Intelligent metasurface assisted user positioning method based on modified reduced-dimension MUSIC

Abstract

The application belongs to the technical field of radio positioning, and particularly relates to a kind of intelligent metasurface assisted user positioning method based on modified dimensionality reduction MUSIC.The method utilizes the phase reconfigurable characteristics of intelligent metasurface, constructs zero-mean orthogonal phase coefficients, separates out base station-user direct channel and base station-intelligent metasurface-user cascade channel, estimates intelligent metasurface-user channel under the assumption that base station-intelligent metasurface has good line-of-sight path, thereby calculates the left and right correlation coefficients of two channel sections, and reconstructs left and right pseudo-covariance matrix using the Toeplitz matrix of correlation coefficient, further obtains pseudo-covariance matrix, then performs eigenvalue decomposition on pseudo-covariance matrix, constructs modified dimensionality reduction MUSIC spatial spectrum using the orthogonality of signal space and noise space, thereby estimates the azimuth of base station-user and intelligent metasurface-user, and then calculates the user position using least square method.The application trades off higher estimation accuracy for complexity.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of radio positioning, and particularly relates to a user positioning method assisted by an intelligent metasurface based on modified reduced-dimension MUSIC. BACKGROUND

[0002] In the Internet of Things era, satellite positioning, Bluetooth, ultra-wideband technology and the like are rapidly developing with the growth of terminal demand for different application scenarios indoors and outdoors. With the advent of the fifth-generation (5G) mobile communication system, the support of large bandwidth, multiple antennas and high-precision synchronization technology enables the positioning accuracy of 5G to reach the meter level, which can make up for the deficiency of satellite positioning in indoor, underground and tunnel environments. Among them, the angle-based positioning method has attracted more and more attention due to its high accuracy, high noise immunity and no need for synchronization. However, in order to accurately measure the incident angle of electromagnetic waves, the receiver must be equipped with a high spatial resolution antenna array; in addition, the multipath effect in areas with dense buildings is also one of the problems affecting the positioning accuracy.

[0003] The emergence of the intelligent metasurface (RIS) makes it possible to solve the landing problem of the angle-based positioning scheme: the RIS is a two-dimensional new type of artificial material composed of a large number of electromagnetic parameter adjustable sub-wavelength units, which can change the amplitude, phase, polarization mode and the like of the reflected or transmitted signal. With the characteristics of reconfigurable electromagnetic properties and flexible deployment and the advantage of not affecting the existing wireless network infrastructure architecture, RIS has been widely studied in the field of mobile communication in recent years, aiming to use its additional degrees of freedom to improve communication quality, eliminate blind areas and expand coverage. In the field of radio positioning, the RIS has the characteristics of easy formation of a super large array structure and naturally has a high spatial resolution; on the other hand, the RIS can also convert the originally harmful multipath effect into a useful virtual line-of-sight path, playing an additional anchor role. However, there is currently little research on RIS-assisted positioning, and directly moving the conventional two-dimensional multiple signal classification (2D-MUSIC) technology cannot fully utilize the large-scale array characteristics of the RIS, losing the spatial resolution of the RIS. SUMMARY

[0004] In view of the research gap in the field of RIS-assisted user positioning and the limitation that the conventional 2D-MUSIC technology cannot fully utilize the aperture of the RIS array, the application proposes a positioning scheme based on modified reduced-dimension MUSIC.

[0005] To better illustrate the present application, first introduce the terms and system structure used in the technical scheme of the present application.

[0006] BS: Base Station, base station

[0007] LoS, Line-of-Sight, LoS.

[0008] MIMO: Multiple Input Multiple Output, multiple input multiple output.

[0009] MUSIC: Multiple Signal Classification, multiple signal classification.

[0010] NLoS, non Line-of-Sight, non-LoS.

[0011] UE: User Equipment, user equipment.

[0012] ULA: Uniform LinearArray, uniform linear array.

[0013] Figure 1 The intelligent metasurface assisted user positioning system involved in the present application is shown in the schematic diagram:

[0014] Without loss of generality, the present application considers the number of antennas of base station BS and user UE to be N B and N U MIMO scene, assuming that BS and RIS are fixed, UE is movable, the number of intelligent metasurface RIS units is N R , and the Rician channel model under two-dimensional scenario is adopted, that is, the matrix and respectively represent the downlink channel of BS-UE, BS-RIS and RIS-UE, and the specific representation is:

[0015]

[0016] Wherein α IJ and κ IJ are the corresponding path loss and Rician factor, respectively; is the LoS component, specifically

[0017]

