Air interface calibration method for network phase error of analog domain phase shifter

Abstract

The invention relates to an air interface calibration method for a network phase error of an analog domain phase shifter, which is applied to an MIMO (Multiple Input Multiple Output) communication system and comprises the following steps of: establishing a channel model H from a signal transmitting end to a signal receiving end; keeping the channel static state unchanged, and enabling a signal receiving end to periodically send a set number of orthogonal pilot signals S; and constructing an observation model Yk of the pilot frequency receiving signal according to the channel model H, performing conjugate transpose and scaling on the observation model Yk by utilizing orthogonality of the pilot frequency signal S to obtain an equivalent observation model, constructing an optimization problem based on the equivalent observation model, and calculating phase shifter phase error estimation which enables the total noise energy of the pilot frequency receiving signal to be minimum. According to the method, air interface phase error calibration under the condition of no channel state information can be realized, the adverse effect of the phase error on a wireless communication system is reduced, the capacity of the wireless communication system is greatly improved, and the method is suitable for various communication scenes.

Description

Technical Field

[0001] This invention relates to the field of wireless communication, and in particular to an air interface calibration method for phase error in analog domain phase shifter networks. Background Technology

[0002] Multiple-input multiple-output (MIMO) technology has been a core technology in the field of wireless communication since the advent of fourth-generation (LTE) wireless communication, playing a crucial role in full-duplex wireless communication and other applications such as heterogeneous networks. Whether for massive MIMO communication or strong interference suppression, analog phase shifter networks (PSNs) are typically deployed between the antenna array and the radio frequency (RF) chain. The phase of the PSN needs to be precisely, even stringently, matched to the nominal value. However, actual phase shifters always have phase deviations due to manufacturing defects, which can lead to system performance degradation.

[0003] As a related research area, the calibration of phased array radar systems has been extensively explored. Existing calibration methods can be divided into two main categories: methods based on single-cell measurements and methods based on array-level synchronous measurements. Among them, single-cell measurement methods require repeated operations on a cell-by-cell basis, resulting in lower calibration timeliness; array-level synchronous measurement methods have better timeliness, but most of them need to be implemented in a (near) noise-free laboratory environment.

[0004] Antenna array calibration research is also involved in fields outside of radar. However, existing methods either assume that the phase shifter phase deviation remains constant across different settings, or they are only applicable to transmit (Tx) antenna arrays or only consider multiple single-input multiple-output (SIMO) systems and cannot be extended to large-scale MIMO systems. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method that can realize air interface phase error calibration under the condition of no channel state information, reduce the adverse effects of phase error on wireless communication system, greatly improve the capacity of wireless communication system, and is applicable to various communication scenarios.

[0006] The technical solution adopted by this invention to solve its technical problem is: to provide an air interface calibration method for phase error of analog domain phase shifter networks, applied to MIMO communication systems, wherein the MIMO communication system includes a signal transmitter and a signal receiver configured with phase shifters, comprising the following steps:

[0007] Establish a channel model H from the signal transmitter to the signal receiver;

[0008] Keeping the channel static, the signal receiver periodically sends a set number of orthogonal pilot signals S;

[0009] Based on the channel model H, construct the observation model Y of the pilot received signal. k Then, the orthogonality of the pilot signal S is used to analyze the observation model Y.k Perform conjugate transpose and scaling to obtain the equivalent observation model.

[0010] Based on equivalent observation model An optimization problem is constructed to calculate the phase shifter phase error estimate that minimizes the total noise energy of the pilot received signal.

[0011] Furthermore, the optimization problem is expressed as:

[0012]

[0013] in, Let W be the phase tensor of the phase shifter, K be the set quantity, and W be the phase tensor of the phase shifter. k The k-th simulated beamforming matrix is ​​given.

[0014] Furthermore, the optimization problem is to solve the phase tensor by... The parameters of the channel model H are each treated as a block variable, and the block coordinate descent algorithm is used to calculate and solve them.

[0015] Furthermore, the optimization problem is solved through the following steps:

[0016] S1 keeps the parameters of the channel model H unchanged and performs phase error estimation to calculate the phase tensor that minimizes the total noise energy of the pilot received signal. estimate;

[0017] S2 is based on the obtained phase tensor The channel parameters are estimated, and the parameters of the channel model H that minimizes the total noise energy of the pilot received signal are calculated.

