-band wireless MIMO channel of the present invention;

[0062] Figure 2 is the delay power spectrum density of different test frequency bands of the present invention;

[0063] Figure 3-1It is a schematic diagram of the power spectrum density of the delay before matching of the present invention;

[0064] Figure 3-2 It is a schematic diagram of the delay power spectrum density after matching of the present invention;

[0065] Figure 4 It is the cumulative distribution function (CDF) of the delay power spectral density correlation coefficient in the line of sight (LoS) scenario of the present invention.

[0066] Figure 5 is the CDF of the angular power spectrum density correlation coefficient under LoS of the present invention. DETAILED DESCRIPTION

[0067] In order to better understand the purpose, structure and function of the present invention, a similarity calculation method for cross-band wireless MIMO channels of the present invention is further described in detail below with reference to the accompanying drawings.

[0068] Step S1: Build a cross-band wireless MIMO channel measurement system, plan test cases, and calculate channel impulse response;

[0069] Step S2, calculating the delay power spectral density of each frequency band according to the channel impulse response;

[0070] Step S3, after preprocessing the delay power spectrum density, calculate the delay power spectrum correlation coefficient as a measurement index of frequency domain similarity;

[0071] Step S4, calculating the angular power spectra of channels in different frequency bands and quantizing them;

[0072] Step S5, calculating the angular power spectrum correlation coefficient of channels in different frequency bands as a measure of spatial domain similarity;

[0073] Step S6: Calculate the small-scale fading correlation coefficients of channels in different frequency bands as a measure of time domain similarity.

[0074] Step S7: assign different weights to the similarity indices in the space, time and frequency domains, and then merge them to generate a final cross-band channel similarity index.

[0075] Furthermore, the step S1 specifically includes the following steps:

[0076] Step S101: perform direct connection calibration on the channel detection system to obtain a direct connection calibration signal.

[0077] The transmitted signal is denoted as x(t), the impulse response of the measurement system and cable is denoted as g(t), and the impulse response of the channel is denoted as h(t). X(f), G(f), and H(f) represent the frequency domain transfer functions obtained by Fourier transform of x(t), g(t), and h(t), respectively. th (f) represents the channel transfer function when the transmitting and receiving ends are directly connected: th (t) = x(t)*g(t), where * represents convolution. r (f) represents the channel transmission function when using the antenna for wireless transmission: Y th (f) = X(f)·G(f).

[0078] The time domain signal and frequency domain signal obtained by antenna air interface transmission are as follows:

[0079] y r (t) = x(t)*g(t)*h(t)

[0080] Y r (f) = X(f)·G(f)·H(f)

[0081] Step S102, use the antenna array to obtain the received signal y(t, τ) propagated through the wireless channel. Note that cross-band MIMO wireless channel measurement requires that the channel acquisition environment of multiple frequency bands be as consistent as possible, and the channel measurement equipment and antenna configuration should also be as similar as possible, so that the propagation channels of the two frequency bands are comparable. The requirements to be met are summarized as follows: 1. The bandwidth used by each frequency band is exactly the same, 2. Each frequency band should use an independent measurement system, the dynamic range of each measurement system is the same, and different measurement systems use the same trigger signal, 3. Different frequency bands should use the same type of antenna as much as possible, 4. The length of a single snapshot in each frequency band should be kept as similar as possible, which requires that the switching mode of the RF switch system at both ends of the transmitter and receiver be kept as consistent as possible, or the difference is small.

[0082] Step S103: Calculate the measurement channel impulse response h(t, τ) based on the inverse Fourier transform IFFT(·), expressed as

[0083] h m，n (t, τ) = IFFT{H m，n (t, f)}=IFFT{Y r (t,f) / Y th (t, f)}, m=1, ..., M r , n=1,...,M t

[0084] where Y r (t, f) and Y th (t, f) respectively r (t, τ) and yth The frequency domain transfer function of (t, τ). The dimension of the channel impulse response in the expression is M t ×M r ×M f ×M s . Where M t is the number of transmitting antennas, M r is the number of receiving antennas, M f is the number of subcarriers, M s is the channel snapshot number, and H(t, f) is the channel transmission matrix.

