阅读说明：本技术 Nr中基于trs的pdp估计方法及系统 (TRS-based PDP estimation method and system in NR ) 是由 王志旭 于 2021-01-20 设计创作，主要内容包括：本发明提供了一种NR中基于TRS的PDP估计方法及系统,涉及功率时延谱估计技术领域,该方法包括：步骤1：根据系统指示,获取频域信道估计值；步骤2：得到频域信道估计值后,再获取观测矩阵Π；步骤3：用N维DFT矩阵作为稀疏矩阵,通过压缩感知算法得到具体的多径位置和多径功率,从而得到准确的PDP估计；步骤4：通过步骤2所选取观测矩阵Π之后,其余的导频可以再次通过随机生成的方法得到观测矩阵Π,再一次重复步骤2和步骤3,得到PDP估计,并对两次估计结果平均,得到更好的估计结果；最终步骤4选取观测矩阵Π,来实现PDP估计次数视性能要求而定。本发明能够提高PDP估计的精确性,且具有更强的鲁棒性,降低了复杂度。(The invention provides a TRS-based PDP estimation method and system in NR, relating to the technical field of power delay spectrum estimation, wherein the method comprises the following steps: step 1: acquiring a frequency domain channel estimation value according to system instructions; step 2: obtaining an observation matrix pi after obtaining a frequency domain channel estimation value; and step 3: an N-dimensional DFT matrix is used as a sparse matrix, and specific multipath positions and multipath powers are obtained through a compressed sensing algorithm, so that accurate PDP estimation is obtained; and 4, step 4: after the observation matrix Π is selected in the step 2, the rest pilot frequencies can obtain the observation matrix Π again through a random generation method, the step 2 and the step 3 are repeated again to obtain PDP estimation, and the estimation results of the two times are averaged to obtain a better estimation result; and finally, selecting an observation matrix pi in the step 4 to realize that the PDP estimation times are determined according to the performance requirement. The invention can improve the accuracy of PDP estimation, has stronger robustness and reduces the complexity.)

1. A TRS based PDP estimation method in NR, characterized by comprising:

step 1: acquiring a frequency domain channel estimation value according to system instructions;

step 2: obtaining an observation matrix pi after obtaining a frequency domain channel estimation value;

and step 3: an N-dimensional DFT matrix is used as a sparse matrix, and specific multipath positions and multipath powers are obtained through a compressed sensing algorithm, so that accurate PDP estimation is obtained;

and 4, step 4: after the observation matrix Π is selected in the step 2, the rest pilot frequencies can obtain the observation matrix Π again through a random generation method, the step 2 and the step 3 are repeated again to obtain PDP estimation, and the estimation results of the two times are averaged to obtain a better estimation result; and finally, selecting an observation matrix pi in the step 4 to realize that the PDP estimation times are determined according to the performance requirement.

2. The method of claim 1, wherein step 1 comprises:

step 1-1: according to the system indication, a local pilot frequency sequence is generated and correlated with the received pilot frequency sequence to obtain a frequency domain channel estimation H_ls；

Step 1-2: since the initial pilot location may be at subcarriers {0,1,2,3}, the channel estimate needs to be compensated for as follows:

H_ls＝H_lsexp { j2 pi i/N }, where i is the subcarrier offset number.

3. The method of claim 1, wherein the step 2 comprises:

in the NR system, the interval of each subcarrier is 30kHz, the total subcarrier number is 3276, each 12 subcarriers are called 1 rb, each rb has three trs pilot frequencies, and then the total pilot frequency is 819; the compressed sensing algorithm needs observation number M version 4-6 times of sparsity K, channel multipath number is small, M is selected to be 50, and the position is random and generated in advance.

