An OTFS downlink channel estimation method based on uplink assistance in ultra-large-scale MIMO systems

By adopting uplink-assisted OTFS downlink channel estimation method in a super-large-scale MIMO system, time-varying non-stationary channel expression is constructed and partial reciprocity is used to solve the complexity problem of high-speed mobile downlink channel estimation, and the non-stationary characteristics are achieved The low-complexity channel estimation under the following conditions improves communication efficiency and quality.

CN116094875BActive Publication Date: 2024-11-01ANHUI JIUTAI ELECTRIC CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310092851.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-02-09
Publication Date
2024-11-01
Estimated Expiration
2043-02-09

AI Technical Summary

Technical Problem

In high-speed mobile scenarios, OFDM is severely affected by the Doppler expansion of time-varying channels in ultra-large-scale MIMO systems, and traditional downlink channel estimation methods are complex in calculations and difficult to effectively utilize channel resources, especially in non-stationary characteristics and high-dimensionality It is difficult to achieve effective downlink channel estimation under the channel.

Method used

A downlink channel estimation method based on uplink assistance is proposed. By constructing a time-varying non-stationary channel expression, the channel parameters independent of the downlink frequency are obtained, and the user side estimates Doppler shift, delay and angle, and passes the Partial reciprocity obtains downlink channel gain, and the base station reconstructs the channel model under the delay Doppler frequency shift domain to reduce the computational complexity.

Benefits of technology

In high-speed mobile and non-stationary scenarios, the complexity of downlink channel estimation and user terminal hardware requirements are reduced, good communication performance is maintained, pilot resource requirements are reduced, and efficiency and quality requirements of ultra-large-scale MIMO systems are met.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116094875B_ABST
    Figure CN116094875B_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of wireless communication, and relates to an OTFS downlink channel estimation method based on uplink assistance in a very large-scale MIMO system, comprising the steps of constructing a time-varying non-stationary channel expression in a downlink link, obtaining a channel parameter set independent of a downlink frequency and sending the set to a user end; the user end oversamples a Doppler frequency shift domain, a time delay domain and an angle domain, and obtains estimated values ​​of Doppler frequency shift, time delay and angle in uplink data transmission; the user end obtains parameters independent of the uplink frequency through partial reciprocity, and estimates a downlink channel gain through the parameters; the user end sends the estimated values ​​of Doppler frequency shift, time delay, angle and downlink channel gain to a base station; the base station reconstructs a channel model in the delay Doppler frequency shift domain; the scheme of the invention reduces the computational complexity in high-speed mobile scenarios and very large-scale MIMO non-stationary scenarios, and satisfies the communication efficiency and quality in very large-scale MIMO complex scenarios.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of wireless communications and relates to an OTFS downlink channel estimation method based on uplink assistance in a super large-scale MIMO system. Background Art

[0002] Orthogonal Frequency Division Multiplexing (OFDM) has been widely used in the fifth generation mobile communication technology (5G) due to its high spectrum utilization efficiency and strong anti-multipath interference ability. However, in high-speed mobile scenarios, due to the Doppler spread of time-varying channels, OFDM will experience significant inter-carrier interference (ICI) and cannot meet the communication requirements of current scenarios. In order to solve the communication problems in high-mobility scenarios, relevant scholars have designed a new two-dimensional modulation technology called Orthogonal Time Frequency Space (OTFS) modulation. As a multiplexing technology designed in the two-dimensional delay-Doppler domain, OTFS modulation shows better performance in high-mobility scenarios than OFDM.

[0003] In the OTFS ultra-large-scale MIMO (Multiple Input Multiple Output, MIMO) system, in the frequency division duplexing (FDD) mode, since the channel is not reciprocal, in the downlink channel estimation, the downlink channel state information (CSI) is often obtained through downlink training and feedback. However, since the base station is equipped with a large number of antennas, the use of traditional downlink channel estimation methods requires a large amount of pilot resources, and it is difficult to use limited channel resources to feedback high-dimensional complex matrices. At the same time, as the number of antennas increases, the size of the antenna array increases, and the propagation signals of different paths can only reach part of the antenna array, that is, one path can only see part of the antenna array area, and the channel exhibits non-stationary characteristics. Therefore, downlink channel estimation is challenging in the OTFS ultra-large-scale MIMO system, and the low-complexity OTFS downlink channel estimation algorithm in the ultra-large-scale MIMO system has become a hot topic for researchers.

