A channel estimation interpolation method based on simplified MSL

Abstract

The present invention relates to the field of wireless communication technology, and in particular to a channel estimation interpolation method based on simplified MSL, comprising the following steps: segmenting the extracted channel response data, setting the original node coordinates and the original node number according to the length of the original pilot sequence; calculating the number of pilot sequence points after interpolation expected from the original pilot sequence; setting the expected node coordinates according to the expected number of pilot sequence points; setting the support radius of each expected node scale; constructing the shape function of the moving least squares MSL at each expected node based on a basis function and a weight function; and multiplying the shape function of each expected node with the original pilot data within the support radius to obtain the interpolated pilot sequence. The present invention can improve the interpolation accuracy of channel parameters and reduce channel response errors.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and more particularly to a channel estimation interpolation method based on simplified MSL. Background Art

[0002] The performance of wireless communication systems is mainly affected by the wireless channel. The wireless channel between the transmitter and the terminal is mainly affected by small-scale fading and multipath propagation. The amplitude and phase of the signal will be distorted during the transmission process, and it will also cause inter-code interference and time-selective fading, making it very difficult for the terminal to recover the transmitted signal.

[0003] Linear interpolation is the simplest and most intuitive interpolation algorithm. The core idea of ​​linear interpolation is to assume that the channel parameters of the data points between the two adjacent pilots are linearly changed. When the channel parameters of the two pilot points are calculated, the slope of the linear function can be calculated, and then the channel parameters of the data points in the middle can be calculated. The channel parameters obtained by the linear interpolation algorithm are interfered by the noise superimposed in the transmission channel, the accuracy of the estimation result is not high, and the error of the interpolation result is large.

[0004] Therefore, how to improve the interpolation accuracy of channel parameters is a problem that those skilled in the art need to solve urgently. Summary of the invention

[0005] In view of this, the present invention provides a channel estimation interpolation method based on simplified MSL, which can improve the interpolation accuracy of channel parameters and reduce channel response errors.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A channel estimation interpolation method based on simplified MSL, characterized by comprising the following steps:

[0008] Segment the extracted channel response data, and set the original node coordinates and the original node number according to the length of the original pilot sequence;

[0009] Insert a corresponding number of new points between every two adjacent original nodes, and calculate the number of pilot sequence points after interpolation expected for the entire original pilot sequence;

[0010] Setting the expected node coordinates according to the expected number of pilot sequence points;

[0011] Set the support radius for each desired node size;

[0012] The shape function of the moving least squares method MSL at each desired node is constructed based on the basis function and the weight function;

[0013] The shape function of each desired node is multiplied by the original pilot data within the support radius to obtain the interpolated pilot sequence.

[0014] Furthermore, the support radius d of each desired node size m The calculation method is:

[0015] d m =scale×c i

[0016] Among them, scale represents the scale radius, c i represents the scale factor, and i represents the index of the scale factor.

[0017] Furthermore, suppose the interpolation function u(x) to be solved has n nodes x in the solution domain Ω l The function value at u l =u(x l ) is known, l = 1, 2, 3, ..., n, then u(x) is locally approximated in the domain Ωx of the computation point x as:

[0018]

[0019] in, Represents the spatial coordinates of each point in the neighborhood Ωx of the calculation point x; basis function is an m-th order complete monomial, where m is the number of terms in the basis function; a i (x) is the coefficient to be determined containing the spatial coordinate x.

[0020] Furthermore, the solution formula for the coefficient a(x) is:

[0021] a(x)=A -1 (x)B(x)u

[0022] Among them, A -1 (x) represents the feature matrix, B(x) represents the channel characteristics, and u represents the coefficient vector;

[0023] Assume m = 3, then A(x) is a symmetric 3×3 matrix, and B(x) is a 3×n matrix, expressed as:

[0024]

[0025] Among them, w l (x) represents the weight function, x l represents the eigenvalue, y l Indicates the channel characteristic value.

[0026] Furthermore, the shape function of the moving least squares method is expressed as:

[0027] N T(x) = p T (x)A -1 (x)B(x).

[0028] Furthermore, the relationship between the channel response data and the pilot signal is:

[0029] Y P =X P H P +N

[0030] Among them, X P Indicates the pilot signal sent, Y P Indicates the received pilot signal, H P represents the frequency domain channel response at the pilot, and N represents the channel noise.

