A Design Method for ReIMV-OMP Channel Estimation Based on MIMO OFDM System

Abstract

The present invention discloses a design method for ReIMV - OMP channel estimation based on MIMO OFDM systems. This method proves that a new matrix can be constructed by taking one frame from the IMV model, which contains the effective information for recovering the non - zero positions. After reducing the dimension to the MMV model and further combining with a random Gaussian vector to reduce the dimension to the SMV model, the non - zero positions can be recovered from the simplified SMV model. The simulation results verify the effectiveness of this algorithm, which not only helps to improve the estimation accuracy of the non - zero positions of the channel, but also has better MSE and BER performance than the BOOMP algorithm that also utilizes the common sparsity of MIMO channels.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital signal processing, and provides a design method for ReIMV-OMP channel estimation based on a MIMO OFDM system. Background Art

[0002] With the explosive growth of mobile data, the wireless communication system has an increasing demand for spectrum utilization. However, currently, the number of antennas configured at the base station of the MIMO OFDM system is small, usually only two or four. At present, the performance improvement of the wireless communication system has reached a bottleneck. In order to break through the bottleneck of traditional MIMO technology, the concept of massive MIMO is proposed, that is, more receiving antennas are configured on the base station side to further deeply explore spatial resources and provide services for multiple single-antenna mobile users in the same frequency band within the cell. When the number of antennas approaches infinity, the determinacy of the channel matrix is enhanced, and the transmission power at the mobile user end can be arbitrarily small. However, due to the substantial increase in the number of base station antennas and mobile users, the number of channel parameters to be estimated becomes extremely large, and the wireless channel state becomes more and more complex, which brings new challenges to improving the accuracy of channel estimation in the MIMO OFDM system. Thus, it can be seen that how to improve the channel estimation performance has become one of the key problems to be solved.

[0003] Currently, a frame is taken from the IMV model to construct a new matrix, which contains the effective information for recovering the non-zero positions. After being reduced to the MMV model, it is further combined with a random Gaussian vector and reduced to the SMV model, and the non-zero positions can be recovered from the simplified SMV model. Therefore, we combine this idea with the channel estimation of the MIMO OFDM system, and utilize the common sparsity of the uplink channel of the MIMO OFDM system to propose a sparse channel estimation algorithm based on ReIMV-OMP, thereby reducing the computational complexity. Summary of the Invention

[0004] The object of the present invention is to design and implement a reduction in the complexity of operations, improve the channel estimation performance and spectrum utilization rate, and provide a new design method - a method for designing ReIMV-OMP channel estimation based on a MIMO OFDM system.

[0005] The design method for ReIMV-OMP channel estimation based on a MIMO OFDM system provided by the present invention specifically comprises the following steps:

[0006] First, write the signals received by all antennas at the base station end of the MIMO system in matrix form, and transform the sparse channel estimation problem into a sparse channel reconstruction problem in the theory of compressive sensing;

[0007] Second, combine the sparse signal reconstruction problem with the compressive sensing IMV model. Extract one frame from the measurement matrix Y with extremely high dimensions to construct an MMV model, and then perform dimensionality reduction on it to further simplify it into an SMV model;

[0008] Third, use the OMP algorithm to find the non-sparse positions and estimate the obtained channel matrix;

[0009] (Take the following ReIMV-OMP sparse channel estimation as an example)

[0010] (1) We consider writing the signals received by all antennas at the base station end of the MIMO system in the following matrix form:

[0011] Y = AH + N (1)

[0012] Among them, is the observation matrix, which represents all the signals received at the base station end; is the channel matrix, which represents all the sub-channels through which the signals sent by mobile users reach the base station, is the noise matrix. At this time, the sparse channel estimation problem in the MIMO OFDM system in formula (1) can be transformed into a sparse signal reconstruction problem in compressive sensing theory, that is:

[0013]

[0014] Among them, the symbol |I(H)| represents the number of non-zero rows in matrix H.

[0015] (2) Input the measurement matrix Y, the sensing matrix A, the signal sparsity K, and the threshold σ.