[0004] The existing downlink channel estimation schemes for very large-scale MIMO in FDD mode often use traditional downlink training and feedback to obtain downlink CSI. In order to reduce the computational complexity of downlink channel estimation, some studies have used the sparsity of the channel to improve the downlink channel estimation algorithm, or used the partial reciprocity of the uplink and downlink channels to perform uplink-assisted downlink channel estimation. However, there are few literatures that reduce the complexity of downlink channel estimation while considering the non-stationary characteristics in very large-scale MIMO channels. Therefore, in the high-speed mobile and non-stationary scenarios of very large-scale MIMO, it is of practical significance to consider using uplink assistance to improve the downlink channel estimation performance, and to use the non-stationary characteristics and channel sparsity characteristics to reduce the computational complexity and user terminal hardware requirements. Summary of the invention

[0005] In order to reduce the complexity of downlink channel estimation in a very large-scale MIMO system, the present invention proposes an uplink-assisted OTFS downlink channel estimation method in a very large-scale MIMO system, which specifically includes the following steps:

[0006] Construct a time-varying non-stationary channel expression in the downlink, obtain a set of channel parameters that are independent of the downlink frequency, and send these parameters to the user end. The channel parameters that are independent of the downlink frequency include Doppler shift, delay, angle, and channel gain.

[0007] The user end oversamples the Doppler frequency shift domain, time delay domain and angle domain to obtain the estimated values ​​of Doppler frequency shift, time delay and angle in uplink data transmission;

[0008] The user end obtains parameters that are independent of the uplink frequency through partial reciprocity, and estimates the downlink channel gain through these parameters;

[0009] The user end sends the estimated values ​​of Doppler frequency shift, delay, angle and downlink channel gain to the base station; the base station reconstructs the channel model in the delay-Doppler frequency shift domain and completes the channel estimation.

[0010] Furthermore, the ultra-large-scale MIMO system is composed of a cascade of a transmitter and a receiver. During the modulation process at the receiver, the data sequence is converted into the delay-Doppler frequency shift domain, and then converted into a time domain signal for transmission through the antenna; at the receiver, during the process of demodulation of the received signal by the antenna, the time domain received signal is converted into the delay-Doppler frequency shift domain.

[0011] Furthermore, the time-varying non-stationary channel expression in the downlink is expressed as:

[0012]

[0013] Among them, h k,l(t) represents the channel with index l in the delay domain of the kth user at time t; P is the number of scattering paths; g k,p is the channel gain of the kth user on the pth scattering path; v k,p is the Doppler frequency shift of the kth user on the pth scattering path; l represents the index of the delay domain; r represents the time index; T s is the sampling period of the system; δ(·) is the Dirkat function; a(θ k,p (t)) is the antenna array response; ⊙ represents the Hadamard product; q(Ψ s ) represents the selection vector of non-stationarity, Ψ s represents the set of paths on which the propagation signal can be received on the subarray s. For the subarray s, if the pth path belongs to,Ψ s Then q(Ψ s ) is 1, otherwise q(Ψ s ) is 0.

[0014] Furthermore, on the subarray s, the antenna array response is expressed as:

[0015]

[0016] Where s represents the index of the current subarray, S represents the total number of subarrays; d is the antenna element spacing on the base station, λ is the downlink subcarrier wavelength; θ k,p represents the angle of the k-th user on the p-th path.

[0017] Further, the process of obtaining the estimated values ​​of Doppler frequency shift, delay, and angle in uplink data transmission includes the following steps:

[0018] Sampling the Doppler frequency shift domain, time delay domain, and angle domain;

[0019] The enhanced sparse OMP algorithm constructs a sensor matrix, and selects the element with the largest inner product with the residual by traversing and searching the elements in the entire sensor matrix. The sampling value corresponding to the element is used as the estimated value of Doppler frequency shift, delay and angle.

