A dynamic air interface matrix fast signal processing method based on a MIMO system

Abstract

A fast signal processing method for dynamic air interface matrix based on MIMO systems includes the following steps: Step 1: Update the core matrix by acquiring the dynamic channel state information of users on the air interface in the MIMO system, identify three dynamic scenarios: user access, user channel state update, and user disconnection, and determine the number of dynamic users; Step 2: Based on the three dynamic scenarios, construct a unified computational framework for the core matrix and select the corresponding algorithm response; Step 3: Calculate and obtain the inverse of the dynamic regularized matrix of the user based on the signal processing steps of channel estimation, precoding, and channel equalization; Step 4: Based on the obtained inverse of the regularized matrix and the corresponding input information, call the closed-form solution of the MMSE algorithm to output the channel estimation, precoding, and channel equalization results. This invention improves the real-time processing capability and resource utilization efficiency of the system, achieving better performance with lower system overhead and lower computational complexity.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and more specifically to a fast signal processing method for a dynamic air interface matrix based on a MIMO system. Background Technology

[0002] In MIMO wireless communication systems, channel estimation, precoding, and channel equalization are the three core modules, all of which can be modeled and optimized under the MMSE criterion. However, existing solutions often design separate algorithms for each module, lacking a unified computational framework across different scenarios. Furthermore, current technologies fail to effectively address the computational redundancy caused by user dynamics, thus limiting the efficient scalability of the system.

[0003] In scenarios involving dynamic user access / update / disconnection, traditional MMSE algorithms require complete reconstruction of the channel response matrix. and its Gram matrix construction core operator .

[0004] In scenarios involving dynamic user access / update / disconnection, the traditional MMSE algorithm requires a new matrix inversion process: The complexity of matrix inversion This leads to a surge in computational latency, making it impossible to meet the real-time requirements of the air interface. Furthermore, when only a small number of users' channel response matrices change dynamically, it results in a large amount of redundant computation.

[0005] Existing technologies employ independent processing procedures and recalculate channel estimation, precoding, and channel equalization (MMSE), as well as user state changes (new user access, CSI update, user churn). User access: Obtain the new channel matrix through column augmentation, reconstruct the Gram matrix, recalculate the inverse of the regularized Gram matrix according to the MMSE algorithm, and obtain the solution to the optimization problem through the closed-form solution of MMSE; User update: The new channel matrix is ​​obtained through column update, the Gram matrix is ​​reconstructed, the inverse of the regularized Gram matrix is ​​recalculated according to the MMSE algorithm, and the solution to the optimization problem is obtained through the closed-form solution of MMSE. User disconnection: Obtain the new channel matrix by column reduction, reconstruct the Gram matrix, recalculate the inverse of the regularized Gram matrix according to the MMSE algorithm, and then obtain the solution to the optimization problem through the closed-form solution of MMSE.

[0006] However, the existing technology has the following drawbacks; The MMSE algorithm for channel estimation, precoding, and channel equalization lacks a unified computational framework, making it impossible to achieve algorithm acceleration design across different scenarios; and it suffers from high computational redundancy in scenarios with dynamic user and channel states.

[0007] In the three dynamic scenarios of user access, user update, and user disconnection, all three algorithms update the channel response matrix by column augmentation, then reconstruct the Gram matrix, and then perform regularized Gram matrix inversion. Since only a small number of users' channel responses change dynamically, users whose responses have not changed are also recalculated during the reconstruction and inversion process, resulting in a large amount of redundant calculation, wasting computing resources and reducing algorithm efficiency.