[0018] S3 uses the solved phase tensor The total noise energy of the pilot received signal is calculated by estimating the parameters of the channel model H. If the difference between the total noise energy of the current round and the previous round is less than a set threshold, then the current phase tensor is used to calculate the noise energy. Estimate the output phase shifter phase error estimate; otherwise, update the parameters of channel model H to the parameter estimate of channel model H for the current round and return to step S1.

[0019] Furthermore, the phase error estimation is performed to calculate the phase tensor that minimizes the total noise energy of the pilot received signal. Estimates, including:

[0020] Obtain the nominal phase set of the phase shifter;

[0021] For any radio frequency link n xBased on the nominal phase set of the phase shifter, the phase of each antenna element on the link is configured, and then the first optimization subproblem is constructed to calculate the phase tensor that minimizes the total noise energy of the pilot received signal on the current link. estimate.

[0022] Furthermore, the first optimization subproblem is expressed as:

[0023]

[0024] in, For radio frequency link n x The equivalent observation model of the pilot received signal. For radio frequency link n x The selection matrix for each antenna element, This is the nominal phase set of the phase shifter.

[0025] Furthermore, the first optimization sub-problem is solved using the Riemann conjugate gradient algorithm.

[0026] Furthermore, the obtained phase tensor Channel parameter estimation is performed to calculate the parameter estimates of the channel model H that minimizes the total noise energy of the pilot received signal, including:

[0027] Based on the obtained phase tensor It is estimated that the optimization problem is transformed into a second optimization sub-problem, and the parameter estimates of the channel model H of the path that minimizes the total noise energy of the pilot received signal in the single-path scenario are calculated.

[0028] The optimization problem in the multipath scenario is decomposed into multiple second optimization subproblems in the single-path scenario. These subproblems are solved successively, and the estimated path components are recursively removed until the parameter estimates of the channel model H for all paths are obtained.

[0029] Furthermore, the parameter estimation of the channel model H for the path that minimizes the total noise energy of the pilot received signal in the single-path scenario includes:

[0030] The second optimization subproblem is vectorized and then Fourier transform is used to calculate an approximate solution for the parameters of the channel model H.

[0031] The approximate solution is optimally estimated using the gradient descent method to obtain the parameter estimate of the channel model H for the single-path scenario that minimizes the total noise energy of the pilot received signal.

[0032] Furthermore, the channel matrix is ​​established based on the Saleh-Valenzuela channel model.

[0033] Beneficial effects

[0034] Due to the adoption of the above technical solutions, this invention has the following advantages and positive effects compared with the prior art: This invention constructs an optimization problem based on the equivalent observation model of the pilot signal and uses the block coordinate descent algorithm to calculate the phase shifter phase error estimate that minimizes the total noise energy of the pilot received signal. It can realize the air interface calibration of the phase shifter network phase error under the condition of no channel state information, saving air interface resources and having universality; This invention ensures the calibration accuracy of the phase shifter phase error from three aspects, including: 1) accurate real-time channel estimation; 2) accurate phase error estimation; 3) the mutual iteration process of channel estimation and phase error further improves the phase error calibration accuracy. Attached Figure Description

[0035] Figure 1 This is a flowchart of an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the phase shifter air interface calibration system according to an embodiment of the present invention;

[0037] Figure 3 This is a flowchart of the phase calibration process according to an embodiment of the present invention;

[0038] Figure 4 This is a graph showing the relationship between the root mean square error of phase error calibration and the system signal-to-noise ratio in an embodiment of the present invention.

[0039] Figure 5 This is a graph showing the relationship between the root mean square error of phase error calibration and the number of measurements in an embodiment of the present invention. Detailed Implementation

[0040] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0041] This invention relates to an over-the-air calibration method for phase error in an analog domain phase shifter network, applied to a MIMO communication system. The system includes a signal transmitter and a signal receiver configured with a phase shifter. Specifically, it includes the following steps:

[0042] Establish a channel model H from the signal transmitter to the signal receiver;

[0043] Keeping the channel static, the signal receiver periodically sends a set number of orthogonal pilot signals S;

[0044] Based on the channel model H, construct the observation model Y of the pilot received signal. k Then, the orthogonality of the pilot signal S is used to analyze the observation model Y. k Perform conjugate transpose and scaling to obtain the equivalent observation model.