[0020] Furthermore, when oversampling the Doppler frequency shift domain, the time delay domain and the angle domain, a sampling codebook is set for sampling, and the sampling codebook is expressed as:

[0021]

[0022] The sampling values ​​obtained after oversampling in the Doppler frequency shift domain, time delay domain and angle domain are expressed as:

[0023]

[0024]

[0025]

[0026] Among them, γ1, γ2, and γ3 are the oversampling factors of the Doppler frequency shift domain, time delay domain, and angle domain steering vector, respectively. O1={0,1,...,γ1N t -1}, O2={0,1,...,γ2L D -1}, O3={0,1,...,γ3sM / S-1}, N t is the number of effective sampling points, L D is the dimension of the delay domain, M is the number of antennas; are the sampling sets in the Doppler frequency shift domain, time delay domain, and angle domain respectively; [·] H It means to find the conjugate transposed matrix.

[0027] Furthermore, the calculation process of the estimated values ​​of Doppler frequency shift, time delay and angle includes:

[0028]

[0029]

[0030] According to the obtained codebook elements The relationship between the estimated uplink parameter value and the estimated uplink parameter set is obtained, a codebook element The corresponding uplink parameters, namely uplink Doppler shift, delay, angle The relationships between them include:

[0031]

[0032]

[0033]

[0034] in, is the codebook element corresponding to the estimated value of uplink Doppler frequency shift, delay and angle; a i,j,o is an element in the sensing matrix, which corresponds to the sampling set in the Doppler frequency shift domain The sampling value of the i-th Doppler frequency shift in In the delay domain sampling set The sample value of the jth delay Angle domain sampling set The sampling value of the oth angle in Represents element a i,j,o The value of represents the residual of the uplink received signal initially received by the base station; represents the Kronecker product, |·| represents the absolute value, and ||·|| represents the Frobenius norm; d represents the antenna element spacing on the base station; λ represents the downlink subcarrier wavelength; T s is the sampling time.

[0035] Furthermore, the user end obtains parameters that are independent of the uplink frequency through partial reciprocity, and estimates the downlink channel gain through these parameters. The process includes:

[0036] At the user end, the received pilots are stacked as a vector, expressed as:

[0037]

[0038] The downlink channel gain is obtained by solving the least square method. The solution process is expressed as:

[0039]

[0040] Among them, y dl Represents the received signal after stacking; G dl represents the downlink channel gain; z dl represents the noise vector matrix; is the matrix constructed by obtaining the uplink frequency-independent parameters through partial reciprocity, represents the pseudo-inverse matrix of matrix B; W is the transmit power.

[0041] Furthermore, the uplink frequency-independent parameter construction matrix B is obtained through partial reciprocity and is expressed as:

[0042]

[0043] Submatrices in matrix B It is expressed as:

[0044]

[0045] and They represent the frequency-independent parameters of the downlink, namely the downlink Doppler shift, angle, delay and visible path area, and the conversion method between them and the frequency-independent parameters of the uplink is:

[0046]

[0047] Where P represents the number of scattering paths; N p Indicates the number of downlink pilots; n N represents the number of subcarriers; Δf represents the subcarrier spacing; and They represent the frequency-independent parameters Doppler shift, angle, delay and visible path area of ​​the downlink end respectively; and Respectively represent the frequency-independent parameters Doppler shift, angle, delay and visible path area of ​​the uplink end; λ ul Indicates the wavelength of the upstream channel; λ dl Indicates the wavelength of the downstream channel.

[0048] Furthermore, the base station reconstructs the channel model in the delay-Doppler frequency shift domain, which is expressed as:

[0049]

[0050] in, represents the gain of the uplink and downlink channels of the pth scattering path; ΔF is the frequency difference between the uplink and downlink channels; and is the estimated value of downlink Doppler frequency shift, delay and angle; Respectively represent the steering vector estimation values ​​corresponding to the downlink Doppler frequency shift, delay, and angle; is the selection vector of non-stationarity in the Doppler frequency shift domain, represents the set of paths on which the propagation signal can be received on subarray s in the Doppler frequency shift domain. For subarray s, if the pth path belongs to but is 1, otherwise is 0.

