A multi-state sensing method based on orthogonal time-frequency-air modulation

Abstract

The present invention discloses a multi-state sensing method based on orthogonal time-frequency-air modulation, which belongs to the field of wireless communications. The method includes the following steps: a user sends an uplink OTFS modulated signal; a base station estimates the delay and Doppler of the first antenna through the OMP algorithm; and jointly estimates the delay difference between antennas. The user places a reference bit sequence in the delay-Doppler domain, transforms it into the time domain through OTFS modulation, and transmits it through a high-speed multipath channel; the base station estimates the delay and Doppler information of the first antenna through the OMP algorithm and distinguishes the number of paths; the base station constructs a joint dictionary matrix for the remaining antennas using the multiple relationship of the delay difference between adjacent antennas, thereby estimating the delay difference between adjacent antennas, and then constructs a small dictionary matrix based on the estimated delay difference, which is then converted into an estimated angle. Finally, the specific location of the user is determined based on the estimated Doppler, delay difference, and angle information. The present invention realizes channel estimation and precise positioning of the user, reducing the computational complexity of angle estimation.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communications, and in particular relates to a multi-state sensing method based on orthogonal time-frequency-air modulation. Background Art

[0002] With the increasing popularity of fifth-generation (5G) wireless communication networks and the increasing maturity of technology and hardware, more optimization algorithms have been proposed to improve communication reliability and quality. The OTFS waveform is gaining widespread adoption due to its inherent ability to withstand high Doppler frequencies. However, the Orthogonal Frequency Division Multiplexing (OFDM) waveform used in fourth- and fifth-generation wireless systems is not suitable for high-speed mobility scenarios. Furthermore, because OTFS places symbols in the delay-Doppler domain, the channel representation in the delay-Doppler domain is relatively sparse. Radar signals are also generally processed in the delay-Doppler domain. Therefore, combining OTFS with synaesthesia integration technology has advantages.

[0003] The Orthogonal Matching Pursuit algorithm, as one of the methods for sparsely decomposing a signal, decomposes the signal on a complete dictionary. The improvement of the OMP algorithm lies in orthogonalizing all selected atoms at each decomposition step. This allows the OMP algorithm to converge faster while maintaining the same accuracy requirements.

[0004] Currently, angle estimation is often performed using methods such as MUSIC and OMP. However, the traditional OMP algorithm, when operating at high resolution, results in a large dictionary matrix and high computational complexity. Furthermore, the OMP algorithm, when combined with OTFS modulation, only estimates delay and Doppler parameters and does not exploit the sparsity of the OTFS channel for angle estimation. Therefore, this method considers how to exploit the sparsity of OTFS channels, combining it with the OMP algorithm to achieve joint estimation of multiple antennas by designing a dictionary matrix. Furthermore, this method uses a step-by-step approach to reduce computational complexity. Summary of the Invention

[0005] To solve the above problems, the present invention discloses a multi-state perception method based on Orthogonal Time Frequency Space (OTFS), which combines the orthogonal matching pursuit algorithm with the OTFS signal to realize angle estimation in high Doppler channels, reduce the computational complexity, and achieve precise vehicle positioning.

[0006] A multi-state sensing method based on orthogonal time-frequency-space modulation includes the following steps: a user sends an uplink signal, a base station estimates the delay and Doppler of the first antenna using an OMP (Orthogonal Matching Pursuit) algorithm, and jointly estimates the delay difference between antennas; a base station and user model is constructed, and the time domain signal sent by the user is known to the base station; the user's reference bit sequence is placed in the delay-Doppler domain, converted to the time-frequency domain through an inverse sigmoid Fourier transform, and then converted to the time domain through an inverse Fourier transform before being sent; the user moves at high speed and passes through a multipath channel.

[0007] The user places the reference bit sequence in the delay-Doppler domain, converts it to the time-frequency domain through the inverse sigmoid Fourier transform, and then converts it to the time domain through the IFFT and sends it through the high-speed multipath channel;

[0008] The base station estimates the delay and Doppler information of the received signal from the first antenna using the OMP algorithm. The number of paths can be distinguished by the differences in delay and Doppler between different paths.