4. The method of claim 1, wherein step 3 comprises:

step 3-1: the generated sparse matrix is a 1024-point DFT matrix, and the sensing matrix is a matrix generated by M rows with corresponding positions different from 0;

step 3-2: selecting a proper compressed sensing algorithm, using a gOMP algorithm in order to save the computational complexity, and enabling the selected atomic number to be S each time;

step 3-3: and the position information obtained by estimation is the multipath position, and the multipath power value is obtained by taking the absolute value of the estimated amplitude value.

5. The method of claim 1, wherein the step 4 comprises:

if the number of times of selecting the observation matrix is 1, skipping the step, otherwise, after excluding the M positions selected in the step 2, selecting M positions again from the rest positions, and performing the step 2 and the step 3 to obtain another group of estimated values theta₁To get a rightAnd all the estimated values are averaged, so that the estimation accuracy is further improved.

6. A TRS based PDP estimation system in NR, comprising:

module M1: acquiring a frequency domain channel estimation value according to system instructions;

module M2: obtaining an observation matrix pi after obtaining a frequency domain channel estimation value;

module M3: an N-dimensional DFT matrix is used as a sparse matrix, and specific multipath positions and multipath powers are obtained through a compressed sensing algorithm, so that accurate PDP estimation is obtained;

module M4: after the observation matrix Π is selected by the module M2, the rest of the pilot frequencies may obtain the observation matrix Π again by a random generation method, the module M2 and the module M3 are repeated again to obtain PDP estimation, and the estimation results of the two times are averaged to obtain a better estimation result; the final module M4 selects the observation matrix Π to achieve PDP estimation times depending on performance requirements.

7. The system according to claim 6, characterized in that said module M1 comprises:

according to the system indication, a local pilot frequency sequence is generated and correlated with the received pilot frequency sequence to obtain a frequency domain channel estimation H_ls；

Since the initial pilot location may be at subcarriers {0,1,2,3}, the channel estimate needs to be compensated for as follows:

H_ls＝H_lsexp { j2 pi i/N }, where i is the subcarrier offset number.

8. The system according to claim 6, characterized in that said module M2 comprises:

in the NR system, the interval of each subcarrier is 30kHz, the total subcarrier number is 3276, each 12 subcarriers are called 1 rb, each rb has three trs pilot frequencies, and then the total pilot frequency is 819; the compressed sensing algorithm needs observation number M version 4-6 times of sparsity K, channel multipath number is small, M is selected to be 50, and the position is random and generated in advance.

9. The system according to claim 6, characterized in that said module M3 comprises:

the generated sparse matrix is a 1024-point DFT matrix, and the sensing matrix is a matrix generated by M rows with corresponding positions different from 0;

selecting a proper compressed sensing algorithm, using a gOMP algorithm in order to save the computational complexity, and enabling the selected atomic number to be S each time;

and the position information obtained by estimation is the multipath position, and the multipath power value is obtained by taking the absolute value of the estimated amplitude value.

10. The system according to claim 6, characterized in that said module M4 comprises:

if the number of times of selecting the observation matrix is 1, skipping the module, otherwise, after excluding the M positions selected by the module M2, selecting M positions again from the rest positions, and performing the module M2 and the module M3 to obtain another set of estimated values theta₁And averaging all the obtained estimated values, and further improving the estimation accuracy.

Technical Field

The invention relates to the technical field of power delay spectrum estimation, in particular to a TRS-based PDP estimation method and system in NR.

Background

In Orthogonal Frequency Division Multiplexing (OFDM) -based NR downlink communications, estimation of channel state information is crucial.

Based on the sparse characteristic of the channel, the TRS signal distributed in the full bandwidth in the NR system is utilized, and the accurate channel power delay spectrum can be estimated by utilizing a compressed sensing algorithm, so that frequency domain correlation information is obtained through Fourier transform and is used for frequency domain filtering and frequency domain channel estimation, and the performance is excellent.

The relationship between PDP and frequency domain correlation is briefly described below.