[0008] With the widespread adoption of IoT and industrial automation, systems need to support the dynamic access (e.g., bursty connection requests), real-time movement (e.g., high-speed vehicle terminals), and instantaneous disconnection (e.g., device hibernation) of massive numbers of user devices, resulting in highly time-varying Channel State Information (CSI). This dynamism leads to two major bottlenecks: First, the high complexity prevents timely response, as core modules such as channel estimation, precoding, and equalization all rely on high-dimensional matrix operations. Traditional methods require global matrix reconstruction when users change, resulting in excessive computational complexity that fails to meet millisecond-level response requirements over the air interface. Second, the algorithms lack generalization ability: existing optimization schemes (e.g., block inversion, Neumann series approximation) are mostly designed for static scenarios, lacking a unified mathematical framework for the entire "access-update-disconnection" scenario, leading to computational redundancy during cross-scenario switching. Summary of the Invention

[0009] To overcome the shortcomings of existing technologies, this invention proposes a fast signal processing method for dynamic air interface matrices based on MIMO systems. This method avoids the repeated construction of Gram matrices through multiplexing and reduces redundant calculations in regularized Gram inversion operations by utilizing fast matrix computation, thereby achieving efficient acceleration under a unified computing framework. This invention can significantly reduce the computational complexity of channel estimation, precoding, and channel equalization processes in scenarios such as dynamic user access, disconnection, and changes in channel conditions, improving the system's real-time processing capabilities and resource utilization efficiency. It provides feasible technical support for the engineering application of MIMO systems in high-speed mobile and large-scale user scenarios, achieving better performance with lower system overhead and lower computational complexity.

[0010] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A fast signal processing method for dynamic air interface matrix based on MIMO system includes the following steps; Step 1: By acquiring the dynamic channel state information of users on the air interface in the MIMO system, the core matrix is ​​updated to determine three types of dynamic scenarios: user access, user channel state update, and user disconnection events, and the number of dynamic users is determined. This step reflects the modeling and response to the dynamic air interface environment. Step 2: Based on the three types of dynamic scenarios, construct a unified computational framework for the core matrix and select the corresponding algorithm response; Step 3: Calculate and obtain the post-regularization matrix based on the signal processing methods of channel estimation, precoding, and channel equalization for user dynamics (user access, user channel state update, and user disconnection). The reverse ; Step 4: Based on the inverse of the obtained regularization matrix and the corresponding input information (such as pilot matrix, received signal, channel matrix), call the MMSE algorithm to output the channel estimation, precoding and channel equalization results in closed form.

[0011] The specific method steps for step 1 are as follows: The base station acquires the instantaneous channel state information (iCSI) of the user. When the user changes dynamically, the channel state information matrix acquired by the base station will also change accordingly, which is called dynamic channel state information. The actual system parameters are obtained, and the dynamic user scenario and number of dynamic users are determined based on the actual system parameter information. It is worth noting that the proposed fast signal processing method for dynamic air interface matrix is ​​applicable to three dynamic scenarios: user access, user channel state update, and user disconnection, with three types of algorithms: channel estimation, channel equalization, and precoding. To avoid redundancy, only three scenarios are introduced as examples: channel estimation algorithm in the user access column augmentation scenario, channel equalization algorithm in the user / channel state update column update scenario, and precoding algorithm in the user disconnection column reduction scenario. For user access augmentation scenarios, channel estimation algorithms obtain pilot matrices. New pilot Let the total pilot matrix after dynamics be denoted as After the user connects, the system receives a signal. The number of new users is determined based on the number of user pilot access signals received by the base station. Before user access , Cholesky decomposition matrix and its reverse ; For user / channel state update column update scenarios, the channel equalization algorithm obtains the original channel matrix. Update the channel matrix ,, , , Receive signal ; For user offline queue reduction scenarios, the precoding algorithm obtains the channel matrix. Subtract the channel matrix before the user's dynamic operation to obtain the non-channel matrix. The row and column indices are the same as the column reduction indexes of the user channel matrix. , , ; For three types of dynamic user scenarios, the channel estimation, precoding, and channel equalization algorithms are adapted by replacing parameters through a unified computing framework.