[0085] The step S2 specifically includes the following steps:

[0086] Calculate the frequency band f based on the channel impulse response c The delayed power spectral density is expressed as:

[0087]

[0088] In the formula The dimension is M f ×1 vector, |·| means taking the absolute value of each element in the matrix, and m and n represent the serial numbers of the receiving and transmitting antennas respectively. s The data of the snapshots are averaged to improve the signal-to-noise ratio and ensure that M s Each snapshot is measured in a static scene. The delay power spectral density of the two test frequency bands is shown in Figure 2 shown.

[0089] The step S3 specifically includes the following steps, S301-S303 are performed separately for each compared frequency band:

[0090] Step S301: extract all delay points within the effective range of the delay power spectrum density. The effective range is generally 50 points before the peak of the delay power spectrum density to 150 delay points after the peak, and can also be adjusted according to actual conditions.

[0091] Step S302: Determine all effective paths of the delay power spectrum density by searching for the local maximum value. @R In the findpeaks function, the prominence parameter is controlled to screen the effective diameters so that the final number of effective diameters is similar to the number of effective diameters observed by the naked eye.

[0092] Step S303, select the delay power spectrum density of the frequency band with fewer effective paths as the benchmark, and match the effective paths of the delay power spectrum density of all frequency bands: for each effective path of the benchmark delay power spectrum density, select the path with the closest relative distance in the delay power density of another frequency band. If there is no such multipath within a certain range, set the power of the delay point in the other frequency band to 0; if it exists, set the matched multipath delay to infinity to prevent it from being searched again in subsequent matching processes. The matching of the delay power spectrum density of the two frequency bands is to eliminate the influence of different channel measurement equipment on the results. The delay power spectrum density before and after matching is as follows: Figure 3-1 and Figure 3-2 shown.

[0093] Step S304: Calculate the delay power spectrum density correlation coefficient, which is expressed as:

[0094]

[0095] in{·} T Represents a transpose operation.

[0096] Step S305: Calculate ρ for all m and n τ,m,n , and obtain ρ τ,m,n The statistical information of the delay power spectrum density under LoS is as follows: Figure 4 As shown, Figure 4 A comparison of the frequency domain similarity indicators between the measured data and two standardized models is also shown, which can be used for the evaluation and correction of the channel model.

[0097] The step S4 specifically includes the following steps, which are performed separately for each compared frequency band:

[0098] Step S401: Calculate the angle power spectrum density of the channel to reach the angle power spectrum density And the departure angle power spectrum Here we use the Bartlett method, and its expressions are:

[0099]

[0100]

[0101] in,{·} H represents the Hermit transpose, and c(θ,φ) represents the steering vector when the elevation angle is θ and the azimuth angle is φ. Since this patent is mainly carried out for stable channels, the angular power spectrum density of different antenna pairs is close. Therefore, when calculating the angular power spectrum density of the transmitting and receiving ends, the antenna pair with the highest signal-to-noise ratio can be selected for calculation.

[0102] Step S402: merge the departure and arrival angle power spectrum density matrices and quantize them, that is:

[0103]

[0104] where vec{·} represents the operation of vectorizing a matrix.

[0105] The step S5 specifically comprises the following steps:

[0106] Step S501, calculate the angular power spectrum density correlation coefficient of two frequency bands, the expression is as follows:

[0107]

[0108] Step S502: Calculate ρ for all f B,f , and obtain ρ B,f The statistical information of the angle power spectrum density in the LoS scenario is as follows: Figure 5 shown. Figure 5 A comparison of spatial similarity indicators between the channel measurement data and two standardized models is also shown, which can be used for the evaluation and correction of the channel model.