PDP correlation with frequency domain:

according to the 3GPP TS 38.101-4 protocol, the NR channel is a Tapped Delay Line (TDL) model in the time domain, with the multipath power falling on the corresponding sample points. May be based on TR 38.901 for different sampling rates

The power adjustment is performed in the manner shown in (a) so that the multipath power falls on the sample point. Without loss of generality we express the channel PDP as:

wherein, N is the total sampling point number, theta is the multipath position set, the size is L, which means that only L non-0 values exist, and according to the 3GPP TS 38.101-4 protocol, 12 non-0 values exist at the sampling rate of 200M.

Then the frequency domain correlation is expressed as:

in practical applications, since the PDP cannot be directly obtained by observation, frequency domain filtering and estimation are often performed in a manner of presetting a power spectrum. When the preset spectrum is a rectangular spectrum, the estimation steps are as follows:

obtaining a time domain value by IFFT conversion of the frequency domain channel coarse estimation value;

taking an absolute value of the time domain value, and then searching to obtain the length of the rectangular window through a preset threshold, snr and the like;

and performing FFT (fast Fourier transform) on the obtained rectangular window to obtain frequency domain correlation.

The disadvantages seen by the above procedure are mainly: 1. the robustness is not strong due to the fact that empirical values such as a preset threshold and the like exist; 2. window length search complexity is high; 3. the performance is less good in this case, which is greatly affected by non-ideal factors.

Disclosure of Invention

Aiming at the defects in the prior art, the invention aims to provide a TRS-based PDP estimation method and system in NR, which can improve the accuracy of PDP estimation, have stronger robustness and reduce the complexity.

According to the TRS-based PDP estimation method and system in NR provided by the invention, the scheme is as follows:

in a first aspect, a TRS-based PDP estimation method in NR is provided, the method including:

step 1: acquiring a frequency domain channel estimation value according to system instructions;

step 2: obtaining an observation matrix pi after obtaining a frequency domain channel estimation value;

and step 3: an N-dimensional DFT matrix is used as a sparse matrix, and specific multipath positions and multipath powers are obtained through a compressed sensing algorithm, so that accurate PDP estimation is obtained;

and 4, step 4: after the observation matrix Π is selected in the step 2, the rest pilot frequencies can obtain the observation matrix Π again through a random generation method, the step 2 and the step 3 are repeated again to obtain PDP estimation, and the estimation results of the two times are averaged to obtain a better estimation result; and finally, selecting an observation matrix pi in the step 4 to realize that the PDP estimation times are determined according to the performance requirement.

Preferably, the step 1 comprises:

according to the system indication, a local pilot frequency sequence is generated and correlated with the received pilot frequency sequence to obtain a frequency domain channel estimation H_ls；

Since the initial pilot location may be at subcarriers {0,1,2,3}, the channel estimate needs to be compensated for as follows:

H_ls＝H_lsexp { j2 pi i/N }, where i is the subcarrier offset number.

Preferably, the step 2 comprises:

in the NR system, the interval of each subcarrier is 30kHz, the total subcarrier number is 3276, each 12 subcarriers are called 1 rb, each rb has three trs pilot frequencies, and then the total pilot frequency is 819; the compressed sensing algorithm needs observation number M version 4-6 times of sparsity K, channel multipath number is small, M is selected to be 50, and the position is random and generated in advance.

Preferably, the step 3 comprises:

the generated sparse matrix is a 1024-point DFT matrix, and the sensing matrix is a matrix generated by M rows with corresponding positions different from 0;

selecting a proper compressed sensing algorithm, using a gOMP algorithm in order to save the computational complexity, and enabling the selected atomic number to be S each time;

and the position information obtained by estimation is the multipath position, and the multipath power value is obtained by taking the absolute value of the estimated amplitude value.

Preferably, the step 4 comprises:

if the number of times of selecting the observation matrix is 1, skipping the step, otherwise, after excluding the M positions selected in the step 2, selecting M positions again from the rest positions, and performing the step 2 and the step 3 to obtain another group of estimated values theta₁And averaging all the obtained estimated values, and further improving the estimation accuracy.