[0045] Based on equivalent observation model An optimization problem is constructed to calculate the phase shifter phase error estimate that minimizes the total noise energy of the pilot received signal.

[0046] by Figure 2 Taking a millimeter-wave MIMO system based on a uniform planar array (UPA) as an example, this system includes an M t The transmitter with one antenna and a configuration section connect to the Phase Shifter Network (PCPSN) M r A receiver with one antenna. This PCPSN can be represented as a matrix W, defined as:

[0047]

[0048] Where blkdiag(·) denotes a block diagonal matrix, and each vector gather And N x With N y These represent the number of RF links and the number of antennas connected to each RF link (satisfying M...). r =N x N y When the transmitter sends the pilot signal S (L is the pilot length), the signal in the receiver's digital domain can be represented as:

[0049] Y = WHS + Z, (1)

[0050] in This is the transmitter-to-receiver channel matrix. The matrix is ​​a complex Gaussian noise matrix whose elements follow the distribution CN(0,σ). 2 ).

[0051] When performing channel modeling, the Saleh-Valenzuela channel model can be used, therefore H can be represented as...

[0052]

[0053] Where N l Indicates the number of multipaths; γ n ~CN(0,1) represents the complex gain of the nth path; a UPA (θ n ,φ n ) indicates a size of N x ×Ny The UPA array channel response, and has in

[0054]

[0055] This represents the Kronecker product; while Let ω and M represent the channel response of a linear uniform array. t These represent the angle of arrival and the number of antennas, respectively, where the spacing between antennas is assumed to be half the carrier wavelength.

[0056] It should be noted that the method described in this application is not limited to uniform planar arrays, but can also be applied to uniform linear arrays, simply by modifying the channel model H. Then, the phase error estimate of the phase shifter can be solved using the method described in this application.

[0057] For a phase shifter (PS), the nominal phase set of its components is: Where N b =2 b And b represents bit resolution. However, in practice, due to manufacturing defects, temperature drift, or aging, the phase shifter may experience issues at N bits. b There is a phase deviation between each gear, which causes the nth gear to... b The actual phase of the gear is Here This represents the phase deviation, and the deviation varies depending on the component and its position. Assuming the phase shifter in a PCPSN has a b-bit resolution, the nth phase shifter in the PCPSN is defined as... x ,n y ) phase shifter nth b The phase of the gear is Collect it as a tensor Right now:

[0058]

[0059] The next goal is to estimate To calibrate Figure 2 PCPSN in the MIMO system shown.

[0060] The method described in this application is not limited to partially connected phase shifter networks, but can also be applied to fully connected phase shifter networks, simply by modifying matrix W to... Then, the phase error estimate of the phase shifter can be solved using the method described in this application.

[0061] Assume the transmitter periodically sends pilot signals K times and the channel H remains static during this process. According to equation (1), the kth transmission can be expressed as:

[0062] Yk =W k HS+Z k (6)

[0063] Among them W k For the k-th analog beamformer; Y k and Z k Let S be the received signal and noise from the k-th transmission, respectively. Assume the pilot signal S satisfies... Then we can get

[0064]

[0065] in Its elements follow the distribution CN(0,Lσ) 2 From this, we can construct information about... And the optimization problem of H:

[0066]

[0067] Due to the unit phase constraint, this problem is a non-convex optimization problem, which is difficult to solve directly. This implementation uses a block coordinate descent algorithm to solve this optimization problem, by... H and H are treated as blocks. In each iteration, only one block variable is optimized while the other blocks are fixed. The process is repeated until convergence is achieved and the optimal solution is obtained.

[0068] like Figure 3 As shown, the air interface calibration scheme for phase error in phase shifter networks mainly consists of three parts: estimating the phase error value, estimating the channel parameter value, and a conditional judgment part. Specifically, it includes:

[0069] Under the condition that the channel model H remains unchanged, phase error estimation is performed, and the phase tensor that minimizes the total noise energy of the pilot received signal is calculated.

[0070] In phase tensor Under given conditions, channel parameters are estimated, and the channel parameters that minimize the total noise energy of the pilot received signal are calculated.