[0031] Furthermore, the least square method is used to estimate the channel of the pilot position. Assuming Y p (k) is the value of the kth received pilot in the frequency domain of any port-any OFDM symbol, X p (k) is the corresponding transmitted pilot value, then the channel H of the pilot position p (k) is expressed as

[0032]

[0033] Further, after obtaining the channel value of the pilot position, the method further includes:

[0034] Perform time domain interpolation. If there are two OFDM symbols and the residual frequency offset is controlled within 100 Hz and there is no phase rotation between the two symbols, then directly perform time domain copying.

[0035] It can be seen from the above technical solutions that, compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention first calculates the channel parameters of all pilot points using all inserted pilots, then regards the channel parameters of all pilot points as a two-dimensional plane, and then uses the moving least squares (MLS) method to interpolate and fit it to obtain the channel response of each data point, and uses the received data to restore the transmitted signal.

[0037] Moving least squares is an ideal method for establishing fitting curves for a large number of discrete data. When the channel environment is relatively harsh and the distribution of a large number of discrete data corresponding to the channel response is relatively messy, the traditional least squares method needs to be used to fit the data in segments. The data fitting is one-dimensional and the fitting effect is often not good. The MLS method uses a multi-dimensional processing method to deal with the same problem. While improving the error performance, it refers to the segmented fitting characteristics of the least squares method to reduce the complexity of the system and avoid the problem of discontinuous and non-smooth fitting curves on adjacent segments. The present invention implements high-precision interpolation fitting of the channel response by simplifying the MLS processing of the discrete channel response, effectively reducing the channel response error and improving the reliability of the channel response transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0039] Figure 1 Provide a flow chart of a channel estimation interpolation method based on simplified MSL for the present invention;

[0040] Figure 2 Schematic diagram of performance comparison between the method of the present invention and the linear interpolation method in Gaussian channel interpolation;

[0041] Figure 3 The figure is a schematic diagram showing the performance comparison between the method of the present invention and the linear interpolation method in ETU channel interpolation. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] like Figure 1 As shown, the embodiment of the present invention discloses a channel estimation interpolation method based on simplified MSL, comprising the following steps:

[0044] S1. Segment the extracted channel response data, and set the original node coordinates and the original node number according to the length of the original pilot sequence;

[0045] S2, inserting a corresponding number of new points between every two adjacent original nodes, and calculating the number of pilot sequence points after interpolation expected to be obtained for the entire original pilot sequence;

[0046] S3. Setting the expected node coordinates according to the expected number of pilot sequence points;

[0047] S4. Setting the support radius of each desired node size;

[0048] S5, constructing the shape function of the moving least squares method MSL at each desired node based on the basis function and the weight function;

[0049] S6. Multiply the shape function of each desired node by the original pilot data within the support radius to obtain an interpolated pilot sequence.

[0050] The above steps are further explained below.

[0051] S0. When the OFDM system inserts pilot information into the transmitted data, the relationship between the channel response data and the pilot signal is:

[0052] Y P =X P H P +N (1-1)

[0053] Among them, X P Indicates the pilot signal sent, Y P Indicates the received pilot signal, H P represents the frequency domain channel response at the pilot, and N represents the channel noise.

[0054] The least square method is used to estimate the channel of the pilot position. Assuming Y p (k) is the value of the kth received pilot in the frequency domain of any port-any OFDM symbol, X p (k) is the corresponding transmitted pilot value, then the channel H of the pilot position p (k) is expressed as

[0055]

[0056] After obtaining the channel value of the pilot position, time domain interpolation is performed first. If there are two OFDM symbols, the residual frequency offset is controlled within 100 Hz, and there is no phase rotation between the two symbols, time domain copying is performed directly. Knowing the pilot information of the first OFDM symbol, the pilot information of the second OFDM symbol can be obtained.

[0057] After time domain interpolation, frequency domain interpolation is required to further perform channel estimation on the entire time-frequency domain. The interpolation algorithm uses a simplified least moving squares method for interpolation. The specific interpolation process is shown in S1-S6 below.