[0051] The present invention aims at downlink channel estimation of FDD ultra-large-scale MIMO system in high-speed mobile scenarios, taking into account the non-stationary characteristics of ultra-large-scale antenna arrays and the partial reciprocity of uplink and downlink channels, and estimates uplink channel parameters through an improved enhanced sparse OMP algorithm and sends them to the downlink terminal, solving the problem of ultra-large-scale MIMO non-stationary characteristics while reducing computational complexity; in the scheme of the present invention, the downlink terminal only needs to estimate frequency-related parameters and feed them back to the base station for channel reconstruction, greatly reducing the pilot signal and downlink hardware requirements required for downlink channel estimation. The scheme of the present invention can still maintain good performance in high-speed mobile scenarios and ultra-large-scale MIMO non-stationary scenarios, reducing computational complexity while meeting the efficiency and quality of communication in ultra-large-scale MIMO complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a block diagram of the ultra-large-scale MIMO system used in the present invention;

[0053] Figure 2 It is the spatial model of the non-stationary characteristics of ultra-large-scale MIMO of the present invention;

[0054] Figure 3 Schematic diagram of NMSE performance comparison of different estimation schemes in OTFS scenario and OFDM scenario of the present invention;

[0055] Figure 4 The present invention is a flowchart of an implementation of an OTFS downlink channel estimation method based on uplink assistance in a very large-scale MIMO system. DETAILED DESCRIPTION

[0056] 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.

[0057] The present invention proposes an OTFS downlink channel estimation method based on uplink assistance in a very large-scale MIMO system, which specifically includes the following steps:

[0058] Construct a time-varying non-stationary channel expression in the downlink, obtain a set of channel parameters that are independent of the downlink frequency, and send these parameters to the user end. The channel parameters that are independent of the downlink frequency include Doppler shift, delay, angle, and channel gain.

[0059] The user end oversamples the Doppler frequency shift domain, time delay domain and angle domain to obtain the estimated values ​​of Doppler frequency shift, time delay and angle in uplink data transmission;

[0060] The user end obtains parameters that are independent of the uplink frequency through partial reciprocity, and estimates the downlink channel gain through these parameters;

[0061] The user end sends the estimated values ​​of Doppler frequency shift, delay, angle and downlink channel gain to the base station; the base station reconstructs the channel model in the delay-Doppler frequency shift domain and completes the channel estimation.

[0062] This embodiment provides a specific implementation method from four aspects: OTFS modulation and demodulation, channel modeling of ultra-large-scale MIMO non-stationary characteristics, uplink channel parameter estimation, and downlink channel reconstruction. Specifically, it includes:

[0063] 1.OTFS modulation and demodulation

[0064] like Figure 1As shown in Figure 1, the ultra-large-scale MIMO-OTFS system consists of a two-stage cascade structure of the transmitter and the receiver. At the transmitter, the information data symbols in the DD domain are transformed into the TF domain using the Inverse Symplectic Finite Fourier Transform (ISFFT), and then the TF codewords are converted to the time domain through the Heisenberg transform for transmission on the channel; at the receiver, the time domain received signal is transformed into the TF symbol by the Wigner transform, and then the TF symbol is mapped back to the DD domain for demodulation using the Symplectic Finite Fourier Transform (SFFT).

[0065] In the process of OTFS modulation, at the kth user in the uplink, the data sequence is converted to the DD domain to become a two-dimensional data symbol x k [l,n], where L D 、N D are the dimensions in the Delay and Doppler domains, respectively, and l and n are the corresponding Delay and Doppler domain indexes; specifically:

[0066] First, the data block in the DD domain is converted from the DD domain to the TF domain by ISFFT, which is expressed as follows:

[0067]

[0068] Among them, l' and n' are index numbers in the TF domain.

[0069] Then, by transmitting a pulse g tx The Heisenberg transform of (t) transforms the time-frequency domain signal Converted to time domain signal, the expression is as follows:

[0070]

[0071] Where Δf is the subcarrier spacing, T s is the sampling time.