In a second aspect, there is provided a TRS-based PDP estimation system in NR, the system comprising:

module M1: acquiring a frequency domain channel estimation value according to system instructions;

module M2: obtaining an observation matrix pi after obtaining a frequency domain channel estimation value;

module M3: an N-dimensional DFT matrix is used as a sparse matrix, and specific multipath positions and multipath powers are obtained through a compressed sensing algorithm, so that accurate PDP estimation is obtained;

module M4: after the observation matrix Π is selected by the module M2, the rest of the pilot frequencies may obtain the observation matrix Π again by a random generation method, the module M2 and the module M3 are repeated again to obtain PDP estimation, and the estimation results of the two times are averaged to obtain a better estimation result; the final module M4 selects the observation matrix Π to achieve PDP estimation times depending on performance requirements.

Preferably, the module M1 includes:

according to the system indication, a local pilot frequency sequence is generated and correlated with the received pilot frequency sequence to obtain a frequency domain channel estimation H_ls；

Since the initial pilot location may be at subcarriers {0,1,2,3}, the channel estimate needs to be compensated for as follows:

H_ls＝H_lsexp { j2 pi i/N }, where i is the subcarrier offset number.

Preferably, the module M2 includes:

in the NR system, the interval of each subcarrier is 30kHz, the total subcarrier number is 3276, each 12 subcarriers are called 1 rb, each rb has three trs pilot frequencies, and then the total pilot frequency is 819; the compressed sensing algorithm needs observation number M version 4-6 times of sparsity K, channel multipath number is small, M is selected to be 50, and the position is random and generated in advance.

Preferably, the module M3 includes:

the generated sparse matrix is a 1024-point DFT matrix, and the sensing matrix is a matrix generated by M rows with corresponding positions different from 0;

selecting a proper compressed sensing algorithm, using a gOMP algorithm in order to save the computational complexity, and enabling the selected atomic number to be S each time;

and the position information obtained by estimation is the multipath position, and the multipath power value is obtained by taking the absolute value of the estimated amplitude value.

Preferably, the module M4 includes:

if the number of times of selecting the observation matrix is 1, skipping the module, otherwise, after excluding the M positions selected by the module M2, selecting M positions again from the rest positions, and performing the module M2 and the module M3 to obtain another set of estimated values theta₁And averaging all the obtained estimated values, and further improving the estimation accuracy.

Compared with the prior art, the invention has the following beneficial effects:

1. the invention can improve the accuracy of PDP estimation;

2. the robustness of the PDP estimation is enhanced;

3. reducing the window length search complexity.

Drawings

Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:

FIG. 1 is a flow chart of a method of the present invention;

FIG. 2 is a diagram illustrating an embodiment of the present invention.

Detailed Description

The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that it would be obvious to those skilled in the art that various changes and modifications can be made without departing from the spirit of the invention. All falling within the scope of the present invention.

The embodiment of the invention provides a TRS-based PDP estimation method in NR, the existing defects can be overcome by using the PDP estimation of a compressed sensing algorithm, and compressed sensing is as follows:

the principle of compressed sensing is that a signal x with Nx1 dimensions exists, and an observation vector y with Mx1 dimensions, M < N, satisfies the following conditions:

y＝Πx

wherein the observation matrix Π dimension is mxn. x is non-sparse, but sparse in the transform domain Φ, which, in combination with the above equation, can be expressed as:

y＝ΠΦθ＝Aθ

where the measurement matrix Φ dimension is N × K and θ dimension is K × 1, then the sensing matrix a dimension is M × K.

For y ═ Π x, x is an indefinite equation solved by y, an unknown number is larger than the number of equations, a unique solution cannot be obtained, and an accurate solution can be obtained by finding a proper sparse matrix to obtain y ═ a θ.