[0012] The specific steps of step 2 are as follows: For user access column augmentation scenarios, the channel estimation algorithm calculates the augmentation block; like at this time It is a scalar; like at this time For a matrix, further calculations are needed. Cholesky decomposition matrix and its reverse ; ; Channel equalization algorithm for user / channel state update column update scenarios; First calculate ,pass Use the index on the non-zero column to obtain the index for the updating user. Then, by looping from 1 to... Assemble auxiliary matrix by column .

[0013] Specifically: , , , , , Obtain the auxiliary matrix ; For users leaving the network, the precoding algorithm is used to reduce the number of users leaving the network. When leaving the network, construct the permutation matrix. ,in For the first A column vector with 1 element and 0 elements; calculate ,Depend on Assemble the block matrix .

[0014] The specific steps of step 3 are as follows: For the channel estimation algorithm in the user access column augmentation scenario, recursive calculation is performed based on the Cholesky column augmentation scheme: ; Assembly Matrix ; calculate ; For the channel equalization algorithm in the user / channel state update column update scenario, it is calculated based on the Sherman-Morrison formula: ; Precoding algorithm calculation for user offline column deletion scenario: ,Depend on Assemble the block matrix , Calculation based on Shur complement theory .

[0015] The specific method steps for step 4 are as follows: For user access augmentation scenarios, the channel estimation algorithm is calculated using the closed-form solution of MMSE: ; The channel estimation matrix after the system undergoes dynamic changes is obtained; For user / channel state update column update scenarios, the channel equalization algorithm is calculated using the closed-form solution of MMSE: ; The estimated transmitted signal vector is obtained after the system undergoes dynamic changes; For scenarios involving user offline column reduction, the precoding algorithm is calculated using the closed-form solution of MMSE: ; The precoded vector after the system dynamically changes is obtained.

[0016] The beneficial effects of this invention are: This invention focuses on the scenario of dynamic user changes in large-scale MIMO systems. It establishes a unified computational framework for the MMSE algorithm, which has three core functions: channel estimation, precoding, and channel equalization, providing a model foundation for rapid algorithm design.

[0017] Within a unified computing framework, for three typical dynamic scenarios—user access, updates, and disconnection—a dynamic fast algorithm with theoretical guarantees was developed in step three, based on Cholesky column augmentation, Sherman-Morrison formula, and Shur complement theory, respectively.

[0018] In steps two and three, redundant matrix reconstruction calculations are avoided by reusing the Gram matrix before dynamic changes. The repeated calculations of the regularized Gram inversion part are reduced by fast matrix calculation, thereby accelerating the algorithm under a unified computing framework and improving the algorithm efficiency.

[0019] The algorithm proposed in this invention can be implemented in all three scenarios while maintaining the same estimation accuracy as traditional methods. The complexity is reduced by a factor of two. This invention provides an efficient solution for real-time signal processing in dynamic user environments, and its unified framework and dynamic scenario design concept can be extended to other wireless communication scenarios. Attached Figure Description

[0020] Figure 1 This diagram illustrates a performance comparison between a specific embodiment of the present invention in a user access scenario and existing methods.

[0021] Figure 2 This diagram illustrates a comparison of the computational overhead of a specific embodiment of the present invention in a user access scenario with existing methods.

[0022] Figure 3 This diagram illustrates a performance comparison between a specific embodiment of the present invention in a user update scenario and existing methods.

[0023] Figure 4 This diagram illustrates a comparison of the computational overhead of a specific embodiment of the present invention in a user update scenario with existing methods.

[0024] Figure 5 This is a schematic diagram comparing the performance of a specific embodiment of the present invention in a user offline scenario with existing methods.

[0025] Figure 6 This diagram illustrates a comparison of computational overhead between a specific embodiment of the present invention in a user offline scenario and existing methods.

[0026] Figure 7 This is a simplified flowchart illustrating a specific embodiment of the present invention.

