Low-complexity terahertz channel estimation method based on smart reflective surface

Abstract

The present invention relates to a low-complexity channel estimation method based on intelligent reflective surface-assisted terahertz, belonging to the field of communication technology. The method comprises: S1: constructing a system model of IRS-assisted terahertz communication; S2: designing a corresponding IRS phase shift matrix and modeling the received signal as a three-dimensional tensor according to different blocks and different time slots; S3: decomposing the received signal using CP decomposition to obtain a factor matrix; S4: estimating the factor matrix obtained by decomposition according to a related estimator to solve the channel parameters; S5: reconstructing the channel matrix according to the solved channel parameters. The present invention can realize channel estimation using IRS in a THz communication system, and can effectively improve the accuracy of channel estimation in the THz communication system by designing the IRS phase shift matrix and modeling the received signal.

Description

Technical Field

[0001] The invention belongs to the technical field of communications and relates to a low-complexity channel estimation method based on intelligent reflecting surface-assisted terahertz. Background Art

[0002] As the key candidate technologies for the next generation of mobile communications, terahertz (THz) communication and intelligent reflecting surface (IRS) have become hot topics for experts and scholars. In short, compared with millimeter wave (mmWave), THz communication can provide richer bandwidth (from 0.1THz to 10THz), higher data rate (Tbps level) and lower latency (microsecond level). However, due to the short wavelength of THz signals, the signal attenuation is serious and the diffraction ability is poor. These shortcomings seriously limit the application of THz frequency bands. IRS can effectively make up for these shortcomings. IRS is a planar array composed of a large number of reflective elements. Each reflective element can control the amplitude and phase of the incident signal, and reconstruct the channel environment by adjusting and reflecting the signal at the transmitting end. Combining IRS with THz communication can increase the transmission distance of THz signals, reduce signal blocking and improve the reliability of communication.

[0003] Compared with traditional relays, IRS has won the recognition of experts and scholars for its passive characteristics and the fact that it does not increase the noise of the communication system. By deploying IRS in multi-cell communication scenarios, it can alleviate the problem of interference to users at the edge of the cell. Similarly, IRS can improve the sum rate of all users in the system through hybrid beamforming. In addition, in terms of physical layer security communications, in order to reduce the impact of eavesdroppers on the information reception of legitimate receivers and the amount of useful information received by eavesdroppers, IRS is used to achieve beam cancellation at the eavesdropping end in wireless power supply communication networks.

[0004] In the scenario of IRS-assisted THz communication, the complex propagation environment brings challenges to channel estimation. How to give full play to the huge communication potential of THz and how to solve the shortcomings of channel instability in THz communication through IRS have become the focus of research. Someone proposed a low-complexity channel estimation method based on compressed sensing, which reduces the computational complexity by eliminating the redundant columns of the sensing matrix during the iteration process. Similarly, the sparsity of the channel transmission process is used to solve the estimation problem of the cascade channel, which well solves the shortcomings of high complexity and excessive training overhead in the channel estimation process. It is also possible to use a beam training scheme based on the collaboration of the transmitter and IRS to perform channel estimation, and further perform hybrid beamforming based on the obtained channel information to reduce the consumption of the RF link. This scheme can achieve performance close to that of full digital beamforming under the condition of perfect channel state information. Although the current method can solve the channel estimation problem in wireless communication systems, with the increase in the number of antennas in THz communication, the large-scale deployment of IRS, and the increase in the number of IRS elements, the dimension of the received signal will also increase greatly, resulting in an increase in the cost of solving the channel estimation problem.

[0005] Therefore, there is an urgent need for a method that can simultaneously improve the accuracy of channel estimation and reduce the complexity of channel estimation to solve the above problems. Summary of the invention

[0006] In view of this, the purpose of the present invention is to provide a low-complexity channel estimation method based on smart reflecting surface-assisted terahertz to improve the accuracy of channel estimation and reduce the complexity of estimation. The present invention designs IRSs in groups based on the analysis of THz channel characteristics, and expresses the channel model of IRS-assisted THz communication as a unified mathematical expression, which effectively avoids the problem of excessive overhead caused by the increase in the number of IRS elements in estimating the IRS cascade channel.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] A low-complexity channel estimation method based on intelligent reflective surface-assisted terahertz, specifically comprising the following steps:

[0009] S1: Construct a system model of intelligent reflecting surface (IRS) assisted terahertz (THz) communication;

[0010] S2: Design the corresponding IRS phase shift matrix and model the received signal as a three-dimensional tensor according to different blocks and different time slots;

[0011] S3: Decompose the received signal using parallel factor (CP) decomposition to obtain a factor matrix;

[0012] S4: Estimate the factor matrix obtained by decomposition according to the correlation estimator to solve the channel parameters;

[0013] S5: Reconstruct the channel matrix according to the solved channel parameters.