Referring to fig. 1 and 2, first, step 1: according to the system instruction, obtaining a frequency domain channel estimation value, specifically as follows:

in the process of estimating PDP by NR downlink TRS, a local pilot frequency sequence is generated according to the system indication and is connected withThe received pilot frequency sequence is correlated to obtain frequency domain channel estimation H_lsSince the initial pilot position may be at subcarrier {0,1,2,3}, channel estimation needs to be compensated, specifically:

H_ls＝H_lsexp { j2 pi i/N }, where i is the subcarrier offset number.

Secondly, step 2: after obtaining the frequency domain channel estimation value, obtaining an observation matrix pi: in the NR system, the interval of each subcarrier is 30kHz, the total subcarrier number is 3276, each 12 subcarriers are called 1 rb, each rb has three trs pilot frequencies, and then the total pilot frequency is 819; the compressed sensing algorithm needs M versions of observation numbers which are 4 to 6 times of sparsity K, the number of channel multipaths is small, M is selected to be 50, the positions of the M are random, and the M can be generated in advance according to re positions to generate an example of {7, 59, 62, 76, 80, 89, 103, 107, 109, 115, 121, 126, 132, 135, 136, 147, 158, 169, 175, 202, 203, 227, 248, 255, 256, 272, 297, 312, 314, 315, 346, 409, 432, 442, 444, 484, 492, 513, 530, 536, 561, 562, 535, 589, 675, 715, 726, 768, 779 }.

Thirdly, step 3: and (3) using the N-dimensional DFT matrix as a sparse matrix, and obtaining specific multipath positions and multipath powers through a compressed sensing algorithm so as to obtain accurate PDP estimation.

The generated sparse matrix is a 1024-point DFT matrix, the sensing matrix is a matrix generated by M rows with corresponding positions not being 0, a proper compressed sensing algorithm is selected, in order to save calculation complexity, a gOMP algorithm is used, the atomic number S selected each time is 3, and the implementation mode of the gOMP is briefly described as follows:

1. initialization: multipath location aggregationCorresponding sparse vector setObserved value r₀＝y；

2. Calculating the inner product of each column of the sensing matrix and the observation matrix, wherein epsilon is less than A_iY > 1. ltoreq. i.ltoreq.N, selectingThe maximum S value;

3. adding the corresponding position and vector to the sets Γ and Ψ;

4. obtaining the estimated value of this time

5. Updating residual errors

6. If the preset condition is met, 7 is reached, otherwise, 2 is returned;

7. of the last iterationIs the final result of the estimation;

and finally, the estimated position information is the multipath position, and the absolute value of the estimated amplitude value is obtained to obtain the multipath power value.

And finally, step 4: after the observation matrix Π is selected in the step 2, the rest pilot frequencies can obtain the observation matrix Π again through a random generation method, the step 2 and the step 3 are repeated again to obtain PDP estimation, and the estimation results of the two times are averaged to obtain a better estimation result; and finally, selecting an observation matrix pi in the step 4 to realize that the PDP estimation times are determined according to the performance requirement.

If the number of times of selecting the observation matrix is 1, skipping the step, otherwise, after excluding the M positions selected in the step 2, selecting M positions again from the rest positions, and performing the step 2 and the step 3 to obtain another group of estimated values theta₁And averaging all the obtained estimated values, and further improving the estimation accuracy.

The embodiment of the invention provides a TRS-based PDP estimation method in NR, which can improve the accuracy of PDP estimation, enhance the robustness of PDP estimation and reduce the complexity of window length search.

Those skilled in the art will appreciate that, in addition to implementing the system and its various devices, modules, units provided by the present invention as pure computer readable program code, the system and its various devices, modules, units provided by the present invention can be fully implemented by logically programming method steps in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Therefore, the system and various devices, modules and units thereof provided by the invention can be regarded as a hardware component, and the devices, modules and units included in the system for realizing various functions can also be regarded as structures in the hardware component; means, modules, units for performing the various functions may also be regarded as structures within both software modules and hardware components for performing the method.

The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention. The embodiments and features of the embodiments of the present application may be combined with each other arbitrarily without conflict.