Low-overhead positioning method assisted by intelligent metasurface under hybrid field beam deviation effect

Abstract

A low-overhead positioning method assisted by an intelligent metasurface under the mixed-field beam offset effect belongs to the field of perception in wireless communications. The present invention uses a polar coordinate domain dictionary to add distance factors to the dictionary elements, optimizes the mismatch problem between the traditional discrete Fourier transform dictionary and the near-field spherical wave, and reduces the adverse effects of spectrum leakage on positioning performance; by using a frequency-dependent dictionary, different subcarriers are considered separately in the dictionary, optimizes the beam offset effect, and improves positioning accuracy; with the assistance of an intelligent metasurface, the arrival time difference is used to solve the problem of inaccurate user positioning caused by the common timing deviation, and at the same time locks the user on the hyperbola, reducing the computational complexity brought by the full-domain search; the polar coordinate domain gradient descent algorithm is used to optimize the angle at which the user arrives at the base station, and optimizes the problem of being unable to accurately estimate the angle when the distance estimate is not completely accurate. The present invention is applicable to fields such as communications and emergency rescue, and is used to improve user positioning accuracy.

Description

Technical Field

[0001] The present invention relates to a low-overhead positioning method assisted by an intelligent metasurface under a mixed-field beam deviation effect, and belongs to the field of perception in wireless communications. Background Art

[0002] Future 6G communication networks will have target identification, location, detection, and tracking capabilities. Communication systems will be able to perceive all services, networks, and terminals, with positioning being a key component of this perception. 6G's ultra-large bandwidth and ultra-large arrays significantly improve both delay and angle estimation accuracy. Consequently, the IMT-2030 (6G) Promotion Group's white paper, "Research Report on Integrated Communication and Perception Technologies," states that future 6G communication networks will achieve centimeter-level positioning accuracy, enabling them to perceive the physical world 24 / 7 and everywhere.

[0003] Compared to open-air locations, where satellite navigation provides positioning, complex urban and indoor environments rely on communication networks for positioning. Communication network positioning can generally be categorized into three types, based on received signal strength, time of arrival, and angle of arrival. Due to the ultra-large bandwidth and array aperture of 6G communication networks, positioning using extremely high delay and angular resolution has attracted significant attention in both academia and industry. Existing research includes research on directly locating users using massive MIMO across multiple base stations, replacing compressed sensing that derives user positions from the observation angles of multiple base stations. Research also includes joint beam training and positioning, using the assistance of multiple intelligent metasurfaces to jointly locate users from multiple observation angles. Research also includes estimating user positions and the departure and arrival angles of base stations and users using simultaneous orthogonal matching pursuit and expectation maximum algorithms. Furthermore, literature has derived Cramer-Rao lower bounds for some positioning parameter estimation, providing theoretical guidance for the deployment of intelligent metasurfaces and algorithm design.

[0004] However, the above methods do not take into account the simultaneous existence of far-field and near-field (i.e., mixed field) caused by the extremely large antenna array of the 6G communication network. At the same time, the problem of how to achieve high-precision user positioning under the beam deviation effect caused by the extremely large bandwidth of the 6G communication network has not been solved. Therefore, further research is urgently needed on how to use smart metasurfaces to assist in high-precision user positioning under the mixed field beam deviation effect. Summary of the Invention

[0005] In response to the problem of difficult positioning under the mixed field beam deviation effect, the main purpose of the present invention is to provide a low-overhead positioning method assisted by an intelligent metasurface under the mixed field beam deviation effect, which improves the accuracy and efficiency of user positioning through the arrival time difference and polar coordinate domain gradient descent method.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] The low-overhead positioning method assisted by an intelligent metasurface under the hybrid field beam offset effect disclosed in the present invention uses the base station and the intelligent metasurface as anchor points and also as the focus of the hyperbola. The arrival time difference obtained by the multiple signal classification algorithm is used to lock the user on the hyperbola. The larger search area in the entire space is reduced to the smaller area where the hyperbola is located, reducing the computational complexity of the full-space search. After roughly obtaining the user position on the hyperbola, the angle between the user and the base station is super-resolved through the polar coordinate domain gradient descent algorithm, thereby obtaining a more accurate user position and improving the accuracy of user positioning.