[0071] Using the calculated phase tensor The total noise energy of the pilot received signal is calculated based on the channel parameters. If the difference between the total noise energy of the current round and the previous round is less than a set threshold, then the current phase tensor is used to calculate the noise energy. Estimate the phase error of the output phase shifter; otherwise, repeat the above steps.

[0072] The following is based on Figure 2 Taking the MIMO system as an example, we will further explain each step.

[0073] Step 1 (Phase Error Estimation): First, assume H is fixed, and express H as... in

[0074]

[0075] From equation (7), we can obtain

[0076]

[0077] in and They represent and The nth x List, The nth y The element is from The selected option here needs to be ensured. All elements will be selected, but the order in which they are selected is random.

[0078]

[0079] Therefore, we can obtain The definition is in n y =1,2,…,N y ,as well as Substituting equation (11) into equation (10), we can obtain

[0080]

[0081] Further statement The expression can be obtained

[0082]

[0083] in as well as Therefore, in order to accurately estimate the phase, solving equation (8) can be transformed into solving equation (8) for...

[0084]

[0085] The problem can be solved here using the Riemann conjugate gradient (RCG) method (14).

[0086] Step 2 (Channel Parameter Estimation): Assuming Given that the solution to problem (8) can be transformed into solving the following problem:

[0087]

[0088] First, consider the solution to problem (15) when there is only one path, i.e., N l =1. Therefore, problem (15) can be transformed into

[0089]

[0090] Use formula It can be obtained as well as

[0091]

[0092] in as well as Therefore, problem (16) can be further expressed as

[0093]

[0094] make and Problem (18) can be further transformed into

[0095]

[0096] What is easy to obtain is Then, substituting γ into problem (19) can transform it into

[0097]

[0098] Problem (20) can be solved by Fourier transform (FFT) to obtain an approximate solution, and then a more accurate solution can be obtained from the approximate solution by using gradient descent.

[0099] Next, consider the multipath scenario, i.e., N l >1. In multipath scenarios, problem (15) can be represented as

[0100]

[0101] in as well as Here we can decompose problem (21) into N1 subproblems with single paths: First assume If there is only one path, then problem (21) degenerates into problem (16), at which point γ1, θ1, φ1, ω1 can be solved; then, its position in the middle can be deduced from γ1, θ1, φ1, ω1. The components in and from which the part Subtract from the middle, and then continue in this manner until all N are found. l Channel parameters for each path.

[0102] Step 3 (determining whether the phase air interface calibration process is complete): Since the optimization problems in equations (14) and (15) can both make the objective function value in equation (8) remain monotonically decreasing, after the second step is completed, calculate the value of equation (8). If the change in the value of equation (8) is less than 10... -5 If the change is greater than 10, the process ends, and the estimated phase error is obtained; -5 If the estimated phase and channel parameter values ​​have not yet converged, then we should return to the first step and continue with phase estimation.

[0103] The performance of this implementation is demonstrated below through simulation examples. Without loss of generality, a 4-bit phase shifter network (PSN) is used in the following simulations, with its phase deviation following a uniform distribution between [-20°, 20°]. By selecting an appropriate Ensure that all positions of each phase shifter are measured an equal number of times. Complex gain γ n n = 1, 2, ..., N l It follows a complex Gaussian distribution CN(0,1), and the parameter θ n ,φ n ,ω n Generated uniformly in the intervals

[0104] The first simulation study investigates the performance of the root mean square error (RMSE), assuming L = M t =N x =4, N1=3. The phase of a phase shifter is measured 20 times at each position, and the number of pilot signals transmitted is K=20×24=320. Figure 4 This demonstrates the RMSE performance as a function of signal-to-noise ratio, at N y In the cases of =8, 16, and 32, the values ​​are very close to the corresponding Cramé-Rao boundary (CRB) across the entire range from -5 dB to 20 dB. However, for N... y =64, the RMSE performance of the algorithm decreases slightly and deviates from CRB, because the increase in the number of unknown variables leads to convergence to a local optimum.

[0105] like Figure 5 The second simulation shown provides the RMSE performance of the proposed phase calibration algorithm under conditions where the number of measurements for each phase shifter ranges from 18 to 30, where the number of pilot transmissions K ranges from 18 × 24 = 288 to 30 × 24 = 480. Simulation results demonstrate that, given a 4-bit phase shifter bit accuracy, increasing the number of pilot transmissions improves phase estimation accuracy, even for N... y =64, under conditions of increasing measurement frequency, the RMSE of phase calibration will decrease.

