Massive MIMO-OFDM alternate iterative beam structure channel estimation method and system

Abstract

The application discloses a kind of large-scale MIMO-OFDM alternate iteration beam structure channel estimation method and system.The method comprises: the non-zero element in equivalent double beam domain channel is orthogonally grouped according to the result of its corresponding beam direction cosine index and time delay index modulo oversampling factor;Convert spatial-frequency domain receiving signal to beam domain;According to the interference element set with cross characteristics corresponding to the non-zero element in the equivalent double beam domain channel of each user, the beam domain receiving signal is extracted in the element in the same group, and estimation is carried out in parallel, and the estimation value of each beam domain channel element is obtained by using alternate iteration processing mode for different groups;According to the equivalent beam set of each user, the beam domain channel of each user is recovered, and the spatial-frequency domain channel of each user is reconstructed.The application makes full use of the structural characteristics of beam domain channel, effectively guarantees the convergence and estimation performance of estimation process while significantly reducing the complexity of channel estimation.

Description

Technical Field

[0001] This invention relates to the field of channel estimation technology, and in particular to a method and system for channel estimation of large-scale MIMO-OFDM alternating iterative beam structure. Background Technology

[0002] Massive Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) is a key enabling technology for fifth-generation (5G) mobile communications. With the further increase in the number of antennas and user terminals, this technology will also play a crucial role in sixth-generation (6G) mobile communications. In a massive MIMO system, the base station can simultaneously provide services to multiple user terminals on the same time-frequency resources by deploying a large number of antennas. Therefore, massive MIMO can effectively utilize spatial radio resources, thereby significantly improving spectrum efficiency.

[0003] The potential capacity gain of massive MIMO-OFDM depends on accurate channel state information. However, in massive MIMO-OFDM systems, with the increase in the number of antennas, subcarriers, and user terminals, the computational complexity of traditional estimation methods such as least squares (LS) and minimum mean square error (MMSE) estimation becomes unbearable due to the need for large-dimensional matrix inversion. Some statistical inference algorithms, such as approximate message passing and channel estimation algorithms based on minimizing constrained Bethe free energy, still have high overall computational complexity due to the large number of iterations, and they often rely on pre-set damping factors to guarantee convergence, which is not conducive to their application in practical systems.

[0004] For large-scale MIMO-OFDM systems, wireless channels typically exhibit sparse multipath structures with relatively small angular and delay spreads. Therefore, a beam-domain channel model can be constructed based on the physical model of the channel. This model exhibits significant sparsity, thereby drastically reducing the number of parameters required for estimation. Furthermore, when two beams are sufficiently far apart in space, they asymptotically tend to be orthogonal. Therefore, designing a low-complexity, high-performance channel estimation algorithm that converges without relying on damping factors, based on these characteristics, has become a key problem to be solved in large-scale MIMO-OFDM systems. Summary of the Invention

[0005] Purpose of the invention: To address the shortcomings of existing technologies, this invention proposes a channel estimation method and system for large-scale MIMO-OFDM alternating iterative beam structures. This method achieves channel estimation with fewer iterations, lower overall computational complexity, and convergence independent of damping factors, while ensuring the accuracy of the acquired channel information.

[0006] Technical Solution: To achieve the above objectives, the present invention provides the following technical solution:

[0007] Large-scale MIMO-OFDM alternating iterative beam structure channel estimation methods include:

[0008] The non-zero elements in the equivalent double beam domain channel are orthogonally grouped according to the modulo result of the oversampling factor based on the direction cosine index and time delay index of their corresponding beams.

[0009] It receives pilot sequences sent by each user and converts the space-frequency domain received signal to the beam domain through beam conversion;

[0010] For elements within the same group, the received signal in the beam domain is extracted and estimated in parallel based on the set of interference elements with cross-shaped characteristics corresponding to the non-zero elements in the equivalent double beam domain channel of each user. For elements in different groups, an alternating iterative processing method is adopted to obtain the estimated value of each beam domain channel element through iteration. The beam domain channel of each user is recovered based on the equivalent beam set of each user, and the space-frequency domain channel of each user is reconstructed based on the double beam base channel model.

