Signal transmission method, signal reception method, device and signal processing system

Abstract

The signal sending method, the signal receiving method, the device and the signal processing system provided by the embodiments of the present application relate to the technical field of communication, and jointly iteratively solve the first precoding matrix and the decoding matrix obtained based on the above minimum multi-cell mean square error function between the multi-cell terminal device and the receiving device. The terminal device of each cell first precodes the source signal using the first precoding matrix and the second precoding matrix to obtain the precoded signal of the source signal; the precoded signal is sent; and the receiving device of the cell decodes the received uplink signal according to the decoding matrix to obtain the converged signal of the source signal transmitted by each terminal device of the cell for over-the-air computation. During the decoding process, the sum of the source signals transmitted by each terminal device of the cell is retained, and the rest of the signals is suppressed, so that the receiving device can accurately converge the source signals transmitted by the terminal devices of the cell, and the accuracy of over-the-air computation is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, and in particular to a signal sending method, a signal receiving method, a device and a signal processing system. BACKGROUND

[0002] With the development of information processing technologies represented by artificial intelligence, wireless communication networks will change from data-centric to compute-centric, for example, aggregating distributed data (such as sensor observations or artificial intelligence model updates) to perform function calculations (such as summation, averaging, etc.) to support large-scale sensors and distributed machine learning. In traditional wireless communication networks, a communication and computation separation architecture is adopted, that is, communication is followed by computation. After transmitting data using a wireless network, a large amount of individual data is recovered before computation. Therefore, the communication and computation separation architecture of the traditional wireless communication network puts a huge pressure on the wireless channel.

[0003] In order to cope with the problem of computation under communication restriction caused by massive node access, wireless networks will change from a communication and computation separation architecture to a communication and computation integration. Communication and computation integration takes over-the-air computation (AirComp) as the core. AirComp applies function computation to signals of multiple terminal devices. AirComp can superimpose waveforms of signals of multiple terminal devices to achieve over-the-air computation.

[0004] A receiving device can be set in each cell. The receiving device can receive an uplink signal sent by a terminal device. Since the terminal device and the receiving device use a wireless communication network, in a multi-cell scenario, the receiving device of a cell will receive uplink signals of terminal devices in the cell and uplink signals of terminal devices in other cells. The uplink signals sent by terminal devices in other cells are uplink interference signals for the receiving device of the cell.

[0005] For a multi-cell over-the-air computation network, due to the influence of inter-cell uplink interference and channel noise received by the receiving device of each cell, the accuracy of over-the-air computation needs to be improved. SUMMARY

[0006] Embodiments of the present application provide a signal sending method, a signal receiving method, a device and a signal processing system, which are applied to the technical field of communication. The problem of low accuracy of over-the-air computation in a multi-cell scenario in the prior art is solved.

[0007] To achieve the above object, the technical scheme adopted by the present application is as follows:

[0008] In a first aspect, the embodiments of the present application provide a signal sending method applied to a terminal device. A double-layer precoding matrix of the terminal device and a decoding matrix of a receiving device are designed, a minimum multi-cell mean square error function is constructed, and the minimum multi-cell mean square error function represents a sum of squares of F-norms of differences between decoding signals of the receiving devices of each cell and target signals. The minimum multi-cell mean square error function is solved by the terminal devices and the receiving devices of the multi-cells to obtain a first precoding matrix and a decoding matrix corresponding to each cell. A plurality of first terminal devices in each cell generate source signals for over-the-air computation according to an over-the-air computation instruction. The first terminal devices perform first precoding on the source signals by using the first precoding matrix to obtain first precoded signals. The first precoded signals are further precoded based on a second precoding matrix to obtain precoded signals of the source signals. The second precoding matrix of each first terminal device is an inverse matrix of a channel matrix between the first terminal device and the receiving device of the cell. The precoded signals are sent by each first terminal device in uplink. The receiving device of the cell (the cell) where the first terminal device is located decodes the received uplink signals by using the decoding matrix.

[0009] In the embodiments of the present application, the minimum multi-cell mean square error function is solved by the terminal devices and the receiving devices to obtain the first precoding matrix and the decoding matrix corresponding to each cell in the multi-cells. When over-the-air computation is performed in the multi-cell scenario, the source signals for over-the-air computation are processed by the terminal devices in each cell by using the first precoding matrix and the second precoding matrix, so that the source signals for over-the-air computation of the plurality of different terminal devices in the same cell are aligned, thereby realizing over-the-air computation. Meanwhile, when the receiving device decodes by using the corresponding decoding matrix, the interference signals and the noise signals sent by the terminal devices of the adjacent cells can be suppressed, and the accuracy of over-the-air computation in the multi-cell scenario can be improved.

[0010] In a possible implementation manner of the first aspect, in each cell, when the first terminal device of the cell detects that a channel between the first terminal device and the receiving device of the cell changes, the first terminal device sends a pilot signal to the receiving device and receives a channel matrix calculated by the receiving device based on the pilot signal. When the channel matrix between the terminal devices of each cell and the receiving devices is known, the terminal devices in each cell can randomly generate a group of initial first precoding matrices satisfying a quasi-unitary constraint. Each first terminal device in each cell sends the initial first precoding matrix to the receiving device of the cell. The first terminal device obtains the decoding matrix sent by the receiving device of each cell. After receiving the decoding matrix of the current iteration sent by the receiving device of each cell, the first terminal device substitutes the decoding matrix corresponding to each cell into an expression of the first precoding matrix to calculate the first precoding matrix of the cell.

[0011] In the embodiments of the present application, each cell receiving device feeds back interference information (implicit in the decoding matrix) through a local decoding matrix without centralized global channel state information, which can reduce the overhead of the backhaul link. The decoding matrix is calculated from the initial precoding matrix of the terminal device, and actually reflects the actual interference coupling relationship among multiple cells. In the iteration process, the decoding matrix transmits interference information, so that the final precoding matrix naturally has the interference alignment characteristic, which can improve the system efficiency. In addition, through the iterative algorithm, the optimal solution is gradually approached, and an acceptable solution can be reached after a limited number of iterations. Compared with directly solving the high-dimensional non-convex optimization problem, the computational complexity is significantly reduced.

[0012] In a possible implementation form of the first aspect, a Lagrange function is constructed according to the minimum multi-cell mean square error function and the terminal device transmit power constraint; and each terminal device of each cell takes the partial derivative of the Lagrange function with respect to the first precoding matrix to obtain an expression of the first precoding matrix of the cell.

[0013] In the embodiments of the present application, the optimization problem with equality constraints is converted into an unconstrained optimization problem by constructing a Lagrange function according to the minimum multi-cell mean square error function and the transmit power constraint. By taking the partial derivative of the Lagrange function with respect to the first precoding matrix, the explicit expression of the first precoding can be derived, which is convenient for subsequent iteration. In addition, the Lagrange function can balance the signal quality and power efficiency control, improve the signal-to-interference-and-noise ratio by minimizing the multi-cell mean square error, and avoid terminal device overload by power constraint, which conforms to the actual terminal device limit and avoids terminal overload through power constraint. The interference signals of adjacent cells can be effectively suppressed, further improving the accuracy of air computing. It conforms to the actual device limit.

[0014] In a possible implementation form of the first aspect, the first terminal device calculates a Lagrange multiplier according to the decoding matrix corresponding to each cell; then updates the first precoding matrix according to the Lagrange multiplier and the decoding matrix corresponding to each cell; then replaces the initial first precoding matrix with the updated first precoding matrix, and sends the updated first precoding matrix to the receiving device of each cell, so that each cell receiving device iteratively decodes the matrix according to the first precoding matrix; and repeats the above process to iteratively calculate the first precoding matrix.

[0015] In the embodiments of the present application, in the process of iteratively calculating the first precoding matrix, the Lagrange multiplier is calculated, and the Lagrange multiplier is used to update the first precoding matrix. Since the Lagrange multiplier dynamically adjusts the weight of the minimum multi-cell mean square error and the power constraint, the first precoding matrix can adapt to different channel conditions.

[0016] In a possible implementation of the first aspect, for each first terminal device in a cell, a first equivalent channel matrix between the first terminal device and a receiving device of a neighboring cell of the first terminal device, a decoding matrix of the neighboring cell, a conjugate transpose of the decoding matrix of the neighboring cell, and a first equivalent channel matrix of the neighboring cell are multiplied; and a second equivalent channel matrix between the first terminal device and a receiving device of the cell, a decoding matrix of the cell, a conjugate transpose of the decoding matrix of the cell, and the second equivalent channel matrix are multiplied; the two products are first products corresponding to the first terminal device; the first products corresponding to each first terminal device in the cell are calculated, the first products corresponding to each first terminal device in the cell are accumulated, and a first matrix is obtained; the first terminal device determines a second equivalent channel matrix according to the second channel matrix between the first terminal device and the receiving device of the cell, calculates a second product of a conjugate transpose of the second equivalent channel matrix and a decoding matrix corresponding to the first terminal device, and accumulates the second products corresponding to each first terminal device in the cell to obtain a second matrix; the first matrix and the second matrix are substituted into a Lagrange multiplier equation, and a Lagrange multiplier of the cell is obtained by using a dichotomy or a Newton iteration algorithm. The terminal devices of the plurality of cells can calculate the Lagrange multipliers of the respective cells in parallel.