[0008] The low-overhead positioning method assisted by an intelligent metasurface under the hybrid field beam deviation effect disclosed in the present invention comprises the following steps:

[0009] Step 1: Using the received signal in the frequency domain, the delay of the user reaching the base station and the smart metasurface is estimated through a multiple signal classification algorithm;

[0010] Turn off the smart metasurface to get the uplink receiving signal in the frequency domain Turn on the smart metasurface to get the uplink receiving signal in the frequency domain Spectral decomposition using uplink received signals

[0011] (Y NRIS ) H Y NRIS =E NRIS Λ NRIS (E NRIS ) H ,

[0012] (Y RIS ) H Y RIS =E RIS Λ RIS (E RIS ) H ,

[0013] Among them, P NRIS and P RIS N is the number of uplink pilot time slots when the intelligent metasurface is turned off and on; RF is the number of RF links; M is the number of subcarriers; E NRIS and E RIS They are (Y NRIS ) H Y NRIS and (Y RIS ) H Y RIS The matrix composed of the eigenvectors of NRIS and ΛRIS They are (Y NRIS ) H Y NRIS and (Y RIS ) H Y RIS The diagonal matrix is ​​composed of the eigenvalues ​​of , where the eigenvalues ​​are arranged in descending order. Since the non-direct path loss in the terahertz band is relatively serious, the eigenvectors other than the eigenvector corresponding to the largest eigenvalue are used as the noise subspace For each subcarrier frequency generate To estimate the delay

[0014]

[0015]

[0016] in, and The delays are the direct path delays from the user directly to the base station and the delays from the user to the base station via the smart metasurface. The smart metasurface provides a degree of freedom in delay to assist in user positioning, without involving channel estimation via the smart metasurface. Specifically, only the direct path delay from the user to the smart metasurface is estimated, without estimating the angle information in the direct path or the scatterer information in the indirect path, including angles, delays, and gains. Therefore, this low-overhead smart metasurface-assisted positioning method, utilizing the hybrid field beam shift effect, can reduce pilot overhead.

[0017] Step 2: Using the two delay information obtained in step 1, the arrival time difference between the user and the base station and the smart metasurface is obtained, and based on the positions of the base station and the smart metasurface, a branch of the hyperbola where the user is located is obtained;

[0018] According to the two delays obtained in step 1, the delay difference τ between the user to the base station and the user to the smart metasurface is obtained. TDoA The equation of the hyperbola is expressed as:

[0019]

[0020] The hyperbola focuses on base stations and smart metasurfaces. c is the speed of light; r B2R is the distance from the center of the base station antenna to the center of the smart metasurface; without loss of generality, If the distance from the user to the base station is greater than or equal to the distance from the user to the smart metasurface, then a hyperbola close to the smart metasurface is taken; if the distance from the user to the base station is less than the distance from the user to the smart metasurface, then a hyperbola close to the base station is taken.

[0021] Step 3: Multiply the phase shift network by the partial dictionary generated on the hyperbola, and correlate the multiplication result with the received signal to obtain the initial user location formula;

[0022] Generate polar domain frequency-dependent dictionary on a hyperbola That is, the angle between the base station and the hyperbola is divided evenly, and the straight line starting from the base station will have a series of intersections with the hyperbola. N, Q and M are the number of array elements of the base station antenna, the number of atoms in the dictionary, and the number of subcarriers, respectively. Each element in the dictionary is composed of the angle and distance of each intersection relative to the base station, which is the meaning of "polar coordinate domain". Since the system is terahertz ultra-wideband, the dictionary of different subcarriers needs to consider the specific subcarrier size, rather than replacing them all with the center carrier frequency, which is the meaning of "frequency dependence". For a frequency of f m , the angle from the user to the center of the base station is θ, which is the real angle The sine value of the element in the dictionary whose distance to the center antenna of the base station is r is expressed as

[0023]

[0024] in, is the distance from the user to the nth antenna element of the base station, δ n =(2×n-N+1) / 2, where n=0,…,N-1 is the index of the base station antenna, and the center element is the zero point of the index; d is the element spacing; is the wave number of the mth subcarrier.

[0025] The phase shift network is multiplied by the polar coordinate domain dictionary on the hyperbola, and the multiplication result is correlated with the received signal to obtain a preliminary angle estimate from the user to the base station.

[0026]

[0027] where e(θ) is the index in the dictionary where θ is located; It's P NRIS The base station phase shift network is composed of time slots. The equation of the joint hyperbola and the position of the base station (x BS ,y BS )Get the user's position formula (x UE ,y UE )

[0028]

[0029] in, y BS Set to 0; a and b are the lengths of the real and imaginary semi-axes of the hyperbola, respectively.

