A positioning method based on delay doppler domain feature fingerprint and WDAKNN in a vehicle networking scene

Abstract

The application discloses a kind of based on time delay doppler domain feature fingerprint and WDAKN N positioning method under vehicle networking scene, comprising the following steps: input N, M, K, P RIS (x i ,y i )、P BS (x,y)、grid vertex coordinate P (x i ,y j );According to the relative position of BS, RIS and grid point coordinate, the transmission azimuth angle of OTFS wireless signal is calculatedAccording to the coordinate relationship, distance length d1d2 is calculated, so time delay information τ BS‑RIS ,τ RIS‑V ;Construction transmission data symbol x DD [k,l];From OTFS transmission characteristics, channel coefficient h i , channel doppler shift v i , channel delay τ i Are estimated using embedded pilot, construct channel feature matrix Θ;Get grid position Q (X i ,Y i ) corresponding total channel feature matrix Θ Ω ;Randomly generate to be positioned point (X v ,Y v ), repeat steps 1 to 4, get to be positioned point channel feature matrix Θ v ;The similarity of Θ Ω And Θ v It is compared using weighted distance and attribute WDAKNN algorithm, and the estimated position information Q (X, Y) is obtained by other neighbor weighted weight vector γ and weight function ω and average, the method realizes the goal of improving positioning accuracy.

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-everything (V2X) edge IoT technology, specifically to a positioning method based on time-delay Doppler domain fingerprint features and WDAKNN in V2X scenarios. Background Technology

[0002] In real-world scenarios, when vehicles are traveling in complex environments, the line-of-sight (LOS) communication links between vehicles may be blocked, which can significantly increase the ranging error between vehicles and ultimately lead to a decrease in the positioning accuracy of the vehicles.

[0003] Meanwhile, Reconfigurable Smart Surfaces (RIS), as an emerging technology, have attracted widespread attention. Their key advantage is the ability to intelligently regulate the electromagnetic wave propagation environment. Based on this characteristic, when the line-of-sight link between a base station and a target is severely blocked by obstacles, the RIS phase can be adjusted according to the actual situation to construct an effective reflection link, thereby solving the problem caused by the lack of a line-of-sight link. In vehicular network systems, RIS is introduced to solve non-line-of-sight (NLOS) problems and construct virtual line-of-sight (VLOS) links to restore positioning capabilities.

[0004] Compared to traditional OFDM signals, OTFS signals offer certain advantages in vehicle-to-everything (V2X) positioning. They are better suited to highly dynamic wireless environments, especially in situations with moving objects, multipath effects, and high-speed motion. Furthermore, OTFS exhibits relatively low sensitivity to delay and Doppler effects, making it generally more robust than traditional methods in complex multipath environments. The modulation mechanism of OTFS also results in unique stable characteristics in the delay-Doppler domain, clearly reflecting the propagation characteristics of the wireless channel and highly correlated with the vehicle's own location. Therefore, vehicle terminals can achieve distributed positioning on the device side without relying on other network equipment, using their own information. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a positioning method based on time-delay Doppler domain feature fingerprint and WDAKNN in vehicle-to-everything (V2X) scenarios, thereby achieving the goal of improving positioning accuracy.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0007] A localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario includes the following steps:

[0008] Step 1: Input N, M, K, P RIS (x i ,y i ), P BS (x,y), grid vertex coordinates P(x)i ,y j )

[0009] Step 2: Calculate the transmission azimuth angle of the OTFS wireless signal based on the relative positions of the BS, RIS, and grid point coordinates.

[0010] Step 3: Calculate the distance lengths d1 and d2 based on the coordinate relationship, and then calculate the time delay information τ. BS-RIS , τ RIS-V

[0011] Step 4: Construct the transmission data symbol x DD [k,l];

[0012] Step 5: Estimate the channel coefficient h using embedded pilot signals based on the OTFS transmission characteristics. i Channel Doppler frequency shift v i Channel delay τ i Construct the channel feature matrix Θ

[0013] Step 6: Obtain the grid position Q(X) i ,Y i The corresponding total channel feature matrix Θ Ω

[0014] Step 7: Randomly generate the point to be located (X) v ,Y v Repeat steps 1 to 4 to obtain the channel feature matrix Θ of the point to be located. v

[0015] Step 8: Compare Θ using weighted distance and attribute-based WDAKNN algorithms. Ω With Θ v Similarity is calculated by averaging the weighted vectors γ and ω from other neighbors.