[0017] In the embodiments of the present application, the first matrix and the second matrix are calculated, and the Lagrange multiplier is calculated according to the first matrix and the second matrix, in the above process, closed-form solution is achieved instead of numerical search, and convergence can be accelerated. The Lagrange multiplier is dynamically updated in the iteration process, and local optimum is avoided.

[0018] In a possible implementation of the first aspect, after the first terminal device in each cell obtains the Lagrange multiplier in each iteration process, the obtained Lagrange multiplier and the obtained decoding matrices of the cells are substituted into an expression of a first precoding matrix to obtain an updated first precoding matrix. The terminal devices of the plurality of cells can calculate the Lagrange multipliers of the respective cells in parallel, and update the first precoding matrices of the respective cells according to the obtained Lagrange multipliers in parallel.

[0019] In the embodiments of the present application, the first precoding matrix is updated according to the Lagrange multiplier and the decoding matrices of the cells, and therefore the first precoding matrix capable of suppressing interference signals can be obtained by an iterative solution method, and the complexity of solving the first precoding matrix is reduced.

[0020] In a second aspect, the embodiments of the present application provide a signal receiving method, applied to a receiving device. The receiving device of each cell receives an uplink signal, the uplink signal comprising precoded signals respectively transmitted by at least one first terminal device in the cell, and interference signals and channel Gaussian white noise signals superimposed in at least one second terminal device in at least one adjacent cell; the precoded signal of each first terminal device is obtained by performing double-layer precoding on a source signal used for over-the-air computation by the first terminal device; and the receiving device of each cell decodes the uplink signal using a decoding matrix of the cell, wherein the decoding matrix suppresses the interference signals and channel Gaussian white noise signals transmitted by the second terminal device in the adjacent cell in a decoding process, thereby obtaining a converged signal of the source signal of each first terminal device in the cell.

[0021] In the embodiments of the present application, the minimum multi-cell mean square error function can be solved jointly by the terminal devices and the receiving device, to obtain the first precoding matrix corresponding to each cell in the multi-cell, and the decoding matrix on the receiving device side; the decoding matrix can converge the source signal in the cell while suppressing the cross-cell interference signals and Gaussian white noise signals. When performing over-the-air computation in the multi-cell scenario, each terminal device in each cell processes the source signal used for over-the-air computation using the first precoding matrix and the second precoding matrix, and then transmits the precoded signal of the source signal; when the receiving device decodes the received uplink signal using the decoding matrix, the interference signals and channel Gaussian white noise signals transmitted by the terminal device in the adjacent cell can be suppressed, the source signal in the cell can be converged with high precision, and the accuracy of over-the-air computation in the multi-cell scenario can be improved.

[0022] In a possible implementation manner of the second aspect, the receiving device constructs a Lagrange function according to the minimum multi-cell mean square error function and a terminal device transmit power constraint, determines the decoding matrix of each cell by finding the extreme value of the Lagrange function with respect to the decoding matrix; the receiving device of each cell obtains the first precoding matrix transmitted by the terminal device of each cell; updates the decoding matrix according to the decoding matrix expression and the first precoding matrix of the terminal device of each cell, and transmits the decoding matrix to the terminal device of each cell, so that the terminal device of each cell updates the first precoding matrix; the above process is repeated and iterated, and the decoding matrix satisfying a preset condition is obtained as the final decoding matrix of the cell.

[0023] In the embodiments of the present application, a closed-form solution of the decoding matrix is directly obtained by taking the extreme value of the Lagrange function, which can accelerate the iterative convergence, the initial precoding matrix satisfies the quasi-unitary constraint, which can ensure the orthogonality of signal transmission and reduce the error propagation in the initial iteration. When the decoding matrix is updated each time, the first precoding matrix reflecting the current interference transformation is automatically adapted, which can realize implicit interference alignment. The above process iteratively optimizes the decoding matrix, and jointly considers the quasi-unitary constraint and power limitation of the first precoding matrix, thereby realizing efficient and robust uplink signal reception and interference management in a multi-cell multiple-input multiple-output system, which is helpful to realize high-precision convergence of the source signal in the current cell and can improve the accuracy of the over-the-air computation in the multi-cell scenario.

[0024] In a possible implementation of the second aspect, the receiving device obtains a third channel matrix of each second terminal device in each neighboring cell to the receiving device of the current cell, calculates a third equivalent channel matrix and a conjugate transpose of the third equivalent channel matrix according to the third channel matrix; calculates a conjugate transpose of the first precoding matrix according to the first precoding matrix of each cell; calculates a second equivalent channel matrix of each first terminal device in the current cell to the receiving device of the current cell; for each neighboring cell of the current cell, calculates a third product corresponding to each second terminal device in the cell, and calculates a third product corresponding to each first terminal device in the current cell; accumulates the third products corresponding to the terminal devices in the plurality of cells respectively to obtain a third matrix; calculates an inverse matrix of a sum of the third matrix and a covariance matrix of Gaussian white noise; calculates a second product of the second equivalent channel matrix of each first terminal device and the first precoding matrix of the current cell, and sums the second products corresponding to the first terminal devices in the current cell to obtain a second matrix; and calculates a product of the inverse matrix and the second matrix according to the decoding matrix expression to obtain the decoding matrix.

[0025] In the embodiments of the present application, the third matrix (joint covariance matrix of interference signals and useful signals) is constructed by accumulating the third products (interference terms) of the terminal devices in the neighboring cells and the third products (useful signals) of the terminal devices in the current cell: the third matrix can quantize the interference introduced by the terminal devices in the neighboring cells through the third channel matrix; and the effective signal components of the terminal devices in the current cell through the second equivalent channel matrix are reserved. The distributed computing architecture is used to calculate the decoding matrix of each cell, which can improve the efficiency of calculating the decoding matrix. In addition, the above process of solving the decoding matrix can realize the minimization of the multi-cell mean square error, and the theoretical optimal reception performance can be achieved through the decoding matrix. The receiving device of each cell can actively suppress the cross-cell interference signals and Gaussian white noise signals through the corresponding decoding matrix, which is conducive to improving the accuracy of the over-the-air computation in the multi-cell scenario.

[0026] In a third aspect, the embodiments of the present application provide a signal processing system, comprising a plurality of terminal devices located in a plurality of cells, and a receiving device located in each of the cells; a double-layer precoding on the terminal device side and a decoding matrix on the receiving device side are designed, and a minimum multi-cell mean square error function is constructed, which represents a sum of squares of F-norms of differences between decoding signals and target signals of the receiving devices in each cell; the minimum multi-cell mean square error function is solved jointly for the terminal devices and the receiving devices in the plurality of cells, so that a first precoding matrix corresponding to each cell and a decoding matrix are obtained; a first terminal device in each cell uses the first precoding matrix of the cell to perform first precoding on a source signal used for over-the-air computation, so as to obtain a first precoded signal, and uses a second precoding matrix to perform second precoding on the first precoded signal, so as to obtain a precoded signal of the source signal; the precoded signal is sent uplink; and the receiving device in each cell receives an uplink signal, and uses the decoding matrix to decode the uplink signal, so as to obtain a converged signal of the source signal of each first terminal device in the cell.

[0027] In the embodiments of the present application, the minimum multi-cell mean square error function is solved jointly for the terminal devices in different cells and the receiving devices in different cells, so that the first precoding matrix corresponding to each cell in the plurality of cells and the decoding matrix on the receiving device side are obtained; the decoding matrix can realize the convergence of the source signal in the cell, and can suppress the cross-cell interference signal and the Gaussian white noise signal. When over-the-air computation is performed in the multi-cell scenario, after each terminal device in each cell processes the source signal used for over-the-air computation using the first precoding matrix and the second precoding matrix, when the receiving device decodes the received uplink signal using the decoding matrix, the interference signal transmitted by the terminal device in the adjacent cell and the channel Gaussian white noise signal can be suppressed, the high-precision convergence of the source signal in the cell can be realized, and the accuracy of over-the-air computation in the multi-cell scenario can be improved.