[0030] Step 4: Fix the distance between the user and the base station and use the polar coordinate domain gradient descent algorithm to calculate the direct path angle from the user to the base station. Optimize;

[0031] After obtaining a rough estimate of the user's location, the polar coordinate domain gradient descent algorithm is used to optimize the angle of the direct path from the user to the base station. The loss function of the polar coordinate domain gradient descent algorithm is

[0032]

[0033] in, is the processed received signal, is the column vector corresponding to its m-th subcarrier; It's P NRIS A base station phase-shift network composed of time slots; It is the virtual channel from user to base station. is the column vector corresponding to the mth subcarrier, and its expression is

[0034]

[0035] in, and are the estimated channel gain, angle, and distance from the user to the base station, respectively. The initial values ​​of distance and angle are the results from step 3, and the channel gain is calculated using the Friis transmission formula. The actual channel expression is

[0036]

[0037] Where L is the number of indirect paths. When l is 0, it is the direct path from the user to the base station. When l is other values, it is the indirect path from the user to the base station. and are the actual channel gain of the lth path, the angle from the user to the base station, and the distance from the user to the base station. In the loss function of the polar coordinate domain gradient descent, the processed received signal is a variant of the original received signal.

[0038]

[0039] in, The premise for this is that the base station's phase shift network needs to be specially designed as follows, that is, for

[0040]

[0041] and Each element from the second row to the last row of is randomly set and obeys Gaussian distribution. The purpose of this is to The influence of is removed in the received signal, and when generating the loss function, the virtual channel No This can be used to estimate the angle between the user and the base station. will not be affected by the inaccurate distance between the user and the base station. to optimize the angle based on an imperfect distance.

[0042] The partial derivative of the loss function of the gradient descent in the polar coordinate domain with respect to the estimated angle from the user to the base station is

[0043]

[0044] Where Re{·} represents the real part; tr{·} represents the trace of the matrix. The estimated angle from user to base station The partial derivative of

[0045]

[0046] in, In the iterative process of gradient descent in polar coordinates, the step size η is determined using the Armijo-Goldstein criterion. Update current estimate This update process continues until The change before and after the update is less than the given threshold κ or the number of iterations reaches the set maximum number of iterations N it .

[0047] Step 5: After obtaining the equation of the hyperbola and using the polar coordinate domain gradient descent algorithm to optimize the direct path angle from the user to the base station Then, combine the location of the base station and Substitute this into the intersection formula described in step 3 to obtain the optimized user position, thereby improving positioning accuracy.

[0048] Beneficial effects:

[0049] 1. The low-overhead positioning method assisted by intelligent metasurface under the hybrid field beam deviation effect disclosed in the present invention uses a polar coordinate domain dictionary to add distance factors to the dictionary elements, thereby optimizing the mismatch problem between the traditional discrete Fourier transform dictionary and the near-field spherical wave, and reducing the adverse effects of spectrum leakage on positioning performance.

[0050] 2. The low-overhead positioning method assisted by intelligent metasurface under the mixed-field beam offset effect disclosed in the present invention uses a frequency-dependent dictionary, in which different subcarriers are considered separately instead of being replaced by the center frequency, thereby optimizing the beam offset effect and improving positioning accuracy.

[0051] 3. The low-overhead positioning method assisted by intelligent metasurface under the mixed-field beam deviation effect disclosed in the present invention, with the assistance of intelligent metasurface, uses the arrival time difference to solve the problem of inaccurate user positioning caused by common timing deviation, and at the same time locks the user on the hyperbola, reducing the computational complexity brought by the global search.

[0052] 4. The low-overhead positioning method assisted by intelligent metasurface under the hybrid field beam deviation effect disclosed in the present invention optimizes the angle at which the user reaches the base station through the polar coordinate domain gradient descent algorithm, and optimizes the problem of being unable to accurately estimate the angle when the distance estimation is not completely accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of the low-overhead positioning method assisted by intelligent metasurface under the hybrid field beam deviation effect disclosed in the present invention;

[0054] Figure 2 Schematic diagram of intelligent metasurface assisted positioning under the hybrid field beam deviation effect disclosed in this embodiment;

[0055] Figure 3 A schematic diagram comparing the angle estimation performance of this embodiment with three comparison schemes under near-field conditions, using the root mean square error as the evaluation metric;