[0011] Furthermore, the dual-beambase channel model is a dual-beam matrix multiplied by a dual-beamdomain channel vector. The dual-beam matrix is ​​a matrix composed of a selected set of spatial-frequency domain direction vectors corresponding to the direction cosine and the time delay sampling grid. Each spatial-frequency domain direction vector is called a dual beam. The operation of multiplying the dual-beam matrix or its conjugate transpose by the vector is implemented by FFT. The beam transformation is to convert the spatial-frequency domain signal vector into a beamdomain signal vector using the dual-beam matrix.

[0012] Furthermore, the number of points uniformly sampled for direction cosine and time delay is the number of antennas and the number of equivalent time delay spread points multiplied by the oversampling factor. The number of equivalent time delay spread points is obtained by multiplying the ratio of the number of effective subcarriers to the total number of subcarriers by the cyclic prefix length.

[0013] Furthermore, the pilot sequences transmitted by each user are adjustable phase-shift pilots, and the pilot sequences assigned to users with different phase-shift factors ensure that their double beam domain channels do not overlap; the equivalent beam set of each user is the double beam set corresponding to the non-zero elements of the equivalent double beam domain channel of each user, and the equivalent double beam domain channel of each user is the channel formed by combining the phase-shift pilots of each user with the double beam domain channel of each user.

[0014] Furthermore, when orthogonally grouping the non-zero elements in the equivalent double beam domain channel, the elements in the same group are guaranteed to have the same modulo value of the direction cosine and time delay of their corresponding beams with respect to the oversampling factor.

[0015] Furthermore, the set of interference elements with cross-shaped characteristics for each non-zero beam domain channel element consists of elements that have the same direction cosine or time delay as the current element and whose interference to the current element is greater than a preset threshold.

[0016] Furthermore, the space-frequency domain estimator of each equivalent double beam domain channel non-zero element has a beam structure of double beam matrix multiplied by beam domain estimator, and uses double beam matrix, beam set corresponding to the non-zero element, prior mean, and prior variance to design the corresponding beam domain estimator to obtain the estimate of posterior mean and posterior variance. The beam domain estimator is obtained by solving an optimization problem that minimizes the mean square error.

[0017] Furthermore, the method of alternating iterative processing for elements in different groups is as follows: beam domain estimators are designed in parallel for elements in the same group to estimate the posterior mean and posterior variance. After all elements in the same group have been estimated, the estimated posterior mean and posterior variance are used to update the prior mean and prior variance. Then, the next group is estimated, and the above process is repeated until the iteration converges.

[0018] Furthermore, at the start of the iteration, the prior mean is initially zero, and the prior variance is initially the beam domain statistical channel information. When updating the prior mean and prior variance using the posterior mean and posterior variance, the prior mean of the current estimated element itself remains zero, and the prior variance remains the beam domain statistical channel information.

[0019] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the computer program, when loaded onto the processor, implements the alternating iterative beam structure channel estimation method.

[0020] A large-scale MIMO-OFDM communication system includes a base station and multiple user terminals. Each user terminal sends pilot signals to the base station in the uplink. The base station uses the received pilot signals to obtain channel state information according to the alternating iterative beam structure channel estimation method.

[0021] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0022] 1. This invention proposes an alternating iterative beam structure channel estimation method. This estimation algorithm only requires a few iterations to obtain accurate statistical channel information, with low overall computational complexity, which is beneficial for its application in practical systems.

[0023] 2. This invention proposes an orthogonal grouping method for channel estimation of alternating iterative beam structures. By grouping the non-zero elements of the equivalent beam domain channel according to the values ​​of the oversampling factor modulo their corresponding beam direction cosine index and time delay index, the parallel estimation of elements within the same group does not affect each other, thereby ensuring the convergence of the algorithm. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the overall process of the channel estimation method based on alternating iterative beam structure according to an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram illustrating the implementation framework of the channel estimation method based on alternating iterative beam structure according to an embodiment of the present invention.

[0026] Figure 3 for Figure 2 A schematic diagram of each estimator group;

[0027] Figure 4 This is a performance diagram of the channel estimation method based on alternating iterative beam structure in an embodiment of the present invention;

[0028] Figure 5 This is a schematic diagram illustrating the convergence of the channel estimation method based on alternating iterative beam structure in an embodiment of the present invention.

