Near-field MIMO broadband OFDM (Orthogonal Frequency Division Multiplexing) flux-sensing integrated system and signal processing method

Abstract

The invention relates to a near-field MIMO (Multiple Input Multiple Output) broadband OFDM (Orthogonal Frequency Division Multiplexing) communication and sensing integrated system and a signal processing method, and the near-field MIMO broadband OFDM communication and sensing integrated system comprises a base station and K single-antenna users serving a BS (Base Station) and located in a near field, the signal processing method comprises the following steps: step A, processing a user angle and distance sensing signal; step B, processing a user speed sensing signal; and step C, initial sensing, design of a beam former, data transmission, speed sensing, and beam prediction and tracking signal processing of setting an i variable. The method has the advantages that the proposed FD-PI algorithm and FD-BC algorithm can significantly improve the accuracy of user position sensing, and the FD-PI algorithm and the FD-BC algorithm have obvious advantages in comparison with the PI algorithm and the BC algorithm which are not subjected to FD sampling under the condition of low SNR (Signal to Noise Ratio).

Description

Technical Field

[0001] The present invention belongs to the technical field of mobile communications, and particularly relates to a MIMO broadband OFDM integrated communication and sensing system and a signal processing method Technical Background

[0002] The future 6G integrated communication and sensing (ISAC) technology deeply integrates communication and sensing functions, and realizes the coordination of communication and sensing by sharing hardware and wireless resources. With the rapid development of mobile communication technology, the demand for higher data rates and more efficient communication is increasing day by day. To meet these demands, communication systems are gradually developing towards higher frequencies, larger bandwidths, and larger-scale antenna arrays. However, in communication scenarios with high frequencies, wide bandwidths, and large-scale arrays, the near-field effect and the broadband beam squint effect are prevalent. At this time, the traditional far-field MIMO model is no longer applicable, which brings new challenges to communication and sensing

[0003] As one of the important functions of ISAC, the perception of the spatial target's position and velocity is crucial for optimizing communication beamforming strategies and improving communication anti-jamming performance. Most traditional spatial target perception algorithms consider far-field scenarios. At this time, the signal is modeled based on the plane wave model. Target position perception is the estimation of the target azimuth angle, and target velocity perception only includes the projection component of the target's true velocity on the line connecting the target and the BS. Moreover, in the far-field scenario, the signal energy of beamforming is concentrated in a specific direction. However, with the future mobile communication's demand for higher data rates, higher frequency bands and larger antenna arrays will be used, and the Rayleigh distance will increase accordingly, resulting in a higher proportion of the near field within the BS coverage range. The spherical wave signal model needs to be considered. At this time, new degrees of freedom in the distance domain are introduced. Therefore, target position perception includes both the estimation of the target angle and distance, and target velocity perception can obtain the target's true velocity, including its radial velocity and transverse velocity. The signal energy during beamforming is focused on specific points in space. Based on the differences in the above signal models, traditional far-field-based perception algorithms are no longer applicable in the near-field scenario. Moreover, existing target position perception algorithms applicable to the near-field scenario, such as 2D-MUSIC, 2D-Capon, etc., have the problem of excessive complexity due to the need for multi-dimensional spatial search. In addition, in the high-frequency broadband scenarios prevalent in future mobile communications, relying solely on phase shifters (PS) to control beamforming, the signal energy focus point is a function of frequency, resulting in a beam squint effect. Most traditional perception algorithms are based on the narrowband assumption, so the broadband beam squint effect also poses challenges to spatial target perception. To address this problem, a True Time Delay (TTD) module can be used to achieve programmable phase shift control. Some existing studies have achieved the control of the beam squint focus points of different carriers of broadband OFDM signals by combining TTD and PS and connecting them in series with each element in the antenna array, and realized the perception of the spatial target angle and distance according to the frequency of the maximum gain signal feedback by the user in space. Although this method solves the problem of spatial target perception in the broadband scenario, the utilization rate of frequency domain carrier resources and the perception accuracy still need to be improved.

[0004] Based on the above existing technical problems, the present invention addresses the problem of spatial target perception under the near-field broadband effect. Considering a large-scale MIMO OFDM integrated communication and sensing system, low-complexity calibration algorithms FD-PI (Forced Degressive Parabolic Interpolation) and FD-BC (Forced Degressive Barycenter Calibration) are proposed to improve the accuracy of the base station's perception of the user's angle and distance, and a signal processing method based on the Expectation Maximization (EM) algorithm is used to achieve the base station's perception of the user's speed. Summary of the Invention

[0005] Aiming at the technical problems existing in the prior art, the present invention proposes a near-field MIMO broadband OFDM integrated communication and sensing system and a signal processing method.

[0006] The near-field MIMO broadband OFDM integrated communication and sensing system of the present invention includes:

[0007] A base station (BS) and K single-antenna users in the near field served by the BS.

[0008] Furthermore, a uniform linear array (ULA) with N = 2L + 1 antennas is deployed on the BS side. The carrier frequency and signal bandwidth of the OFDM signal transmitted by the BS are f c and W respectively, and the number of subcarriers is M + 1. Then the subcarrier spacing is Δf = W / M, and the frequency of the m-th subcarrier is f m = f c - W / 2 + mW / M , and the baseband frequency is where m = 0, 1,..., M.

[0009] Furthermore, each antenna of the BS is connected in series with a true time delay (TTD) and a phase shifter (PS) to control the signal energy focusing point of different frequency subcarriers, that is, the beam squint point. Then the beamforming vector of the BS array is expressed by the following formula (1):

[0010]

[0011] In formula (1), represents the baseband frequency of the signal, φ n represents the phase shift of the n-th (n = -L,..., 0,..., L) PS, and t n represents the time delay of the n-th TTD, and its expression is as follows formula (2):

[0012]

[0013] In Equation (2), r start，n and r end，n represent the distances from the starting point of beam squint (the beam squint point of the 0th sub - carrier (r start , θ start) ) and the end point (the beam squint point of the Mth sub - carrier (r end , θ end )) to antenna n. Let the beam squint point of sub - carrier m be represented as Then the following equations (3) and (4) are obtained:

[0014]

[0015] Let the position of user k be (x k , y k ), and the corresponding polar coordinates be (r k , θ k ). The moving speed of user k is v k . Let v r，k and v θ，k represent the radial speed and the transverse speed of the user relative to the center of the ULA of the BS respectively. r n，k represents the distance between antenna n of the BS and user k, and v n，k represents the projection component of the user's speed v k on the line connecting the BS antenna n and the user.