[0109] The step S6 specifically includes the following steps, S601-S603 are performed separately for each compared frequency band:

[0110] Step S601: Calculate the channel gain after removing small-scale fading:

[0111]

[0112] Where Δτ and Δt represent the delay resolution of the channel matrix and the duration of a single snapshot, respectively. W represents the window length, which is generally 40 wavelengths.

[0113] Step S602: Path loss is obtained by fitting the antenna gain and receiving and sending power of the transmitting and receiving ends:

[0114]

[0115] where d = τ LOS c, τ LOS Obtained from the time-delay power spectral density, c represents the speed of light in vacuum; Indicates the least squares fitting method and expressions of d.

[0116] Step S603: remove the path loss from the channel gain to obtain a small-scale fading value:

[0117]

[0118] Step S604: Calculate the Pearson correlation coefficient of small-scale fading in two frequency bands:

[0119]

[0120] in ρ S When it is a negative number, it means that the small-scale fading of the two frequency bands is negatively correlated, a positive number indicates a positive correlation, and 0 indicates linear independence.

[0121] Step S605: Calculate ρ for all mobility measurement routes S , and obtain ρ S Statistical information, such as CDF, mean, variance, etc.

[0122] Step S701: merge the three indicators in the following manner:

[0123] ρ a =α 1 ρ τ,m,n +α 2 ρ B,f +α 3 ρ S

[0124] where α 1 , α 2 and α 3 Respectively represent the weights of the similarity index in the frequency domain, spatial domain, and time domain, satisfying α 1 +α 2 +α 3 =1. 1 , α 2 and α 3 The value of can be set according to the requirements. For example, when using the cross-band channel similarity to compress the multi-frequency codebook of the FDD system, we are mainly concerned with the similarity of the space-frequency dimension of the channels in different frequency bands, so we set α 1 =0.5,α 2 =0.5,α 3 =0.

[0125] In summary, the present invention proposes to analyze the similarity index of cross-band wireless MIMO channels from three dimensions: space, time, and frequency, and proposes three indicators: delay power spectrum density correlation coefficient, angle power spectrum density correlation coefficient, and small-scale fading correlation coefficient. The present invention quantifies the similarity of channels, supports the study of the similarity of channels in different frequency bands, provides guidance for the deployment of cross-band communication systems, and provides a judgment standard for measuring the accuracy of cross-band wireless MIMO channel models.

[0126] It can be understood that the present invention is described by real channel measurement data, and those skilled in the art know that various changes or equivalent substitutions can be made to these features and embodiments without departing from the present invention. In addition, under the guidance of the present invention, these features and embodiments can be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application belong to the scope protected by the present invention.