[0009] The base station first constructs dictionary matrices for each of the remaining antennas using the multiples of the delay differences between adjacent antennas. Because the multiples are included in each dictionary matrix, the projections of different antennas on their respective dictionary matrices are identical. This allows a joint dictionary matrix for all remaining antennas to be constructed. The OMP algorithm is then used to estimate the delay differences between adjacent antennas. A small-number dictionary matrix is ​​then constructed based on the estimated delay differences to improve their accuracy, which is then converted into an estimated angle.

[0010] After building a model of a vehicle traveling on a straight road with a roadside unit, we can construct a vehicle topology map using the time delay, Doppler, and angle estimates between different paths in the previous steps. Assuming a road width of 10 meters, simply estimating the angle of the direct path cannot determine the specific position of the vehicle on the road. However, this information can reduce the vehicle's position error to within 1 meter.

[0011] The user uplink signal is an OTFS signal. Before using multiple base station antennas to perform channel estimation on the signal passing through the multipath channel, the user generates a reference signal in the delay-Doppler domain. This reference signal is known at the base station. An inverse sigmoid Fourier transform and an IFFT are performed to convert the signal to the time domain. The specific steps are as follows:

[0012] Consider an uplink user communicating with a base station (BS) receiver. The user is equipped with one transmitting antenna and uses OTFS modulation for uplink transmission. The receiver at the base station is equipped with N r The information bits from different users are multiplexed on the N×M layered Doppler grid Γ as follows:

[0013]

[0014] in, and are the resolutions of Doppler shift and delay shift respectively, N is the number of Doppler grid points, and M is the number of delay grid points. Let τ max represents the maximum delay spread of the multi-user channel, ν max represents the maximum Doppler spread. Then, ∆f is chosen to satisfy x[k,l], k = 0, 1, ..., N-1, l = 0, 1, ..., M-1 represents the delay-Doppler domain information symbol transmitted by the user.

[0015] The information symbols of user x[k,l] are mapped from the delay-Doppler domain to the time-frequency domain using the inverse symplectic finite Fourier transform (I SFFT). The modulated time-frequency domain signal corresponding to the user is:

[0016]

[0017] Where N is the number of grid points in the Doppler domain, and M is the number of grid points in the delay domain.

[0018] Then, the obtained time-frequency domain signal is transformed into the time domain using the Heisenberg transform and transmitted over the time-varying channel. The user's transmission time domain signal is:

[0019]

[0020] Among them, g tx (t) represents the shape of the transmitted pulse, T is a symbol period, and △f is the subcarrier spacing.

[0021] At the base station, after obtaining the received signals of multiple antennas, the main task is to design a dictionary matrix for two fractional estimates.

[0022] Since the number of paths is estimated and distinguished for the first antenna, for 2-N r The structure of the dictionary matrix constructed for different paths of the root antenna is as follows:

[0023] For the p-th path and the i-th antenna, construct the fractional dictionary matrix:

[0024]

[0025] Among them, the value range of k in the kth column in the above formula is 1-M,

[0026] is the estimated delay of the first antenna, Doppler value estimated for the first antenna.

[0027] F MN is the FFT matrix, Γ is the delay matrix, Δ is the Doppler matrix;

[0028] N is the number of grid points in the Doppler domain, and M is the number of grid points in the delay domain.

[0029] The Doppler values ​​of different antennas on the same path are the same, so the Doppler value estimated by the first antenna can be used to reduce the dimension of the dictionary matrix.

[0030] The projection of the received signal of the i-th antenna on its dictionary matrix can be expressed as: i =Ψ p,i h

[0031] Due to the particularity of the dictionary matrix construction, the projection h of the received signals of different antennas on the dictionary matrix is ​​the same. r The dictionary matrices of the root antennas are concatenated into a joint dictionary matrix:

[0032]

[0033] Then, r = Ψ p h, where

[0034] use By drawing the peak curve, we can obtain the delay difference at the peak, which can be accurate to the percentile of the delay domain grid point.