[0018] Wherein is the array response vector, θ IJ represents the corresponding azimuth angle, represents the corresponding array tilt angle. for NLoS components, each element of which is independently and identically distributed in complex Gaussian distribution Assume that the far power of the LoS component of the BS-RIS channel is stronger than that of the NLoS component, that is, At the same time, it is assumed that the RIS has no power loss to the reflected signal, and the phase offset in the nth(=1, 2, …, N T ) time slot is

[0019]

[0020] where N T is the total number of time slots. The coefficients of each unit of the RIS satisfy the zero mean orthogonal characteristic, that is,

[0021]

[0022] and

[0023]

[0024] where k, l∈{1, 2, …, N R}. At the same time, it is assumed that the BS transmits the same positioning assistance signal in each time slot where N X ≥N B is the number of symbol vectors, so the received signal matrix in the nth time slot can be expressed as

[0025]

[0026] where is a Gaussian white noise, each element of which is independently and identically distributed in complex Gaussian distribution where N0 is the noise power.

[0027] The technical scheme adopted by the present application is:

[0028] S1, in each time slot, the UE right multiplies the right inverse of the received signal to perform channel estimation, that is,

[0029]

[0030] S2, take the time average of as the estimation of H BU , that is,

[0031]

[0032] S3, calculate the pseudo-covariance matrix of , and perform eigenvalue decomposition to obtain the MUSIC spatial spectrum of the BS-UE path, and take the angle corresponding to the spatial spectrum peak as the estimation of BU , specifically:

[0033] S31, calculating the left pseudo-covariance matrix L of BU , specifically

[0034]

[0035] wherein

[0036]

[0037] S32, calculating the right pseudo-covariance matrix R of BU , specifically

[0038]

[0039] wherein

[0040]

[0041] S33, reconstructing the pseudo-covariance matrix C BU , and performing eigenvalue decomposition, i.e.

[0042]

[0043] wherein Λ is an N B N U N B N U diagonal matrix whose diagonal elements are the eigenvalues of C BU arranged in descending order; the eigenvector matrix is composed of the corresponding eigenvectors;

[0044] S34, dividing the noise space, specifically: deleting the first column of the eigenvector matrix U to obtain U N ;

[0045] S35, calculating the reduced-dimension MUSIC spatial spectrum of the BS-UE path:

[0046]

[0047] S36, taking the peak value of P BU (θ) as the estimation of θ BU :

[0048]

[0049] S4, performing correlation operation between and the coefficients of each unit of the RIS to estimate the BS-RIS-UE channel, thereby obtaining the LoS component of the BS-RIS channel, specifically:

[0050] S41, extract the k∈{1, 2, ..., N R} BS-RIS-UE cascade channel units, specifically:

[0051]

[0052] S42, As [H BR ] k，： Estimates, yes Right multiplication The right inverse of [H RU ] k，： The estimate of

[0053]

[0054] S5. Calculation The pseudo-covariance matrix of the RIS-UE path is decomposed by eigenvalue to obtain the MUSIC spatial spectrum of the RIS-UE path, and the angle corresponding to the peak of the spatial spectrum is used as θ RU The estimates are as follows:

[0055] S51, calculation The left pseudo-covariance matrix L SR , specifically

[0056]

[0057] in

[0058]

[0059] S52, calculation The right pseudo-covariance matrix R RU , specifically

[0060]

[0061] in

[0062]

[0063] S53, reconstruct the pseudo covariance matrix C RU , and perform eigenvalue decomposition, that is

[0064]

[0065] Where Γ is N R N U ×N R N U The diagonal matrix of RU The eigenvalues ​​of are arranged in descending order; the characteristic matrix is composed of corresponding characteristic vectors;

[0066] S54, dividing the noise space, specifically: deleting the first column of the feature matrix V to obtain V N ;

[0067] S55, calculating the reduced dimension MUSIC spatial spectrum of RIS-UE path:

[0068]

[0069] S56, taking the peak value of P RU (θ) as the estimation of θ RU :

[0070]

[0071] S6, according to the results of S36 and S56, calculating the UE coordinates by the least square method, specifically:

[0072]

[0073] Where (x I , y I ), I=B, R, U represent the coordinates of BS, RIS and UE; the above method is also suitable for the uplink channel, and only the order of subscripts B and U needs to be exchanged.