[0028] In a possible implementation of the third aspect, a Lagrange function is constructed by the minimum mean square error function and the terminal device transmit power constraint, and an expression of the first precoding matrix corresponding to each cell is obtained by taking the extreme value of the Lagrange function with respect to the first precoding matrix of each cell, and an expression of the decoding matrix corresponding to each cell is obtained by taking the extreme value of the Lagrange function with respect to the decoding matrix of each cell; the terminal device of each cell generates an initial first precoding matrix, and sends the initial first precoding matrix to the receiving device of the cell; the terminal device of each cell and the receiving device of each cell perform the following joint interference alignment iteration operation: the receiving device of each cell receives the first precoding matrix sent by the terminal device of the cell, and the first precoding matrix of each cell is shared by the receiving devices of the cells through the central unit connected to the receiving devices; the receiving device of each cell sends the shared first precoding matrix of each cell to the terminal device of the cell; the receiving device of each cell calculates the decoding matrix according to the expression of the decoding matrix of the cell and the first precoding matrix, and obtains the decoding matrix for iterative calculation; the receiving devices of the cells share the decoding matrix of each cell through the central unit connected to the receiving devices, and send the decoding matrix of each cell obtained to the terminal device of the cell; the terminal device of each cell calculates the Lagrange multiplier based on the decoding matrix of each cell; the first precoding matrix is updated according to the Lagrange multiplier, and the first precoding matrix for iterative calculation is obtained; the first precoding matrix is sent to the receiving device of the cell; the first precoding matrix and the decoding matrix obtained when the iterative calculation is stopped and a preset condition is met are taken as the final first precoding matrix and the decoding matrix.

[0029] In the embodiments of the present application, the expression of the first precoding matrix and the expression of the decoding matrix are obtained by using the preset constraint and the Lagrange function constructed by the minimum multi-cell mean square error function, and the initial first precoding matrix is randomly generated to jointly and iteratively calculate the first precoding matrix and the decoding matrix of each cell by the terminal devices and the receiving devices of the multi-cells. Since the first precoding matrix and the decoding matrix are obtained by solving the above Lagrange function, the source signal of the terminal device is precoded by using the first precoding matrix, the receiving device decodes the precoded signal by using the decoding matrix, and the signals received by the receiving devices of each cell satisfy the minimum multi-cell mean square error function, so that the receiving device of each cell can effectively detect the converged signal of the source signal sent by the terminal device of the cell and suppress the interference signal and the noise signal of the adjacent cell, thereby improving the accuracy of the air calculation.

[0030] In a fourth aspect, an embodiment of the present application provides a signal sending apparatus, which can be an electronic device, or a chip or chip system in the electronic device. The signal sending apparatus can include a display unit and a processing unit. When the signal sending apparatus is an electronic device, the display unit can be a display screen. The display unit is configured to perform the step of displaying, so that the electronic device implements a signal sending method described in the first aspect or any possible implementation of the first aspect. When the signal sending apparatus is an electronic device, the processing unit can be a processor. The signal sending apparatus can further include a storage unit, which can be a memory. The storage unit is configured to store instructions, and the processing unit executes the instructions stored in the storage unit, so that the electronic device implements a signal sending method described in the first aspect or any possible implementation of the first aspect. When the signal sending apparatus is a chip or chip system in the electronic device, the processing unit can be a processor. The processing unit executes the instructions stored in the storage unit, so that the electronic device implements a signal sending method described in the first aspect or any possible implementation of the first aspect. The storage unit can be a storage unit (for example, a register, a cache, etc.) in the chip, or a storage unit (for example, a read-only memory, a random access memory, etc.) outside the chip in the electronic device. The electronic device can be a terminal device in multiple cells.

[0031] For example, the processing unit of the first terminal device in each cell is configured to: generate a source signal for over-the-air computation of the first terminal device according to the over-the-air computation instruction; perform first precoding on the source signal using a first precoding matrix to obtain a first precoded signal; and perform second precoding on the first precoded signal based on a second precoding matrix to obtain a precoded signal of the source signal; the second precoding matrix of each first terminal device is an inverse matrix of a channel matrix between the first terminal device and a receiving device in the cell; and the precoded signal is sent uplink; and the receiving device in the cell of the first terminal device uses a decoding matrix to decode the received uplink signal.

[0032] In an embodiment of the present application, the minimum multi-cell mean square error function is solved by the terminal device and the receiving device, to obtain the first precoding matrix and the decoding matrix corresponding to each cell in the multi-cell. In the multi-cell scenario, the source signal for over-the-air computation is processed by the first precoding matrix and the second precoding matrix by each terminal device in each cell, so that the source signals of multiple different terminal devices in the same cell are aligned for over-the-air computation, thereby realizing over-the-air computation, and meanwhile, the interference signal and the noise signal sent by the terminal devices in the adjacent cells can be suppressed when the receiving device decodes using the corresponding decoding matrix, and the accuracy of over-the-air computation in the multi-cell scenario can be improved.

[0033] In a possible implementation of the fourth aspect, the processing unit is further configured to: send a pilot signal to the receiving devices in the cells when detecting that the channel between the receiving devices and the cell changes, and receive a channel matrix calculated according to the pilot signal and sent by the receiving devices; randomly generate a set of initial first precoding matrices satisfying the quasi-unitary constraint, when the channel matrix between each receiving device and each cell is known; send the initial first precoding matrices to the receiving devices in the cells; and obtain the decoding matrices sent by the receiving devices in the cells, and after receiving the decoding matrices of the current iteration sent by the receiving devices in the cells, substitute the decoding matrices corresponding to the cells into the expression of the first precoding matrix to calculate the first precoding matrix of the cell.

[0034] In the embodiments of the present application, the receiving devices in the cells feed back interference information (implicit in the decoding matrix) through the local decoding matrix, without centralized global channel state information, which can reduce the overhead of the backhaul link. The decoding matrix is calculated from the initial precoding matrix of the terminal device, and actually reflects the actual interference coupling relationship among the cells. The interference information is transmitted through the decoding matrix in the iteration process, so that the final precoding matrix has the interference alignment characteristic, and the system efficiency can be improved. In addition, the optimal solution is gradually approached through the iterative algorithm, and an acceptable solution can be obtained after a limited number of iterations. Compared with directly solving the high-dimensional non-convex optimization problem, the computational complexity is significantly reduced.

[0035] In a possible implementation of the fourth aspect, the processing unit is further configured to: construct a Lagrange function according to the minimum multi-cell mean square error function and the terminal device transmit power constraint; and obtain an expression of the first precoding matrix of the cell by finding the extreme value of the first precoding matrix of the Lagrange function.

[0036] In the embodiments of the present application, the Lagrange function is constructed according to the minimum multi-cell mean square error function and the transmit power constraint, and the optimization problem with equality constraints is converted into an unconstrained optimization problem by using the Lagrange multiplier method. The explicit expression of the first precoding matrix can be derived by taking the partial derivative of the Lagrange function with respect to the first precoding matrix, which is convenient for subsequent iteration. In addition, the Lagrange function can balance the signal quality and power efficiency control, improve the signal-to-interference-and-noise ratio by minimizing the multi-cell mean square error, and avoid terminal overload by power constraint, which conforms to the actual device limit. The interference signals of the adjacent cells can be effectively suppressed, and the accuracy of the air calculation is further improved. It conforms to the actual device limit.

[0037] In a possible implementation of the fourth aspect, the processing unit is further configured to: calculate a Lagrange multiplier according to the decoding matrix corresponding to each cell; update the first precoding matrix according to the Lagrange multiplier and the decoding matrix corresponding to each cell; replace the initial first precoding matrix with the updated first precoding matrix, and send the updated first precoding matrix to the receiving device of each cell, so that the receiving device of each cell iteratively decodes the matrix according to the first precoding matrix; and repeat the above process to iteratively calculate the first precoding matrix.

[0038] In the embodiments of the present application, the first precoding matrix is updated using the Lagrange multiplier in the process of iteratively calculating the first precoding matrix. Since the Lagrange multiplier dynamically adjusts the weight of the minimum mean square error minimization and the power constraint of the multi-cell implicitly, the first precoding matrix can adapt to different channel conditions.

[0039] In a possible implementation of the fourth aspect, the processing unit is further configured to: calculate, for the cell, a first product corresponding to each first terminal device in the cell, and accumulate the first product corresponding to each first terminal device in the cell to obtain a first matrix; wherein the first product corresponding to each first terminal device in the cell is the product of the conjugate transpose of the first equivalent channel matrix between the first terminal device and the receiving device of the adjacent cell, the decoding matrix of the adjacent cell, the conjugate transpose of the decoding matrix of the adjacent cell, and the first equivalent channel matrix of the adjacent cell, and the product of the conjugate transpose of the second equivalent channel matrix between the first terminal device and the receiving device of the cell, the decoding matrix of the cell, the conjugate transpose of the decoding matrix of the cell, and the second equivalent channel matrix; determine the second equivalent channel matrix according to the second channel matrix between each first terminal device in the cell and the receiving device of the cell, calculate the product of the conjugate transpose of the second equivalent channel matrix and the decoding matrix of the second equivalent channel matrix corresponding to each first terminal device in the cell, and accumulate the product corresponding to each first terminal device to obtain a second matrix; and solve a Lagrange multiplier equation based on the first matrix and the second matrix to obtain the Lagrange multiplier.