[0072] In order to avoid inter-symbol interference between data blocks, the modulator usually adds a cyclic prefix (CP) to each symbol. After adding the cyclic prefix, a one-dimensional transmission signal in the time domain can be obtained, which is expressed as:

[0073]

[0074] in, The sampling point after each symbol is copied to the front of the symbol, ensuring that the number of waveform periods included in the delay of the symbol is an integer in the Fast Fourier Transform (FFT) period, thus ensuring the orthogonality of the subcarriers.

[0075] Transmitting signals The signal transmitted through the time-varying channel is the received signal Through a CP-free matrix, the received signal without CP is obtained, and the received signal r is received on the corresponding m-th antenna. m (t), the expression is as follows:

[0076]

[0077] in, The unvec{} operator is used to reconstruct a vector into a matrix form; v(t) is additive white Gaussian noise, and the channel impulse response h in the DD domain is km (τ,v) is:

[0078]

[0079] Among them, δ(·) is the Dirac function, P is the total number of resolvable paths, and g p , τ p and v p are the channel gain, delay and Doppler shift of the pth path respectively.

[0080] In the OTFS demodulation process, the received waveform g is first used through the Wigner transform rx (x) The received signal is converted into the TF domain. The received signal after conversion is expressed as follows:

[0081]

[0082] in, represents the receiving pulse shaping function for conjugate operation; the corresponding time and frequency are sampled at t = l'T and f = n'Δf respectively to obtain Y m [l',n'].

[0083] Then, the received signal in the DD domain is obtained according to the SFFT:

[0084]

[0085] 2. Channel Modeling of Non-Stationary Characteristics of Ultra-Large-Scale MIMO

[0086] like Figure 2As shown, when the size of the array becomes very large, different terminals can be seen in different parts of the array, because the energy of each terminal is concentrated on a part of the array, which is called the visible area (VR). In this embodiment, it is assumed that there are P scattering paths. For path p, due to the large aperture of the antenna array, the path cannot reach the entire antenna array surface, but can only reach one or several sub-arrays. Therefore, the visible area of ​​path p is defined, that is, the sub-array set that can receive the propagation channel along path p. The visible area VR index is as follows:

[0087] Φ p ={s p,start ,s p,start +1,...,s p,end}

[0088] Among them, s p,start 、s p,end The first and last subarray indices in the visible area, respectively, and satisfy 1≤s p,start ≤s p,end ≤S, where S is the total number of subarrays.

[0089] Correspondingly, the visible path area of ​​subarray s, that is, the set of paths on which the propagation signal can be received on the subarray, is expressed as:

[0090]

[0091] Among them, 1≤l s,i ≤P,i=1,...,L s , L s is the set size, 0≤L s ≤P.

[0092] The wireless signal can reach the user side along the line-of-sight path or be reflected by multiple scattering points, which means that the channel link is affected by frequency selective fading. And due to the mobility of the user, the channel changes over time, so the channel link experiences time selective fading. Along the downlink, each scattering path corresponds to a direction of departure (DOD), a Doppler shift, and a time delay. Therefore, the time-varying non-stationary channel in the downlink can be expressed as:

[0093]

[0094] Among them, k is the user, g k,p is the channel gain, v k,p is the Doppler shift, τ k,p is the delay, l is the index of the delay domain, r is the time index, δ(·) is the Dirac function, Ts is the sampling period of the system, a(θ k,p ) is the antenna array response, which is expressed as follows:

[0095]

[0096] Where d is the antenna element spacing on the base station, and λ is the downlink subcarrier wavelength. p ) is the selection vector considering non-stationarity, which is expressed as follows:

[0097]

[0098] Therefore, at subarray s, the antenna array response is rewritten as:

[0099] In high-speed movement, since the signal is transmitted at the speed of light, the transmission time is short. Therefore, during the change of time index r, the distance moved by the user can be ignored, and the change of its angle can also be ignored. Then it can be obtained that in the corresponding visible path area, h k,l Only with parameter set related.