[0016] The present invention further provides a signal processing method for a near - field MIMO broadband OFDM integrated communication and sensing system, including the following steps:

[0017] Step A, user angle and distance perception signal processing: In the user position, i.e., the angle and distance perception stage, the BS first sends an all - 1 pilot signal modulated by OFDM and controlled by the beam squint path to the user, and then the BS obtains the final angle and distance perception result through the uplink signal frequency fed back by the user;

[0018] Step B, user speed perception signal processing: Assume that the BS realizes the sharing of the antenna array between the transmitter and the receiver through a circulator and full - duplex technology. For user k, at this time, the BS knows the current position of user k and transmits an OFDM signal with the signal energy focused on the user position to the user, that is, both the starting point and the end point of the beam squint are set to Use the EM algorithm to obtain the perception of the user's speed through the echo signal of the user;

[0019] Step C, including initial perception, beamformer design, data transmission and speed perception, and beam prediction and tracking signal processing steps for setting the i variable.

[0020] Further, in step A, the Angle and Distance Perception Phase of the user location, the BS first sends a full-1 pilot signal modulated by OFDM and controlled by the beam squint path to the user. The user generates a corrected spectrum using the received OFDM signal and applies the calibration algorithms FD-PI or FD-BC proposed in the present invention to obtain the signal frequency with the maximum amplitude and feedback it to the BS. Then, the BS obtains the final angle and distance perception results through the uplink signal frequency fed back by the user, which specifically includes:

[0021] Step A1, taking user k as an example, in the user angle perception phase, set the starting point and ending point of the beam squint to be (r mid , θ min ) and (r mid , θ max ), respectively. Among them, r min ≤ r mid ≤ r max , and obtain the corrected received signal spectrum of user k, expressed as

[0022] Step A2, calibrate the corrected spectrum using the following calibration algorithm:

[0023] Step A2.1, perform a spectral peak search on the discrete corrected spectrum, expressed as the following formula (5):

[0024]

[0025] Let the number of observation points for calibration be The amplitudes of the observation points for calibration are represented as a vector The corresponding indices are represented as a vector m′. The amplitudes of the observation points on the left and right sides of the peak point are represented by vectors and respectively, and the corresponding indices are represented as vectors m′ l and m′ r . Since the number of observation points may be less than due to the peak point being close to the edge, the above vectors for storing the amplitudes and indices of the observation points are initialized to be empty;

[0026] Step A2.2, if the FD-PI calibration algorithm is selected for use, assume that the number of observation points for calibration is is odd:

[0027] Step A2.2.1, use m′ l to represent the last valid point on the left side of the peak point and initialize it to m′ l = m′, and set the variable i = m′ - 1;

[0028] Step A2.2.2, if the conditions are simultaneously met If \(i\geq1\), then proceed to execute steps A2.2.3 to A2.2.4; otherwise, skip steps A2.2.3 to A2.2.4 and directly execute step A2.2.5;

[0029] Step A2.2.3, if then update the vector m′ l and the last valid point \(m'\) to the left of the peak point l , expressed as the following formula (6):

[0030]

[0031] If the above conditions are not met, continue to execute the next step;

[0032] Step A2.2.4, update the variable \(i\) to \(i = i - 1\) and return to execute step A2.2.2;

[0033] Step A2.2.5, use \(m'\) r to represent the last valid point to the right of the observed peak point and initialize it as \(m'\) r = \(m'\), set the variable \(i = m' + 1\);

[0034] Step A2.2.6, if both conditions and are satisfied, then proceed to execute steps A2.2.7 to A2.2.8; otherwise, skip steps A2.2.7 to A2.2.8 and directly execute step A2.2.9;

[0035] Step A2.2.7, if then update the vector m′ r , and the last valid point \(m'\) to the right of the peak point r , expressed as the following formula (7):

[0036]

[0037] If the above conditions are not met, continue to execute the next step;

[0038] Step A2.2.8, update the variable \(i\) to \(i = i + 1\) and return to execute step A2.2.6;

[0039] Step A2.2.9, obtain the amplitude of the observation point and the corresponding index for the FD - PI algorithm, expressed as the following formula (8):

[0040]

[0041] Let And \(m'\) are used as the vertical and horizontal coordinates respectively for the regression fitting of the quadratic function \(y = p_2x^2 + p_1x + p_0\) to obtain the fractional multiple index of the calibrated peak. 2 The expression is as follows in Equation (9):

[0042]

[0043] Step A2.3, if the FD - BC calibration algorithm is selected for use, then set the number of observation points for calibration to be an even number:

[0044] Step A2.3.1, if then execute Steps A2.3.2 to A2.3.9; otherwise, jump to Step A2.3.10;

[0045] Step A2.3.2, initialize the index of the last valid point on the left side of the peak as \(m' = m'\), and initialize the variable \(i = m' - 1\); l

[0046] Step A2.3.3, if both conditions and \(i\geq1\) are satisfied, then proceed to execute Steps A2.3.4 to A2.3.5; otherwise, skip Steps A2.3.4 to A2.3.5 and directly execute Step A2.3.6;

[0047] Step A2.3.4, if then similar to Equation (6), update the vector \(m'\) l and the last valid point \(m'\) on the left side of the peak point l ; if the above conditions are not met, then continue to the next step;

[0048] Step A2.3.5, update the variable \(i\) to \(i = i - 1\) and return to execute Step A2.3.3;

[0049] Step A2.3.6, initialize the index of the last valid point on the right side of the peak as \(m' = m'\), and initialize the variable \(i = m' + 1\); r

[0050] Step A2.3.7, if both conditions and are satisfied, then proceed to execute Steps A2.3.8 to A2.3.9; otherwise, jump to Step A2.3.18;