[0058] S1. Segment the extracted channel response data according to the original pilot sequence N ori The length of the original node coordinates and the number of original nodes are set. The segmentation basis is to reduce the number of points for each operation as much as possible while ensuring that the bit error performance is not reduced, so as to reduce the complexity during FPGA implementation. Since the pilot sequence is distributed equally according to the frequency interval, the node coordinates can directly take the frequency coordinates of the pilot point.

[0059] S2. Calculate the number of pilot sequence points after interpolation expected from the original pilot sequence. Assume that the pilot points in the original pilot sequence are equally spaced, with a frequency interval of Δf. The expected number of interpolation points is determined based on the target frequency resolution. Insert k new points between each original pilot point. If the channel is complex and the expected number of pilot points needs to be increased, the expected factor (E∈(0,1)) needs to be taken. Then N exp =E×N ori .

[0060] S3. Set the expected node coordinates according to the expected number of pilot sequence points.

[0061] S4, set the support radius dm of each desired node size,

[0062] The support radius for each desired node size is calculated as:

[0063] d m =scale×c i (1-3)

[0064] Among them, scale represents the scale radius, c i represents the scale factor, and i represents the index of the scale factor.

[0065] The larger the scale radius, the better the signal filtering effect and the better the error performance, but it cannot be too large, otherwise it will drown the signal. The disadvantage is that the larger the scale, the greater the amount of calculation. i The recommended setting is 0.5.

[0066] S5, constructing the shape function of the moving least squares method MSL at each desired node based on the basis function and the weight function, specifically including:

[0067] S51, calculating the weight w, and different weight functions can be set according to different situations.

[0068] The distance r between the calculation point (i.e., the expected node) and the target node (i.e., the original node) is normalized to:

[0069] r=|d i | / d m(1-4)

[0070] Among them, d i is the actual distance from the i-th original node to the desired node to be calculated, d m is the support radius.

[0071] S51, suppose the interpolation function u(x) to be solved has n nodes x in the solution domain Ω (support function) l The function value at u l =u(x l ) is known (u(x) is the interpolation function to be solved, x1 is the calculation point, and u1 is the value of u(x) at x1), l = 1, 2, 3, ..., n, then u(x) is locally approximated in the domain Ωx of the calculation point x as:

[0072]

[0073] in, Represents the spatial coordinates of each point in the neighborhood Ωx of the calculation point x; basis function is an m-th order complete monomial, where m is the number of terms in the basis function; a i (x) is the coefficient to be determined containing the spatial coordinate x.

[0074] The basis functions for the one-dimensional case can be chosen to be

[0075] Linear basis:

[0076] Quadratic basis:

[0077] S52, coefficient a in formula (1-5) i (x) can be determined by the condition that the L2 norm J(x) takes a minimum value. The specific form of J(x) is:

[0078]

[0079] make We can get:

[0080] A(x)a(x)=B(x)u (1-9)

[0081] in,

[0082]

[0083] B(x)=[w1(x)p(x1)w2(x)p(x2)L w n (x)p(x n )] (1-11)

[0084] w l (x) = w(xl ) (1-12)

[0085] u={u1 u2 L u n} T (1-13)

[0086] S53. In a two-dimensional problem, if the number of monomials m = 3, then A is a symmetric 3×3 matrix and B is a 3×n matrix, which can be expressed as:

[0087]

[0088] Among them, w l (x) represents the weight function, x l represents the eigenvalue, y l Indicates the channel characteristic value.

[0089] From formula (1-9), the solution formula for the coefficient vector a(x) is:

[0090] a(x)=A -1 (x)B(x)u (1-16)

[0091] Among them, A -1 (x) represents the feature matrix, B(x) represents the channel characteristics, and u represents the coefficient vector;

[0092] The approximate function of MLS is expressed as:

[0093]

[0094] In formula (1-17), N(x) is used as the MLS shape function of n nodes in the support domain corresponding to point x. Substituting a(x) into formula (1-5), we can get

[0095] u h (x) = P T (x)A -1 (x)B(x)u (1-18)

[0096] Combining equation (1-17) and equation (1-18), we can get:

[0097] N T (x) = p T (x)A -1 (x)B(x) (1-19)

[0098] According to the communication signal requirements, it is recommended to set m to 1. Increasing m will increase the signal fitting accuracy and will also fit the noise in the signal, so m should not be too large.