[0100] 3. Uplink channel parameter estimation

[0101] When estimating uplink frequency-independent parameters, this scheme designs an oversampling codebook for the enhanced sparse OMP algorithm to improve sampling accuracy. The elements in the codebook are expressed as follows:

[0102]

[0103] Where O1=0,1,...,γ1N t -1,O2=0,1,...,γ2L D -1, O3 = 0, 1, ..., γ3sM / S-1; γ1, γ2, γ3 are the oversampling factors of their respective steering vectors.

[0104] The entire Angel, Delay and Doppler domains are sampled by sampling the codebook, and the oversampled Doppler frequency shift, delay and angle are obtained by considering the sparse representation of the non-stationary characteristics of the ultra-large-scale antenna array antenna in the antenna domain:

[0105]

[0106]

[0107]

[0108] The corresponding elements in the codebook are expressed as follows:

[0109]

[0110] After considering the non-stationary characteristics of the ultra-large-scale MIMO system, the dimension of the sampling codebook element is increased from the initial Reduce to That is, the dimension of the sampling codebook elements in this embodiment is reduced, that is, the complexity is reduced after considering the non-stationary characteristics.

[0111] According to the enhanced sparse OMP algorithm, the sensor matrix is ​​constructed. By traversing and searching the elements in the entire sensor matrix, the element with the largest inner product with the residual is selected and its corresponding μ,κ, The sampling value is taken as the estimation result, and the process is expressed as follows:

[0112]

[0113] 4. Downlink channel reconstruction

[0114] The user terminal receives the uplink parameter set sent in the uplink. According to the partial reciprocity of the uplink and downlink channels in the FDD mode, this embodiment uses the uplink channel frequency-independent parameters and the corresponding visible path range to obtain the following through partial reciprocity:

[0115]

[0116] in and They represent the frequency-independent parameters Doppler shift, angle, delay and visible path area of ​​the downlink end respectively.

[0117] At the downlink user end, the received pilots are stacked into a vector, which is expressed as:

[0118]

[0119] y dl , G dl and z dl , represent the stacked received signal, downlink channel gain and noise vector matrix respectively, and the expressions are as follows:

[0120]

[0121] B is the matrix constructed by obtaining uplink frequency-independent parameters through partial reciprocity, which is expressed as follows:

[0122]

[0123] The submatrix of the matrix for:

[0124]

[0125] At the downlink user end, the uplink frequency-independent parameters are obtained through partial reciprocity to construct the matrix B. Therefore, the downlink channel gain can be obtained by the least squares method, and the solution is as follows:

[0126]

[0127] in represents the pseudo-inverse of matrix B, is fed back to the base station, whereby the multi-antenna downlink channel is reconstructed as:

[0128]

[0129] Based on OTFS modulation and demodulation, channel modeling of ultra-large-scale MIMO non-stationary characteristics, uplink channel parameter estimation, and downlink channel reconstruction, the process of downlink channel estimation in this embodiment is as follows: Figure 4 As shown, the specific steps include:

[0130] By using OTFS modulation and demodulation, the uplink receiving signal model in the OTFS system is obtained; considering the non-stationary characteristics of the ultra-large-scale MIMO system, a parameter-separated channel model h is established;

[0131] Using the uplink received signal y, based on its non-stationary and sparse characteristics, a low-complexity uplink frequency-independent parameter estimation algorithm in ultra-large-scale MIMO systems is proposed. The base station informs the downlink user end of the estimated uplink delay and angle, Doppler shift and corresponding beamforming scheme.

[0132] By utilizing partial reciprocity at the downlink end, only a small number of pilot signals are needed to estimate the frequency-independent parameter, the channel gain, which is fed back to the base station end to ultimately complete the downlink channel reconstruction and downlink channel estimation.