[0074] The beneficial effects of the application are:

[0075] The application provides an intelligent metasurface assisted user positioning scheme based on a modified reduced dimension MUSIC. The scheme utilizes the phase reconfigurable characteristics of the intelligent metasurface, constructs zero-mean orthogonal phase coefficients, separates the base station-user direct channel and the base station-intelligent metasurface-user cascade channel, estimates the intelligent metasurface-user channel under the assumption that the base station-intelligent metasurface has a good line-of-sight path, calculates the left and right correlation coefficients of the two channels, reconstructs the left and right pseudo-covariance matrices by using the Toeplitz matrix of the correlation coefficients, further obtains the pseudo-covariance matrix, performs eigenvalue decomposition on the pseudo-covariance matrix, constructs a modified reduced dimension MUSIC spatial spectrum by utilizing the orthogonality of the signal space and the noise space, estimates the azimuth angles of the base station-user and the intelligent metasurface-user, and then calculates the user position by using the least square method. The application fully utilizes the phase reconfigurable characteristics of the intelligent metasurface, compared with the conventional reduced dimension MUSIC scheme, the scheme reconstructs the pseudo-covariance matrix by using the Toeplitz matrix of the estimated channel correlation coefficients, has the effect of spatial smoothing, and does not waste the effective aperture of the array, so that the estimation accuracy is improved at the cost of complexity, and the practicability is higher. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 Fig. 1: The schematic diagram of the intelligent user positioning system assisted by smart surface based on modified reduced-dimension MUSIC;

[0077] Fig. 2: The modified reduced-dimension MUSIC spatial spectrum of BS-UE channel and RIS-UE channel of the proposed scheme, where (a) is BS-UE channel, (b) is RIS-UE channel;

[0078] Fig. 3: The conventional reduced-dimension MUSIC spatial spectrum of BS-UE channel and RIS-UE channel, where (a) is BS-UE channel, (b) is RIS-UE channel;

[0079] Figure 4 Fig. 4: The 2D-MUSIC spatial spectrum of the conventional 2D-MUSIC scheme;

[0080] Fig. 5: The 100 times positioning simulation results of the proposed scheme, where (a) is κ0= κ1= 0 dB, (b) is κ0= κ1= 20 dB. DETAILED DESCRIPTION

[0081] The technical scheme of the present application has been described in detail in the summary section. The practicality of the present application will be illustrated in the following combined with the attached drawings and simulation examples.

[0082] In the simulation examples of Figs. 2-5, the system configuration of N B = 16, N U = 4 and N R = 63 is adopted unless otherwise specified, the BS, UE and RIS are all uniform linear arrays with half-wavelength array element spacing, the positioning assistance signal X is a unit matrix, the signal power is 1 kW, the time slot number N T = N R + 1, the coordinates of BS and RIS are (50 m, 0 m) and (50 m, 100 m), the noise power is -90 dBm, the angle interval is 0.1 degree, and the Rician factor κ2= 40 dB.

[0083] Fig. 2 shows the modified reduced-dimension MUSIC spatial spectrum of the proposed scheme when the UE is located at (100 m, 75 m) and κ0= κ1= 0 dB, where 2(a) is the BS-UE channel and 2(b) is the RIS-UE channel. As can be seen from the figure, the highest peak of the BS-UE channel spatial spectrum appears at 56.3 degrees, and the highest peak of the RIS-UE channel spatial spectrum appears at 153.4 degrees, which coincides with the corresponding azimuth angle.

[0084] Figure 3 gives the conventional reduced dimension MUSIC spatial spectrum when the UE is located at (100m, 75m) and κ0= κ1= 0dB, wherein 2(a) is the BS-UE channel and 2(b) is the RIS-UE channel. As can be seen from the figure, the position of the highest peak of the spatial spectrum coincides with the corresponding azimuth angle. Comparing Figure 2 with Figure 3, it can be seen that the modified reduced dimension MUSIC spatial spectrum of the scheme has a more sharp peak, and the peak-to-average ratio is more than 20dB; while the peak-to-average ratio of the conventional reduced dimension MUSIC spatial spectrum is only about 3dB. Thus, the scheme has higher positioning accuracy.

[0085] Figure 4 Figure 3 gives the conventional reduced dimension MUSIC spatial spectrum when the UE is located at (100m, 75m) and κ0= κ1= 0dB, wherein 2(a) is the BS-UE channel and 2(b) is the RIS-UE channel. As can be seen from the figure, the position of the highest peak of the spatial spectrum coincides with the corresponding azimuth angle. Comparing Figure 2 with Figure 3, it can be seen that the modified reduced dimension MUSIC spatial spectrum of the scheme has a more sharp peak, and the peak-to-average ratio is more than 20dB; while the peak-to-average ratio of the conventional reduced dimension MUSIC spatial spectrum is only about 3dB. Thus, the scheme has higher positioning accuracy.