[0016] Step 9: Output estimated location information Q(X,Y)

[0017] As a preferred option, step 1 is as follows:

[0018] Input N, M, K, P RIS (x i ,y i ), P BS (x,y), grid vertex coordinates P(x) i ,y j )

[0019] Where N represents the number of OTFS symbols, M represents the number of subcarriers, and K is the optimal value for the WDAKNN algorithm.

[0020] As a preferred option, step 2 is as follows:

[0021] In the OTFS-RIS vehicle-to-everything (V2X) system model, the locations of the base station (BS) and the RIS are known. At this point, the total delay (TDOA) and path-related azimuth angle information are available. and pitch angle It can be obtained

[0022]

[0023]

[0024] As a preferred option, step 3 is as follows:

[0025] Find the propagation time τ BS-RIS and τ RIS-V :

[0026]

[0027]

[0028]

[0029] The distance d1 between the base station (BS) and the RIS, and the distance d2 between the RIS and the grid center point, can be calculated using the formula d = tc. Based on the principle of triangulation, d1, d2, and azimuth angle information are used. The coordinates of all grid points can be calculated and solved.

[0030] As a preferred option, step 5 is as follows:

[0031] The channel can be obtained by sampling the original time-delay Doppler domain cascaded channel response:

[0032]

[0033] g q and For the time-delay Doppler domain response and delay shift of the q-th path in the effective link, φ T Indicates phase shift operation, This represents the turning vector of the RIS. Represents the Kronecker product:

[0034]

[0035] and These represent the horizontal array response and the vertical array response, respectively, where λ is the carrier wavelength and d is the antenna spacing that satisfies d = λ / 2.

[0036]

[0037]

[0038] The cascaded channel response is a two-dimensional convolution of the responses of two separate channels, representing the collection of all effective channel delay and effective Doppler shift pairs. δ(.) represents the Dirac function. Therefore, the effective channel can be simplified to:

[0039]

[0040] Channel estimation yields the cascaded channel coefficients h for each grid point. i Cascaded channel Doppler frequency shift v i and cascaded Doppler channel delay τ i Channel estimation results.

[0041]

[0042]

[0043]

[0044] The channel feature matrix is ​​concatenated along the third dimension to obtain the Θ = N×M×I feature matrix.

[0045] As a preferred option, step 6 is as follows:

[0046] Repeat step 5 to traverse all points within the grid area, obtaining the total location database as Θ. Ω A matrix, where Ω is the total number of grid points.

[0047] As a preferred option, step 8 is as follows:

[0048] The WDAKNN algorithm, which combines weighted distance and attribute size, is used for matching and localization. The weighted vector is denoted as γ, and d(γ) is the distance between any two online matching data and offline data.

[0049]

[0050] Then the distance d is updated to the weighted distance:

[0051]

[0052] in This indicates feature weighting, i.e.:

[0053]

[0054] A = (a1, a2, ... a i )

[0055] B = (b1, b2, ... b) i )

[0056] Therefore, the vehicle terminal location is obtained by averaging the weighted vector γ and weight function ω from other neighbors:

[0057]

[0058] The weight allocation is achieved through the weight function ω(d) i To achieve the transformation, and to overcome the problem of excessively large weights due to excessively small distances, a normal distribution is used. Therefore, the weight function is given by [formula missing] and satisfies:

[0059]

[0060] In the formula, Q represents the coordinates (X,Y) of the vehicle terminal to be located. i This represents the coordinates (X, X) of the sampling point corresponding to the i-th nearest neighbor data. i ,Y i ).

[0061] This invention has the following characteristics and beneficial effects:

[0062] 1. This invention provides a technical solution for vehicle-to-everything (V2X) positioning in a connected vehicle scenario based on time-delay Doppler domain feature fingerprinting and WDAKNN matching. It addresses the problem that in real-world scenarios where vehicles are traveling in complex environments, the line-of-sight (LOS) communication link between vehicles may be obstructed, leading to a significant increase in ranging errors and ultimately a decrease in vehicle positioning accuracy. This invention offers a novel solution.