[0040] In the embodiments of the present application, the first matrix and the second matrix are calculated, the Lagrange multiplier is calculated based on the first matrix and the second matrix, and then the first precoding matrix is updated according to the Lagrange multiplier and the decoding matrix of each cell, so that the first precoding matrix capable of suppressing the interference signal can be obtained by the iterative calculation method. In the above process, the closed-form solution is realized instead of the numerical search, which can accelerate the convergence. The Lagrange multiplier is dynamically updated in the iterative process, which avoids falling into a local optimum.

[0041] In a possible implementation of the fourth aspect, the processing unit is further configured to: substitute the Lagrange multipliers and the decoding matrix of each cell into an expression of the first precoding matrix to obtain an updated first precoding matrix.

[0042] In a fifth aspect, an embodiment of the present application provides a signal receiving apparatus. The signal receiving apparatus can be an electronic device, or a chip or chip system in the electronic device. The signal receiving apparatus can include a display unit and a processing unit. When the signal receiving apparatus is an electronic device, the display unit can be a display screen. The display unit is configured to perform the display step, so that the electronic device implements a signal receiving method described in the second aspect or any possible implementation of the second aspect. When the signal receiving apparatus is an electronic device, the processing unit can be a processor. The signal receiving apparatus can further include a storage unit, which can be a memory. The storage unit is configured to store instructions. The processing unit executes the instructions stored in the storage unit, so that the electronic device implements a signal receiving method described in the second aspect or any possible implementation of the second aspect. When the signal receiving apparatus is a chip or chip system in the electronic device, the processing unit can be a processor. The processing unit executes the instructions stored in the storage unit, so that the electronic device implements a signal receiving method described in the first aspect or any possible implementation of the first aspect. The storage unit can be a storage unit (for example, a register, a cache, etc.) in the chip, or a storage unit (for example, a read-only memory, a random access memory, etc.) in the electronic device and located outside the chip. The electronic device can be a receiving device corresponding to each cell in a multi-cell.

[0043] For example, the processing unit in the receiving device of each cell is configured to: receive an uplink signal, the uplink signal including precoding signals respectively transmitted by at least one first terminal device in the cell, and interference signals respectively transmitted by at least one second terminal device in at least one adjacent cell and superimposed with Gaussian white noise signals in a channel; the precoding signal of each first terminal device is obtained by performing double-layer precoding on a source signal used by the first terminal device for over-the-air calculation; and the receiving device of each cell performs decoding processing on the uplink signal using a decoding matrix of the cell. During the decoding processing of the uplink signal by the decoding matrix, the interference signals respectively transmitted by the second terminal devices in the adjacent cells and the Gaussian white noise signals in the channel are suppressed, so that a converged signal of the source signals of the at least one first terminal device in the cell is obtained.

[0044] In the embodiments of the present application, the minimum multi-cell mean square error function can be solved in combination with the terminal device and the receiving device to obtain the first precoding matrix corresponding to each cell in the multi-cell and the decoding matrix on the receiving device side. The decoding matrix can realize the convergence of the source signal in the cell while suppressing the cross-cell interference signal and the Gaussian white noise signal. When the over-the-air computation is performed in the multi-cell scenario, after each terminal device in each cell processes the source signal used for over-the-air computation using the first precoding matrix and the second precoding matrix, the terminal device transmits the precoded signal of the source signal, and when the receiving device decodes the received uplink signal using the decoding matrix, the interference signal transmitted by the terminal device in the adjacent cell and the channel Gaussian white noise signal can be suppressed, the high-precision convergence of the source signal in the cell can be realized, and the accuracy of the over-the-air computation in the multi-cell scenario can be improved.

[0045] In a possible implementation of the fifth aspect, the processing unit is further configured to: construct a Lagrange function according to the minimum multi-cell mean square error function and the terminal device transmit power constraint, and determine the decoding matrix of each cell by finding the extreme value of the Lagrange function with respect to the decoding matrix; the receiving device of each cell acquires the first precoding matrix transmitted by the terminal device of each cell; the decoding matrix is updated according to the decoding matrix expression and the first precoding matrix of the terminal device of each cell, and the decoding matrix is transmitted to the terminal device of each cell to update the first precoding matrix by the terminal device of each cell, and the above process is repeated and iterated, and the decoding matrix satisfying the preset condition is obtained as the final decoding matrix of the cell.

[0046] In the embodiments of the present application, the closed-form solution of the decoding matrix is directly obtained by finding the extreme value of the Lagrange function, which can accelerate the iterative convergence, the initial precoding matrix satisfies the quasi-unitary constraint, which can ensure the orthogonality of signal transmission and reduce the error propagation in the initial iteration. When the decoding matrix is updated each time, the first precoding matrix reflecting the interference transformation is automatically adapted, which can realize the implicit interference alignment. The above process iteratively optimizes the decoding matrix and realizes the efficient and robust uplink signal reception and interference management in the multi-cell multiple-input multiple-output system in combination with the quasi-unitary constraint of the first precoding matrix and the power limit, which is helpful to realize the high-precision convergence of the source signal in the cell and improve the accuracy of the over-the-air computation in the multi-cell scenario.

[0047] In a possible implementation manner of the fifth aspect, the processing unit is further configured to: obtain a third channel matrix of each second terminal device in each neighboring cell to the receiving device of the cell, calculate a third equivalent channel matrix and a conjugate transpose of the third equivalent channel matrix according to the third channel matrix; calculate a conjugate transpose of the first precoding matrix according to the first precoding matrix of each cell; calculate a second equivalent channel matrix of each first terminal device in the cell to the receiving device of the cell; for each neighboring cell of the cell, calculate a third product corresponding to each second terminal device in the cell, and calculate a third product corresponding to each first terminal device in the cell; accumulate the third products corresponding to the terminal devices in the plurality of cells respectively to obtain a third matrix; calculate an inverse matrix of a sum of the third matrix and a covariance matrix of Gaussian white noise; calculate a second product of the second equivalent channel matrix of each first terminal device and the first precoding matrix of the cell, and sum the second products corresponding to the first terminal devices in the cell to obtain a second matrix; and calculate a product of the inverse matrix and the second matrix according to the decoding matrix expression to obtain the decoding matrix.

[0048] In the embodiments of the present application, the third matrix (interference-signal joint covariance matrix) is constructed by accumulating the third product items (interference items) of the terminal devices in the neighboring cells and the third product (useful signal) of the terminal in the cell. The third matrix can quantize the interference introduced by the terminal devices in the neighboring cells through the third channel matrix explicitly, and the effective signal component of the terminal devices in the cell through the second equivalent channel matrix is reserved. The distributed computing architecture is used to calculate the decoding matrix of each cell respectively, so that the efficiency of calculating the decoding matrix can be improved. In addition, the decoding matrix is solved according to the above process, and the solution can minimize the multi-cell mean square error. The theoretical optimal receiving performance can be achieved through the decoding matrix. The receiving device of each cell can realize active suppression of the cross-cell interference signal and the Gaussian white noise signal through the corresponding decoding matrix, which is beneficial to improving the accuracy of the over-the-air computation in the multi-cell scenario.

[0049] In a sixth aspect, the embodiments of the present application provide an electronic device, including a processor and a memory. The memory is configured to store computer execution instructions. The processor is configured to execute the computer execution instructions stored in the memory to execute the method described in the first aspect, the second aspect, or any possible implementation manner of the first aspect and the second aspect.

[0050] In a seventh aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program or instructions. When the computer program or instructions run on a computer, the computer is caused to execute the method described in the first aspect, the second aspect, or any possible implementation manner of the first aspect and the second aspect.

[0051] In an eighth aspect, an embodiment of the present application provides a computer program product including a computer program, which, when executed by a computer, causes the computer to perform the method described in the first aspect, the second aspect, or any possible implementation manner of the first aspect and the second aspect.

[0052] In a ninth aspect, the present application provides a chip or a chip system, which includes at least one processor and a communication interface, the communication interface and the at least one processor are interconnected through a line, the at least one processor is configured to execute a computer program or an instruction to perform the method described in the first aspect, the second aspect, or any possible implementation manner of the first aspect and the second aspect. The communication interface in the chip can be an input / output interface, a pin, or a circuit, etc.

[0053] In a possible implementation, the chip or the chip system described above in the present application further includes at least one memory, and the at least one memory stores the instruction. The memory can be a storage unit inside the chip, for example, a register, a cache, etc., or a storage unit of the chip (for example, a read-only memory, a random access memory, etc.).

[0054] It should be understood that the second aspect to the sixth aspect of the present application correspond to the technical solution of the first aspect of the present application, and the beneficial effects obtained by each aspect and the corresponding possible implementation manner are similar, which will not be described herein again. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A communication system schematic diagram for implementing a multi-cell over-the-air computing network is provided for an embodiment of the present application;

[0056] Figure 2 A step flowchart of a signal sending method is provided for an embodiment of the present application;

[0057] Figure 3 A three-cell over-the-air computing system block diagram based on interference alignment is provided for an embodiment of the present application;

[0058] Figure 4 A step flowchart of a first precoding matrix calculation is provided for an embodiment of the present application;

[0059] Figure 5 A step flowchart of a first precoding matrix calculation at a terminal device end is provided for an embodiment of the present application;

[0060] Figure 6 A step flowchart of a signal receiving method is provided for an embodiment of the present application;

[0061] Figure 7 A step flowchart of a decoding matrix calculation is provided for an embodiment of the present application;

[0062] Figure 8 The decoding matrix provided in the present application is calculated at the receiving device end. DETAILED DESCRIPTION

[0063] In the embodiments of the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration. Any embodiment or design scheme described as "exemplary" or "for example" in the present application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Rather, the use of the words "exemplary" or "for example" is intended to present concepts in a particular manner.