[0014] Further, in step S1, the communication system model constructed is: assuming that the direct line of sight path is blocked, the multipath fading and blocking in the communication process are alleviated by introducing IRS; the signal emitted by the transmitter reaches the receiver after being scattered by scatterers in the environment or reflected by the IRS; wherein the IRS can be used as an element that controls the incident signal and can reflect the signal, and can also be regarded as a controllable scatterer in the communication link.

[0015] Further, in step S1, the terahertz (THz) downlink communication system model is constructed as follows: the base station is equipped with Antenna, user end equipped antennas, the IRS has N reflective array elements; the base station and the user are equipped with and RF links; then the signal y received by the user can be expressed as

[0016] y=W H Hs+n

[0017] The superscript H indicates the matrix conjugate transpose; s = F RF F B x, s represents the signal after hybrid beamforming at the transmitting end, is the data stream sent by the base station, r represents the number of sent data streams, and denote the digital and analog precoding matrices at the transmitting end, respectively; represents the channel matrix; represents the precoding matrix of the user end, and denote the digital and analog precoding matrices at the transmitting end, respectively; represents additive white Gaussian noise;

[0018] The general IRS cascade channel model is expressed as a unified mathematical expression based on the geometric channel model. For the THz channel, the geometric channel model is used and only one reflection is considered.

[0019]

[0020] Among them, L1 represents the number of paths; Represents the path loss. In the THz frequency band, the present invention mainly considers the free space loss and molecular absorption loss of the signal, satisfying

[0021]

[0022] Where c represents the speed of light, τ(f) represents the molecular absorption loss, and d represents the signal transmission distance;

[0023] a BS (θ), denote the array response vectors of the base station and the user respectively; assuming that the base station and the user adopt uniform linear arrays, the array response can be expressed as

[0024]

[0025]

[0026] The superscript T represents the matrix transpose, k = 2π / λ, λ represents the wavelength, and d t and d r Denote the antenna spacing, θ and denote the departure angle and arrival angle respectively;

[0027] Assuming that the IRS has N array elements, the IRS is divided into M sub-arrays, and the cascade channel can be expressed as:

[0028] H I =H t ΦH r

[0029] Among them, H t and H r denote the channel matrices of BS-IRS and IRS-UE respectively; Φ represents the phase shift matrix of IRS, and the amplitude β n ∈{0,1}, phase θ n ∈(0,2π], for simplicity, assume that β n =1; Assuming that each subarray is a linear array, the cascade channel can be further expressed as

[0030]

[0031] make The present invention regards the IRS as a controlled scatterer and re-establishes a channel model according to the propagation path of the signal in space. The above formula can be further expressed as

[0032]

[0033] It can be clearly seen that the role of IRS in the environment is to increase the number of paths from BS to UE, and by controlling the amplitude and phase of the IRS elements, the gain of each path can be effectively adjusted, and the gain is known and controllable, thereby achieving the role of controlling the channel environment. The complete channel can be expressed as

[0034]

[0035] Among them, K=L1+M represents the total number of paths in the entire channel environment; it can be clearly seen that the role of IRS in wireless communication is to increase the signal propagation path and provide a relatively stable propagation environment even in a more complex propagation environment.

[0036] Further, step S2 specifically includes: setting the channel coherence time T s Divided into T blocks, each block has L time slots, so that T s =TL; During the training process, for different blocks and different time slots, the following transmission rules are assumed: 1) The transmitted signal is repeated in each block; 2) The phase shift of the IRS within a block remains unchanged;

[0037] The signal received by the user in the lth time slot of the tth block is modeled as a third-order tensor:

[0038]

[0039] Among them, α t,k represents the phase shift combination of the kth IRS subarray, g k represents the path gain provided by the subarray, F = F RF ×F B represents the precoding matrix at the base station, n[t,l] represents the noise in the lth time slot of the tth block;

[0040] Then the tensor form of the total received signal of the tth block is:

[0041]

[0042] in,

[0043] Y[t]=[y[t,1],y[t,2],...,y[t,l]],

[0044]

[0045]

[0046]

[0047] N[t]=[n[t,1],n[t,2],...,n[t,l]].