[0016] (3) Calculate the coherence matrix:

[0017]

[0018] (4) Construct matrix V according to the following formula:

[0019]

[0020] Among them, E represents the eigenvector matrix of matrix Q, and λ represents the eigenvalue vector of matrix Q.

[0021] (5) Obtain the linear system of the MMV model:

[0022] V = AU (5) where U = (A T A) -1 A T V.

[0023] (6) Optimize the MMV model with the random Gaussian vector a, and calculate the vectors v and u respectively according to the following formula:

[0024] v = Va (6a)

[0025] u = Ua (6b)

[0026] (VII) Obtain the linear system of the SMV model:

[0027] v = Au (7)

[0028] (VIII) Initialization: Residual r t-1 = v, non - zero position set Iteration number t = 1.

[0029] (IX) Find the column index J with the maximum correlation from the sensing matrix A t :

[0030]

[0031] (X) Update the non - zero position set:

[0032] S t = S t-1 ∪J t (9)

[0033] (XI) Solve the least squares:

[0034]

[0035] (XII) Update the residual:

[0036]

[0037] (XIII) Update iteration t = t + 1; when t ≤ K, return to step 8, otherwise go to the next step.

[0038] (XIV) Estimate the obtained channel matrix as:

[0039]

[0040] The present invention has the following beneficial effects:

[0041] 1. The present invention provides a design method for ReIMV - OMP channel estimation based on the MIMO OFDM system for the first time.

[0042] 2. The present invention takes a frame from the IMV model to construct a new matrix, which contains effective information for restoring non - zero positions. After reducing the dimension to the MMV model, it is further combined with a random Gaussian vector and reduced to the SMV model. The non - zero positions can be restored from the simplified SMV model using the OMP algorithm, saving operation time and improving spectrum utilization.

[0043] 3. The simulation part proves the effectiveness of the algorithm, which not only helps to improve the estimation accuracy of the non-zero positions of the channel, but also has better MSE and BER performance than the BOOMP algorithm that also utilizes the common sparsity of the MIMO channel. Description of the Drawings

[0044] Figure 1 is the flow chart of the algorithm of the present invention;

[0045] Figure 2 is the estimated probability diagram of the non-zero positions of the ReIMV-OMP channel estimation based on the MIMO OFDM system;

[0046] Figure 3 is the channel impulse response amplitude diagram of the ReIMV-OMP channel estimation based on the MIMO OFDM system;

[0047] Figure 4 is the MSE performance comparison diagram of the ReIMV-OMP channel estimation based on the MIMO OFDM system

[0048] Figure 5 is the BER performance comparison diagram of the ReIMV-OMP channel estimation based on the MIMO OFDM system Detailed Implementation Manner

[0049] Example 1:

[0050] The specific steps of the design method of the ReIMV-OMP channel estimation based on the MIMO OFDM system provided by the present invention are as follows:

[0051] First, write the signals received by all antennas at the base station end of the MIMO system in matrix form, and transform the sparse channel estimation problem into a sparse channel reconstruction problem in the compressed sensing theory;

[0052] Second, combine the sparse signal reconstruction problem with the compressed sensing IMV model, extract one frame from the measurement matrix Y of extremely large dimension to construct an MVV model, and then perform dimensionality reduction processing on it to further simplify it into an SMV model;

[0053] Third, use the OMP algorithm to find the non-sparse positions and estimate the obtained channel matrix;

[0054] To verify the effectiveness of the sparse channel estimation design method, computer simulation was carried out on this method.

[0055] Design requirements: In this part, MATLAB software is used to implement the design of ReIMV-OMP channel estimation based on the MIMO OFDM system. The number of subcarriers of the OFDM symbol is 256, the number of transmitting antennas is 1, and the number of receiving antennas is 32, 48, and 64 respectively. The cyclic prefix length is 64. It is assumed that the channel is a frequency-selective slow fading channel, the length of the channel impulse response is 60, the channel is sparse and follows a Rayleigh distribution, and the channel sparsity is 12. The positions of the non-zero elements are set as [2, 13, 21, 24, 29, 33, 41, 42, 43, 53, 54, 60]. The modulation method adopted is 16QAM constellation mapping, the frequency interval of the subcarriers is 15 kHz, the convolutional encoder with a rate of 1 / 2 and a constraint length of 7 [133, 171] is used for encoding at the transmitting end, and the number of inserted pilots is 16. In order to maximize the channel estimation performance, the coherence of the sensing matrix is minimized.