[0133] Figure 3This is a simulation comparison diagram of the proposed method and other methods, which records the normalized mean square error (NMSE) performance curve of the proposed method and the traditional LS downlink channel estimation scheme. It can be seen from the simulation results that as the signal-to-noise ratio increases and the noise power decreases, the NMSE of both schemes decreases, and the estimation performance gradually improves; in the OTFS scenario and the OFDM scenario, the NMSE of the uplink assisted downlink channel estimation scheme is lower than the traditional LS downlink channel estimation scheme; in the OFDM scenario, the proposed uplink assisted estimation scheme is slightly better than the LS estimation scheme, but the gap between the two schemes is small; in the OTFS scenario, the proposed uplink assisted estimation scheme has a NMSE reduction of about 10 to 15 dB compared with the LS estimation scheme after OTFS modulation. The NMSE performance curves in the above scenarios show that the NMSE performance of the uplink assisted enhanced sparse OMP downlink channel estimation scheme proposed in the present invention is much better than the LS estimation algorithm, and the complexity of the enhanced sparse OMP algorithm is much lower than traditional channel estimation algorithms such as LS.

[0134] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for estimating downlink channel based on OTFS based on uplink assistance in a very large-scale MIMO system, characterized in that: The specific steps include: Construct a time-varying non-stationary channel expression in the downlink, obtain a set of channel parameters that are independent of the downlink frequency, and send these parameters to the user end. The channel parameters that are independent of the downlink frequency include Doppler shift, delay, angle, and channel gain. The time-varying non-stationary channel expression in the downlink is expressed as: Among them, h k,l (t) represents the channel with index l in the delay domain of the kth user at time t; P is the number of scattering paths; g k,p is the channel gain of the kth user on the pth scattering path; v k,p is the Doppler frequency shift of the kth user on the pth scattering path; l represents the index of the delay domain; r represents the time index; T s is the sampling period of the system; δ(·) is the Dirkat function; a(θ k,p (t)) is the antenna array response; ⊙ represents the Hadamard product; q(Ψ s ) represents the selection vector of non-stationarity, Ψ s represents the set of paths on which the propagation signal can be received on the subarray s. For the subarray s, if the pth path belongs to,Ψ s Then q(Ψ s ) is 1, otherwise q(Ψ s ) is 0; The user end oversamples the Doppler frequency shift domain, time delay domain and angle domain to obtain the estimated values ​​of Doppler frequency shift, time delay and angle in uplink data transmission; The user end obtains parameters that are independent of the uplink frequency through partial reciprocity, and estimates the downlink channel gain through these parameters; The user end sends the estimated values ​​of Doppler frequency shift, delay, angle and downlink channel gain to the base station; the base station reconstructs the channel model in the delay-Doppler frequency shift domain and completes the channel estimation.

2. The method for estimating downlink channel based on uplink-assisted OTFS in a very large-scale MIMO system according to claim 1, characterized in that: The ultra-large-scale MIMO system consists of a cascade of a transmitter and a receiver. During the modulation process at the receiver, the data sequence is converted into the delay-Doppler frequency shift domain, and then converted into a time domain signal for transmission through the antenna. At the receiver, the antenna receives the signal and demodulates it, converting the time domain received signal into the delay-Doppler frequency shift domain.

3. The uplink-assisted OTFS downlink channel estimation method in a very large-scale MIMO system according to claim 1, characterized in that: On subarray s, the antenna array response is expressed as: Where s represents the index of the current subarray, S represents the total number of subarrays; d is the antenna element spacing on the base station, λ is the downlink subcarrier wavelength; θ k,p represents the angle of the k-th user on the p-th path.

4. The method for estimating downlink channel based on uplink-assisted OTFS in a very large-scale MIMO system according to claim 1, characterized in that: The process of obtaining the estimated values ​​of Doppler frequency shift, delay, and angle in uplink data transmission includes the following steps: Sampling the Doppler frequency shift domain, time delay domain, and angle domain; The enhanced sparse OMP algorithm constructs a sensor matrix, and selects the element with the largest inner product with the residual by traversing and searching the elements in the entire sensor matrix. The sampling value corresponding to the element is used as the estimated value of Doppler frequency shift, delay and angle.