[0063] 2. The present invention provides a technical solution for positioning in the vehicle-to-everything (V2X) scenario based on time-delay Doppler domain feature fingerprint and WDAKNN matching method. It proposes a DD-WDAKNN matching algorithm for effective positioning, analyzes the positioning accuracy under different grids and different K values, and finds that the optimal solution has significant advantages compared with traditional methods.

[0064] 3. By adopting the above technical solution, this invention applies OTFS modulation technology and the significant advantages of RIS to vehicle networking technology. It also proposes a technical solution for positioning based on time-delay Doppler domain feature fingerprint and WDAKNN matching method to address the problem of decreased positioning accuracy caused by missing links in vehicle networking scenarios. This achieves the goal of effectively improving positioning accuracy in vehicle networking scenarios. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a block diagram of the DD-WDAKNN localization algorithm in this invention;

[0067] Figure 2 This is the OTFS-RIS positioning system model for the vehicle-to-everything (V2X) scenario in this invention;

[0068] Figure 3 This is a schematic diagram of the channel characteristics under different Doppler frequency shifts and delays in this invention;

[0069] Figure 4 This is a graph showing the cumulative probability distribution of errors in the DD-WDAKNN localization algorithm of this invention. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0071] The present invention provides a positioning solution based on time-delay Doppler domain feature fingerprints and the WDAKNN matching method. This solution utilizes the characteristics of OTFS in the time-delay Doppler domain, which clearly reflects the propagation characteristics of the wireless channel and is highly correlated with its own location. It also constructs a VLOS with the assistance of RIS and finally solves the positioning problem using the WDAKNN matching algorithm, thereby effectively improving the positioning accuracy.

[0072] The technical solution in this embodiment is implemented as follows:

[0073] A localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario, such as Figure 1 As shown, it includes the following steps:

[0074] Step 1: Input N, M, K, P RIS (x i ,y i ), P BS (x,y), grid vertex coordinates P(x) i ,y jIn this embodiment, N represents the number of OTFS symbols, M represents the number of subcarriers, and K is the preferred value of the WDAKNN algorithm. In this embodiment, N = 128, M = 128, K = 6, and RIS is the same as the grid vertex coordinates (0,0), (0,30), (30,30), (30,0). The base station position is (15,45).

[0075] Step 2: Calculate the transmission azimuth angle of the OTFS wireless signal based on the relative positions of the BS, RIS, and grid point coordinates.

[0076] like Figure 2 As shown, in the OTFS-RIS vehicle-to-everything (V2X) system model, the locations of the base station (BS) and the RIS are known. At this point, the total delay (TDOA) and path-related azimuth angle information are... and pitch angle It can be obtained

[0077]

[0078]

[0079] Step 3: Calculate the distance lengths d1 and d2 based on the coordinate relationship, and then calculate the time delay information τ. BS-RIS , τ RIS-V .

[0080] Find the propagation time τ BS-RIS and τ RIS-V :

[0081]

[0082]

[0083]

[0084] The distance d1 between the base station (BS) and the RIS, and the distance d2 between the RIS and the grid center point, can be calculated using the formula d = tc. Based on the principle of triangulation, d1, d2, and azimuth angle information are used. This allows for the calculation and solution of the coordinate positions of all grid points. BS This refers to the coordinate information of the base station (BS), p RIS This refers to the location information of the RIS, where c is the speed of light, and the first formula calculates the time delay information between the base station and the RIS.

[0085] Step 4: Construct the transmission data symbol x DD [k,l].

[0086] It should be noted that, according to the OTFS transmission principle, the data symbols are randomly generated by the transmitter. k represents the coordinates of the data symbol along the Doppler frequency shift direction in the DD domain, and l represents the coordinates of the data symbol along the time delay direction in the DD domain.

[0087] Step 5: Estimate the channel coefficient h using embedded pilot signals based on the OTFS transmission characteristics. i Channel Doppler frequency shift v i Channel delay τ i Construct the channel feature matrix Θ.

[0088] The channel can be obtained by sampling the original time-delay Doppler domain cascaded channel response:

[0089]

[0090] g q and For the time-delay Doppler domain response and delay shift of the q-th path in the effective link, φ T Indicates phase shift operation, This represents the turning vector of the RIS. Represents the Kronecker product:

[0091]

[0092] and These represent the horizontal array response and the vertical array response, respectively, where λ is the carrier wavelength and d is the antenna spacing that satisfies d = λ / 2.