[0051] Step A2.3.8, if then similar to Equation (7), update the vector \(m'\) r and the last valid point \(m'\) on the left side of the peak point r ​​​; If the above conditions are not met, continue to the next step;

[0052] Step A2.3.9, update the variable i to i = i + 1, and return to execute Step A2.3.7;

[0053] Step A2.3.10, initialize the index of the last valid point on the left side of the peak as m′ l = m′, initialize the variable i = m′ - 1;

[0054] Step A2.3.11, if both conditions and i ≥ 1 are satisfied, then proceed to execute Steps A2.3.12 to A2.3.13; otherwise, skip Steps A2.3.12 to A2.3.13 and directly execute Step A2.3.14;

[0055] Step A2.3.12, if then similar to Equation (6), update the vector m′ l and the last valid point m′ on the left side of the peak point l ; If the above conditions are not met, continue to the next step;

[0056] Step A2.3.13, update the variable i to i = i - 1, and return to execute Step A2.3.11;

[0057] Step A2.3.14, initialize the index of the last valid point on the right side of the peak as m′ r = m′, initialize the variable i = m′ + 1;

[0058] Step A2.3.15, if both conditions and are satisfied, then proceed to execute Steps A2.3.16 to A2.3.17; otherwise, skip Steps A2.3.16 to A2.3.17 and directly execute Step A2.3.18;

[0059] Step A2.3.16, if then similar to Equation (7), update the vector m′ r and the last valid point m′ on the left side of the peak point r ; If the above conditions are not met, continue to the next step;

[0060] Step A2.3.17, update the variable i to i = i + 1, and return to execute Step A2.3.15;

[0061] Step A2.3.18, obtain the amplitude of the observation point for the FD - BC algorithm and the corresponding index m′, with the expression the same as Equation (8), where, Indicates the actual number of observation points, satisfying Let and m′ be used as the ordinate and abscissa respectively, and the fractional multiple index expression of the calibrated peak is obtained as follows in Equation (10):

[0062]

[0063] Step A3: According to the fractional multiple index of the calibrated peak Obtain the signal frequency corresponding to the corrected spectrum maximum The expression is: Next, the user feeds back this frequency information to the BS, and the BS will θ start = θ min θ end = θ max Substitute into (3) to calculate the angle information of user k, expressed as

[0064] Step A4: After obtaining the angle estimation result of user k After that, sense the distance of user k. At this time, the starting point and the ending point of the beam squint are set as and Obtain the corrected received signal spectrum of user k, expressed as

[0065] Step A5: Refer to Step A2, apply the calibration algorithm to the corrected spectrum to obtain the fractional multiple index of the peak

[0066] Step A6: Refer to Step A3, map the obtained fractional multiple index of the corrected spectrum peak to the frequency The user feeds back this frequency information to the BS, and the BS will r start = r min r end = r max , Substitute into Equation (4) to calculate the distance information of the user, expressed as

[0067] Furthermore, in Step B, the signal processing flow for user speed sensing. Assume that the BS realizes the sharing of the antenna array by the transmitter and the receiver through the circulator and the full-duplex technology. Still taking user k as an example, the BS knows the current position of user k and transmits an OFDM signal with the signal energy focused on the user position to the user, that is, set the starting point and the ending point of the beam squint to be both And obtain the speed perception of the user through the user's echo signal, specifically including:

[0068] Step B1, construct the echo signal model of the BS receiving the user k. Let the total echo signal received by the BS from the user k at the sampling time t0 be expressed as Its expression is the following formula (11):

[0069]

[0070] Step B2, construction of the optimization problem:

[0071] Unknown parameter v k =[v r，k , v θ，k T The estimation of β and can be transformed into the following optimization problem:

[0072]

[0073] Step B3, let Use the Expectation-Maximization (EM) algorithm to solve the optimization problem in formula (12):

[0074] Step B3.1, randomly initialize the initial speed of user k For the first iteration of the EM algorithm, initialize the constant h0 for the condition judgment of the loop end;

[0075] Step B3.2, assume that the speed of user k is fixed and known as Then, the estimation result of the channel fading coefficient is obtained by the least squares method, as shown in the following formula (13):

[0076]

[0077] In formula (13), Denotes the pseudo-inverse,

[0078] Step B3.3, substitute the obtained estimation result of the channel fading coefficient Into formula (12), then for the fixed channel fading coefficient The speed estimation of user k is transformed into the following optimization problem, as shown in the following formula (14):

[0079]

[0080] For this problem, through the gradient-based method, the objective function G(v k ) is used to derive the user speed v k Finally, the user speed estimation result is obtained through the following gradient descent process, as shown in the following formula (15): ​

[0081]

[0082] In Equation (15), γ (i) represents the learning rate of the i-th iteration, and the obtained user speed estimation result is expressed as

[0083] Step B3.4, if the user speed estimation result obtained in the previous step satisfies then is used as the final estimation result of the user speed, and the user speed perception is completed; otherwise, set and jump to Step B3.2

[0084] Furthermore, Step C: Beam prediction and tracking signal processing flow. For user k, the specific steps of beam prediction and tracking are as follows:

[0085] Step C1, Initial perception. It is known that within a coherent processing interval (CPI), the position and speed of the user are considered unchanged. Let the duration of one CPI be T0. In the first CPI, set i = 1. At this time, the BS does not perform data transmission but only focuses on the perception of the position and speed of user k. Specifically, the BS first obtains the user position perception result according to the process of Step A within the first CPI, which is expressed as Then set both the starting point and the ending point of the beam squint to Substitute into Equations (1) and (2) to obtain the beamforming vector of the BS at this time and transmit an OFDM signal to the user. Based on the echo signal of the user, the estimation of the initial speed of user k is obtained according to the process of Step B, which is expressed as

[0086] Step C2, Beamformer design. Given the estimation results of the user position and speed within the i-th CPI and then the position of the user within the next CPI is estimated as: Design the corresponding beamformer according to the user position within the (i + 1)-th CPI, that is, set both the starting point and the ending point of the beam squint to Substitute into Equations (1) and (2) to obtain the beamforming vector, which is expressed as

[0087] Step C3, Data transmission and speed perception: Within the (i + 1)-th CPI, use the beamformer obtained within the previous CPI to perform effective data transmission, and based on the process of Step B, obtain the radial and transverse speeds of the user under the current (i + 1)-th CPI according to the echo signal reflected by the user, which are expressed as

[0088] Step C4: Set \(i = i + 1\) and jump to Step C2.