[0029] Figure 6 This is a schematic diagram illustrating the low-complexity design performance of the channel estimation method based on alternating iterative beam structure in an embodiment of the present invention.

[0030] Figure 7 This is a schematic diagram illustrating the complexity of the channel estimation method based on alternating iterative beam structure in an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions provided by the present invention will be described in detail below with reference to specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention.

[0032] This invention discloses a method for channel estimation using a large-scale MIMO alternating iterative beamstructure. After receiving the pilot sequences transmitted through the channel from each user, the base station uses a space-frequency domain alternating iterative beamstructure channel estimator to estimate the channel for each user. Figure 1 and Figure 2 As shown, the specific steps include: orthogonally grouping the non-zero elements in the equivalent double-beam domain channel according to the modulo result of the oversampling factor based on the direction cosine index and time delay index of their corresponding beams; receiving the pilot sequences sent by each user and converting the space-frequency domain received signal to the beam domain through beam transform; for elements in the same group, extracting the beam domain received signal according to the set of interference elements with cross characteristics corresponding to the non-zero elements in the equivalent double-beam domain channel of each user, estimating them in parallel, and using an alternating iterative processing method for elements in different groups to obtain the estimated value of each beam domain channel element through iteration; recovering the beam domain channel of each user based on the equivalent beam set of each user, and reconstructing the space-frequency domain channel of each user based on the double-beam base channel model.

[0033] Figure 3 The diagram illustrates the structure of the estimator for each group of elements in the alternating iterative beamform channel estimation: This beam domain estimator, based on a double beam matrix and the corresponding beam set, prior mean and variance, obtains estimates of the posterior mean and variance by solving a problem that minimizes the mean square error. During iteration, the prior information is updated based on the posterior values ​​estimated by other groups. Then, each group of elements estimates its corresponding posterior value in parallel, and the prior value used by each element to estimate itself remains unchanged from its initial state (mean is 0, variance is beam domain statistical channel information).

[0034] The method of this invention is mainly applicable to large-scale MIMO-OFDM systems equipped with large-scale antenna arrays on the base station side to simultaneously serve multiple users. The specific implementation process of the channel information acquisition method involved in this invention will be described in detail below with reference to a specific communication system example. It should be noted that the method of this invention is not only applicable to the specific system model illustrated in the following example, but also to other system models with different configurations.

[0035] I. System Configuration

[0036] In this embodiment, a large-scale MIMO-OFDM system is considered. The base station is configured with a uniform linear array, and the number of antennas is... ,Serve Single-antenna user. The spacing between adjacent antennas is... It means that among them carrier frequency The corresponding wavelength. In OFDM modulation, the number of carriers is... The length of the cyclic prefix is The subcarrier spacing is The sampling interval is The number of effective subcarriers used for transmitting data and pilot signals is: Its index set is denoted as .

[0037] Large-scale MIMO-OFDM systems operate in Time Division Multiplexing (TDD) mode, utilizing the reciprocity of TDD systems to acquire channel information during uplink transmission.

[0038] II. Channel Model

[0039] First, the dual-beambase channel model can be established as follows. During uplink transmission, at the base station... The continuous baseband signal received by the antenna can be expressed as

[0040] (1)

[0041] In the formula For user terminals With the The time-varying channel impulse response between the root antennas For user terminals The transmitted OFDM signal, For the first Additive white Gaussian noise at the root antenna. Assuming the CSI remains constant within one OFDM symbol during uplink transmission, the user terminal... With the The channel impulse response between the root antennas can be expressed as

[0042] (2)

[0043] in For users The number of paths between the base station and the ground station. and They are users The Complex gain and direction cosine of the stripe diameter, User The first antenna between the base station and the first antenna The time delay of the stripe, In equation (3), the direction cosine is defined as... ,in and These are the azimuth and elevation angles, respectively. So, the user... and base station The first between the root antennas The first OFDM symbol The channel frequency response on each subcarrier is expressed as:

[0044] (3)

[0045] make Let the scaled direction cosine be represented as follows: Then the rudder vector in the spatial and frequency domains can be expressed as...