[0064] In the embodiments of the present application, the words "first", "second", and the like are used to distinguish between similar items or similar items with substantially the same function and effect. For example, the first chip and the second chip are merely used to distinguish between different chips, and do not limit the order. Those skilled in the art can understand that the words "first", "second", and the like do not limit the quantity and execution order, and the words "first", "second", and the like do not necessarily mean different.

[0065] It should be noted that in the embodiments of the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration. Any embodiment or design scheme described as "exemplary" or "for example" in the present application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Rather, the use of the words "exemplary" or "for example" is intended to present concepts in a particular manner.

[0066] In the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. The association relationship between the associated objects is described, which means that there can be three relationships, for example, A and / or B, which can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects. "At least one of the following" or similar expressions means any combination of these items, including any combination of single item or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0067] In order to clearly describe the technical solutions of the embodiments of the present application, the following briefly introduces some terms and technologies involved in the embodiments of the present application:

[0068] 1、Terminal device (also referred to as edge device), the terminal device of the embodiment of the application can include handheld devices, vehicle-mounted devices, etc. with wireless communication function. For example, some terminal devices are: mobile phones, tablet computers, palm computers, notebook computers, mobile internet devices (MID), wearable devices, VR devices, AR devices, wireless terminals in industrial control, wireless terminals in self driving, wireless terminals in remote medical surgery, wireless terminals in smart grid, wireless terminals in smart city, wireless terminals in smart home, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDA), handheld devices with wireless communication function, computing devices or other processing devices connected to wireless modems, vehicle-mounted devices, wearable devices, terminal devices in 5G networks, or terminal devices in future evolved public land mobile networks (PLMN), etc. The embodiment of the application is not limited thereto.

[0069] As an example but not limited, in the embodiment of the application, the terminal device can also be a wearable device. The wearable device can also be referred to as a wearable smart device, which is a general term for devices that are designed and developed by applying wearable technology to daily wear, such as glasses, gloves, watches, clothing, and shoes. The wearable device is a portable device that can be directly worn on the body or integrated into the user's clothes or accessories. The wearable device is not only a hardware device, but also a powerful function realized through software support and data interaction, cloud interaction. The general wearable smart device includes full functions, large size, and can realize complete or partial functions without relying on a smart phone, such as smart watches or smart glasses, etc., and focuses on a certain application function and needs to cooperate with other devices such as smart phones, such as various smart wristbands, smart jewelry, etc. for monitoring physical signs. In addition, in the embodiment of the application, the terminal device can also be a terminal device in an internet of things (IoT) system, and the IoT is an important part of future information technology development, and its main technical feature is to connect objects through communication technology and network, so as to realize the intelligent network of man-machine interconnection and object-object interconnection.

[0070] The terminal device in the embodiments of the present application can also be referred to as a user equipment (UE), a mobile station (MS), a mobile terminal (MT), an access terminal, a subscriber unit, a subscriber station, a mobile station, a mobile terminal, a remote station, a remote terminal, a mobile device, a user terminal, a terminal, a wireless communication device, a user agent, or a user device, etc.

[0071] In the embodiments of the present application, the electronic device or each network device includes a hardware layer, an operating system layer running on the hardware layer, and an application layer running on the operating system layer. The hardware layer includes a central processing unit (CPU), a memory management unit (MMU), a memory (also referred to as a main memory), and the like. The operating system can be any one or more computer operating systems that implement business processing through processes, such as a Linux operating system, a Unix operating system, an Android operating system, an iOS operating system, or a windows operating system, etc. The application layer includes a browser, an address book, word processing software, instant messaging software, and the like.

[0072] In the embodiments of the present application, the device for implementing the function of the terminal can be a terminal; or can be a device capable of supporting the terminal to implement the function, such as a chip system, which can be installed in the terminal.

[0073] 2、The receiving device in the embodiment of the present application is an access point (AP) device serving a cell, including a base station (BS) disposed in the cell for connecting the wireless network in the cell with the wired network, which can also be referred to as a base station device, and is a device deployed in a radio access network (RAN) to provide wireless communication functions. For example, the device providing the base station function in the 2G network includes a base transceiver station (BTS), the device providing the base station function in the 3G network includes a NodeB, the device providing the base station function in the 4G network includes an evolved NodeB (eNB), in the wireless local area network (WLAN), the device providing the base station function is an access point (AP), the device providing the base station function in the 5G NR is a gNB, and the device providing the base station function in the continued evolution of the NodeB (ng-eNB), wherein the gNB and the terminal device communicate with each other using the NR technology, the ng-eNB and the terminal device communicate with each other using the evolved universal terrestrial radio access network (E-UTRA) technology, and the gNB and the ng-eNB can be connected to the 5G core network. The receiving device in the embodiment of the present application also includes a device providing the base station function in the future new communication system and the like.

[0074] In the embodiment of the present application, the device for implementing the function of the receiving device can be a network device, or a device capable of supporting the network device to implement the function, such as a chip system, which can be installed in the network device.

[0075] 3、Uplink, refers to the process of sending data from the terminal device to the receiving device.

[0076] 4、Over-the-Air Computation (AirComp), refers to a communication technology that directly completes data aggregation or function calculation in the transmission process by using the physical characteristics of the wireless channel (such as electromagnetic wave superposition). It skips the traditional "transmit first, then calculate" mode, and through multiple devices sending signals at the same time, the signals are naturally superimposed in the wireless channel, so that the receiving end directly obtains the calculation result (such as summation, average, maximum / minimum value, etc.). AirComp includes linear calculation (weighted sum and average, etc.), nonlinear calculation (maximum, minimum, and logical operation, etc.), and function calculation (such as polynomial and exponential function, etc.). Among them, function calculation needs to jointly design the pre-processing strategy of the transmitting end and the post-processing strategy of the receiving end. In this implementation example, it is assumed that the AirComp scheme to be performed is to sum all the source signals of the terminal devices (transmitting end) in the cell.

[0077] 5、Multi-Cell Over-the-Air Computation (MC-AirComp), is a communication-computation fusion architecture that extends the AirComp technology to the multi-base station cooperation scenario, aiming to solve the cross-cell data aggregation problem in large-scale distributed systems (such as Internet of Things, federated learning). The core challenge lies in inter-cell interference, resource allocation, and collaborative computation optimization.

[0078] 6、d-stream signal, the edge device sends signals in parallel through d independent data streams, and the d-stream signal can share a precoding matrix.

[0079] In order to better understand the method provided in the embodiments of the present application, first, the communication system architecture that may be involved in the embodiments of the present application will be described.

[0080] Figure 1 The communication system schematic diagram for implementing the multi-cell AirComp network. As shown in Figure 1 , the communication system includes terminal devices 11, …, and terminal devices 1N in cell 1, terminal devices 21, …, and terminal devices 2N in cell 2; terminal devices 31, …, and terminal devices 3N in cell 3; a receiving device 101 arranged in cell 1, a receiving device 102 arranged in cell 2, a receiving device 103 arranged in cell 3, and a core network device 104.

[0081] Each terminal device has a wireless transceiver function, and can be deployed indoors or outdoors, handheld or vehicle-mounted. The terminal device can be a user equipment (UE), where the UE includes a handheld device, a vehicle-mounted device, a wearable device, or a computing device with a wireless communication function. Illustratively, the UE can be a mobile phone, a tablet computer, or a computer with a wireless transceiver function. The terminal device can also be a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a mixed reality (MR) terminal device, a wireless terminal in industrial control, a wireless terminal in unmanned driving, a wireless terminal in remote medical treatment, a wireless terminal in smart power grids, a wireless terminal in smart cities, a wireless terminal in smart homes, etc. In the embodiments of the present application, the apparatus that implements the function of the edge device can be a terminal device, or an apparatus that can support the terminal device to implement the function, such as a chip system installed in the terminal device.

[0082] Each receiving device in the embodiments of the present application can be a key device connecting the terminal device and the core network, and is an intelligent node that integrates communication, computing, and sensing capabilities. For example, the receiving device can be a small base station, a macro base station, and a multi-access edge computing access point (MEC-AP), etc.

[0083] In the embodiments of the present application, the device for implementing the function of the receiving device can be a receiving device, or an apparatus that supports the receiving device to implement the function, such as a chip system, which can be installed in the receiving device.