[0056] Figure 4 A schematic diagram comparing the distance estimation performance of this embodiment and three comparison schemes using the root mean square error as the evaluation metric under near-field conditions;

[0057] Figure 5 A schematic diagram comparing the angle estimation performance of this embodiment with three comparison schemes under far-field conditions, using the root mean square error as the evaluation metric;

[0058] Figure 6 A schematic diagram comparing the distance estimation performance of this embodiment and three comparison schemes is shown in the far-field condition, using the root mean square error as the evaluation indicator. DETAILED DESCRIPTION

[0059] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. The technical problems solved by the technical solution of the present invention and the beneficial effects thereof are also described. It should be noted that the described embodiments are only intended to facilitate understanding of the present invention and do not serve to limit the present invention in any way.

[0060] Due to the complex occlusion relationship between outdoor buildings, the accuracy of satellite positioning is limited, and satellites cannot be used for high-precision positioning indoors. Therefore, the present invention is suitable for indoor and outdoor scenes in urban areas with high-precision user positioning requirements in future 6G communication networks. And because of the extremely large bandwidth and ultra-large-scale antenna array of the 6G communication network, it produces a mixed field beam offset effect that was not found in previous communication networks. In response to this scenario, the present invention provides a low-overhead positioning method assisted by an intelligent metasurface under the effect of a mixed field beam offset. The basic idea is to first obtain the delay between the user arriving at the base station and arriving at the intelligent metasurface with the assistance of the intelligent metasurface, and use the assistance of the delay information to narrow the user's area to the hyperbola, thereby reducing the computational complexity brought by the full-domain search. Secondly, the polar coordinate domain gradient descent algorithm is used to optimize the arrival angle of the user to the base station on the hyperbola, and then obtain the user's final position estimate. Since the present invention uses a polar coordinate domain dictionary, it can work both under far-field plane waves and near-field spherical waves. And because the beam offset effect is considered in the polar coordinate domain dictionary, different frequencies correspond to the same angle and distance, so multi-vector observations can be used to model the problem.

[0061] like Figure 1 As shown, this embodiment is used in a complex outdoor urban environment, where high-precision positioning of users is required to assist in communication. The base station serves users within a radius of 50m with the central array element of the base station antenna as the center, and the angle range of the service sector is -45° to 45°. Both the base station and the smart metasurface have a large number of array elements, which increases the Rayleigh distance in the terahertz band, and the communication scenario becomes a mixed field where near-field and far-field coexist. At the same time, the terahertz band has a large bandwidth, and different subcarriers can no longer share the same dictionary elements, that is, the steering vector during modeling will be related to the subcarrier. The specific parameters are: the base station antenna adopts a uniform linear array with the number of array elements N = 256; the smart metasurface adopts a uniform linear array with the number of array elements N RIS =256; Number of radio frequency links of the base station N RF =4; carrier frequency f c = 0.1THz; bandwidth B = 10GHz; the orientation of the base station antenna and the smart metasurface is The number of uplink pilot time slots is P NRIS =8,P RIS =16; in the near-field simulation, the position of the base station center, the center position of the smart metasurface, and the position of the user are set to (5.96m, -10.1m); in the far-field simulation, the position of the base station center, the center position of the smart metasurface, and the position of the user are set as (11.82m, -20.1m). Specifically, the spherical waveguide vector is expressed as follows

[0062]

[0063] where f m is the frequency of the mth subcarrier; is the distance from the user to the nth array element of the base station / intelligent metasurface; r is the distance from the user to the central array element of the base station / intelligent metasurface, which is sampled using a non-uniform inverse proportional function; δ n =(2×n-N+1) / 2, n=0,…,N-1 is the index representing the nth array element, and the center array element of the base station / smart metasurface array is the reference point; is the angle from user to base station / smart metasurface The sine value of is uniformly sampled between -1 and 1; d is the array element spacing of the base station / smart metasurface; is the wave number. Since the wavelength in the wave number is no longer simplified to the carrier wavelength, different subcarriers will correspond to the same angle and distance, eliminating the beam offset effect. After establishing the expression of the steering vector in the dictionary, the channel of the mth subcarrier in this patent can also be established as follows,

[0064]

[0065] Among them, there are L diameters in total; α l,m , b m,l are the channel gain and true steering vector on the lth path and the mth subcarrier respectively; θ l and r l are the angle and distance of the user to the center of the base station / intelligent metasurface on the lth path. Since the scenario of this patent is a mixed field scenario, when modeling the channel, the scatterer can no longer be regarded as a point source, but the area of ​​the scatterer must be considered. In this embodiment, the area of ​​the scatterer is 1m 2 The channel model is a multipath cluster sparse model. In this embodiment, the number of scatterers between the user and the base station and the number of scatterers between the user and the intelligent metasurface are both set to 5. Their positions are randomly generated in the area served by the base station using the rand function in MATLAB, and the average performance is obtained by multiple simulations. Based on the establishment of the channel model, the frequency domain received signal model of this paper can be derived.