[0048] Further, step S3 specifically includes: expressing the received signal in the form of CP:

[0049]

[0050] in, represents the outer product of matrices, and

[0051]

[0052]

[0053]

[0054] Among them, A, B, and C represent factor matrices. Since the THz signal is sparse, that is, the number of paths L1 is very small, and the number of IRS subarrays M is manually controllable, the total number of paths can be controlled within a certain range during operation. At this time, the received signal tensor It has a low-rank structure, which can ensure the uniqueness of the decomposition of the tensor. Then the problem is converted into a least squares problem to solve the factor matrix;

[0055] Assuming that the number of THz channel paths and the number of IRS reflectors are known a priori, the goal of CP decomposition is

[0056]

[0057] in, The above optimization problem can be solved by the alternating least squares method. In this method, the other variable matrix is ​​solved by fixing two variable matrices, and the process is repeated until the convergence criterion is met. Under the condition of CP decomposition uniqueness, the tensor Expand into the form of mode-n (n=1,2,3)

[0058]

[0059]

[0060]

[0061] In the formula, ⊙ represents the Khatri-Rao product of the matrix. According to the above expansion formula, the alternating least squares method is used to solve it:

[0062]

[0063]

[0064]

[0065] Among them, |||| F represents the F-norm of the matrix, ⊙ represents the Khatri-Rao product of the matrix, Representing a tensor The mode-n (n=1,2,3) expansion, They represent the factor matrices solved during the t-th iteration.

[0066] Further, in step S4, after estimating the factor matrices A, B, and C, the specific channel parameters θ k ,g k Under the condition that the tensor decomposition satisfies the uniqueness, the factor matrix obtained by decomposition has a certain error compared with the cause matrix, and the relationship between them is as follows:

[0067]

[0068]

[0069]

[0070] Where {Λ1, Λ2, Λ3} and Π are unknown non-singular diagonal matrices and unknown permutation matrices, respectively, satisfying Λ1Λ2Λ3=I. {E1, E2, E3} correspond to the estimated errors of the three factor matrices. Estimated factor matrix and Each column contains the angle information of the transmitting and receiving antennas and θ k , can be estimated by a correlation-based estimator. The correlation estimator can be expressed as:

[0071]

[0072]

[0073] Among them, ||||2 represents the 2-norm of the matrix, and Represents the factor matrix and The lth column of , J represents the search dimension, and denote the estimated values ​​of departure angle and arrival angle respectively, and respectively denote the reference values ​​of the response vector in the array.

[0074] Further, step S5 specifically includes: according to the channel parameters obtained in step S4 and θk Reconstruct the factor matrix and Then estimate the factor matrix C

[0075]

[0076] According to the known IRS phase shift matrix α t,k The path gain can be solved with the following formula:

[0077]

[0078] Among them, the upper right represents the pseudo-inverse of the matrix; at this point, the channel parameters and factor matrix estimation are completed, and the complete channel matrix can be restored.

[0079] The beneficial effects of the present invention are:

[0080] 1) With the increase in the number of antennas at the transmitting and receiving ends and the number of IRS array elements, the overhead caused by channel estimation cannot be ignored. The present invention effectively reduces the problem of increased channel overhead caused by the increase in the number of IRS elements by grouping the IRS.

[0081] 2) Smart reflective surfaces, as the key enabler of smart reconfigurable radio environments, reflective elements are characterized by small size, low cost, and low energy consumption. No special RF processing is required. By properly adjusting the phase shift of the reflective unit of the smart reflective surface, its reflected signal can be consistently combined with the signals of other paths, which is beneficial to assist THz signals in communication.

[0082] 3) The present invention considers using IRS to assist channel estimation, increasing the transmission distance of THz signals, reducing signal blocking and improving communication reliability. In the present invention, by grouping IRS, the problem of increased complexity caused by the increase in the number of IRS elements is effectively reduced.

[0083] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0085] Figure 1 This is a flow chart of a low-complexity channel estimation method based on IRS-assisted terahertz according to the present invention;

[0086] Figure 2 This is a system model diagram of the low-complexity channel estimation method based on IRS-assisted terahertz of the present invention;

[0087] Figure 3 Schematic diagram of CP decomposition tensor;

[0088] Figure 4 This is the pilot transmission structure diagram. DETAILED DESCRIPTION

[0089] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0090] See also Figure 1 to Figure 4 ,like Figure 1 As shown, the present invention provides a low-complexity method for terahertz channel estimation based on intelligent reflective surface, which specifically includes the following steps:

[0091] S1: Construct a system model of IRS-assisted THz communication;

[0092] S2: Design the corresponding IRS phase shift matrix and model the received signal as a three-dimensional tensor according to different blocks and different time slots;

[0093] S3: Decompose the received signal using CP decomposition to obtain a factor matrix;

[0094] S4: Estimate the factor matrix obtained by decomposition according to the correlation estimator to solve the channel parameters;

[0095] S5: Reconstruct the channel matrix according to the solved channel parameters.