[0084] The core network (CN) device 104 includes a user plane function (UPF) network element, an access and mobility management function (AMF) network element, a session management function (SMF) network element, a policy control function (PCF) network element, etc. The UPF network element is mainly responsible for the transmission of user data, and other network elements can be referred to as control plane function network elements, which are mainly responsible for authentication, authorization, registration management, session management, mobility management, and policy control, etc., to ensure reliable and stable transmission of user data.

[0085] In the embodiments of the present application, the device implementing the function of the core network device can be a core network device, or a device, such as a chip system, supporting the core network device to implement the function, which can be installed in the core network device.

[0086] The technical scheme provided by the embodiments of the present application can be applied to a 5G communication system or a communication system after 5G, such as a 6G communication system. In addition, the technical scheme provided by the embodiments of the present application can also be applied to any other wireless communication system with similar structure and function, and the embodiments of the present application do not make any limitation in this regard.

[0087] In Figure 1 , the terminal devices 11~1N in the cell 1 and the receiving device 101 can perform wireless communication, the terminal devices 21~2N in the cell 2 and the receiving device 101 can perform wireless communication, and the terminal devices 31~3N in the cell 3 and the receiving device 101 can perform wireless communication; the terminal devices 11~1N in the cell 1 and the receiving device 102 in the cell 2 can perform wireless communication, the terminal devices 21~2N in the cell 2 and the receiving device 102 can perform wireless communication, and the terminal devices 31~3N in the cell 3 and the receiving device 102 can perform wireless communication; the terminal devices 11~1N in the cell 1 and the receiving device 103 in the cell 3 can perform wireless communication, the terminal devices 21~2N in the cell 2 and the receiving device 103 can perform wireless communication, and the terminal devices 31~3N in the cell 3 and the receiving device 103 can perform wireless communication.

[0088] The receiving devices 101, 102 and 103 can perform wireless communication with the core network device 104.

[0089] The terminal devices in the same cell can perform wireless communication, and the terminal devices in different cells can perform wireless communication.

[0090] The receiving devices in different cells can perform wireless communication.

[0091] Figure 1 In the above-mentioned embodiment, only three cells are taken as an example, and it can be understood that the air computing network composed of multiple cells can be applied to multiple adjacent cells. In the above-mentioned embodiment, the terminal devices and the receiving devices between any two cells can perform wireless communication according to the communication architecture as shown in Figure 1 .

[0092] For the air computing network composed of multiple cells, the task of each cell is to converge the data sent by the terminal devices in the cell and to implement air computing. In the following, the air computing network composed of three cells is taken as an example, and please refer toFigure 1 Taking cell 1 as an example, the receiving device 101 in cell 1 can receive the uplink signals sent by the terminal devices 11, 12, …, 1N in cell 1, and can also receive the uplink signals sent by the terminal devices 21, 22, …, 2N in cell 2 and the uplink signals sent by the terminal devices 31, 32, …, 3N in cell 3.

[0093] For the receiving device 101, the receiving device 101 needs to converge the uplink signals sent by the terminal devices 11, 12, …, 1N in cell 1 to realize the in-air calculation. The uplink signals sent by the terminal devices 21, 22, …, 2N in cell 2 and the uplink signals sent by the terminal devices 31, 32, …, 3N in cell 3 are interference signals. In addition, the receiving device 101 can also receive noise signals. Both the interference signals and the noise signals will adversely affect the in-air calculation of the receiving device 101 on the uplink signals of the terminal devices 11, 12, …, 1N in the cell, and reduce the accuracy of the in-air calculation.

[0094] Based on the above problems, the embodiments of the present application propose a signal sending method and a signal receiving method, and the main invention idea is as follows:

[0095] A minimum multi-cell mean square error function is constructed, and a joint iteration is performed between the multi-cell terminal devices and the multi-cell receiving devices to obtain an expression of a first precoding matrix and an expression of a decoding matrix based on the minimum multi-cell mean square error function, so as to obtain the first precoding matrix and the decoding matrix corresponding to each cell. When the terminal devices in each cell send source signals for in-air calculation, the terminal devices can first use the first precoding matrix corresponding to the cell and a second precoding matrix to perform precoding on the source signals to obtain precoded signals, and then send the precoded signals. The receiving device in the cell decodes the received uplink signals according to the decoding matrix to obtain the converged signals of the source signals sent by the terminal devices in the cell for in-air calculation. In the decoding process, the sum of the source signals sent by the terminal devices in the cell is retained, and the interference signals and the noise signals sent by the terminal devices in the other cells are suppressed, so that the receiving device in the multi-cell in-air calculation network can accurately converge the source signals sent by the terminal devices in the cell, which is beneficial to improving the accuracy of the in-air calculation. The problem of low accuracy of in-air calculation caused by the receiving device receiving the interference signals and the noise signals sent by the terminal devices in other cells is solved.

[0096] The technical solutions shown in the present application will be described in detail below through specific embodiments. It should be noted that the following embodiments can exist independently or in combination. For the same or similar content, such as the explanation of terms or nouns, and the explanation of steps, etc., reference can be made to different embodiments, and the explanation will not be repeated.

[0097] Reference Figure 2 , Figure 2 The above signal sending method can be applied to a terminal device in some embodiments of the present application, and includes:

[0098] S201: The first terminal device generates a source signal for over-the-air computation according to an over-the-air computation instruction.

[0099] The first terminal device here can be any one of a plurality of first terminal devices located in a cell. The cell here can be any one of a plurality of adjacent cells.

[0100] The first terminal device can realize wireless communication with a receiving device of one or more cells through a wireless network.

[0101] The above over-the-air computation can first aggregate the source signals of a plurality of terminal devices in the cell to realize over-the-air computation corresponding to each terminal device in the cell.

[0102] The source signal of the first terminal device can be a signal to be subjected to over-the-air computation. In some examples, the signal can be generated by data collected by a sensor in the first terminal device, can be generated by user input data, or can be generated by the first terminal device according to a data generation rule.

[0103] In some embodiments, the source signal is a signal obtained by preprocessing original data according to a target function for over-the-air computation.

[0104] For example, if the over-the-air computation target function is to find the sum of squares of a plurality of data, the above source signal is obtained by squaring the original data;

[0105] If the over-the-air computation target function is to find the arithmetic mean of a plurality of data, the above source signal is obtained by dividing the original data by the total number of terminal devices participating in the computation in the cell.

[0106] In these embodiments, the data participating in over-the-air computation is first preprocessed by the first terminal device to obtain the source signal, and the waveforms of the source signals of a plurality of first terminal devices are aggregated at the receiving device, so that the purpose of over-the-air computation can be achieved according to the source signal.

[0107] The source signal can be a d-flow complex signal, where d is an integer greater than or equal to 1.

[0108] In S202, the first terminal device performs first precoding on the source signal using a first precoding matrix to obtain a first precoded signal.

[0109] In S203, the first terminal device performs second precoding on the first precoded signal using a second precoding matrix to obtain a precoded signal of the source signal.

[0110] The first precoding matrix is determined based on solving a minimum multi-cell mean square error function, which represents a minimum sum of squares of F-norms of differences between decoded signals of receiving devices in each cell and target signals; and the second precoding matrix is an inverse matrix of a channel matrix between the first terminal device and the receiving device in the cell.

[0111] In some embodiments, the decoded signal of each receiving device is obtained by processing a received uplink signal using a decoding matrix, the uplink signal corresponding to each receiving device includes interference signals transmitted by terminal devices in neighboring cells of the receiving device, precoded signals transmitted by the first terminal devices in the cell of the receiving device, and a Gaussian white noise signal superimposed in the channel; and the target signal corresponding to each receiving device is a converged signal of source signals of the first terminal devices in the cell of the receiving device.

[0112] In this application, the terminal device can be equipped with multiple antennas, and the receiving device can be configured with multiple antennas, forming a multiple-input multiple-output (MIMO) system between the terminal device and the receiving device. Each terminal device can transmit signals to each receiving device using the channel between the terminal device and each receiving device.

[0113] In this application, the wireless communication channel between the terminal device and the receiving device can be a block fading channel. The channel characteristics do not change within a certain time period. Within the time period when the channel does not change, the first precoding matrix, the second precoding matrix, and the decoding matrix also do not change.

[0114] After obtaining the first precoding matrix and the second precoding matrix, the terminal device can use the first precoding matrix and the second precoding matrix to perform precoding on the source signal within the period when the channel characteristics do not change. Then, the receiving device in the cell decodes the received uplink signal using the decoding matrix corresponding to the first precoding matrix.

[0115] In S204, the uplink precoded signal is transmitted.

[0116] To ensure that the terminal device uses the first precoding matrix and the second precoding matrix to precode the source signal, the receiving device can use the decoding matrix corresponding to the first precoding matrix to detect the sum of the source signals of the terminal devices in the cell, it is required that the first precoding matrix and the decoding matrix of the receiving device in the cell satisfy the minimum multi-cell mean square error function.