[0066]

[0067]

[0068] in, and are the received signals of the mth subcarrier with the smart metasurface turned off and on respectively; and are the observation matrices when the smart metasurface is turned off and on, respectively. Since the channel between the base station and the smart metasurface is a broadband channel when the smart metasurface is turned on, the observation matrix Frequency related; and They are the channels from user to base station and smart metasurface respectively; and The Gaussian noise on the mth subcarrier when the smart metasurface is turned off and on, respectively. Since the pilot signal is known to both the transmitter and receiver, without loss of generality, the pilot signal is set to 1. Therefore, the transmitted pilot signal is hidden in the frequency domain received signal model.

[0069] The low-overhead positioning method assisted by an intelligent metasurface under the hybrid field beam deviation effect disclosed in this embodiment includes the following steps:

[0070] Step 1: Use the received signal in the frequency domain to estimate the delay of the user reaching the base station and the smart metasurface through the multiple signal classification algorithm.

[0071] By turning the smart metasurface on and off, the time of arrival (ToA) is estimated using the received signal in the frequency domain and the Multiple Signal Classification (MUSIC) algorithm. (For details about the MUSIC algorithm, see the reference "Multiple emitter location and signal parameter estimation," by R. Schmidt, "Multiple emitter location and signal parameter estimation," in IEEE Transactions on Antennas and Propagation, vol. 34, no. 3, pp. 276-280, March 1986.) Specifically, the uplink received signal in the frequency domain is obtained by turning the smart metasurface off. Turn on the smart metasurface to get the uplink receiving signal in the frequency domain Spectral decomposition using uplink received signals

[0072] (Y NRIS ) H Y NRIS =E NRIS Λ NRIS (E NRIS ) H , (5)

[0073] (Y RIS ) H Y RIS =E RIS Λ RIS (E RIS ) H , (6)

[0074] Among them, P NRIS and P RIS is the number of uplink pilot time slots when the smart metasurface is turned off and on; E NRIS and E RIS They are (Y NRIS ) H Y NRIS and (Y RIS ) H Y RIS The matrix composed of the eigenvectors of NRIS and Λ RIS They are (Y NRIS ) H Y NRIS and (Y RIS ) H Y RIS The diagonal matrix is ​​composed of the eigenvalues ​​of , where the eigenvalues ​​are arranged in descending order. Since the non-direct path loss in the terahertz band is relatively serious, we use the eigenvectors other than the eigenvector corresponding to the largest eigenvalue as the noise subspace For each subcarrier frequency generate To estimate the delay

[0075]

[0076]

[0077] in, and The delays for the direct path from the user directly to the base station and the delays for reaching the base station via the smart metasurface are respectively. The smart metasurface here only provides the degree of freedom for delay to assist in user positioning. Channel estimation via the smart metasurface is not involved. This means that only the delay information for the direct path from the user to the smart metasurface is estimated, without estimating the scatterer information (angle, delay, gain, etc.) contained in the indirect path. Therefore, this low-overhead positioning method assisted by the smart metasurface under the hybrid field beam offset effect can reduce pilot overhead compared to other positioning methods used in conjunction with channel estimation.

[0078] Step 2: The arrival time difference between the user and the base station and the smart metasurface is obtained by using the two delay information obtained in step 1, and a branch of the hyperbola where the user is located is obtained according to the positions of the base station and the smart metasurface.

[0079] The hyperbola where the user is located is obtained through the delay information. According to the two delays obtained, the delay difference τ between the user to the base station and the user to the smart metasurface is obtained. TDoA , and write the equation of the hyperbola

[0080]

[0081] The hyperbola takes the center of the array element of the base station and the smart metasurface as the focus. c is the speed of light; r B2R is the distance from the center of the base station antenna to the center of the smart metasurface; in this embodiment, the distance from the user to the base station is greater than the distance from the user to the smart metasurface, that is,

[0082] Step 3: Multiply the phase shift network by the partial dictionary generated on the hyperbola, and correlate the multiplication result with the received signal to obtain the initial user location formula.