[0089] Compared with the prior art in the technical field of the present invention, the following superior technical effects are achieved:

[0090] 1. For the proposed FD-PI and FD-BC algorithms in the near-field MIMO broadband OFDM integrated communication and sensing system and signal processing method of the present invention, the accuracy of user position perception can be significantly improved. At low SNR, compared with the PI and BC algorithms without FD sampling, the FD-PI algorithm and FD-BC algorithm have obvious advantages. When the SNR is low, the larger it is, the higher the perception accuracy of FD-PI. The rule of the FD-BC algorithm changing with the number of sampling points is the larger it is, the higher the perception accuracy.

[0091] 2. For the proposed FD-BC algorithm in the near-field MIMO broadband OFDM integrated communication and sensing system and signal processing method of the present invention, the accuracy of user position perception can be significantly improved. At low SNR, compared with the BC algorithm without FD sampling, the FD-BC algorithm has obvious advantages. The rule of the FD-BC algorithm changing with the number of sampling points is the larger it is, the higher the perception accuracy.

[0092] 3. The proposed EM algorithm in the near-field MIMO broadband OFDM integrated communication and sensing system and signal processing method of the present invention is used for user speed perception, which can make full use of frequency domain resources and effectively save time domain resources, laying a foundation for subsequent beam prediction and tracking signal processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] FIG. 1 is a schematic diagram of the system model in an embodiment of the present invention;

[0094] Figure 2 is a schematic diagram of user position and speed perception in an embodiment of the present invention;

[0095] FIG. 3 is a schematic diagram of RMSE-SNR simulation of angle perception in an embodiment of the present invention. DETAILED IMPLEMENTATION MANNER

[0096] In order to more clearly understand the principles and features of the present invention, the specific implementation manners of the near-field MIMO broadband OFDM integrated communication and sensing system and signal processing method of the present invention will be described in detail below with reference to FIGS. 1-3 of the specification.

[0097] Embodiment

[0098] The near-field MIMO broadband OFDM communication and sensing integrated system model of the present invention is shown in Figure 1, and includes a base station (BS) and K single-antenna users in the near field served by the BS. Let the carrier frequency and signal bandwidth of the OFDM signal transmitted by the BS be f c and W respectively, and the number of subcarriers be M + 1, then the subcarrier spacing is Δf = W / M, and the frequency of the m-th subcarrier is f m = f c - W / 2 + mW / M, and the baseband frequency is Each antenna of the BS is connected in series with a true time delay device (TTD) and a phase shifter (PS) to control the signal energy focusing point of different subcarriers, that is, the beam squint point. Then the beamforming vector of the BS array is expressed as:

[0099]

[0100] Among them, represents the baseband frequency of the signal, and φ n represents the phase shift of the n-th (n = -l,..., 0,..., L) PS, and t n represents the time delay of the n-th TTD, and its expression is as follows:

[0101]

[0102] Among them, r start，n and r end，n represent the distances from the starting point of beam squint (the beam squint point of the 0-th subcarrier (r start , θ start )) and the end point (the beam squint point of the M-th subcarrier (r end , θ end )) to the antenna n.

[0103] The schematic diagram of user position and speed perception of the present invention is shown in Figure 2 . Let the position of user k be (x k , y k ), and the corresponding polar coordinates be (r k , θ k ). The moving speed of user k is v k , and v r，k and v θ，k are used to represent the radial speed and transverse speed of the user relative to the center of the ULA of the BS respectively. r n，k represents the distance between the n-th antenna of the BS and user k, and v n，k represents the speed v kThe projection component on the line connecting the BS antenna n and the user, and the signal gain of the m-th subcarrier received by the user k is expressed as where the frequency offset caused by mobility relative to the subcarrier frequency f m can be ignored; α m = c / 4πf m r represents the channel gain on the m-th subcarrier. For the convenience of analysis, multiply both sides of the equation by to eliminate the influence of channel fading of different subcarriers, and the corrected received signal gain is obtained as the following formula (16):

[0104]

[0105] where

[0106]

[0107] Based on the above description of the near-field MIMO broadband OFDM integrated communication and sensing system, the steps of the signal processing method of the present invention are as follows:

[0108] Step A, user angle and distance perception signal processing: At the user position, that is, the angle and distance perception stage, the BS first sends an OFDM-modulated all-1 pilot signal controlled by the beam squint path to the user, and then the BS obtains the final angle and distance perception result through the uplink signal frequency fed back by the user. Let the range where the user position is located be {(r, θ)|r min ≤ r ≤ r max , θ min ≤ θ ≤ θ max};

[0109] Step A1, taking the user k as an example, in the user angle perception stage, set the starting point and ending point of the beam squint to be (r mid , θ min ) and (r min , θ max ), respectively, where r min ≤ r mid ≤ r max , and substitute it into the above formula (16) to obtain the corrected received signal spectrum of the user k where:

[0110]

[0111] In the above formula,

[0112] At this time, the beam squint path is similar to Figure 1a the T3 or T4 in the "beam squint path diagram" of

[0113] Step A2: Apply the calibration algorithm proposed in this work to the corrected spectrum in Equation (17):

[0114] Step A2.1: Perform a spectral peak search on the discrete corrected spectrum, expressed as follows:

[0115]

[0116] Let the number of observation points for calibration be The amplitudes of the observation points for calibration are represented as a vector The corresponding indices are represented as a vector m′, and the amplitudes of the observation points to the left and right of the peak point are represented by vectors and respectively, and the corresponding indices are represented as vectors m′ l and m′ r , since the number of observation points may be less than due to the peak point being close to the edge, the above vectors for storing the amplitudes and indices of the observation points are initialized as empty;

[0117] Step A2.2: If the FD-PI calibration algorithm is selected for use, assume that the number of observation points for calibration is is odd;

[0118] Step A2.2.1: Use m′ l to represent the last valid point to the left of the peak point and initialize it as m′ l = m′, and set the variable i = m′ - 1;