[0117] In this embodiment, the minimum multi-cell mean square error function is solved by the terminal device and the receiving device, and the first precoding matrix corresponding to each cell in the multi-cell and the decoding matrix on the receiving device side are obtained; when the over-the-air calculation is performed in the multi-cell scenario, the terminal device in each cell uses the first precoding matrix and the second precoding matrix to process the source signal used for over-the-air calculation, so that the source signals of multiple different terminal devices in the same cell participating in the over-the-air calculation are aligned, thereby realizing over-the-air calculation, and at the same time, it can be ensured that when the receiving device decodes using the corresponding decoding matrix, the interference signals and noise signals transmitted by the terminal devices in the adjacent cells can be suppressed, and the accuracy of over-the-air calculation in the multi-cell scenario can be improved.

[0118] The calculation process of the first precoding matrix expression is described in detail below.

[0119] To clearly describe the technical solutions of the embodiments of the present application, it is assumed in this embodiment that the number of terminal devices in each cell is equal, and is Ka; it is assumed that there are Ku adjacent cells in total. Ka is an integer greater than or equal to 1, and Ku is an integer greater than or equal to 2. Figure 3 is a system block diagram of a three-cell over-the-air calculation system based on interference alignment.

[0120] As shown in Figure 3 , the kth terminal device in cell 1 precodes the source signal x [k,1] using the first precoding matrix V1 of cell 1 to obtain the first precoding signal, and then precodes the first precoding signal using the second precoding matrix F [k,1] to obtain the precoding signal of the kth terminal device in cell 1. k is 1, 2, …, Ka. The kth terminal device in cell 1 can send the precoding signal to the receiving device 1 of cell 1, the receiving device 2 of cell 2, and the receiving device 3 of cell 3 through a wireless network.

[0121] The terminal devices in cell 2 and cell 3 send the precoding signals to the receiving devices in the cell and the receiving devices in other adjacent cells, respectively.

[0122] The source signal x sent by each terminal device in each cell satisfies ( , which represents a unit matrix, Let E{∙} denote the conjugate transpose of the source signal x, and E{∙} denote the expected value. Assume the first precoding matrix of the i-th cell is... The second precoding matrix of the k-th terminal device in the i-th cell is . i is greater than or equal to 1, less than or equal to Ku, and i is an integer. ; For the k-th terminal device in the i-th cell and the AP receiving device in the i-th cell i The channel matrix between them. For the k-th terminal device in the j-th cell and the AP receiving device in the i-th cell i The channel matrix between. j is an integer greater than or equal to 1 and less than or equal to Ku. The matrix is ​​represented as an M×M complex matrix. M represents the number of transmitting antennas of the terminal device, where it is assumed in this application that the number of transmitting antennas of the terminal device is the same as the number of receiving antennas of the receiving device. M is an integer greater than or equal to 1.

[0123] based on Figure 3 As shown, the uplink signal received by the receiving device of the i-th cell is characterized by the following formula (1).

[0124] (1);

[0125] in, , The matrix is ​​represented as an M×d-dimensional complex matrix. The vector of Gaussian white noise superimposed in the channel has a mean of The covariance matrix is d represents the number of data streams that the terminal device sends independently in parallel, and d is an integer greater than or equal to 1.

[0126] The above formula (1) can be further represented as the following formula (2):

[0127] (2);

[0128] in, For the k-th device in the j-th cell and the receiving device AP in the i-th cell i The channel matrix between them. The second precoding matrix. For the k-th device in the j-th cell and the receiving device AP in the j-th cell j The inverse of the channel matrix between them.

[0129] The signal obtained by the receiving device of the i-th cell through the decoding matrix processing of the received uplink signal can be represented by the following formula (3):

[0130] (3);

[0131] wherein, is the conjugate transpose of the decoding matrix .

[0132] The AP of the i-th cell i receives a target signal (a converged signal of source signals of the plurality of first terminal devices of the i-th cell, i.e., a target signal) as follows: .

[0133] Taking as an equivalent channel matrix, the decoded received signal can be written as the following formula (4):

[0134] (4).

[0135] Based on the minimum mean square error between the signal obtained after the decoding matrix of the i-th cell and the target signal of the cell , a minimum mean square error function of the following formula (5) is constructed:

[0136] (5);

[0137] is the square of the F-norm of the difference between the decoded signal obtained by the receiving device of the i-th cell using the decoding matrix to decode the received uplink signal and the target signal of the i-th cell.

[0138] The above formula (5) is expanded to obtain the following formula (6):

[0139] (6).

[0140] From the above formula (6), the following formula (7) can be derived:

[0141] (7).

[0142] The above is the mean square error at the AP i of the i-th cell. represents taking the mathematical expectation of the random vector and the random vector . is the signal distortion error of the sum of the source signals sent by the terminal devices of the i-th cell, which measures the difference between the decoded signal and the sum of the source signals of the terminal devices of the i-th cell. Minimizing the objective function makes the signal distortion error tend to 0, i.e., makes approximately equal to The item minimizes the sum of the source signals transmitted by the terminal devices in the i-th cell, and guarantees that the source signals are correctly received; The item represents the interference signal transmitted by the terminal devices from other cells, and the minimization of the objective function makes approximately equal to 0, so that the power of the decoded interference signal is minimized, and interference alignment is achieved. The minimization of the item guarantees that the interference is minimized; The minimization of the item guarantees that the influence of the noise is minimized.

[0143] Further, the problem is extended to minimize the total mean square error of all the air computing cells, and the final minimization multi-cell mean square error function is represented by the following formula (8):

[0144] (8);

[0145] The constraint condition is: ; wherein P is the transmission power of each terminal device. Tr{} is a trace function.

[0146] Since the above optimization problem has an equality constraint, the Lagrange multiplier method can be used to convert it into an unconstrained optimization problem. The general form of the Lagrange function is: = objective function + (constraint term). Wherein λ j is the Lagrange multiplier (scalar) corresponding to each power constraint.

[0147] The constraint term is .

[0148] Therefore, the Lagrange function can be represented by the following formula (9):

[0149] (9);

[0150] Expanding the above formula (9), the following formula (10) can be obtained:

[0151] (10);

[0152] Taking the extreme value of the above Lagrange function with respect to the decoding matrix to solve , the specific process is as follows:

[0153]

[0154] Solving the above formula (11), the decoding matrix of the i-th cell can be obtained, which is represented by the following formula (12):

[0155] (12). Wherein i=1,……,Ku.

[0156] Similarly, the first precoding matrix V is solved by taking the extreme value of the above Lagrange function with respect to the first precoding matrix j , and the specific process is as follows:

[0157]

[0158] According to the above formula, when , and are independent of each other, it can be obtained that

[0159] (14);

[0160] That is:

[0161] (15);

[0162] Therefore, the first precoding matrix of the jth cell is obtained, which is represented by the following formula (16):

[0163] (16); wherein j = 1, …, Ku.

[0164] Wherein, is an unknown scalar.

[0165] According to the above constraint condition, , thus: .

[0166] Let A j = (17);

[0167] B j = (18);

[0168] Then there is the following formula:

[0169] (19);

[0170] Solve this equation by using the bisection method or Newton iteration method, and the can be obtained, wherein,

[0171] ;

[0172] After obtaining , the first precoding matrix can be calculated.

[0173] Through the above process, the expression of the first precoding matrix of the terminal device of each cell and the expression of the decoding matrix of the receiving device of each cell can be obtained. For the terminal device of each cell, when the channel between the terminal device and the receiving device of the cell changes, the first precoding matrix of the terminal device of the cell can be iteratively calculated through the expression of the first precoding matrix.

[0174] Specifically, refer to Figure 4 , Figure 4 The flowchart of the calculation steps of the first precoding matrix provided by the present disclosure. In some embodiments of the present application, the calculation steps of the first precoding matrix are performed by the terminal device, and the above steps include:

[0175] S401: The first terminal device detects whether the channel between the terminal device and the receiving device in the cell changes.

[0176] It should be noted that the plurality of terminal devices of the plurality of cells and the plurality of receiving devices of the plurality of cells can be jointly performed by the plurality of terminal devices of the plurality of cells and the receiving devices in parallel in the process of calculating the first precoding matrix. The terminal device can periodically detect whether the channel between the terminal device and the receiving device in the cell changes.

[0177] S402: In response to the change of the channel, the first terminal device sends a pilot signal to the receiving device, and receives a channel matrix calculated according to the pilot signal sent by the receiving device.

[0178] Wherein, the receiving device receives the receiving signal generated by the channel through the pilot signal, and calculates the channel matrix between the terminal device and the receiving device according to the receiving signal and the pilot signal.

[0179] The pilot signal is a specific signal known to the terminal device and the receiving device in the same cell. For a multiple-input multiple-output system, the receiving signal model of the receiving device can be represented by the following formula (21):

[0180] (21) ;

[0181] Wherein, ∈C Nr×L The received pilot matrix (N r is the number of receiving antennas, and L is the number of pilot symbols); H∈C Nr×Nt ; Nt is the number of transmitting antennas. ∈C Nt×L , the sending pilot matrix, (L≥Nt). N is the noise matrix.