[0083] Generate a small number of dictionaries on the hyperbola to roughly estimate the user's position. Generate a frequency-dependent dictionary in the polar coordinate domain on the hyperbola (For details on the polar coordinate domain dictionary, see the paper "Translation: Extremely Large-Scale Multiple Input Multiple Output Channel Estimation: Far-Field or Near-Field?" by M. Cui and L. Dai, "Channel Estimation for Extremely Large-Scale MIMO: Far-Field or Near-Field?" in IEEE Trans. Commu., vol. 70, no. 4, pp. 2663-2677, April 2022.) This means that the angle between the base station and the hyperbola is evenly divided, and a straight line from the base station will have a series of intersections with the hyperbola. N, Q, and M are the number of array elements in the base station antenna, the number of elements in the dictionary, and the number of subcarriers, respectively. Each element in the dictionary is composed of the angle and distance of each intersection point relative to the base station, which is the meaning of the "polar coordinate domain." Furthermore, due to terahertz ultra-wideband systems, the dictionaries for different subcarriers need to consider the specific subcarrier size, rather than uniformly using the center carrier frequency. This is the meaning of "frequency dependence."

[0084] The correlation between the received signal and the observation matrix after multiplying the polar coordinate domain dictionary on the hyperbola

[0085]

[0086] Get a rough estimate of the angle from the user to the base station Then, the user's position (x UE ,y UE )

[0087]

[0088] in, And y BSThe parameter settings are the same as those during simulation, all of which are 0.

[0089] Step 4: Fix the distance between the user and the base station and use the polar coordinate domain gradient descent algorithm to optimize the angle between the user and the base station;

[0090] The proposed polar coordinate domain gradient descent algorithm is applied to optimize the angle of arrival from the user to the base station. After obtaining a rough estimate of the user's position, the rough estimate is used as the initial condition and the polar coordinate domain gradient descent algorithm is used to optimize the angle of the direct path from the user to the base station. The loss function of the polar coordinate domain gradient descent algorithm is

[0091]

[0092] in, is the processed received signal, is the column vector corresponding to its m-th subcarrier; It's P NRIS The base station phase shift network composed of time slots is also the observation matrix; It is the virtual channel from user to base station. is the column vector corresponding to the mth subcarrier, and its expression is

[0093]

[0094] in, and are the estimated channel gain, angle, and distance from the user to the base station. The initial values ​​of distance and angle are the results from step 3, and the channel gain is calculated using the Friis transmission formula. The actual channel expression is

[0095]

[0096] Where L is the number of indirect paths. When l is 0, it is the direct path from the user to the base station. When l is other values, it is the indirect path from the user to the base station. and are the actual channel gain of the lth path, the angle from the user to the base station, and the distance from the user to the base station. The partial derivative of the proposed loss function with respect to the estimated angle from the user to the base station is

[0097]

[0098] Where Re{·} represents the real part; tr{·} represents the trace of the matrix. The estimated angle from user to base station The partial derivative of

[0099]

[0100] in, In the iterative process of gradient descent in polar coordinates, the step size η is updated using the Armijo-Goldstein criterion. Update current estimate This update process continues until The change before and after the update is less than the given threshold κ or the number of iterations reaches the set maximum number of iterations N it The processed received signal in the loss function is a variant of the original received signal

[0101]

[0102] in, The premise for this is that the base station's phase shift network needs to be specially designed as follows, that is, for

[0103]

[0104] and Each element from the second row to the last row of is randomly set and obeys Gaussian distribution. The purpose of this is to The influence of is removed in the received signal, and when generating the loss function, the virtual channel No This can be used to estimate the angle between the user and the base station. will not be affected by the inaccurate distance between the user and the base station. to optimize the angle based on an imperfect distance.

[0105] Step 5: Combine the polar coordinate domain gradient descent algorithm and the equation of the hyperbola to obtain an accurate estimate of the user's location. After obtaining the equation of the hyperbola and optimizing the direct path angle from the user to the base station using the polar coordinate domain gradient descent algorithm, the user's location is accurately estimated. Then, the location of the combined base station will be Substitute this into the previous intersection point formula to get the final estimated user location.