[0027] Figure 8 This is a detailed flowchart illustrating a specific embodiment of the present invention. Detailed Implementation

[0028] The present invention will now be described in further detail with reference to the accompanying drawings.

[0029] Terminology Explanation: Massive MIMO systems Definition: A wireless communication architecture in which the base station is equipped with dozens to hundreds of antennas, simultaneously serving multiple user equipments (UEs) through spatial multiplexing technology, significantly improving system capacity and spectrum efficiency. The application carrier of the dynamic air interface algorithm proposed in this invention.

[0030] Air Interface Computational Bottleneck Definition: In large-scale MIMO systems, insufficient real-time channel data processing capacity due to the dynamic access of massive numbers of users manifests as high-dimensional matrix operation delays exceeding the system's tolerance threshold, specifically in channel estimation, precoding, and equalization. This invention aims to address this bottleneck with a fast matrix computation method.

[0031] Channel Response Matrix Definition: A complex matrix characterizing the transmission properties of a wireless channel. Where N is the number of base station antennas, M is the number of users, and the matrix elements are... This represents the channel gain from the j-th user to the i-th antenna. It is the core input data for algorithm optimization.

[0032] Minimum Mean Square Error (MMSE) Algorithm Definition: A signal processing technique based on statistical optimality criteria, which achieves optimal estimation of channel parameters or signals by minimizing the mean square error (MSE) between the estimated and true values. This invention improves the basic algorithm framework in the channel estimation, precoding, and channel equalization modules.

[0033] Dynamic pilot signal Definition: Known reference signals transmitted on demand by user equipment, and the base station reconstructs the channel response matrix in real time by analyzing their space-time-frequency domain characteristics. This supports the channel estimation module in handling critical signaling for dynamic user access.

[0034] Multi-dimensional Beamforming Matrix Definition: Complex weight matrix generated by precoding techniques By spatially weighting the transmitted signal, the main lobe of the beam is aligned with the target user, and the null is directed towards the interference source. The transmitter optimization matrix in this invention is based on fast matrix calculation.

[0035] Inter-User Interference (IUI) Definition: In a multi-user MIMO system, crosstalk between signals from different users within the same resource block caused by the non-orthogonality of the spatial channel. It is an object that the precoding and channel equalization modules need to jointly suppress.

[0036] Channel equalization Definition: The signal processing procedure at the receiving end that performs inverse channel response compensation on the distorted signal to recover the original transmitted signal. This invention presents an uplink interference suppression module based on the fast MMSE algorithm.

[0037] Spatial Degrees of Freedom Definition: The available antenna array dimensional resources in a massive MIMO system, the number of which is equal to the number of base station antennas. It is used to separate multi-user signals or suppress interference.

[0038] Matrix Fast Computation Definition: A low-complexity numerical method designed for high-dimensional matrix operations (such as inversion, decomposition, and multiplication), utilizing the sparsity, low rank, or structural characteristics of the channel matrix to reduce computational latency. The core innovation of this invention is its application in accelerating the implementation of MMSE-type algorithms.

[0039] like Figure 7 , Figure 8 As shown, the purpose of this invention is to provide a fast signal processing method for dynamic air interface matrix based on MIMO system, which solves the problems existing in the prior art and achieves better performance with lower system overhead and lower computational complexity.

[0040] Building a unified computing framework: First, a specific MIMO system model is given, and parameters are defined to explain the parameter acquisition step in the technical solution. Consider a large-scale multi-user MIMO system, where the base station (BS) in the cell is equipped with... One antenna, and through time division duplex (TDD) to provide the same time and frequency resources. Single-antenna user service.

[0041] In a time-division duplex massive MIMO system, the base station uses the pilot signals transmitted by the user in the uplink (UL) to estimate the uplink channel, and then uses the channel reciprocity in the TDD system to obtain the downlink (DL) channel state information.