[0046] (4)

[0047] (5)

[0048] Pointing to the scaled direction cosine respectively Delay superscript This indicates transposition. Therefore, the user... The space-frequency domain channel vector can be represented as , of which The elements are .parameter and The ranges are respectively and It can be evenly divided into and A subset, such as

[0049] (6)

[0050] (7)

[0051] in , , and This represents the oversampling factor, used to adjust the oversampling density. (Scaled direction cosine) and delay Each is located in a sampling interval and We approximate the true parameters located within the sampling interval using the left endpoint of this interval. Therefore, the SF domain channel can be approximated as...

[0052] (8)

[0053] in It is the channel vector in the double beam domain. , Denotes the Kronecker product, and The column sum The The columns are respectively

[0054] (9)

[0055] (10)

[0056] in and These are the direction vectors sampled in the spatial and frequency domains, respectively. That is, the dual-beambase channel model is represented as the product of the dual-beam matrix and the dual-beamdomain channel vectors. The dual-beam matrix consists of a selected set of spatial-frequency domain direction vectors, each corresponding to a direction cosine and a time-delay sampling grid. Since these direction vectors are all columns of a Discrete Fourier Transform (DFT) matrix, therefore... and Both can be generated from a portion of the DFT matrix. The product of the double beam matrix or its conjugate transpose with the vector is achieved through the Fast Fourier Transform.

[0057] Because the number of local scatterers is finite, the two-dimensional beam domain channel vector is sparse in both the angle and delay domains, greatly reducing the number of elements that need to be estimated. We assume... The elements in the matrix follow independent cyclic symmetric complex Gaussian distributions with different variances and zero mean. Therefore, the two-dimensional beam domain channel covariance matrix... It is a diagonal matrix.

[0058] III. Alternating Iterative Beam Structure Channel Estimation

[0059] Using the beam-based channel model, the pilot signal received by the base station is

[0060] (11)

[0061] in Each element is independently and identically distributed as The noise vector, and , For users Adjustable phase-shift pilot, Indicates the diagonal is diagonal matrix, This represents an M-dimensional identity matrix. The adjustable phase-shift pilot is defined as...

[0062] (12)

[0063] in For Hadamard product, It is the same root sequence for all users. It is a phase-shifted sequence with a phase-shift factor of . .

[0064] Typically, the root sequence can be generated from the Zadoff-Chu (ZC) sequence, for example...

[0065] (13)

[0066] Indicates to Mold taking. Because... It is sparse; utilizing the properties of tunable phase-shift pilots, the received signal can be re-represented as...

[0067] (14)

[0068] in The sum of the equivalent double-beam domain channel vectors, For users The equivalent double beam domain channel vector, and Since non-zero elements from different users can not overlap with appropriate phase shifts, they can be derived from... Recovery Non-zero elements, such as ,in for The ordered set composed of non-zero element indices, i.e., the equivalent beam set of each user, satisfy and This indicates the number of elements in the set. Then, the empty frequency domain channel of the corresponding user can be recovered as... ,middle Since the root sequence has unity modulus and its noise distribution remains unchanged, its influence can be removed without affecting the optimality of channel estimation.

[0069] (15)

[0070] in Therefore, it is estimated that... Converted into an estimate .

[0071] For the received signal model in (15), The MMSE estimates of the posterior mean are respectively

[0072] (16)

[0073] in For large-scale MIMO-OFDM systems, when the number of antennas and subcarriers is large, The dimension becomes very high, making the computational complexity of the MMSE estimation unacceptable. Because The elements in the array are independent of each other, therefore they can be... The estimator design is transformed into an estimator design for each beam domain channel element, thereby finding an estimator that is suitable for... Channel estimation is sufficient for low-dimensional observation vectors.

[0074] First, regarding the received signal Beamforming involves using a double beam matrix to convert a space-frequency domain signal vector into a beam domain signal vector, thereby obtaining the beam domain received signal.

[0075] (17)

[0076] in and . The The element is The inner product of the two columns can be expressed as

[0077] (18)

[0078] in , , , and , This indicates rounding down to the nearest integer. Quantified Beam domain observations The magnitude of the influence. This indicates the matrix It can be used as a beam domain element in the received signal The interference matrix in the image. Note the received signal. and They have the same dimensions, and their elements correspond one-to-one. Therefore, for , This can be considered as the interference level between the nth beam element and the mth beam element. Specifically, when... The Column and number Orthogonal, we have Therefore, the first The element and the first There is no interference between the elements.