[0106] We tested the low-overhead positioning method assisted by intelligent metasurfaces under the mixed field beam deviation effect in the near field and far field respectively. The delay estimation of the comparison scheme and this embodiment uses the same method, but the angle estimation uses the MUSIC and ESPRIT methods. The simulation results are shown in Figure 2. Figures 3 to 6 As shown. Figure 3 It can be seen that although the embodiment adopts the hardware structure of hybrid beamforming, its angular positioning accuracy is improved by 1 to 2 orders of magnitude compared with the comparison solution using full digital beamforming. Figure 4It can be seen that the present embodiment has a higher angle accuracy, so even if the delay is estimated using the same method, its distance positioning accuracy is better than the comparison solution. Figure 5 It can be seen that the performance of this embodiment is also better than the comparison solution when the far-field transmission power is greater than 7dBm. This is because the polar coordinate domain gradient descent method in this embodiment will not work when the signal-to-noise ratio is too low. Figure 6 It can be seen that the distance positioning accuracy of this embodiment is improved by 3 times compared with the comparison solution in the far field. This is due to the fact that when the far field transmission power is greater than 7dBm, the angle positioning accuracy of this embodiment is improved by 2 orders of magnitude. Therefore, on the same hyperbola, it can bring gains to the distance positioning accuracy. Figure 3 and Figure 5 The comparison, and Figure 4 and Figure 6 From the comparison, it can be seen that the positioning accuracy of the embodiment is not only better than the comparison scheme in the near field, but also better than the comparison scheme when the far field transmission power is large.

[0107] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A low-overhead positioning method assisted by an intelligent metasurface under the hybrid field beam deviation effect, characterized by: The following steps are included: Step 1: Using the received signal in the frequency domain, the delay of the user reaching the base station and the smart metasurface is estimated through a multiple signal classification algorithm; Step 2: Using the two delay information obtained in step 1, the arrival time difference between the user and the base station and the smart metasurface is obtained, and based on the positions of the base station and the smart metasurface, a branch of the hyperbola where the user is located is obtained; Step 3: Multiply the phase shift network by the partial dictionary generated on the hyperbola, and correlate the multiplication result with the received signal to obtain the initial user location formula; The implementation method of step 3 is: Generate polar domain frequency-dependent dictionary on a hyperbola That is, the angle between the base station and the hyperbola is divided equally, and the straight line starting from the base station will have a series of intersections with the hyperbola; N, Q and M are the number of array elements of the base station antenna, the number of atoms in the dictionary, and the number of subcarriers respectively; each element in the dictionary is composed of the angle and distance of each intersection relative to the base station, which is the meaning of "polar coordinate domain"; because the system is terahertz ultra-wideband, the dictionary of different subcarriers needs to consider the specific subcarrier size, rather than replacing them all with the center carrier frequency, which is the meaning of "frequency dependence"; for frequency f m , the angle from the user to the center of the base station is θ, which is the real angle The sine value of the element in the dictionary whose distance to the center antenna of the base station is r is expressed as in, is the distance from the user to the nth antenna element of the base station, δ n =(2×n-N+1) / 2, where n=0,…,N-1 is the index of the base station antenna, and the center element is the zero point of the index; d is the element spacing; is the wave number of the mth subcarrier; The phase shift network is multiplied by the polar coordinate domain frequency-dependent dictionary on the hyperbola, and the multiplication result is correlated with the received signal to obtain a preliminary angle estimate from the user to the base station. where e(θ) is the index in the dictionary where θ is located; It's P NRIS A base station phase-shift network composed of time slots; is the uplink received signal in the frequency domain, P NRIS is the number of uplink pilot time slots of the closed intelligent super surface; the equation of the joint hyperbola and the position of the base station (x BS ,y BS )Get the user's location (x UE ,y UE ) in, y BS Set to 0; a and b are the real semi-axis length and imaginary semi-axis length of the hyperbola respectively; Step 4: Fix the distance between the user and the base station and use the polar coordinate domain gradient descent algorithm to calculate the direct path angle from the user to the base station. Optimize; Step 5: After obtaining the equation of the hyperbola and using the polar coordinate domain gradient descent algorithm to optimize the direct path angle from the user to the base station Then, combine the location of the base station and Substitute this into the user location formula described in step 3 to obtain the optimized user location, thereby improving positioning accuracy.