[0056] Step 1: Write the signals received by all antennas at the base station of the MIMO system in matrix form, and transform the sparse channel estimation problem into a sparse channel reconstruction problem in the theory of compressive sensing:

[0057]

[0058] Step 2: Combine the sparse signal reconstruction problem with the compressive sensing IMV model. Take one frame from the measurement matrix Y with a very large dimension to construct an MMV model, and then perform dimensionality reduction on it to further simplify it into an SMV model.

[0059] Step 3: Use the OMP algorithm to find the non-sparse positions and estimate the obtained channel matrix The sparsity is 12 and the number of pilots is 16.

[0060] The estimated probability map of the non-zero positions obtained is as shown in Figure 2 As shown. According to the estimated probability map of the non-zero positions, it can be seen that with the increase of the number of receiving antennas, the estimation accuracy gradually improves. When the number of receiving antennas is 64, the non-zero positions of the channel can be accurately estimated when the signal-to-noise ratio reaches 18 dB. The amplitude diagram of the frequency channel impulse response is as shown in Figure 3 As shown. For the reconstruction of the amplitudes of the sparse channel taps, it can be seen from the figure that the estimation accuracy of the ReIMV-OMP algorithm for the channel impulse response values is slightly higher than that of the BOOMP algorithm. For the convenience of comparison, the amplitudes of the taps of the actual simulation channel are also given in the figure. The MSE performance comparison diagram and the BER performance comparison diagram are as shown in Figure 4 and Figure 5As shown in the figure. To further objectively evaluate the performance of the sparse channel estimation algorithm based on ReIMV-OMP, in this paper, the performance of the two channel estimation methods is evaluated by the mean square error (MSE) of channel estimation and the bit error rate (BER) of the system. When the number of receiving antennas configured on the base station side is 32, 48, and 64, the MSE and BER performance curves are obtained. As the number of receiving antennas configured on the base station increases, the effect of channel estimation will be better.

Claims

1. A design method for ReIMV-OMP channel estimation based on MIMO OFDM system, characterized in that by means of the compressive sensing theory, the IMV model is simplified and dimension-reduced, and combined with the OMP algorithm for channel estimation, thus greatly reducing the complexity of channel estimation and reducing the operation time. This method is carried out according to the following steps: Step 1: We consider writing the signals received by all antennas at the base station end of the MIMO system in the following matrix form: Y = AH + N(1) Among them, is the observation matrix, which represents all signals received at the base station end; is the channel matrix, which represents all sub-channels through which the signals sent by mobile users reach the base station. is the noise matrix. At this time, the sparse channel estimation problem of the MIMO OFDM system in formula (1) can be transformed into the sparse signal reconstruction problem in the compressed sensing theory, that is: where the symbol |I(H)| represents the number of non-zero rows in matrix H; Step 2: Input the measurement matrix Y, the sensing matrix A, the signal sparsity K, and the threshold σ; Step 3: Calculate the coherence matrix: Step 4: Construct matrix V according to the following formula: where E represents the eigenvector matrix of matrix Q, and λ represents the eigenvalue vector of matrix Q; Step 5: Obtain the linear system of the MMV model: V = AU (5) where U = (A T A) -1 A T V; Step 6: Optimize the MMV model with the random Gaussian vector a, and calculate the vectors v and u respectively according to the following formula: v, = Va(6a) u = Ua (6a) Step 7: Obtain the linear system of the SMV model: v = Au (7) 8. Initialization: residual r t-1 = v, non - zero position set Iteration number t = 1; 9. Find the column index J with the maximum correlation from the sensing matrix A t . Step 10: Update the non-zero position set: S t = S t-1 ∪ J t (9) Step 11: Solve the least squares: Step 12: Update the residual: Step 13: Update the iteration t = t + 1; when t ≤ K, return to Step 8, otherwise go to the next step; Step 14: The estimated channel matrix is:

Images