[0042] The ULBS received signal is: The signal received by the DL user is: ,in For the channel matrix, For the precoding matrix, , These are the transmission signals for UL and DL, respectively. , These are additive white Gaussian noise matrices of UL and DL, respectively, with elements following the... .

[0043] The base station senses the channel state through uplink training, that is, it receives pilot signals transmitted by users. Then receive the signal At this point, channel estimation can be transformed into equivalent MMSE optimization:

[0044] Its closed-form solution is: .

[0045] The MMSE estimator makes full use of the channel statistical characteristics and suppresses pilot pollution and noise interference by using the regularization inverse of the pilot signal covariance matrix, thus achieving the optimal trade-off between pilot pollution and noise.

[0046] Based on estimated channel The channel equalization MMSE optimization problem can be expressed by the following formula:

[0047] Its closed-form solution is: The MMSE detector utilizes the regularized inverse of the channel Gram matrix to achieve spatial separation of interfering users, and achieves an optimal balance between interference and noise enhancement between users through spatial interference alignment.

[0048] The MMSE precoding algorithm designs the precoding matrix based on minimizing the mean square error, thereby minimizing the error between the receiver and transmitter signals. The optimization problem can be expressed as: ; Its closed-form solution is: ; in The power constraint factor is used to balance signal leakage suppression and transmit power efficiency, so that the precoded beam achieves Pareto optimality between user signal leakage and transmit power.

[0049] For establishing a closed-form solution objective of the MMSE algorithm under a unified framework, and defining the symbol and optimization paradigm, the MMSE algorithm for large-scale MIMO systems can be unified into the following optimization paradigm: , The unified framework for closed-form solutions is: ; The parameter mapping tables are shown in Table 1 and Table 2 below. Establishing a unified framework ensures that the algorithm response implementation forms in steps two and three of the technical solution are consistent.

[0050]

[0051]

[0052] This unified framework provides a modular theoretical tool for the design of large-scale MIMO systems, enabling rapid portability and performance analysis of different functional modules through parameter permutation. Through simple induction and proof, it can be seen that the closed-form solutions of the three algorithms share the following core mathematical characteristics: matrix condition number improvement through regularization terms to avoid ill-conditioned inversion; all are based on the core matrix. or The core operator is constructed using the Gram matrix; modules in the closed-form solution satisfy unitary invariance. Based on these properties, a dynamic air interface algorithm based on a unified computing framework was developed.

[0053] Dynamic air interface algorithm based on a unified computing framework This section corresponds to the specific signal processing flow in steps two and three of the technical solution. Considering the dynamic user scenario of a large-scale MIMO system, a unified parameter is defined: It is the instantaneous channel state matrix (considered as a pilot matrix in channel estimation algorithms). For regularized Gram matrices, for The Cholesky decomposition matrix satisfies .

[0054] When a new user connects, the channel matrix expands to The corresponding Gram matrix is ​​updated as follows: ; If the original system's Cholesky decomposition Given that the updated decomposition matrix can be obtained recursively: , ; in Furthermore, the formula for updating its inverse matrix is: .but .

[0055] When multiple new users connect, the channel matrix expands to The corresponding Gram matrix is ​​updated as follows: ; If the original system's Cholesky decomposition It is known that at this point, it is necessary to first calculate the inverse of the augmented block matrix and its Choleskey decomposition matrix;

[0056] ; The updated decomposition matrix can then be obtained through recursive calculation: ; The formula for updating the inverse matrix is: ,but .

[0057] When the k-th user updates or the user channel updates, the k-th column of the channel changes from... Updated to At that time, the Gram matrix changes as follows: from Change to Expanding on this, we can write: It can be derived and simplified to obtain: , in , At this point, the formula for updating the inverse matrix can be obtained by using the Sherman-Morrison-Woodbury formula:

[0058] When there are multiple user updates or multiple user channel updates, the Gram matrix changes as follows: From Change to First, we need to calculate and through Use the index on the non-zero column to obtain the index for the updating user. .