[0119] Step A2.2.2: If both conditions and i ≥ 1 are satisfied, then proceed to execute Steps A2.2.3 to A2.2.4; otherwise, skip Steps A2.2.3 to A2.2.4 and directly execute Step A2.2.5;

[0120] Step A2.2.3: If then update the vector m′ l and the last valid point m′ to the left of the peak point l , expressed as:

[0121] m′ l = [m′ l T i T , m′ l = i,

[0122] If the above conditions are not met, continue to the next step; ​

[0123] Step A2.2.4, update the variable i to i = i - 1, and return to execute Step A2.2.2;

[0124] Step A2.2.5, use m′ r to represent the last valid point on the right side of the observed peak point, and initialize it as m′ r = m′, set the variable i = m′ + 1;

[0125] Step A2.2.6: If both conditions and are satisfied, then proceed to execute Steps A2.2.7 to A2.2.8. Otherwise, skip Steps A2.2.7 to A2.2.8 and directly execute Step A2.2.9;

[0126] Step A2.2.7, if then update the vector m′ r , and the last valid point m′ on the right side of the peak point r , which is expressed as:

[0127] m′ r = [m′ r T i] T , m′ r = i,

[0128] If the above conditions are not satisfied, then continue to execute the next step;

[0129] Step A2.2.8, update the variable i to i = i + 1, and return to execute Step A2.2.6;

[0130] Step A2.2.9, obtain the amplitude of the observation point and the corresponding index for the FD-PI algorithm, which is expressed as:

[0131]

[0132] Take and m′ as the ordinate and abscissa respectively, for the regression fitting of the quadratic function y = p2x 2 + p1x + p0, and obtain the fractional multiple index of the calibrated peak The expression is:

[0133]

[0134] Step A2.3, if the FD-BC calibration algorithm is selected for use, then set the number of observation points used for calibration to be an even number:

[0135] Step A2.3.1, if Then perform steps A2.3.2 to A2.3.9; otherwise, directly jump to step A2.3.10;

[0136] Step A2.3., initialize the index of the last valid point on the left side of the peak as m′ l = m′, initialize the variable i = m′ - 1;

[0137] Step A2.3.3, if both conditions and i ≥ 1 are satisfied, then proceed to execute steps A2.3.4 to A2.3.5; otherwise, skip steps A2.3.4 to A2.3.5 and directly execute step A2.3.6;

[0138] Step A2.3.4, if then step A2.2.3, update the vector m′ l and the last valid point m′ on the left side of the peak point l ; if the above conditions are not met, continue to the next step;

[0139] Step A2.3.5, update the variable i to i = i - 1, and return to execute step A2.3.3;

[0140] Step A2.3.6, initialize the index of the last valid point on the right side of the peak as m′ r = m′, initialize the variable i = m′ + 1;

[0141] Step A2.3.7, if both conditions and are satisfied, then proceed to execute steps A2.3.8 to A2.3.9; otherwise, jump to step A2.3.18;

[0142] Step A2.3.8, if then similar to step A2.2.7, update the vector m′ r and the last valid point m′ on the left side of the peak point r ; if the above conditions are not met, continue to the next step;

[0143] Step A2.3.9, update the variable i to i = i + 1, and return to execute step A2.3.7;

[0144] Step A2.3.10, initialize the index of the last valid point on the left side of the peak as m′ l = m′, initialize the variable i = m′ - 1;

[0145] Step A2.3.11, if both conditions If i≥1, then proceed to steps A2.3.12 to A2.3.13; otherwise, skip steps A2.3.12 to A2.3.13 and directly proceed to step A2.3.14;

[0146] Step A2.3.12, if then, similar to step A2.2.3, update the vector m′ l and the last valid point m′ to the left of the peak point; if the above conditions are not met, then continue to the next step; l ; if the above conditions are not met, then continue to the next step;

[0147] Step A2.3.13, update the variable i to i = i - 1 and return to execute step A2.3.11;

[0148] Step A2.3.14, initialize the index of the last valid point to the right of the peak as m′ r = m′, and initialize the variable i = m′ + 1;

[0149] Step A2.3.15, if both conditions and are satisfied, then proceed to steps A2.3.16 to A2.3.17; otherwise, skip steps A2.3.16 to A2.3.17 and directly proceed to step A2.3.18;

[0150] Step A2.3.16, if then, similar to step A2.2.7, update the vector m′ r and the last valid point m′ to the left of the peak point; if the above conditions are not met, then continue to the next step; r ; if the above conditions are not met, then continue to the next step;

[0151] Step A2.3.17, update the variable i to i = i + 1 and return to execute step A2.3.15;

[0152] Step A2.3.18, similar to step A2.2.9, obtain the amplitudes of the observation points for the FD - BC algorithm and the corresponding index m′, where represents the actual number of observation points, satisfying Take and m′ as the ordinate and abscissa respectively, and obtain the fractional - multiple index expression of the calibrated peak as:

[0153]

[0154] Step A3, according to the fractional - multiple index of the calibrated peak obtain the signal frequency corresponding to the corrected spectral maximum The expression is: Next, the user feeds back this frequency information to the BS, and the BS will θ start = θ min θ end = θ max Substitute into the following formula:

[0155]

[0156] Calculate the angle information of user k, denoted as

[0157] Step A4, after obtaining the angle estimation result of user k then, sense the distance of user k. At this time, the starting point and the ending point of the beam squint are respectively set as and Substitute it into Equation (16) to obtain the corrected received signal spectrum of user k where:

[0158]

[0159] Ignoring the error of angle sensing, we get:

[0160]

[0161] Then the beam squint path at this time is similar to Figure 1a T1 or T2 in the "beam squint path diagram" of

[0162] Step A5, referring to Step A2, apply the calibration algorithm proposed in this work to the corrected spectrum in Equation (18) to obtain the fractional multiple index of the calibrated corrected power spectrum peak

[0163] Step A6, referring to Step A3, map the obtained fractional multiple index of the corrected spectrum peak to the frequency Next, user k feeds back this frequency information to the BS, and the BS will r start = r min r end = r max , Substitute into the following formula:

[0164] Calculate to obtain the distance information of the user, denoted as

[0165] Step B, User speed perception signal processing flow. Assume that the BS realizes the sharing of the antenna array by the transmitter and the receiver through a circulator and full-duplex technology. Still taking user k as an example, at this time, the BS knows the current position of user k and transmits an OFDM signal with the signal energy focused on the user's position to the user, that is, the starting point and the ending point of the beam squint are both set to The beam squint path is similar to Figure 1a T7 in the "Beam squint path diagram" of

[0166] Step B1, Construct the echo signal model of the BS receiving user k: Let the total echo signal received by the BS from user k at the sampling time t0 be expressed as:

[0167] where Its expression is:

[0168]

[0169] In formula (19), G t and G r are the gains of the transmitting and receiving antennas, σ RCS represents the radar cross section, ⊙ represents the Hadamard product, and the matrix The vector a m (r, θ) and d m (v r , v θ ) are the array response vector and the Doppler shift vector respectively, and the expression is as formula (20) below:

[0170]

[0171] where, represents the OFDM baseband signal of the m-th subcarrier transmitted by the l-th antenna of the BS, and the expression is as formula (21) below:

[0172]

[0173] is the noise vector. Let Then the total echo signal matrix received by the BS from user k at the sampling time t0 is expressed as:

[0174]

[0175] where, The vector β = [β 0 , …, β M T ​;

[0176] Step B2, construction of the optimization problem: unknown parameter v k = [v r，k , v θ，k T The estimation of and β can be transformed into the following optimization problem:

[0177]

[0178] Step B3, set Use the Expectation-Maximization (EM) algorithm to solve the optimization problem in Step B2:

[0179] Step B3.1, randomly initialize the initial speed of user k For the first iteration of the EM algorithm, initialize the constant h0 for the condition judgment of the loop end;

[0180] Step B3.2: Assume that the speed of user k is fixed and known as The estimation result of the channel fading coefficient is obtained by the least squares method as follows:

[0181]

[0182] where, represents the pseudo-inverse,

[0183] Step B3.3, substitute the obtained estimation result of the channel fading coefficient into the optimization problem in Step B2, then for the fixed channel fading coefficient the speed estimation of user k is transformed into the following optimization problem:

[0184]

[0185] For this problem, through the gradient-based method, take the derivative of the objective function G(v k ) with respect to the user speed v k and obtain the final user speed estimation result through the following gradient descent process:

[0186]

[0187] where, γ (i) represents the learning rate of the i-th loop. The obtained user speed estimation result is expressed as

[0188] Step B3.4, if the user speed estimation result obtained in the previous step satisfies then ​As the final estimated result of the user's speed, the user speed perception is completed; otherwise, set and jump to step B3.2;

[0189] Step C, Beam Prediction and Tracking Signal Processing Flow. Taking user k as an example, the following are the steps of the beam prediction and tracking framework:

[0190] Step C1, Initial Sensing: It is known that within a coherent processing interval (CPI), the position and speed of the user can be regarded as unchanged. Let the duration of one CPI be T0. In the first CPI, set i = 1. At this time, the BS does not perform data transmission but only focuses on the sensing of the position and speed of user k. Specifically, in the first CPI, the BS first obtains the user position sensing result according to the process of step A, denoted as Then set both the starting point and the ending point of the beam squint to Substitute the above equations (1) and (2) to obtain the beamforming vector of the BS at this time And transmit an OFDM signal to the user. According to the echo signal of the user, based on the process of step B, the estimation of the initial speed of user k is obtained, denoted as:

[0191]

[0192] Step C2, Beamformer Design: Given the estimation results of the user position and speed within the i-th CPI, and Then the position of the user within the next CPI can be estimated, denoted as:

[0193]

[0194] Design the corresponding beamformer according to the user position within the (i + 1)-th CPI, that is, set both the starting point and the ending point of the beam squint to Substitute into equations (1) and (2) to obtain the beamforming vector, denoted as

[0195] Step C3, Data Transmission and Speed Sensing: Within the (i + 1)-th CPI, use the beamformer obtained in the previous CPI to perform effective data transmission, and based on the process of step B, the radial and transverse speeds of the user under the current (i + 1)-th CPI can be obtained according to the echo signal reflected by the user, denoted as

[0196] Step C4: Set i = i + 1 and jump to step C2.

[0197] In summary, in order to verify the superiority of the low-complexity calibration algorithm proposed in this embodiment, FIG. 3 shows the curves of the angle-aware root mean square error (RMSE) of the FD-PI algorithm and the FD-BC algorithm varying with the signal-to-noise ratio (SNR). In the simulation, the number of antennas is set to N = 128, the number of subcarriers is M = 2048, the angles of the users are uniformly distributed within the sensing range, the element spacing is half a wavelength, the signal bandwidth is W = 3 GHz, f0 = 30 GHz, and the number of Monte Carlo experiments is 1000 times. Among them, Figure 3a the curve of the RMSE of the FD-PI algorithm varying with the SNR is simulated, and the uncalibrated case, i.e., the "Original" curve, and the PI algorithm without forced descent (FD) sampling are used as comparisons. It can be obtained that, compared with the uncalibrated case, the FD-PI algorithm can significantly improve the user angle sensing accuracy; compared with the PI algorithm without FD sampling, especially at low SNR, the PD-PI algorithm proposed in the present invention has obvious advantages. In addition, Figure 3a the performance of the algorithm with different numbers of sampling points is also simulated. It can be obtained that under the simulation conditions of the present invention, when the SNR is low, the larger the value, the higher the sensing accuracy of the FD-PI; however, at high SNR, there is an optimal number of sampling points. Figure 3b The curve of the RMSE of the FD-BC algorithm varying with the SNR is simulated, and its law is similar to that of the FD-PI algorithm. Only under the simulation conditions of the present invention, the law of the FD-BC algorithm varying with the number of sampling points is the larger the value, the higher the sensing accuracy.

[0198] The present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims.