[0079] because The structure is similar to , It exhibits sparsity. This sparsity is referred to as the interference sparsity among beam-domain elements. Therefore, based on the matrix... The property that a set of interfering elements can be obtained by containing only elements... Constructing with elements that cause significant interference, i.e.

[0080] (19)

[0081] in It is a pre-set threshold.

[0082] Note that when When relatively large, for example, when the threshold is At that time, the interference beam set can be constructed as At this point, the interference is dominated by only a few elements, and the interference beam set has a cross-shaped characteristic. All elements in the set satisfy or , This refers to the set of interfering elements, which consists of elements that share the same direction cosine or time delay as the current element and significantly interfere with it. Using this set of interfering elements, we can... Extract elements to construct a model for estimation Low-dimensional observations ,in It is a selection matrix. and This indicates the number of elements in the set.

[0083] For this low-dimensional observation vector ,have

[0084] (20)

[0085] in Let the mutual information of two random variables be represented. The above inequality holds true when the following conditions are met: and Therefore, when it is the first When the beam selected for beam estimation is orthogonal to beams outside the set, It is used for estimation Sufficient observations. Note that due to the sparsity of the beam domain channel, most elements are 0, so the resulting interference is also 0. Therefore, under the beam-based channel model, a low-dimensional observation vector with small mutual information loss can be found to achieve sufficient observations. Estimates of each element in the equation.

[0086] At this time, the observed vector It can be represented as

[0087] (twenty one)

[0088] in and Therefore, a linear estimator can be designed. Make the estimate The mean square error is minimized, and its mean square error can be expressed as:

[0089] (twenty two)

[0090] Therefore, by solving the optimization problem of minimizing the mean square error, the beam domain estimator can be obtained. The solution is

[0091] (twenty three)

[0092] in and It is used for estimation A low-dimensional beam domain estimator. Then, for... The estimate can be expressed as

[0093] (twenty four)

[0094] in, It is the time-frequency domain estimator corresponding to the beam domain channel elements, which can be expressed as:

[0095] (25)

[0096] It reveals the beam structure of the spatial frequency domain estimator, namely, by selecting specific beams to design a low-dimensional beam domain estimator for each beam domain channel element, and the estimator for the equivalent double beam domain channel non-zero elements has the structure of a double beam matrix multiplied by the beam domain estimator. Therefore, (24) is called beam structure channel estimation. Note that only non-zero elements need to be considered. Estimate the elements. For each non-zero element Channel estimation for repeating beam structures can be obtained. The estimation and spatial frequency domain channel can be recovered.

[0097] As mentioned above, beam structure channel estimation can be achieved by utilizing low-dimensional observations. To reduce due to The high dimensionality of beamforming results in computational complexity that is difficult to bear. However, this strategy also carries risks. Even if most beams experience only weak interference, ignoring too many beams can degrade the estimation performance of beamstructured channel estimation. Therefore, we introduce the Turbo principle into beamstructured channel estimation to improve estimation performance.

[0098] The Turbo principle utilizes iterative feedback of external information to improve performance and is widely used in channel decoding and signal detection. Based on this principle, the aforementioned beamform channel can be evolved into an iterative form called alternating iterative beamform channel estimation. In alternating iterative beamform channel estimation, elements are divided into groups, and elements within the same group are estimated in parallel while alternating iterative processing is performed on different groups. At the start of the iteration, the prior mean is initialized to zero, the prior variance is initialized to beam-domain statistical channel information, and the iteration process uses the feedback of elements... The estimated posterior information is used to update the prior mean and covariance. This not only helps to estimate other elements more accurately, but also ultimately improves the estimation performance within an iterative framework. All elements are divided into... Group, Indicates containing the first The set of indices corresponding to the group elements. Let... and They are respectively In the The posterior mean and variance at the nth iteration. Then the... Group 1 in The prior mean and covariance after the next iteration can be expressed as:

[0099] (26)

[0100] (27)

[0101] in Including the Beam indexes in groups and subsequent groups, No. The beam index within the group preceding the group. Additionally, for According to the Turbo principle, when estimating the current element, its own contribution to the prior information needs to be removed. That is, when updating the prior information using the posterior mean and posterior variance, the prior mean of the currently estimated element remains zero, and the prior variance remains the beam domain statistical channel information. For example...