[0059] Furthermore, a matrix can be assembled by referring to the above-mentioned individual user cases. That's it, at this time Each pair of columns can be considered as information changed by a single user. Furthermore, since matrix multiplication operations are performed repeatedly at the intersections of matrix indices, this needs to be eliminated. The general form can be written as:

[0060]

[0061] The formula for updating the inverse matrix can be obtained by using the Sherman-Morrison-Woodbury formula:

[0062] When the k-th user leaves the network, the k-th user... With column channels removed, the Gram matrix changes as follows: From Change to Assuming It is known that if the user at this time To go offline, we first construct the permutation matrix. ,in For the first A unit column vector with elements all equal to 1 is obtained through symmetric permutation in the following form: ; in:

[0063]

[0064] Right now The matrix represents the users The reduced regularized Gram block matrix, This represents the relevant block matrix for deleting users. Based on the assumption... Given that, we can rearrange them using the same symmetric permutation as described above, as follows: ; because Both are symmetric positive definite matrices, thus invertibility is guaranteed. By the Shur complement formula: ; in yes Schur complement is defined as Compare From the block-based form, we can obtain: , .

[0065] At this point, the inverse of the target matrix can be obtained from... get.

[0066] For multiple users When leaving the network, simply construct a permutation matrix of the following form: , and set The inverse of the target matrix can be obtained from... get.

[0067] Algorithm Steps After completing the above steps, the final target output can be obtained by calling the MMSE closed-form solution. Therefore, this section summarizes all the processing steps. For the three dynamic scenarios of user access, user update, and user disconnection, the algorithm step flowcharts are given for channel estimation, precoding, and channel equalization, respectively. The core algorithm steps are as given in the previous section.

[0068]

[0069]

[0070]

[0071] Simulation content Numerical results for the dynamic MMSE algorithm are presented under three scenarios: user access column augmentation for channel estimation, user update column correction for precoding, and user off-network column pruning for channel equalization. Performance is compared with that of the conventional MMSE algorithm. Considering the OFDM system includes... Each subcarrier (each subcarrier channel exhibits flat fading) is modeled in numerical experiments, with each subcarrier being processed independently. The base station is equipped with... One antenna, for the community Single-antenna user service. These numerical experimental results are in The results were obtained and averaged across 1000 Monte Carlo trials.

[0072] For user access augmentation scenarios involving channel estimation, consider multiple users sending pilot signals to the base station in the uplink of wireless communication. The base station estimates channel state information by receiving these signals, and the pilot length is taken as... . Figure 1 The results show the normalized Frobenius norm error of the conventional MMSE algorithm and the dynamic MMSE algorithm under the condition of augmented user access column in channel estimation. This indicates that both algorithms perform well at a 20dB SNR, and the dynamic algorithm does not introduce new errors. Figure 2 The results show a comparison of the running time of the conventional MMSE algorithm and the dynamic MMSE algorithm under the condition of augmented user access in channel estimation. It can be seen that the complexity of the conventional MMSE algorithm increases rapidly with the number of user accesses due to a large number of repeated calculations, while the dynamic MMSE algorithm is more gradual in the case of column augmentation, and its running time is consistently lower than that of the conventional MMSE.

[0073] For the precoded user update column update scenario, this invention considers that in the downlink of wireless communication, the base station performs precoding based on the estimated channel state information and calculates the bit error rate at the receiver. Figure 3 The results demonstrate the bit error rate performance of the receiver when using the traditional MMSE algorithm and the dynamic MMSE algorithm for precoding in a user update scenario with a signal-to-noise ratio of 10dB. Both algorithms exhibit excellent bit error rate performance, and the dynamic MMSE algorithm does not introduce additional errors, indicating that it has a significant advantage in maintaining system performance stability. Figure 4 This shows a comparison of runtime between the traditional MMSE algorithm and the dynamic MMSE algorithm for precoding in a user update scenario. In this case, the dynamic MMSE algorithm performs better when the number of users updating is less than the total number of users. At that time, the running time is shorter than that of the conventional MMSE algorithm.