[0035] Based on the first decimal estimation, a second decimal estimation is performed, accurate to the ten-thousandth place. The dictionary matrix for the second decimal estimation can be constructed in the same way as the dictionary matrix for the i-th antenna of the p-th path in the first decimal estimation, that is:

[0036] in, is the estimated delay of the first antenna, is the Doppler value estimated for the first antenna, L frac,p It is the first decimal estimate of the delay difference of the p-th path.

[0037] F MN is the FFT matrix, Γ is the delay matrix, Δ is the Doppler matrix;

[0038] N is the number of grid points in the Doppler domain, and M is the number of grid points in the delay domain.

[0039] The dictionary matrix constructed in this way contains the estimated delay of the first antenna and the fractional delay difference of the first fractional estimate. By drawing the peak curve, we can obtain the delay difference at the peak, which can be accurate to the ten-thousandth of the delay domain grid point.

[0040] 1. The specific operations of step S105 are as follows:

[0041] When there are two paths in the multipath channel, the incident angles of the two paths at the base station are According to the vehicle topology diagram, the angle between the vehicle and the base station and the normal line can be known:

[0042] because The angle between the second diameter of the vehicle and the normal

[0043] Due to the delay difference between the two paths Known, the distance between the vehicle and the base station It can be calculated that:

[0044]

[0045] Since the incident angle of the direct path has been estimated, the position of the vehicle on the straight road can be determined. If a rectangular coordinate system is established with the base station as the origin, the estimated coordinates of the vehicle are As shown below:

[0046]

[0047]

[0048] Beneficial effects of the present invention:

[0049] The OTFS waveform is combined with the orthogonal matching pursuit algorithm to fully leverage their technical advantages. By designing a dictionary matrix, multiple antennas can be jointly estimated when using the OMP algorithm, reducing estimation errors. By estimating the delay and Doppler of the first antenna in step one, the subsequent small-number dictionary matrix reduces the Doppler domain dimensionality and computational complexity. This enables channel estimation and vehicle positioning in connected vehicle scenarios, improving the effectiveness and real-time performance of wireless communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flowchart of a multi-state sensing method based on orthogonal time-frequency-air modulation according to an embodiment of the present invention;

[0051] Figure 2 A model diagram of a multi-state perception method based on orthogonal time-frequency-air modulation according to an embodiment of the present invention;

[0052] Figure 3This is a graph showing how the angle estimation error changes with the signal-to-noise ratio.

[0053] Figure 4 This is a graph of the peaks on the normalized delay-Doppler grid. This simulation estimates the delay and Doppler values ​​using the first antenna. The graph shows two distinct peaks, indicating two paths. The coordinates of the two peaks correspond to the delay and Doppler values ​​of the two paths, respectively.

[0054] Figure 5 This is a graph showing the vehicle positioning error changing with the signal-to-noise ratio. DETAILED DESCRIPTION

[0055] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are intended only to illustrate the present invention and are not intended to limit the scope of the present invention. It should be noted that the terms "front," "rear," "left," "right," "up," and "down" used in the following description refer to directions in the accompanying drawings, and the terms "inward" and "outward" refer to directions toward or away from the geometric center of a particular component, respectively.

[0056] Figure 1 The present invention provides a flowchart of a multi-state sensing method based on orthogonal time-frequency-air modulation according to an embodiment of the present invention.

[0057] like Figure 1-2 As shown, the multi-state sensing method based on orthogonal time-frequency-air modulation includes the following steps:

[0058] In step S101, consider an uplink user communicating with a base station receiver. The user is equipped with one transmitting antenna and uses OTFS modulation for uplink transmission. The receiver at the base station is equipped with N r The information bits from different users are multiplexed on the N×M layered Doppler grid Γ as follows:

[0059]

[0060] in, and are the resolutions of Doppler shift and delay shift respectively, N is the number of Doppler grid points, and M is the number of delay grid points. Let τ max represents the maximum delay spread of the multi-user channel, ν max represents the maximum Doppler spread. Then, ∆f is chosen to satisfy x[k,l], k = 0, 1, ..., N-1, l = 0, 1, ..., M-1 represents the delay-Doppler domain information symbol transmitted by the user.