Claims

1. A near-field MIMO broadband OFDM synaesthesia integrated system, characterized in that: include: A base station (BS) and K single-antenna users in the near field serving the BS. A uniform linear array of N = 2L + 1 antennas is deployed on the BS side. The carrier frequency and signal bandwidth of the OFDM signal transmitted by the BS are f c and W, the number of subcarriers is M+1, then the subcarrier spacing is Δf=W / M, and the frequency of the mth subcarrier is f m =f c -W / 2+mW / M, baseband frequency is Wherein, m = 0, 1, ..., M; Each antenna of the BS is connected in series with a true time delay device and a phase shifter to control the signal energy focus point of different frequency subcarriers, that is, the beam squint point. The beamforming vector of the BS array is The expression is as follows (1): In formula (1), Indicates the baseband frequency of the signal φ n represents the phase shift of the nth (n=-L, ..., 0, ..., L) PS, t n represents the delay of the nth TTD, which is expressed as follows: In formula (2), r start，n and r end，n Indicates the beam squint starting point (the beam squint point of the 0th subcarrier (r start ,θ start )) and the distance from the end point to antenna n, let the beam oblique view point of subcarrier m be expressed as Then we get the following equations (3) and (4): Assume that the location of user k is (x k ,y k ), the corresponding polar coordinates are (r k ,θ k ), The moving speed of user k is v k , use v r，k and v θ，k are the radial velocity and lateral velocity of the user relative to the ULA center of the BS, r n，k represents the distance between BS antenna n and user k, v n，k Indicates the user's speed v k The projection component on the line between BS antenna n and the user.

2. The signal processing method based on the near-field MIMO broadband OFDM synaesthesia integrated system according to claim 1 comprises the following steps: Step A, user angle and distance perception signal processing: At the user location, i.e. angle and distance perception stage, the BS first sends an OFDM-modulated all-one pilot signal controlled by a beam squint path to the user, and then the BS obtains the final angle and distance perception result through the uplink signal frequency fed back by the user; Step B, user speed perception signal processing: Assume that the BS uses a circulator and full-duplex technology to realize the sharing of the transmitter and receiver antenna array. For user k, the BS knows the current location of user k and transmits an OFDM signal to the user with the signal energy focused on the user's location, that is, the starting point and end point of the beam squint are set to Use the EM algorithm to perceive the user's speed through the user's echo signal; Step C includes initial perception, beamformer design, data transmission and speed perception, and beam prediction and tracking signal processing steps for setting i variables.