[0074] For the scenario of reducing the number of users leaving the network due to channel equalization, it is considered that in the downlink of wireless communication, the base station performs signal detection and calculates the bit error rate by estimating the channel state information. Figure 5 The results show signal detection using the conventional MMSE algorithm and the dynamic MMSE algorithm, and the bit error rate (BER) calculation under the condition of user off-network column reduction. This indicates that both algorithms perform well at a 20 dB SNR, and the dynamic algorithm does not introduce new errors. Figure 6 The results show a comparison of the running time of the conventional MMSE algorithm and the dynamic MMSE algorithm for signal detection under the condition of user offline and column reduction. In this case, due to column reduction, the running time of both algorithms gradually decreases, but the running time of the dynamic MMSE algorithm is consistently lower than that of the conventional MMSE algorithm. In summary, the numerical results obtained through the above experiments fully demonstrate the effectiveness of the proposed algorithm, and also verify its stability and speedup advantage compared to traditional algorithms.

[0075] Complexity Analysis Algorithm 1 in single-user access Step 2 calculation need Complexity, in step 3 Matrix update only requires When multiple users access ( Step 2 calculation need Complexity, in step 3, based on the Cholesky column augmentation scheme, Matrix updates and calculation of its Cholesky decomposition require The total complexity is In contrast, traditional methods require reconstruction of the matrix. And perform matrix inversion, the total complexity reaches The numerical stability of the proposed algorithm is guaranteed by the Cholesky update mechanism, consistent with the error bound of traditional methods. Based on the scheme design in steps 2 and 3, it brings... The complexity gain.

[0076] The initialization and indexing complexity in step 1 of Algorithm 2 is O(n). Incremental matrix The computational complexity is Step 2, the process of constructing the inner-rank matrix in the loop, consumes a complex time complexity. Step 3, updating the inverse matrix, is achieved using the Sherman-Morrison-Woodbury formula with a complexity of O(n). The complexity of reconstructing the regularized Gram matrix using traditional methods is... and the complexity of direct inversion The total complexity reaches The total complexity of the proposed algorithm is The algorithm preserves the original matrix decomposition structure through incremental updates, and its numerical stability is guaranteed by the Woodbury identity. The upper bound of error propagation is consistent with traditional methods. Based on the scheme design in steps 2 and 3, it brings... The complexity gain.

[0077] In step 2 of Algorithm 3, the complexity of permutation matrix generation and matrix rearrangement is O(n). In the calculation of the block inverse matrix The required complexity is In step 3, the submatrix is ​​calculated based on the Shur complement theory. The complexity of consumption is The total complexity is .

[0078]

[0079] The complexity of reconstructing the regularized Gram matrix using traditional methods is O(n). The time complexity of directly inverting is... Algorithm 3 avoids matrix reconstruction through block matrix inverse updates. Its numerical stability is guaranteed by the block Schur complement theory, and its error propagation bound is consistent with traditional methods. Based on the scheme design in steps 2 and 3, it brings... The complexity gain.

Claims

1. A fast signal processing method for dynamic air interface matrix based on MIMO system, characterized in that, Includes the following steps; Step 1: By acquiring the dynamic channel state information of users on the air interface in the MIMO system, update the core matrix, determine three types of dynamic scenarios: user access, user channel state update, and user disconnection, and determine the number of dynamic users. Step 2: Based on the three types of dynamic scenarios, construct a unified computational framework for the core matrix and select the corresponding algorithm response; Step 3: Calculate and obtain the user dynamic post-regularization matrix based on the signal processing steps of channel estimation, precoding, and channel equalization. The reverse ; Step 4: Based on the inverse of the obtained regularization matrix and the corresponding input information, call the MMSE algorithm in closed form to output the channel estimation, precoding and channel equalization results.