[0102] (28)

[0103] (29)

[0104] in and For use in the first The second iteration The prior information for channel estimation of the element beam structure. At this time, the corresponding beam domain estimator can be designed based on the double beam matrix, the set of interfering beams corresponding to the element, the prior mean and the prior variance, so as to obtain the estimate of the posterior mean and the posterior variance. From (23), it can be seen that the estimator is at this time.

[0105] (30)

[0106] Then the estimated posterior mean is

[0107] (31)

[0108] in for

[0109] (32)

[0110] And the estimated posterior mean is

[0111] (33)

[0112] This process allows for the parallel design of beam domain estimators for elements within the same group to estimate their posterior mean and posterior variance. Once all elements in the group have been estimated, the prior mean and prior variance are updated using the obtained posterior mean and posterior variance. The estimation is then performed on the next group of elements, and this process is repeated until the iteration converges or the maximum number of iterations is reached. .

[0113] The alternating iterative beam structure channel estimation described above can be summarized as follows:

[0114] Step 1: Input , , , , Number of iterations ;

[0115] Step 2: Initialization , , ;

[0116] Step 3: Let ;

[0117] Step 4: Calculated by (32) ;

[0118] Step 5: Designed through (30) ;

[0119] Step 6: Calculated by (31) ;

[0120] Step 7: Calculated by (33) ;

[0121] Step 8: Then judge Is it greater than If it is greater than, continue; if it is not greater than, return to step 4.

[0122] Step 9: Then judge Is it greater than If it is greater than, continue; if it is not greater than, return to step 3.

[0123] Step 10: Output the estimation results .

[0124] IV. Orthogonal Grouping

[0125] For channel estimation in alternating iterative beam structures, if the element grouping is random, then there may be... Make In this case, the beam and They influence each other, therefore they are estimated simultaneously. and Potential damage assessment The optimality of this overall problem can even lead to non-convergence of the algorithm. Therefore, in order to ensure the performance and convergence of the channel estimation for alternating iterative beam structures, it is necessary to finely group the elements.

[0126] The following conditions are given for two elements to be estimated in parallel without affecting each other: For and , making for and for If all elements within a group satisfy this condition, the group can be called an orthogonal group, and the grouping method is called orthogonal grouping.

[0127] According to (18), when and It is possible Similarly, , Sometimes, it is also possible to have At the same time, because when When the relative size is large, the interference beam set has a cross-shaped characteristic, that is, the set... All elements in the set satisfy or , At this point, the conditions for the orthogonal system can be transformed into... , , and .because , , and It is easy to calculate whether it satisfies the condition. This is also easily verifiable. Furthermore, and The conditions can be relaxed to That is, only requires and In this case, all beams can be divided into Orthogonal group, elements within the group and satisfy and That is, the direction cosine index and time delay index of the beam corresponding to the elements in the same group have the same modulus value with respect to the oversampling factor.

[0128] Through the orthogonal grouping, the performance and convergence of the alternating iterative beam structure channel estimation can be guaranteed.

[0129] V. Low-complexity implementation

[0130] Since (26) to (29) only construct prior information using already calculated values, the complexity of alternating iterative beam channel estimation mainly consists of the following three parts: 1. Estimator design (30); 2. Calculation through (32) 3. Estimate the posterior mean (31) and variance (33).

[0131] Here, calculating (31) and (33) are both inner products of low-dimensional vectors, with a complexity of O(n log n). The complexities of (30) and (32) are respectively and ,in Let be the number of non-zero elements in the beam domain channel. Therefore, the complexity of alternating iterative beam structure channel estimation is dominated by (30) and (32). As mentioned above, Most of the elements in the set have values ​​close to 0. Therefore, for the calculation of (30) and (32) It can be simplified to

[0132] (34)

[0133] (35)

[0134] in , And threshold and The beam used to filter out negligible interference. Therefore, the computational complexity of (30) and (32) can be reduced to... and .

[0135] In addition, calculation Need ,because and They can all be generated from partial DFT matrices, therefore It can be implemented using the Fast Fourier Transform (FFT) and only requires one computation.

[0136] Therefore, iteration The total complexity of channel estimation for alternating iterative beam structures is: ,in for

[0137] (36)

[0138] in yes A set consisting of non-zero elements that satisfies .