[0093]

[0094]

[0095] Where λ is the carrier wavelength, d is the antenna spacing satisfying d = λ / 2, and RIS consists of M in total. R It consists of several passive reflective elements, arranged in a uniform rectangular array of M1×M2, M R =M1×M2. And:

[0096] m1=[0,1,…,M1-1],m2=[0,1,…,M2-1]

[0097] like Figure 3 As shown, the cascaded channel response is a two-dimensional convolution of the responses of two separate channels, representing the collection of all effective channel delay and effective Doppler frequency shift pairs. δ(.) represents the Dirac function. Therefore, the effective channel can be simplified to:

[0098]

[0099] Channel estimation yields the cascaded channel coefficients h for each grid point. i Cascaded channel Doppler frequency shift v i and cascaded Doppler channel delay τ i Channel estimation results.

[0100]

[0101]

[0102]

[0103] The channel feature matrix is ​​concatenated along the third dimension to obtain the Θ = N×M×I feature matrix.

[0104] Where I refers to the number of features. When there are I RIS in the system, each signal reflected by the RIS will have a corresponding feature matrix in the DD domain. The feature matrix is ​​of size M*N, and there are I in total. These I feature matrices are concatenated along the third dimension to finally obtain the feature matrix Θ = N×M×I.

[0105] Step 6: Obtain the grid position Q(X) i ,Y i The corresponding total channel characteristic matrix Θ Ω Specifically, by repeating step 5 to traverse all points within the grid area, the total location database is obtained as Θ. Ω A matrix, where Ω is the total number of grid points.

[0106] Step 7: Randomly generate the point to be located (X) v ,Y v Repeat steps 1 to 4 to obtain the channel feature matrix Θ of the point to be located. v .

[0107] Step 8: Compare Θ using weighted distance and attribute-based WDAKNN algorithms. Ω With Θ v Similarity is derived by averaging the weighted vectors γ and weight function ω of other neighbors.

[0108] The WDAKNN algorithm, which combines weighted distance and attribute size, is used for matching and localization. The weighted vector is denoted as γ, and d(γ) is the distance between any two online matching data and offline data.

[0109]

[0110] Then the distance d is updated to the weighted distance:

[0111]

[0112] in This indicates feature weighting, i.e.:

[0113]

[0114] A = (a1, a2, ... a i )

[0115] B = (b1, b2, ... b) i )

[0116] Therefore, the vehicle terminal location is obtained by averaging the weighted vector γ and weight function ω from other neighbors:

[0117]

[0118] The weight allocation is achieved through the weight function ω(d) i To achieve the transformation, and to overcome the problem of excessively large weights due to excessively small distances, a normal distribution is used. Therefore, the weight function is given by:

[0119]

[0120] In the formula, Q represents the coordinates (X,Y) of the vehicle terminal to be located. i This represents the coordinates (X, X) of the sampling point corresponding to the i-th nearest neighbor data. i ,Y i ).

[0121] pass Figure 4 It can be seen that the localization method using OTFS channel estimation features is significantly improved compared to the traditional TDOA localization method. The proposed DD-WDAKNN algorithm achieves a localization error probability distribution of over 90% within 2m, which is about 20% higher than the TDOA-Chan and TDOA-Taylor algorithms.

[0122] Furthermore, when using OTFS channel estimation features for positioning, the classic KNN and WKNN algorithms were also compared. Simulations were conducted under the same condition K=6. Compared with the DD-KNN and DD-WKNN algorithms, the error within 1m was improved by about 10%, and 90% of the positioning errors were within 2m. Moreover, it showed good performance in terms of average error, minimum error, and maximum error, proving that this method has good positioning performance.

[0123] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions, and variations can be made to these embodiments, including components, without departing from the principles and spirit of the present invention, and these variations still fall within the protection scope of the present invention.