2. The fast signal processing method for dynamic air interface matrix based on MIMO system according to claim 1, characterized in that, The specific method steps for step 1 are as follows: The base station acquires the instantaneous channel state information (iCSI) of the user. When the user changes dynamically, the channel state information matrix acquired by the base station will also change accordingly, which is called dynamic channel state information. Obtain actual system parameters and determine the dynamic user scenario and number of dynamic users based on the actual system parameter information; the fast signal processing method for dynamic air interface matrix is ​​applicable to three dynamic scenarios: user access, user channel state update and user disconnection, with three types of algorithms: channel estimation, channel equalization and precoding; channel estimation algorithm under user access column augmentation scenario, channel equalization algorithm under user / channel state update column update scenario, and precoding algorithm under user disconnection column deletion scenario. For user access augmentation scenarios, channel estimation algorithms obtain pilot matrices. New pilot Let the total pilot matrix after dynamics be denoted as After the user connects, the system receives a signal. The number of new users is determined based on the number of user pilot access signals received by the base station. Before user access , Cholesky decomposition matrix and its reverse ; For user / channel state update column update scenarios, the channel equalization algorithm obtains the original channel matrix. Update the channel matrix ,, , , Receive signal ; For user offline queue reduction scenarios, the precoding algorithm obtains the channel matrix. Subtract the channel matrix before the user's dynamic operation to obtain the non-channel matrix. The row and column indices are the same as the column reduction indexes of the user channel matrix. , , ; For three types of dynamic user scenarios, the channel estimation, precoding, and channel equalization algorithms are adapted by replacing parameters through a unified computing framework.

3. The fast signal processing method for dynamic air interface matrix based on MIMO system according to claim 2, characterized in that, The specific method steps for step 2 are as follows: For user access column augmentation scenarios, the channel estimation algorithm calculates the augmentation block; like at this time It is a scalar; like at this time For a matrix, further calculations are needed. Cholesky decomposition matrix and its reverse ; ； Channel equalization algorithm for user / channel state update column update scenarios; First calculate ,pass Use the index on the non-zero column to obtain the index for the updating user. Then, by looping from 1 to... Assemble auxiliary matrix by column .

4. The fast signal processing method for dynamic air interface matrix based on a MIMO system according to claim 3, characterized in that, Specifically: , , , , , Obtain the auxiliary matrix ; For users leaving the network, the precoding algorithm is used to reduce the number of users leaving the network. When leaving the network, construct the permutation matrix. ,in For the first A column vector with 1 element and 0 elements; calculate ,Depend on Assemble the block matrix .

5. The fast signal processing method for dynamic air interface matrix based on MIMO system according to claim 4, characterized in that, The specific steps of step 3 are as follows: For the channel estimation algorithm in the user access column augmentation scenario, recursive calculation is performed based on the Cholesky column augmentation scheme: ； Assembly Matrix ; calculate ; For the channel equalization algorithm in the user / channel state update column update scenario, it is calculated based on the Sherman-Morrison formula: ； Precoding algorithm calculation for user offline column reduction scenario: ,Depend on Assemble the block matrix , Calculation based on Shur complement theory .

6. The fast signal processing method for dynamic air interface matrix based on MIMO system according to claim 5, characterized in that, The specific method steps for step 4 are as follows: For user access augmentation scenarios, the channel estimation algorithm is calculated using the closed-form solution of MMSE: ； The channel estimation matrix after the system undergoes dynamic changes is obtained; For user / channel state update column update scenarios, the channel equalization algorithm is calculated using the closed-form solution of MMSE: ； The estimated transmitted signal vector is obtained after the system undergoes dynamic changes; For scenarios involving user offline column reduction, the precoding algorithm is calculated using the closed-form solution of MMSE: ； The precoded vector after the system dynamically changes is obtained.

Images