Claims

1. An over-the-air calibration method for phase error of an analog domain phase shifter network, applied to a MIMO communication system, the MIMO communication system comprising a signal transmitter and a signal receiver configured with phase shifters, characterized in that, Includes the following steps: Establish a channel model H from the signal transmitter to the signal receiver; Keeping the channel static, the signal receiver periodically sends a set number of orthogonal pilot signals S; Based on the channel model H, construct the observation model Y of the pilot received signal. k Then, the orthogonality of the pilot signal S is used to analyze the observation model Y. k Perform conjugate transpose and scaling to obtain the equivalent observation model. Based on equivalent observation model An optimization problem is constructed to calculate the phase shifter phase error estimate that minimizes the total noise energy of the pilot received signal.

2. The method according to claim 1, characterized in that, The optimization problem is expressed as: in, Let W be the phase tensor of the phase shifter, K be the set quantity, and W be the phase tensor of the phase shifter. k The k-th simulated beamforming matrix is ​​given.

3. The method according to claim 2, characterized in that, The optimization problem is to solve the phase tensor by... The parameters of the channel model H are each treated as a block variable, and the block coordinate descent algorithm is used to calculate and solve them.

4. The method according to claim 1, characterized in that, The optimization problem is solved through the following steps: S1 keeps the parameters of the channel model H unchanged and performs phase error estimation to calculate the phase tensor that minimizes the total noise energy of the pilot received signal. estimate; S2 is based on the obtained phase tensor The channel parameters are estimated, and the parameters of the channel model H that minimizes the total noise energy of the pilot received signal are calculated. S3 uses the solved phase tensor The total noise energy of the pilot received signal is calculated by estimating the parameters of the channel model H. If the difference between the total noise energy of the current round and the previous round is less than a set threshold, then the current phase tensor is used to calculate the noise energy. Estimate the output phase shifter phase error estimate; otherwise, update the parameters of channel model H to the parameter estimate of channel model H for the current round and return to step S1.

5. The method according to claim 4, characterized in that, The phase error estimation is performed, and the phase tensor that minimizes the total noise energy of the pilot received signal is calculated. Estimates, including: Obtain the nominal phase set of the phase shifter; For any radio frequency link n x Based on the nominal phase set of the phase shifter, the phase of each antenna element on the link is configured, and then the first optimization subproblem is constructed to calculate the phase tensor that minimizes the total noise energy of the pilot received signal on the current link. estimate.

6. The method according to claim 5, characterized in that, The first optimization subproblem is expressed as: in, For radio frequency link n x The equivalent observation model of the pilot received signal. For radio frequency link n x The selection matrix for each antenna element, This is the nominal phase set of the phase shifter.

7. The method according to claim 6, characterized in that, The first optimization subproblem is solved using the Riemann conjugate gradient algorithm.

8. The method according to claim 4, characterized in that, The obtained phase tensor Channel parameter estimation is performed to calculate the parameter estimates of the channel model H that minimizes the total noise energy of the pilot received signal, including: Based on the obtained phase tensor It is estimated that the optimization problem is transformed into a second optimization sub-problem, and the parameter estimates of the channel model H of the path that minimizes the total noise energy of the pilot received signal in the single-path scenario are calculated. The optimization problem in the multipath scenario is decomposed into multiple second optimization subproblems in the single-path scenario. These subproblems are solved successively, and the estimated path components are recursively removed until the parameter estimates of the channel model H for all paths are obtained.

9. The method according to claim 8, characterized in that, The parameter estimation of the channel model H for the path that minimizes the total noise energy of the pilot received signal in a single-path scenario includes: The second optimization subproblem is vectorized and then Fourier transform is used to calculate an approximate solution for the parameters of the channel model H. The approximate solution is optimally estimated using the gradient descent method to obtain the parameter estimate of the channel model H for the single-path scenario that minimizes the total noise energy of the pilot received signal.

10. The method according to claim 1, characterized in that, The channel matrix is ​​established based on the Saleh-Valenzuela channel model.

Images