3. Based on the signal processing method based on the near-field MIMO broadband OFDM interawareness integrated system described in claim 1, step A, in the user position, i.e., angle and distance perception stage, the BS first sends an OFDM-modulated all-one pilot signal controlled by a beam squint path to the user, the user uses the received OFDM signal to generate a corrected spectrum, and uses the calibration algorithm FD-PI or FD-BC proposed by the present invention to obtain the maximum amplitude signal frequency and feed it back to the BS, and the BS then obtains the final angle and distance perception results through the uplink signal frequency fed back by the user, specifically including: Step A1, taking user k as an example, in the user angle perception stage, the starting point and end point of the beam squint are set to be (r mid ,θ min ) and (r mid ,θ max ), where r min ≤r mid ≤r max , and the modified spectrum of the received signal of user k is obtained, which is expressed as Step A2: calibrate the corrected spectrum using the following calibration algorithm: Step A2.1, perform a peak search on the discrete corrected spectrum, which is expressed as the following formula (5): Assume that the number of observation points used for calibration is The magnitude of the observation points used for calibration is represented as a vector The corresponding index is represented by vector m′, and the amplitudes of the observation points on the left and right sides of the peak point are represented by vectors and Represents, the corresponding index is represented by vector m′ l and m′ r , since the number of observation points may be less than Therefore, the vector used to store the amplitude and index of the observation point is initialized to empty; Step A2.2: If you choose to use the FD-PI calibration algorithm, assume that the number of observation points used for calibration is For odd numbers: Step A2.2.1, use m′ l Indicates the last valid point on the left side of the peak point and is initialized to m′ l = m', set variable i = m'-l; Step A2.2.2, if the conditions are met at the same time If and i≥1, then proceed to step A2.2.3 to step A2.2.4: Otherwise, skip steps A2.2.3 to step A2.2.4 and directly proceed to step A2.2.5; Step A2.2.3, if Then update the vector m′ l and the last valid point m′ on the left side of the peak point l , expressed as the following formula (6): If the above conditions are not met, proceed to the next step; Step A2.2.4, update variable i to i=i-1, and return to execute step A2.2.2; Step A2.2.5, use m′ r Represents the last valid point on the right side of the observed peak point and is initialized to m′ r = m', set variable i = m' + 1; Step A2.2.6, if the conditions are met at the same time and Then execute step A2.2.7 to step A2.2.8; otherwise, skip step A2.2.7 to step A2.2.8 and directly execute step A2.2.9; Step A2.2.7, if Then update the vector m′ r , and the last valid point m′ on the right side of the peak point r , expressed as the following formula (7): If the above conditions are not met, proceed to the next step; Step A2.2.8, update variable i to i=i+1, and return to execute step A2.2.6; Step A2.2.9, obtain the observation point amplitude and corresponding index used for the FD-PI algorithm, expressed as the following formula (8): Will and m′ as the ordinate and abscissa, respectively, for the quadratic function y=p2x 2 +P1x+P0 regression fitting to obtain the fractional index of the peak after calibration The expression is as follows (9): Step A2.3: If you choose to use the FD-BC calibration algorithm, then set the number of observation points used for calibration For an even number: Step A2.3.1, if Then execute steps A2.3.2 to A2.3.9; otherwise, jump to step A2.3.10; Step A2.3.2, initialize the index of the last valid point on the left side of the peak to m′ l = m', initialize variable i = m'-1; Step A2.3.3, if the conditions are met at the same time If and i≥1, then execute step A2.3.4 to step A2.3.5; otherwise, skip step A2.3.4 to step A2.3.5 and directly execute step A2.3.6; Step A2.3.4, if Then, the update vector is the same as formula (6). m′ l and the last valid point m′ on the left side of the peak point l ; If the above conditions are not met, proceed to the next step; Step A2.3.5, update variable i to i=i-1, and return to execute step A2.3.3; Step A2.3.6, initialize the last valid point index to the right of the peak to m′ r = m', initialize variable i = m' + 1; Step A2.3.7, if the conditions are met at the same time and Then execute step A2.3.8 to step A2.3.9; otherwise, jump to step A2.3.18; Step A2.3.8, if Then, the update vector is the same as formula (7). m′ r and the last valid point m′ on the left side of the peak point r ; If the above conditions are not met, proceed to the next step; Step A2.3.9, update variable i to i=i+1, and return to execute step A2.3.7; Step A2.3.10, initialize the index of the last valid point on the left side of the peak to m′ l = m', initialize variable i = m'-l; Step A2.3.11, if the conditions are met at the same time If and i≥1, then execute step A2.3.12 to step A2.3.13; otherwise, skip step A2.3.12 to step A2.3.13 and directly execute step A2.3.14; Step A2.3.12, if Then, the update vector is the same as formula (6). m′ l and the last valid point m′ on the left side of the peak point l ; If the above conditions are not met, proceed to the next step; Step A2.3.13, update variable i to i=i-1, and return to execute step A2.3.11; Step A2.3.14, initialize the index of the last valid point on the right side of the peak to m′ r = m', initialize variable i = m' + 1; Step A2.3.15, if the conditions are met at the same time and Then execute step A2.3.16 to step A2.3.17; otherwise, skip step A2.3.16 to step A2.3.17 and directly execute step A2.3.18; Step A2.3.16, if Then, the update vector is the same as formula (7). m′ r and the last valid point m′ on the left side of the peak point r ; If the above conditions are not met, proceed to the next step; Step A2.3.17, update variable i to i=i+1, and return to execute step A2.3.15; Step A2.3.18, obtain the observed point amplitude for the FD-BC algorithm And the corresponding index m′, the expression is the same as formula (8), where, Represents the actual number of observation points, satisfying Will and m′ are used as the ordinate and abscissa respectively, and the fractional index expression of the peak after calibration is obtained as follows: Step A3, indexing based on the fractional multiple of the peak value after calibration Get the signal frequency corresponding to the maximum value of the corrected spectrum The expression is: Then, the user feeds back the frequency information to the BS, which θ start =θ min ,θ end =θ max Substitute (3) to calculate the angle information of user k, expressed as Step A4: After obtaining the angle estimation result of user k After that, the distance of user k is sensed, and the starting point and end point of the beam squint are set to and The corrected received signal spectrum of user k is expressed as Step A5, referring to step A2, correct the spectrum Use the calibration algorithm to get the fractional index of the peak value Step A6, referring to step A3, multiply the obtained modified spectrum peak fraction by the index Mapping to frequency The user feeds back the frequency information to the BS, and the BS r start =r min , r end =r max , Substitute into formula (4) to calculate the user's distance information, expressed as 4. Based on the signal processing method based on the near-field MIMO broadband OFDM interawareness integrated system described in claim 1, step B, user speed perception signal processing flow, assuming that the BS realizes the transmitter and receiver sharing the antenna array through a circulator and full-duplex technology, still taking user k as an example, the BS knows the current position of user k, and transmits an OFDM signal to the user with signal energy focused on the user's position, that is, setting the starting point and end point of the beam squint to be And the user's speed perception is obtained through the user's echo signal, including: Step B1: construct the echo signal model received by the BS from user k. Suppose the total echo signal received by the BS from user k at sampling time t0 is expressed as Its expression is as follows (11): Step B2, construction of optimization problem: Unknown parameter v k =[v r，k , v θ，k ] T The estimation of and β can be transformed into the following optimization problem: Step B3, set The expectation maximization (EM) algorithm is used to solve the optimization problem in equation (12): Step B3.1, randomly initialize the initial speed of user k Used for the first iteration of the EM algorithm, the initialization constant h0 is used for the conditional judgment of the end of the loop; Step B3.2, assume that the speed of user k is fixed and known to be Then the estimation result of the channel fading coefficient is obtained by the least square method, as shown in the following formula (13): In formula (13), represents the pseudo-inverse, Step B3.3, the estimated result of the channel fading coefficient is obtained Substituting into equation (12), for a fixed channel fading coefficient The speed estimation of user k is transformed into the following optimization problem, as shown in equation (14): For this problem, we use the gradient-based method to solve the problem with the objective function G(v k ) for user speed v k Derived, the user speed estimation result is finally obtained through the following gradient descent process, as shown in the following formula (15): In formula (15), γ (i) represents the learning rate of the i-th cycle, and the user speed estimation result is expressed as Step B3.4: If the user speed estimation result obtained in the previous step is satisfy Then As the final estimation result of user speed, user speed perception is completed; otherwise, set And skip to step B3.

2.

5. Based on the signal processing method based on the near-field MIMO broadband OFDM interawareness integrated system of claim 1, step C: beam prediction and tracking signal processing flow, for user k, the specific steps of beam prediction and tracking are: Step C1, initial perception. It is known that within a coherent processing interval (CPI), the user's position and speed are considered unchanged. Let a CPI duration be T0. In the first CPI, set i = 1. At this time, the BS does not perform data transmission but only focuses on the perception of the user k's position and speed. Specifically, the BS first obtains the user position perception result according to the process of step A within the first CPI, which is expressed as Then set the starting and ending points of the beam squint to Substituting into equations (1) and (2), we can obtain the beamforming vector of the BS at this time: And transmit OFDM signal to the user, according to the user's echo signal, based on the process of step B, the initial speed of user k is estimated, which is expressed as Step C2, beamformer design, given the estimated results of user position and velocity within the i-th CPI, and Then the user's position in the next CPI is estimated and expressed as: Design the corresponding beam former according to the user position in the i+1th CPI, that is, set the starting point and end point of the beam squint to Substituting into equations (1) and (2), we get the beamforming vector, expressed as Step C3, data transmission and speed perception: In the i+1th CPI, use the beamformer obtained in the previous CPI Perform effective data transmission, and based on the process of step B, obtain the radial and lateral velocities of the user at the current i+1th CPI according to the echo signal reflected by the user, expressed as Step C4: Set i=i+1 and jump to step C2.

Images