[0086] Figure 5 gives the simulation results of the scheme, wherein “x” represents the true coordinates, “o” represents the estimated coordinates, (a) κ0= κ1= 0dB, and (b) κ0= κ1= 20dB. As can be seen from the figure, the scheme has high estimation accuracy, when κ0= κ1= 0dB, the positioning error is in the order of decimeter in most cases; when κ0= κ1= 20dB, the positioning error is in the order of centimeter in most cases. Only when the UE is close to the midpoint of the BS and RIS or close to the horizontal line of the BS / RIS array, there is a large error (in the order of meter or decimeter). This is a common limitation of the azimuth-based positioning scheme: any point on the midpoint of the BS and RIS has the same azimuth angle; the angle estimation accuracy of the ULA array decreases with the increase of the angle away from the central axis. By introducing more RIS and using different types of arrays, the above problems can be solved.

[0087] As can be seen, compared with the conventional 2D-MUSIC and reduced dimension MUSIC schemes, the positioning scheme has higher estimation accuracy, and the complexity is between the two, without the need to determine the characteristics of the receiving and transmitting angle pair, and only by searching the peak values of the two one-dimensional spatial spectrums, the azimuth angles of the BS-UE and the RIS-UE can be determined, and the scheme has higher practicability.

Claims

1. An intelligent metasurface assisted user positioning method based on modified reduced-dimension MUSIC, the positioning system comprising a base station with a root antenna, a user with a root antenna, and an intelligent metasurface with a number of units, defining the downlink channels from the base station to the user, from the base station to the intelligent metasurface, and from the intelligent metasurface to the user as , and , respectively. ， wherein and are the corresponding path loss and Rician factor, respectively; is the LoS component: ， wherein, is the array response vector, denotes the corresponding azimuth angle, denotes the corresponding array tilt angle, is the NLoS component, each element is independently and identically distributed from a complex Gaussian distribution ; set and the intelligent metasurface has no power loss for the reflected signal, the phase shift in the first time slot is: ， wherein , is the total number of slots, and the coefficients of the smart metasurface satisfy the zero-mean orthogonal property: ， ， wherein the base station transmits the same positioning assistance signal in each time slot wherein is the number of symbol vectors, the received signal matrix in the th time slot is ， wherein is a Gaussian white noise; The positioning method comprises the following steps: S1, in each time slot, the user right multiplies the received signal by the right inverse of the positioning auxiliary signal to perform channel estimation, that is ； S2, with the time average of as an estimate of ； S3、Computing the pseudo-covariance matrix, and performing eigenvalue decomposition to obtain the MUSIC spatial spectrum of the base station-to-user path, and taking the angle corresponding to the spatial spectrum peak as the estimation of the direction of arrival, specifically: S31, calculate left pseudo-covariance matrix , specifically ， Wherein ； S32, calculate the right pseudo-covariance matrix , in particular ， Wherein ； S33, reconstructing the pseudo-covariance matrix and performing eigenvalue decomposition, i.e. ， wherein is a diagonal matrix whose diagonal elements are arranged in descending order of their eigenvalues; the eigenvector matrix is composed of the corresponding eigenvectors. S34, dividing the noise space, specifically: delete the first column of the feature matrix , obtaining ; S35, calculate the reduced dimension MUSIC spatial spectrum of the base station to user path: ； S36、with the peak correspondence angle as the estimate of the peak correspondence angle: ； S4. Perform correlation operations with the coefficients of each unit of the smart metasurface to estimate the channel from the base station to the smart metasurface to the user, thereby obtaining the LoS component of the base station to smart metasurface channel, specifically: S41, extract the first The cascade channel from the base station to the smart metasurface to the user is as follows: ； S42, to As the estimate of right-multiply the right inverse of the estimate, i.e. ； S5、Calculate the pseudo-covariance matrix, and perform eigenvalue decomposition to obtain the MUSIC spatial spectrum of the intelligent metasurface to the user path, and take the angle corresponding to the spatial spectrum peak as the estimation, specifically: S51, calculate left pseudo-covariance matrix , in particular ， Wherein ； S52, compute the right pseudo-covariance matrix , in particular ， Wherein ； S53, reconfigure the pseudo-covariance matrix and perform eigenvalue decomposition, i.e. ， wherein is a diagonal matrix whose diagonal elements are arranged in descending order of their eigenvalues; the eigenvector matrix is composed of the corresponding eigenvectors. S54, dividing the noise space, specifically: deleting the first column of the feature matrix , obtaining ; S55, calculate the reduced dimension MUSIC spatial spectrum of the intelligent metasurface to user path: ； S56, to the peak corresponding angle as the estimate of ； S6, according to the results of S36 and S56, calculate the UE coordinates by the least square method, specifically: ， wherein denotes the coordinates of the base station, the intelligent metasurface, the user.

Images