[0061] The information symbol of user x[k,l] is mapped from the delay-Doppler domain to the time-frequency domain using the inverse symplectic Fourier transform (I SFFT). The modulated time-frequency domain signal corresponding to the user is:

[0062]

[0063] Then, the obtained time-frequency domain signal is transformed into the time domain using the Heisenberg transform and transmitted over the time-varying channel. The user's transmission time domain signal is:

[0064]

[0065] Among them, g tx (t) represents the shape of the transmitted pulse.

[0066] The channel is represented in the DD domain, and its complex baseband channel response to the user is represented by h(τ,υ), where τ is the delay and υ is the Doppler shift. The transmitted signal x(t) undergoes a rapidly time-varying channel before being received at the base station. The time domain signal y(t) received at the base station is:

[0067] y(t)=∫ υ ∫ τ h(τ,υ)x(t-τ)e j2 πυ(t-τ)dτ dυ +n(t)

[0068] However, the cyclic delay matrix cannot directly include fractional delay terms. To account for fractional delays, the cyclic delay matrix can be replaced by a diagonal delay matrix, where the cyclic shifts associated with the target-induced delay are converted into phase shifts in the frequency domain with the help of the DFT matrix. The inverse DFT matrix converts the signal back to the time domain. This makes it possible to directly include fractional delays and fractional Doppler shifts without changing the size of the channel matrix. The resulting channel matrix H, which includes non-integer delays and Doppler shifts, is:

[0069]

[0070] We can see that for any integer delay, the product Equivalent to the traditional cyclic delay matrix.

[0071] In step S102, the base station estimates the delay and Doppler information of the received signal of the first antenna using the OMP algorithm, and the number of paths can be distinguished by the differences in delay and Doppler of different paths.

[0072] Let the target in the channel produce integer delay and Doppler frequency shift on the transmitted signal. The OTFS signal vector received in the time domain, r i It can be written as follows:

[0073] r i =H i s+v

[0074] The above formula can be reconstructed into sparse processing as follows:

[0075] r i =Ψh+v

[0076] where Ψ is a MN×MN dictionary matrix that takes into account integer delay and Doppler shift.

[0077]

[0078] Therefore, using the received signal vector and the integer approximation dictionary matrix Ψ (which can be constructed at the radar receiver), the target can be detected and the delay-Doppler information can be estimated using the traditional OMP algorithm.

[0079] Since the delay and Doppler of different paths in the channel are different, the number of multipaths and the delay and Doppler of each path can be determined by observing the number of peaks in the peak curve. Then, each path can be analyzed separately to estimate the incident angle of each path.

[0080] In step S103, the user is sending with a single antenna, and the base station is equipped with N r Since the number of paths is estimated and distinguished for the first antenna, for 2-N r The structure of the dictionary matrix constructed for different paths of the root antenna is as follows:

[0081] For the p-th path and the i-th antenna, construct the fractional dictionary matrix:

[0082]

[0083] in, F MN is the FFT matrix, Γ is the delay matrix, Δ is the Doppler matrix;

[0084]

[0085] The Doppler values ​​of different antennas on the same path are the same, so the Doppler value estimated by the first antenna can be used to reduce the dimension of the dictionary matrix.

[0086] The projection of the received signal of the i-th antenna on its dictionary matrix can be expressed as:

[0087] r i =Ψ p,i h

[0088] Due to the particularity of the dictionary matrix construction, the projection h of the received signals of different antennas on the dictionary matrix is ​​the same. r The dictionary matrices of the root antennas are concatenated into a joint dictionary matrix:

[0089]

[0090] Then, r = Ψ p h, where

[0091] use By drawing the peak curve, we can obtain the delay difference at the peak, which can be accurate to the percentile of the delay domain grid point.

[0092] In step S104, based on the first decimal estimation, a second decimal estimation is performed, accurate to the ten-thousandth place. The dictionary matrix for the second decimal estimation can be constructed in the same way as the dictionary matrix for the first decimal estimation of the p-th path i-th antenna, that is:

[0093]

[0094] in, is the estimated delay of the first antenna, is the Doppler value estimated for the first antenna, L frac,p It is the first decimal estimate of the delay difference of the p-th path.