Claims

1. A similarity quantification method for cross-band wireless MIMO channels, characterized in that: The following steps are involved: Step S1: Build a cross-band wireless MIMO channel measurement system, plan test cases, and calculate channel impulse response; Step S2, calculating the delay power spectral density of each frequency band according to the channel impulse response; Step S3, extracting effective data points of the delay power spectrum density and calculating the delay power spectrum correlation coefficient as a measurement index of frequency domain similarity; Step S4, calculating the Bartlett angle power spectra of the transmitting and receiving ends of the channels in different frequency bands, combining them and then quantizing them; Step S5, calculating the angular power spectrum correlation coefficient of channels in different frequency bands as a measure of spatial domain similarity; Step S6, after removing the large-scale fading, calculate the small-scale fading correlation coefficient of the mobile channels in different frequency bands as a measure of time domain similarity; Step S7: assign different weights to the similarity indices of the three domains, namely, space, time and frequency, and then merge them to generate a final cross-band channel similarity index; The step S1 specifically includes the following steps: Step S101: perform direct connection calibration on the channel detection system to obtain a direct connection calibration signal y th (t,τ); Step S102: Use the antenna array to obtain the received signal y propagated through the wireless channel r (t,τ); Step S103: Calculate the measurement channel impulse response h(t, τ) based on the inverse Fourier transform IFFT(·), expressed as h(t,τ)=IFFT{H(t,f)}=IFFT{Y r (t,f) / Y th (t,f)} where Y r (t,f) and Y th (t,f) are y r (t,τ) and y th The frequency domain transfer function of (t,τ) is: t ×M r ×M f ×M s , where M t is the number of transmitting antennas, M r is the number of receiving antennas, M f is the number of subcarriers, M s is the number of channel snapshots, H(t, f) is the channel transmission matrix; The step S2 specifically includes the following steps: Calculate the frequency band f based on the channel impulse response c The delayed power spectral density is expressed as: In the formula The dimension is M f ×1 vector, |·| means taking the absolute value of each element in the matrix, m and n represent the serial numbers of the receiving and transmitting antennas respectively, M f is the number of subcarriers, M s is the number of channel snapshots, h(t, τ) is the channel impulse response; The step S3 specifically includes the following steps, S301-S303 are performed separately for each compared frequency band: Step S301, extracting all delay points within the effective range of the delay power spectrum density; Step S302, determining all effective paths of the delay power spectrum density by searching for the local maximum value; Step S303: Select the delay power spectrum density of the frequency band with the least effective path as a reference, and match the effective paths of the delay power spectrum densities of all frequency bands; Step S304: Calculate the delay power spectrum density correlation coefficient, which is expressed as: in{·} T represents the transpose operation, Respectively represent the time delay power spectral density of frequency bands f1 and f2; Step S305: Calculate ρ for all m and n τ,m,n , and obtain ρ τ,m,n Statistical information; The step S4 specifically includes the following steps, which are performed separately for each compared frequency band: Step S401: Calculate the angle power spectrum density of the channel to reach the angle power spectrum density And the departure angle power spectrum Using the Bartlett method, the expressions are: in,{·} H represents Hermit transpose, c(θ,φ) represents the steering vector when the pitch angle is θ and the azimuth angle is φ, H m,n represents the channel transmission matrix between the mth transmitting antenna and the nth receiving antenna, M f is the number of subcarriers; Step S402: merge the departure and arrival angle power spectrum density matrices and quantize them, that is: where vec{·} represents the operation of vectorizing the matrix; The step S5 specifically comprises the following steps: Step S501, calculate the angular power spectrum density correlation coefficient of two frequency bands, the expression is as follows: Step S502: Calculate ρ for all f B,f , and obtain ρ B,f Statistical information; The step S6 specifically includes the following steps, S601-S603 are performed separately for each compared frequency band: Step S601: Calculate the channel gain after removing small-scale fading: Where Δτ and Δt represent the delay resolution of the channel matrix and the duration of a single snapshot, respectively, W represents the window length, and M t is the number of transmitting antennas, M r is the number of receiving antennas, M f is the number of subcarriers; Step S602: Path loss is obtained by fitting the antenna gain and receiving and sending power of the transmitting and receiving ends: where d = τ LOS c, τ LOS Obtained by the time-delay power spectral density; Indicates the least squares fitting method and d k Expression of Step S603: remove the path loss from the channel gain to obtain a small-scale fading value: Step S604: Calculate the Pearson correlation coefficient of small-scale fading in two frequency bands: in ρ S When it is a negative number, it means that the small-scale fading of the two frequency bands is negatively correlated, a positive number indicates a positive correlation, and 0 indicates linear independence; Step S605: Calculate ρ for all mobility measurement routes S , and obtain ρ S Statistics.

2. The similarity quantification method of a cross-band wireless MIMO channel according to claim 1, characterized in that: Cross-band MIMO wireless channel measurement meets the following requirements: 1) The bandwidth used by each frequency band is exactly the same; 2) Each frequency band uses an independent measurement system, each measurement system has the same dynamic range, and different measurement systems use the same trigger signal; 3) Different frequency bands use the same type of antenna; 4) The length of a single snapshot in each frequency band remains the same.

Images