[0099] S6. Multiply the shape function of each desired node by the original pilot data to obtain an interpolated pilot sequence within the support radius of each desired node.

[0100] In order to further verify the performance of the method of the present invention, the following experiments were conducted:

[0101] In order to reduce the complexity of engineering implementation, the MLS algorithm is processed in sections, such as Figure 2 As shown in the figure, by comparing the linear interpolation and MLS interpolation algorithms, we can see that the bit error rate is 10 -4 At the same time, the performance gain of the MLS interpolation optimization algorithm is 3.25dB.

[0102] In a multipath channel, Figure 3 As shown in the figure, MLS has a good performance gain at low signal-to-noise ratio, which can improve by about 2dB. As the signal-to-noise ratio increases, the error performance is basically the same as linear interpolation. Therefore, in complex environments, the MLS algorithm has a good error performance gain at low signal-to-noise ratio.

[0103] It can be verified that the MLS algorithm used in the present invention has higher fitting accuracy and robustness in channel estimation, and can provide strong support for wireless communication systems.

[0104] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0105] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A channel estimation interpolation method based on simplified MSL, characterized in that: The following steps are involved: Segment the extracted channel response data, and set the original node coordinates and the original node number according to the length of the original pilot sequence; Insert a corresponding number of new points between every two adjacent original nodes, and calculate the number of pilot sequence points after interpolation expected for the entire original pilot sequence; Setting the expected node coordinates according to the expected number of pilot sequence points; Set the support radius for each desired node size; The shape function of the moving least squares method MSL at each desired node is constructed based on the basis function and the weight function; The shape function of each desired node is multiplied by the original pilot data within the support radius to obtain the interpolated pilot sequence.

2. The channel estimation interpolation method based on simplified MSL according to claim 1, characterized in that: Support radius d for each desired node size m The calculation method is: d m =scale×c i Among them, scale represents the scale radius, c i represents the scale factor, and i represents the index of the scale factor.

3. The channel estimation interpolation method based on simplified MSL according to claim 1, characterized in that: Assume that the interpolation function u(x) to be solved has n nodes x in the solution domain Ω l The function value at u l =u(x l ) is known, l = 1, 2, 3, ..., n, then u(x) is locally approximated in the domain Ωx of the computation point x as: Among them, x = [x, y, z], which represents the spatial coordinates of each point in the neighborhood Ωx of the calculation point x; basis function is an m-th order complete monomial, where m is the number of terms in the basis function; a i (x) is the coefficient to be determined containing the spatial coordinate x.

4. The channel estimation interpolation method based on simplified MSL according to claim 3, characterized in that: The solution formula for the coefficient a(x) is: a(x)=A -1 (x)B(x)u Among them, A -1 (x) represents the feature matrix, B(x) represents the channel characteristics, and u represents the coefficient vector; Assume m = 3, then A(x) is a symmetric 3×3 matrix, and B(x) is a 3×n matrix, expressed as: Among them, w l (x) represents the weight function, x l represents the eigenvalue, y l Indicates the channel characteristic value.

5. The channel estimation interpolation method based on simplified MSL according to claim 4, characterized in that: The shape function of the moving least squares method is expressed as: N T (x)=p T (x)A -1 (x)B(x)。 6. The channel estimation interpolation method based on simplified MSL according to claim 1, characterized in that: The relationship between the channel response data and the pilot signal is: Y P =X P H P +N Among them, X P Indicates the pilot signal sent, Y P Indicates the received pilot signal, H P represents the frequency domain channel response at the pilot, and N represents the channel noise.

7. The channel estimation interpolation method based on simplified MSL according to claim 6, characterized in that: The least square method is used to estimate the channel of the pilot position. Assuming Y p (k) is the value of the kth received pilot in the frequency domain of any port-any OFDM symbol, X p (k) is the corresponding transmitted pilot value, then the channel H of the pilot position p (k) is expressed as 8. The channel estimation interpolation method based on simplified MSL according to claim 7, characterized in that: After obtaining the channel value of the pilot position, it also includes: Perform time domain interpolation. If there are two OFDM symbols and the residual frequency offset is controlled within 100 Hz and there is no phase rotation between the two symbols, then directly perform time domain copying.

Images