[0095] F MN is the FFT matrix, Γ is the delay matrix, Δ is the Doppler matrix;

[0096] N is the number of grid points in the Doppler domain, and M is the number of grid points in the delay domain.

[0097] The dictionary matrix constructed in this way contains the estimated delay of the first antenna and the fractional delay difference of the first fractional estimate. By drawing the peak curve, we can obtain the delay difference at the peak, which can be accurate to the ten-thousandth of the delay domain grid point.

[0098] In step S105, after constructing a model of a vehicle traveling on a straight road with a roadside unit, the vehicle topology map can be constructed using the time delay difference, Doppler, and angle between different paths estimated in the above steps. Assuming the road width is 10 meters, simply estimating the angle of the direct path cannot determine the specific position of the vehicle on the road. However, the above information can reduce the vehicle's position error to within 1 meter. The specific operation is as follows:

[0099] In the case where there are two paths in the multipath channel, the above steps can obtain the incident angles of the two paths at the base station as follows: According to the vehicle topology diagram, the angle between the vehicle and the base station and the normal line can be known:

[0100] because The angle between the second diameter of the vehicle and the normal

[0101] Due to the delay difference between the two paths Known, the distance s between the vehicle and the base station can be calculated as:

[0102]

[0103] The present invention will be further described below with reference to signal examples.

[0104] The OTFS signal with a subcarrier spacing of 750kHz is used. The number of base station receiving antennas is N r = 20, a single user sends an uplink signal, the multipath channel has two paths, with coefficients of 1 and 0.75, respectively. The delay grid points on the first antenna are 1 and 3, the angles are 30° and 60°, and the Doppler grid points are 3 and 6, respectively. The number of delay grid points M = 100, the number of Doppler grid points N = 40, the spacing between adjacent antennas d = 1m, and the signal-to-noise ratio is 20dB.

[0105] like Figure 3 , which is the matlab simulation result of the embodiment, wherein the horizontal axis represents the signal-to-noise ratio, the vertical axis represents the average angular error of the two paths, the red line represents the present solution, and the blue line represents the traditional OMP algorithm.

[0106] like Figure 4 Figure 1 shows the Matlab simulation results of an embodiment, showing a peak curve plotted on the standardized delay-Doppler grid. This simulation estimates the delay and Doppler values ​​using the first antenna. The figure shows two distinct peaks, indicating two paths. The coordinates of the two peaks correspond to the delay and Doppler values ​​of the two paths, respectively.

[0107] like Figure 5 , which is a matlab simulation result of the embodiment, wherein the horizontal axis represents the signal-to-noise ratio, and the vertical axis represents the distance difference between the estimated vehicle position and the actual position.

[0108] The technical means disclosed in the solution of the present invention are not limited to the technical means disclosed in the above-mentioned embodiment, but also include technical solutions composed of any combination of the above technical features.