[0139] V. Implementation Results

[0140] To enable those skilled in the art to better understand the present invention, the performance results of the channel information acquisition method in this embodiment under a specific configuration are given below.

[0141] Considering a large-scale MIMO-OFDM communication system, the system parameters are configured as follows: number of base station antennas. carrier frequency Subcarrier spacing Number of subcarriers Circular prefix length Number of effective subcarriers oversampling factor For ease of description, the alternating iterative beam structure channel estimation is called a-BSCE, and the generalized approximate message passing algorithm is called GAMP.

[0142] Figure 4This paper demonstrates the performance of a-BSCE's normalized mean square error (RMSE) compared to other channel estimation algorithms under different signal-to-noise ratios (SNRs). The number of iterations for a-BSCE was set to 10, while the number of iterations for GAMP was set to 200. Without using a low-complexity implementation, a-BSCE's performance is almost identical to MMSE at low SNRs, and it remains close to MMSE's performance at high SNRs, even outperforming GAMP.

[0143] Figure 5 The relationship between the NMSE performance of the channel estimation algorithm and the number of iterations is shown at signal-to-noise ratios (SNRs) of 0 dB and 20 dB. When the SNR is low, a-BSCE achieves the performance of MMSE in only 3 iterations, while GAMP requires dozens of iterations. When the SNR is high, the results show that a-BSCE achieves better performance than GAMP (50 iterations) and approaches the performance of MMSE in only 5 iterations. Furthermore, Figure 5 The convergence of a-BSCE under orthogonal grouping was also proven.

[0144] Figure 6 The NMSE performance of a-BSCE implemented with low complexity is compared at different thresholds. Simulation results show that when an efficient implementation is used, the performance of a-BSCE is close to that of the implementation without low complexity, especially when the number of iterations is small.

[0145] Figure 7 The relationship between the total complexity of the channel estimation algorithm and the number of user terminals is shown, where the average number of non-zero elements in the two-dimensional beam domain channel vector for each user terminal is set to 40. The number of iterations for a-BSCE is set to 5, and the number of iterations for GAMP and s-IGA is set to 50. Furthermore, and The complexity of a-BSCE is significantly lower than that of MMSE and GAMP.

[0146] The large-scale MIMO-OFDM alternating iterative beam structure channel estimation method proposed in this embodiment utilizes the sparsity characteristics of the large-scale MIMO-OFDM channel in the space-frequency dual beam domain and the structural features of the channel to design the channel estimator. The base station designs a corresponding low-dimensional beam domain estimator for each channel element in the beam domain to be estimated. The resulting beam structure channel estimator can significantly reduce the complexity of channel estimation on the base station side while ensuring channel estimation performance.

[0147] Based on the same inventive concept, an embodiment of the present invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it implements the channel estimation method based on alternating iterative beam structure in the large-scale MIMO-OFDM system.

[0148] In a specific implementation, the device includes a processor, a communication bus, a memory, and a communication interface. The processor can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. The communication bus may include a path for transmitting information between the aforementioned components. The communication interface, using any transceiver-like device, is used for communicating with other devices or communication networks. The memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), read-only optical disc (CD-ROM) or other optical disc storage, disk storage media, or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. The memory can exist independently and be connected to the processor via a bus. The memory can also be integrated with the processor.

[0149] The memory stores application code that executes the present invention and is controlled by a processor. The processor executes the application code stored in the memory to implement the channel acquisition method provided in the above embodiments. The processor may include one or more CPUs, or multiple processors, each of which may be a single-core processor or a multi-core processor. Here, "processor" may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0150] Based on the same inventive concept, this invention discloses a large-scale MIMO-OFDM communication system, including a base station and multiple user terminals. The base station is used to generate a dual-beambase channel model and to perform pilot scheduling for each user using statistical channel information. Each user sends an adjustable phase-shift pilot as a pilot signal to the base station in the uplink. The base station obtains channel state information using the received pilot signals according to the channel estimation method based on the alternating iterative beam structure.

[0151] In the embodiments provided in this application, it should be understood that the disclosed methods can be implemented in other ways without departing from the spirit and scope of this application. The current embodiments are merely exemplary examples and should not be considered limiting, nor should the specific content given limit the purpose of this application. For example, some features may be omitted or not performed.