Claims

1. A localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario, characterized in that, Includes the following steps: Step 1: Obtain , , , , Grid vertex coordinates , among which, among which Represents the number of OTFS symbols. To represent the number of subcarriers, The optimal value for the WDAKNN algorithm is... , These represent the two-dimensional plane coordinates of the RIS reflection module and the base station, respectively. Step 2: Under the OTFS-RIS vehicle-to-everything (V2X) system model, based on the known locations of the base stations (BS) and RIS, obtain the relative positions of the BS, RIS, and grid point coordinates, and calculate the transmission azimuth angle of the OTFS wireless signal. and pitch angle ; Step 3: Distance between base station (BS) and RIS and the distance between the RIS and the grid center point Calculate latency information , ; Step 4: Construct transmission data symbols ; Step 5: Estimate channel coefficients using embedded pilots Channel Doppler frequency shift Channel delay Construct the channel feature matrix ; Step 6: Repeat step 5 to traverse all points within the grid area and obtain the grid position. Corresponding total channel feature matrix ,in This is the total number of grid points; Step 7: Randomly generate the points to be located Repeat steps 1 to 5 to obtain the channel feature matrix of the point to be located. ; Step 8: Compare using weighted distance and attribute-based WDAKNN algorithms and Similarity is calculated by weighting other neighbors, resulting in a weighted vector. and weight function And average to obtain estimated location information ; The specific method for step 8 is as follows: The WDAKNN algorithm, which combines weighted distance and attribute magnitude, is used for matching and localization. The weighted vector is denoted as... , The distance between any two online matching data and offline data: Then distance Updated to weighted distance: in This indicates feature weighting, i.e.: Therefore, the vehicle terminal location is weighted by its other neighbors, and the weighting vector is... and weight function And by averaging: ； The weight allocation is performed through a weighting function. The transformed result is obtained using a normal distribution, with a weight function that satisfies: In the formula, This represents the coordinates of the vehicle terminal to be located. , It indicates the first Coordinates of sampling points corresponding to each nearest neighbor data point .

2. The localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario according to claim 1, characterized in that, The calculation of the transmission azimuth angle of the OTFS wireless signal and pitch angle The expression is as follows: in, It refers to The ordinate of the location coordinates, This refers to the ordinate of the grid points. for The x-coordinate of the location. This refers to the fact that the x-coordinate Z of the grid point is the third dimension coordinate.

3. The localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario according to claim 1, characterized in that, The delay information and The calculation method is as follows: in, For the speed of light, the total time delay It can be obtained from base station equipment. Indicates the first generation built via RIS Path-related delay coefficients Indicates the number of subcarriers. This represents the carrier spacing, the distance between the base station (BS) and the RIS. and the distance between the RIS and the grid center point It can be derived from the formula Calculations show that, based on the principle of triangle positioning, by and and azimuth angle information This allows us to calculate and solve for the coordinates of all grid points.

4. The vehicle-to-everything (V2X) localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario according to claim 3, characterized in that, The specific method for step 5 is as follows: Step 5.1: The channel samples the original time-delay Doppler domain cascaded channel response; Step 5.2: The cascaded channel response is a two-dimensional convolution of the responses of the two separate channels. This represents the collection of all effective channel delay and effective Doppler frequency shift pairs. The effective channel is obtained through the Dirac function, as shown in the following expression: in, Represents the Dirac function; Step 5.3: Channel estimation yields the cascaded channel coefficients for each grid point. Cascaded channel Doppler frequency shift and cascaded Doppler channel delay The channel estimation result is expressed as follows: Step 5.4: Concatenate the channel feature matrix along the third dimension to obtain... The feature matrix, where I refers to the eigenvector.

5. The vehicle-to-everything (V2X) localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario according to claim 4, characterized in that, The sampling method in step 5.1 is as follows: and For the first in the effective link The time-delay Doppler domain response and delay displacement of the path, Indicates phase shift operation, This represents the turning vector of the RIS. Represents the Kronecker product: and These are represented as the horizontal array response and the vertical array response, respectively. For carrier wavelength, To meet Antenna spacing.

6. The vehicle-to-everything (V2X) localization method based on time-delay Doppler domain feature fingerprint and WDAKNN in a vehicle-to-everything (V2X) scenario according to claim 5, characterized in that, The horizontal array response With vertical array response The method for obtaining it is as follows: in Where is the carrier wavelength, and j is the imaginary unit. To meet The antenna spacing, the total RIS consists of Composed of a passive reflective element, organized in a A uniform rectangular array configuration .

Images