[0096] Figure 2 The system model diagram of the present invention is shown in FIG. The smart reflective surface assisted terahertz communication system model considered in the present invention is as follows: Figure 2 As shown, the base station is equipped with Antenna, user end equipped antennas, and the IRS has N reflective array elements. The base station and the user are equipped with and RF links. Then the signal received by the user can be expressed as

[0097] y=WH Hs+n

[0098] The superscript H indicates the matrix conjugate transpose; s = F RF F B x, s represents the signal after hybrid beamforming at the transmitting end, is the data stream sent by the base station, r represents the number of sent data streams, and denote the digital and analog precoding matrices at the transmitting end, respectively; represents the channel matrix; represents the precoding matrix of the user end, and denote the digital and analog precoding matrices at the transmitting end, respectively; represents additive white Gaussian noise;

[0099] In order to avoid loss of generality, in the THz frequency band, the channel is represented by a geometric channel model with limited scattering paths. Since the energy attenuation of the THz signal after multiple reflections exceeds 20 dB, the present invention only considers the case of one reflection. According to the geometric channel model, each scattering path determines a single propagation path, which can be described by the following formula:

[0100]

[0101] Among them, L1 represents the number of paths, Represents the path loss. In the THz frequency band, the present invention mainly considers the free space loss and molecular absorption loss of the signal, satisfying

[0102]

[0103] Where c represents the speed of light, τ(f) represents the molecular absorption loss, and d represents the signal transmission distance. BS (θ), Denote the array response vectors of the base station and the user respectively. Assuming that the base station and the user adopt uniform linear arrays, the array response can be expressed as

[0104]

[0105]

[0106] Where k = 2π / λ, λ represents the wavelength, d t and d r Denote the antenna spacing, θ and represent the departure angle and arrival angle respectively.

[0107] Figure 3 The figure shows a schematic diagram of the CP decomposition tensor of the present invention, which will be described below in conjunction with the accompanying drawings:

[0108] CP decomposition was first proposed by Hitchcock, and its main idea is to decompose a tensor into the sum of a series of low-rank tensor products. is a third-order tensor, which can be expressed as

[0109]

[0110] In the formula, represents the outer product of matrices, Combine all vectors a into a factor matrix A = [a1, a2, ..., a n ], and similarly, we can get factor matrices B and C. The factor matrix can be used to represent the tensor in the following matrix form:

[0111]

[0112]

[0113]

[0114] Where "⊙" represents the Khatri–Rao product of the matrix, Represents the mode-n (n=1,2,3) expansion form of the tensor. In summary, the CP decomposition of the tensor has the following expression:

[0115]

[0116] Figure 4 As shown, it is a schematic diagram of the pilot transmission structure of the present invention, which is described below in conjunction with the accompanying drawings:

[0117] The present invention uses the channel coherence time T s Divided into T blocks, each block has L time slots, so that T s =TL. During the training process, for different blocks and different time slots, the following assumptions are made:

[0118] 1) Assume that the signal is sent repeatedly in each block;

[0119] 2) Assume that the phase shift of IRS within a block does not change.

[0120] That is:

[0121] x[t,l]=x[l],l=1,...,L,

[0122] φ n [t,l]=φ n [t],t=1,...,T.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A low-complexity channel estimation method based on intelligent reflective surface-assisted terahertz, characterized in that: The method specifically comprises the following steps: S1: Construct a system model of IRS-assisted terahertz communication; where IRS is a smart reflective surface; The communication system model constructed is as follows: Assuming that the direct line of sight path is blocked, the multipath fading and blocking in the communication process are alleviated by introducing IRS; the signal emitted by the transmitter reaches the receiver after being scattered by scatterers in the environment or reflected by IRS; among them, IRS can be used as a component that controls the incident signal and can reflect the signal, and can also be regarded as a controllable scatterer in the communication link; The terahertz downlink communication system model is constructed as follows: the base station is equipped with Antenna, user end equipped antennas, the IRS has N reflective array elements; the base station and the user are equipped with and RF links; then the signal y received by the user is expressed as y=W H Hs+n The superscript H indicates the matrix conjugate transpose; s = F RF F B x, s represents the signal after hybrid beamforming at the transmitting end, is the data stream sent by the base station, r represents the number of sent data streams, and denote the digital and analog precoding matrices at the transmitting end, respectively; represents the channel matrix; represents the precoding matrix of the user end, and denote the digital and analog precoding matrices at the transmitting end, respectively; represents additive white Gaussian noise; The general IRS cascade channel model is expressed as a unified mathematical expression based on the geometric channel model. For the THz channel, the geometric channel model is used and only one reflection is considered. Among them, L1 represents the number of paths; Represents the path loss. In the THz frequency band, considering the free space loss and molecular absorption loss of the signal, it satisfies Where c represents the speed of light, τ(f) represents the molecular absorption loss, and d represents the signal transmission distance; a BS (θ), denote the array response vectors of the base station and the user respectively; assuming that the base station and the user adopt uniform linear arrays, the array response is expressed as The superscript T represents the matrix transpose, k = 2π / λ, λ represents the wavelength, and d t and d r Denote the antenna spacing, θ and denote the departure angle and arrival angle respectively; Assuming that the IRS has N array elements, the IRS is divided into M sub-arrays, and the cascade channel is expressed as: H I =H t ΦH r Among them, H t and H r denote the channel matrices of BS-IRS and IRS-UE respectively; Φ represents the phase shift matrix of IRS, and the amplitude β n ∈{0,1}, phase θ n ∈(0,2π]; Assuming that each subarray is a linear array, the cascade channel is further expressed as make Considering IRS as a controlled scatterer, a channel model is re-established according to the propagation path of the signal in space; the above formula can be further expressed as The complete channel representation is Where K = L1 + M represents the total number of paths in the entire channel environment; S2: Design the corresponding IRS phase shift matrix and model the received signal as a three-dimensional tensor according to different blocks and different time slots; S3: Decompose the received signal using CP decomposition to obtain a factor matrix, where CP is a parallel factor; S4: Estimate the factor matrix obtained by decomposition according to the correlation estimator to solve the channel parameters; S5: Reconstruct the channel matrix according to the solved channel parameters.

2. The low complexity channel estimation method according to claim 1, characterized in that: Step S2 specifically includes: setting the channel coherence time T s Divided into T blocks, each block has L time slots, so that T s =TL; During the training process, for different blocks and different time slots, the following transmission rules are assumed: 1) The transmitted signal is repeated in each block; 2) The phase shift of the IRS within a block remains unchanged; The signal received by the user in the lth time slot of the tth block is modeled as a third-order tensor: Among them, α t,k represents the phase shift combination of the kth IRS subarray, g k represents the path gain provided by the kth IRS subarray, F = F RF ×F B represents the precoding matrix at the base station, n[t,l] represents the noise in the lth time slot of the tth block; Then the tensor form of the total received signal of the tth block is: in, Y[t]=[y[t,1],y[t,2],...,y[t,l]], N[t]=[n[t,1],n[t,2],...,n[t,l]].

3. The low complexity channel estimation method according to claim 2, characterized in that: Step S3 specifically includes: expressing the received signal in the form of CP: in, represents the outer product of matrices, and Among them, A, B, and C represent factor matrices. Then the problem is converted into a least squares problem to solve the factor matrix. The problem is described as Among them, |||| F represents the F-norm of the matrix, ⊙ represents the Khatri-Rao product of the matrix, Representing a tensor The mode-n (n = 1, 2, 3) expansion, A (t) , B (t) , C (t) They represent the factor matrices solved during the t-th iteration.

4. The low complexity channel estimation method according to claim 3, characterized in that: Step S4 specifically includes: using the factor matrix obtained in step S3 to estimate the corresponding channel parameters using a correlation estimator and θ k ; The correlation estimator is expressed as: Among them, ||||2 represents the 2-norm of the matrix, and Represents the factor matrix and The lth column of , J represents the search dimension, θ l and denote the estimated values ​​of departure angle and arrival angle respectively, and respectively denote the reference values ​​of the response vector in the array.

5. The low complexity channel estimation method according to claim 4, characterized in that: Step S5 specifically includes: according to the channel parameters obtained in step S4 and θ k Reconstruct the factor matrix and Then according to the IRS phase shift matrix α t,k Solve for the path gain: Among them, the upper right Represents the pseudo-inverse of the matrix; reconstructs the last factor matrix based on the path gain Finally, the entire channel is reconstructed.

Images