Claims

1. A multi-state sensing method based on orthogonal time-frequency-space modulation, characterized by: The specific steps include: Step S101: The user places a reference bit sequence in the delay-Doppler domain, converts it to the time-frequency domain through an inverse sigmoid Fourier transform, and then converts it to the time domain through a Heisenberg transform and sends it through a high-speed multipath channel; Step S102: The base station estimates the delay and Doppler information of the received signal of the first antenna using the OMP algorithm. The number of paths can be distinguished by the differences in delay and Doppler of different paths. Step S103: The base station first constructs a dictionary matrix for each of the remaining antennas using the multiple relationship of the delay difference between adjacent antennas. Since the multiple number is added to each dictionary matrix, the projections of different antennas on their respective dictionary matrices are the same. Therefore, a joint dictionary matrix of all remaining antennas is constructed, and then the OMP algorithm is used to estimate the delay difference between adjacent antennas. Step S103: The user is transmitting with a single antenna, and the base station is equipped with N r Root receiving antennas; Due to the estimation of the first antenna and the distinction of the number of paths, this step is the first decimal estimation, so for 2-N r The structure of the dictionary matrix constructed for different paths of the root antenna is as follows: For the p-th path and the i-th antenna, construct the fractional dictionary matrix: Among them, the value range of k in the kth column in the above formula is 1-M, is the estimated delay of the first antenna, Doppler value estimated for the first antenna; F MN is the FFT matrix, Γ is the delay matrix, Δ is the Doppler matrix; M is the number of grid points in the delay domain, and N is the number of grid points in the Doppler domain; The Doppler values ​​of different antennas on the same path are the same, so the Doppler value estimated by the first antenna can be used to reduce the dimension of the dictionary matrix; The projection of the received signal of the i-th antenna on its dictionary matrix is ​​expressed as: i =Ψ p,i Due to the particularity of the dictionary matrix construction, the projection h of the received signals of different antennas on the dictionary matrix is ​​the same; therefore, 2-N r The dictionary matrices of the root antennas are concatenated into a joint dictionary matrix: Then, r = Ψ p h, where use Draw the peak curve to obtain the delay difference at the peak, accurate to the percentile of the delay domain grid point; Step S104: Based on the first fractional estimation, a dictionary matrix for a second fractional estimation is constructed in the same manner as the dictionary matrix for the i-th antenna of the p-th path in the first fractional estimation, and the second fractional estimation is performed to the ten-thousandth place. Step S105: After constructing a model of a vehicle traveling on a straight road with a roadside unit, a vehicle topology map is constructed using the time delay difference, Doppler, and angle between different paths estimated in the above steps to determine the specific position of the vehicle on the road.

2. The multi-state sensing method based on orthogonal time-frequency-air modulation according to claim 1 is characterized in that: In step S102, basic delay and Doppler information is first obtained through the first antenna, and then the Doppler component does not need to be considered when performing joint estimation using the remaining antennas, thereby reducing the complexity of the calculation.

3. The multi-state sensing method based on orthogonal time-frequency-air modulation according to claim 1, characterized in that: Said step S102: after receiving the signal, the base station estimates the integer delay and Doppler information of the first antenna by using the OMP algorithm. Since the delay or Doppler of different paths is different, the number of paths can be distinguished.

4. The multi-state sensing method based on orthogonal time-frequency-air modulation according to claim 1, characterized in that: The step S103: the fractional OMP algorithm is then applied to each path; the fractional estimation here needs to be accurate to the percentile of the delay domain grid point and the Doppler domain grid point.

5. The multi-state sensing method based on orthogonal time-frequency-air modulation according to claim 1 is characterized in that: Step S104: Based on the first decimal estimation, a second decimal estimation is performed, accurate to the ten-thousandth place; the dictionary matrix for the second decimal estimation can be constructed in the same way as the dictionary matrix for the p-th path i-th antenna in the first decimal estimation, that is: in, is the estimated delay of the first antenna, is the Doppler value estimated for the first antenna, L frac,p is the first decimal estimate of the delay difference of the p-th path; F MN is the FFT matrix, Γ is the delay matrix, Δ is the Doppler matrix; M is the number of grid points in the delay domain, and N is the number of grid points in the Doppler domain; The constructed dictionary matrix contains the estimated delay of the first antenna and the fractional delay difference of the first fractional estimate; it can also be used By drawing the peak curve, we can obtain the delay difference at the peak, which can be accurate to the ten-thousandth of the delay domain grid point.

6. The multi-state sensing method based on orthogonal time-frequency-air modulation according to claim 1, characterized in that: The specific operations of step S105 are as follows: When there are two paths in the multipath channel, the incident angles of the two paths at the base station are According to the vehicle topology diagram, the angle between the vehicle and the base station and the normal line can be known: because The angle between the second diameter of the vehicle and the normal Due to the delay difference between the two paths Known, the distance between the vehicle and the base station It can be calculated that: Since the incident angle of the direct path has been estimated, the position of the vehicle on the straight road can be determined; if a rectangular coordinate system is established with the base station as the origin, the estimated coordinates of the vehicle are As shown below:

Images