2. The low-overhead positioning method assisted by an intelligent metasurface under the hybrid field beam deviation effect according to claim 1, characterized in that: The implementation method of step 1 is: Turn off the smart metasurface to get the uplink receiving signal in the frequency domain Turn on the smart metasurface to get the uplink receiving signal in the frequency domain Spectral decomposition using uplink received signals (AND NRIS ) H AND NRIS =E NRIS L NRIS (AND NRIS ) H , (AND RIS ) H AND RIS =E RIS L RIS (AND RIS ) H , Among them, P RIS N is the number of uplink pilot time slots when the intelligent metasurface is turned off and on; RF is the number of RF links; M is the number of subcarriers; E NRIS and E RIS They are (Y NRIS ) H Y NRIS and (Y RIS ) H Y RIS The matrix composed of the eigenvectors of NRIS and Λ RIS They are (Y NRIS ) H Y NRIS and (Y RIS ) H Y RIS The diagonal matrix is ​​composed of the eigenvalues ​​of , where the eigenvalues ​​are arranged in descending order; due to the serious non-direct path loss in the terahertz frequency band, the eigenvectors other than the eigenvector corresponding to the largest eigenvalue are used as the noise subspace For each subcarrier frequency generate To estimate the delay in, and They are the direct path delay from the user directly to the base station and the delay from the user to the base station via the smart metasurface; the smart metasurface provides the freedom of delay to assist in locating the user, and does not involve channel estimation via the smart metasurface, that is, only the direct path delay information of the user reaching the smart metasurface is estimated, and the angle information in the direct path is not estimated, nor is the scatterer information contained in the non-direct path including angle, delay, and gain; therefore, the low-overhead positioning method assisted by the smart metasurface under the mixed-field beam offset effect can reduce the pilot overhead.

3. The low-overhead positioning method assisted by intelligent metasurface under the hybrid field beam deviation effect according to claim 1, characterized in that: The implementation method of step 2 is: According to the two delays obtained in step 1, the delay difference τ between the user to the base station and the user to the smart metasurface is obtained. TDoA ; The hyperbola equation is expressed as: The hyperbola focuses on base stations and smart metasurfaces; c is the speed of light; r B2R is the distance from the center of the base station antenna to the center of the smart metasurface; without loss of generality, If the distance from the user to the base station is greater than or equal to the distance from the user to the smart metasurface, then a hyperbola close to the smart metasurface is taken; if the distance from the user to the base station is less than the distance from the user to the smart metasurface, then a hyperbola close to the base station is taken.

4. The low-overhead positioning method assisted by intelligent metasurface under the hybrid field beam deviation effect as claimed in claim 1, characterized in that: The implementation method of step 4 is: After obtaining a rough estimate of the user's location, the polar coordinate domain gradient descent algorithm is used to optimize the angle of the direct path from the user to the base station; the loss function of the polar coordinate domain gradient descent algorithm is in, is the processed received signal, is the column vector corresponding to its m-th subcarrier; It's P NRIS A base station phase-shift network composed of time slots; It is the virtual channel from user to base station. is the column vector corresponding to the mth subcarrier, and its expression is in, and are the estimated channel gain, angle, and distance from the user to the base station respectively; the initial values ​​of distance and angle are the results in step 3, and the channel gain is calculated using the Friis transmission formula; the actual channel expression is Where L is the number of indirect paths. When l is 0, it is the direct path from the user to the base station. When l is other values, it is the indirect path from the user to the base station. and r l BU are the true channel gain of the lth path, the angle from the user to the base station, and the distance from the user to the base station; in the loss function of the polar coordinate domain gradient descent, the processed received signal is a variant of the original received signal in, The premise for this is that the base station's phase shift network needs to be specially designed as follows, that is, for and Each element from the second row to the last row is randomly set and obeys Gaussian distribution; the purpose of this is to The influence of is removed in the received signal, and when generating the loss function, the virtual channel No This can be used to estimate the angle between the user and the base station. will not be affected by the inaccurate distance between the user and the base station. to optimize the angle based on the imperfect distance; The partial derivative of the loss function of the gradient descent in the polar coordinate domain with respect to the estimated angle from the user to the base station is Among them, Re{·} represents the real part; tr{·} represents the trace of the matrix; virtual channel The estimated angle from user to base station The partial derivative of in, In the iterative process of gradient descent in the polar coordinate domain, the step size η is determined using the Armijo-Goldstein criterion; then Update current estimate This update process continues until The change before and after the update is less than the given threshold κ or the number of iterations reaches the set maximum number of iterations N it .

Images