5. The method for estimating downlink channel based on uplink-assisted OTFS in a very large-scale MIMO system according to claim 4, characterized in that: When oversampling the Doppler frequency shift domain, delay domain, and angle domain, a sampling codebook is set for sampling. The sampling codebook is expressed as: The sampling values ​​obtained after oversampling in the Doppler frequency shift domain, time delay domain and angle domain are expressed as: Among them, γ1, γ2, and γ3 are the oversampling factors of the steering vector in the Doppler frequency shift domain, delay domain, and angle domain, respectively. O1={0,1,...,γ1Nt-1}, O2={0,1,...,γ2L D -1}, O3={0,1,…,γ3sM / S-1}, N t is the number of effective sampling points, L D is the dimension of the delay domain, M is the number of antennas; are sampling sets in the Doppler frequency shift domain, time delay domain, and angle domain respectively; [·] H It means to find the conjugate transposed matrix.

6. The uplink-assisted OTFS downlink channel estimation method in a very large-scale MIMO system according to claim 4 or 5, characterized in that: The calculation process of the estimated values ​​of Doppler frequency shift, delay, and angle includes: According to the obtained codebook elements The relationship between the estimated uplink parameters and the estimated uplink parameters is obtained, that is, the uplink Doppler shift, delay, angle The relationships between them include: in, is the codebook element corresponding to the estimated value of uplink Doppler frequency shift, delay and angle; a i,j,o is an element in the sensing matrix, which corresponds to the sampling set in the Doppler frequency shift domain The sampling value of the i-th Doppler frequency shift in In the delay domain sampling set The sample value of the jth delay Angle domain sampling set The sampling value of the oth angle in Represents element a i,j,o The value of represents the residual of the uplink received signal initially received by the base station; represents the Kronecker product, |·| represents the absolute value, and ||·|| represents the Frobenius norm; d represents the antenna element spacing on the base station; λ represents the downlink subcarrier wavelength; T s is the sampling time.

7. The method for estimating downlink channel based on uplink-assisted OTFS in a very large-scale MIMO system according to claim 1, characterized in that: The user end obtains parameters that are independent of the uplink frequency through partial reciprocity and estimates the downlink channel gain through these parameters. The process includes: At the user end, the received pilots are stacked as a vector, expressed as: The downlink channel gain is obtained by solving the least square method. The solution process is expressed as: Among them, y dl Represents the received signal after stacking; G dl represents the downlink channel gain; z dl represents the noise vector matrix; is the matrix constructed by obtaining the uplink frequency-independent parameters through partial reciprocity, represents the pseudo-inverse matrix of matrix B; W is the transmit power.

8. The uplink-assisted OTFS downlink channel estimation method in a very large-scale MIMO system according to claim 7, characterized in that: The uplink frequency-independent parameter construction matrix B obtained through partial reciprocity is expressed as: Submatrices in matrix B It is expressed as: and They represent the frequency-independent parameters of the downlink, namely the downlink Doppler shift, angle, delay and visible path area, and the conversion method between them and the frequency-independent parameters of the uplink is: Where P represents the number of scattering paths; N p Indicates the number of downlink pilots; n N represents the number of subcarriers; Δf represents the subcarrier spacing; and They represent the frequency-independent parameters Doppler shift, angle, delay and visible path area of ​​the downlink end respectively; and They represent the frequency-independent parameters Doppler shift, angle, delay and visible path area of ​​the uplink end respectively; ul Indicates the wavelength of the upstream channel; λ dl Indicates the wavelength of the downstream channel.

9. The method for estimating downlink channel based on uplink-assisted OTFS in a very large-scale MIMO system according to claim 1, characterized in that: The base station reconstructs the channel model in the delay-Doppler frequency shift domain, which is expressed as: in, represents the gain of the uplink and downlink channels of the pth scattering path; ΔF is the frequency difference between the uplink and downlink channels; and is the estimated value of downlink Doppler frequency shift, delay and angle; Respectively represent the steering vector estimation values ​​corresponding to the downlink Doppler frequency shift, delay, and angle; is the selection vector of non-stationarity in the Doppler frequency shift domain, represents the set of paths on which the propagation signal can be received on subarray s in the Doppler frequency shift domain. For subarray s, if the pth path belongs to but is 1, otherwise is 0; represents the Kronecker product.