[0152] The technical means disclosed in this invention are not limited to those disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications are also considered within the scope of protection of this invention.

Claims

1. A channel estimation method based on alternating iterative beam structure in a large-scale MIMO-OFDM system, characterized in that, include: The non-zero elements in the equivalent double beam domain channel are orthogonally grouped according to the modulo result of the oversampling factor based on the direction cosine index and time delay index of their corresponding beams. It receives pilot sequences sent by each user and converts the space-frequency domain received signal to the beam domain through beam conversion; For elements within the same group, the received signal in the beam domain is extracted and estimated in parallel based on the set of interference elements with cross characteristics corresponding to the non-zero elements in the equivalent double beam domain channel of each user. For elements in different groups, an alternating iterative processing method is adopted to obtain the estimated value of each beam domain channel element through iteration. The beam domain channel of each user is recovered based on the equivalent beam set of each user, and the space-frequency domain channel of each user is reconstructed based on the double beam base channel model.

2. The channel estimation method based on alternating iterative beam structure according to claim 1, characterized in that, The dual-beambase channel model is represented as the product of a dual-beam matrix and a dual-beamdomain channel vector. The dual-beam matrix consists of a selected set of space-frequency domain direction vectors, each direction vector corresponding to a direction cosine and a time delay sampling grid, which is called a dual beam. The product operation of the dual-beam matrix or its conjugate transpose with the vector is implemented by fast Fourier transform. The beam transformation is to use the dual-beam matrix to convert the space-frequency domain signal vector into a beamdomain signal vector.

3. The channel estimation method based on alternating iterative beam structure according to claim 1, characterized in that, Each user transmits a pilot sequence that is adjustable phase-shift pilot. By assigning pilot sequences with different phase-shift factors to different users, the equivalent double beam domain channels of each user do not overlap in the beam domain. The equivalent beam set of each user refers to the set of double beams corresponding to the non-zero elements in the user's equivalent double beam domain channel. The equivalent double beam domain channel is formed by combining the phase-shift pilots of each user with the double beam domain channels of each user.

4. The channel estimation method based on alternating iterative beam structure according to claim 1, characterized in that, When orthogonally grouping the non-zero elements in the equivalent double beam domain channel, the direction cosine index and time delay index of the beam corresponding to the elements in the same group have the same modulo value of the oversampling factor.

5. The channel estimation method based on alternating iterative beam structure according to claim 1, characterized in that, The set of interference elements with cross-shaped characteristics for each non-zero beam domain channel element consists of elements that have the same direction cosine or time delay as the current element and whose interference to the current element is greater than a preset threshold.

6. The channel estimation method based on alternating iterative beam structure according to claim 1, characterized in that, Each estimator for a non-zero element of an equivalent double beam domain channel has a structure of a double beam matrix multiplied by a beam domain estimator. Based on the double beam matrix, the beam set corresponding to the element, the prior mean, and the prior variance, a corresponding beam domain estimator is designed to obtain estimates of the posterior mean and posterior variance. The beam domain estimator is obtained by solving an optimization problem that minimizes the mean square error.

7. The channel estimation method based on alternating iterative beam structure according to claim 1, characterized in that, The method of alternating iterative processing of elements in different groups is as follows: beam domain estimators are designed in parallel for elements in the same group to estimate their posterior mean and posterior variance. Once all elements in the same group have been estimated, the prior mean and prior variance are updated using the obtained posterior mean and posterior variance. Then, the next group of elements is estimated, and this process is repeated until the iteration converges.

8. The channel estimation method based on alternating iterative beam structure according to claim 7, characterized in that, At the start of the iteration, the prior mean is initialized to zero, and the prior variance is initialized to the beam domain statistical channel information. When updating the prior information using the posterior mean and posterior variance, the prior mean of the currently estimated element itself remains zero, and the prior variance remains the beam domain statistical channel information.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements the channel estimation method based on alternating iterative beam structure according to any one of claims 1-8.

10. A large-scale MIMO-OFDM communication system, comprising a base station and multiple user terminals, characterized in that, Each user terminal sends pilot signals to the base station in the uplink; the base station uses the received pilot signals to obtain channel state information according to the channel estimation method based on alternating iterative beam structure according to any one of claims 1-8.

Images