[0182] The channel matrix can be estimated by the following formula (22):

[0183] (22) ;

[0184] By the above formula, the channel matrix between each terminal device of the ith cell and the receiving device in the cell can be estimated , and the channel matrix between the kth terminal device of the jth cell and the receiving device of the ith cell ; wherein i is not equal to j.

[0185] S403: The first terminal device randomly generates a set of initial first precoding matrices satisfying the quasi-unitary constraint, and sends the initial first precoding matrices to the receiving device in the cell for iterative calculation of the first precoding matrix.

[0186] In the case that the channel matrix between the terminal device and the receiving device of each cell is known, the terminal device in each cell can randomly generate a set of initial first precoding matrices satisfying the quasi-unitary constraint, which are V1, V2, …, V Ku The terminal device in each cell can send the initial first precoding matrix to the receiving device AP in the cell.

[0187] Quasi-unitary constraint: refers to the first precoding matrix V j needs to satisfy a mathematical condition, the core purpose of which is to ensure the power conservation and orthogonality of signal transmission, so as to avoid power leakage or interference amplification.

[0188] Specifically, the initial first precoding matrix satisfies the following condition: .

[0189] That is, the initial first precoding matrix V satisfies column orthogonality. Column orthogonality means that the column vectors are orthogonal to each other and have a norm of 1.

[0190] S404: Obtain the decoding matrix sent by the receiving device of each cell, wherein for each cell, the decoding matrix of the cell is calculated by the receiving device according to the initial first precoding matrix sent by the terminal device of each cell.

[0191] When the receiving device receives the initial first precoding matrix of the terminal device of each cell, it can update according to the decoding matrix expression according to the first precoding matrix of each cell to obtain the updated decoding matrix. Each receiving device can share the decoding matrix of each cell through the central unit connected by wire, and then the receiving device of each cell sends all the updated decoding matrices to each first terminal device in the cell.

[0192] S405: Based on the expression of the first precoding matrix and the decoding matrix corresponding to each cell, the first precoding matrix is iteratively calculated.

[0193] After receiving the decoding matrix of the current iteration sent by the receiving device of each cell, the first terminal device can substitute the decoding matrix corresponding to each cell into the expression of the first precoding matrix to calculate the first precoding matrix of the cell.

[0194] The first precoding matrix obtained in this calculation is sent to the receiving device. The receiving device receives the decoding matrix updated again according to the first precoding matrix, and then sends the updated decoding matrix to the terminal device to update the first precoding matrix. The above process is repeated until the iteration meets the preset condition.

[0195] In the above steps S404 and S405, the receiving device of each cell feeds back the interference information (implicit in the decoding matrix) through the local decoding matrix, without centralized global channel state information, which can reduce the overhead of the backhaul link. The decoding matrix is calculated by the terminal device initial precoding matrix, which actually reflects the actual interference coupling relationship among multiple cells. The interference information is transmitted through the decoding matrix in the iteration process, so that the final precoding matrix has the interference alignment characteristic, which can improve the system efficiency. In addition, the optimal solution is gradually approached through the iterative algorithm, and an acceptable solution can be reached after a limited number of iterations. Compared with directly solving the high-dimensional non-convex optimization problem, the computational complexity is significantly reduced.

[0196] In some embodiments, the expression of the first precoding matrix in the above step S405 is obtained based on the following steps:

[0197] Based on taking the extreme value of the Lagrange function with respect to the first precoding matrix, the expression of the first precoding matrix is obtained; wherein the Lagrange function is constructed according to the minimization of the multi-cell mean square error function and the terminal device transmit power constraint.

[0198] The expression of the Lagrange function can refer to the above formula (9).

[0199] In these embodiments, by constructing the Lagrange function according to the minimization of the multi-cell mean square error function and the transmit power constraint, the optimization problem with equality constraints is converted into an unconstrained optimization problem using the Lagrange multiplier method. By taking the partial derivative of the Lagrange function with respect to the first precoding matrix, the explicit expression of the first precoding can be derived, which is convenient for subsequent iteration implementation and avoids black box optimization. In addition, the Lagrange function can balance the signal quality and power efficiency control, improve the signal-to-interference-and-noise ratio by minimizing the multi-cell mean square error, and avoid terminal device overload through power constraint, which conforms to the actual terminal device limit and avoids terminal overload through power constraint. It can effectively suppress the interference signals of adjacent cells, further improve the accuracy of air computing, and conform to the actual device limit.

[0200] In this embodiment, the initial first precoding matrix is generated when the channel between the terminal device and the receiving device changes, and the receiving device iteratively calculates the first precoding matrix of the terminal device, so that the first precoding matrix of the terminal device of multiple cells can be obtained through the above iteration operation, so that the air calculation result in the multi-cell scenario can be obtained accurately when the channel changes.

[0201] In some embodiments, as shown in Figure 5 Figure 5 The flowchart of the calculation steps of the first precoding matrix at the terminal device side is shown in Figure 5 The calculation steps of the first precoding matrix at the terminal device side are as follows:

[0202] S51: Calculate the Lagrange multiplier according to the decoding matrix corresponding to each cell.

[0203] Specifically, the decoding matrix corresponding to each cell is substituted into formulas (17) and (18) to obtain and Then, the matrices and are substituted into formula (19), and the Lagrange multiplier is calculated by using the bisection method or Newton iteration method on formula (19). In this way, the Lagrange multiplier in the expression of the first precoding matrix corresponding to the current cell can be obtained.

[0204] S52: Update the first precoding matrix according to the Lagrange multiplier and the decoding matrix corresponding to each cell.

[0205] The first precoding matrix expression can be updated according to the above formula (16).

[0206] S53: Replace the initial first precoding matrix with the updated first precoding matrix to iteratively calculate the first precoding matrix.

[0207] The terminal device can send the updated first precoding matrix to the receiving device of the current cell to start the next update of the first precoding matrix and the decoding matrix.

[0208] S54: In response to the iterative calculation satisfying the preset condition, the first precoding matrix obtained when the preset condition is satisfied is taken as the final first precoding matrix.

[0209] In some embodiments, the above-mentioned preset condition can be that the iteration number reaches a predetermined value.

[0210] In these embodiments, the iteration number reaching the predetermined value can be used as the iteration end condition, which can reduce the calculation of judging whether the intermediate data in the iteration meets the iteration stop condition, save resources, and quickly obtain the first precoding matrix.​

[0211] In these embodiments, the Lagrange multiplier is calculated according to the decoding matrix of each cell, and then the first precoding matrix is updated according to the Lagrange multiplier and the decoding matrix of each cell described above, to obtain an updated first precoding matrix for iteration. When the preset condition is met in the iterative calculation, the final first precoding matrix is obtained, which realizes the calculation of the first precoding matrix and further helps to realize the suppression of interference or noise according to the first precoding matrix. In the process of calculating the first precoding matrix, the Lagrange multiplier is used to update the first precoding matrix. Since the Lagrange multiplier dynamically adjusts the weight of the minimum mean square error minimization and the power constraint of multiple cells implicitly, the first precoding matrix can adapt to different channel conditions.

[0212] In some embodiments, the step S51 includes the following sub-steps:

[0213] First, for the current cell, the first product corresponding to each first terminal device in the current cell is calculated, and the first products corresponding to each first terminal device in the current cell are accumulated to obtain a first matrix. The first product corresponding to each first terminal device in the current cell is the product of the conjugate transpose of the first equivalent channel matrix between the first terminal device and the receiving device of its adjacent cell, the decoding matrix of the adjacent cell, the conjugate transpose of the decoding matrix of the adjacent cell, and the first equivalent channel matrix of the adjacent cell; and the product of the conjugate transpose of the second equivalent channel matrix between the first terminal device and the receiving device of the current cell, the decoding matrix of the current cell, the conjugate transpose of the decoding matrix of the current cell, and the second equivalent channel matrix.

[0214] The first matrix can be shown as formula (17). Through the first sub-step described above, the first matrix can be calculated.

[0215] Second, the second equivalent channel matrix is determined according to the second channel matrix between each first terminal device and the receiving device of the current cell, the second product of the conjugate transpose of the second equivalent channel matrix and the decoding matrix corresponding to each first terminal device in the current cell is calculated, and the second products corresponding to each first terminal device are accumulated to obtain a second matrix.

[0216] The second matrix can be shown as formula (18). Through the second sub-step described above, the second matrix can be calculated.

[0217] Third, the Lagrange multiplier equation is solved based on the first matrix and the second matrix to obtain the Lagrange multiplier.

[0218] The Lagrange multiplier equation is shown as formula (19). The first matrix and the second matrix can be substituted into formula (19) to obtain the Lagrange multiplier of the current cell by using the bisection method or the Newton iteration algorithm.

[0219] In the above process, the first matrix and the second matrix are calculated, and then the Lagrange multiplier is calculated using the first matrix and the second matrix. The solving process of the Lagrange multiplier realizes closed-form solution instead of numerical search, which can accelerate convergence. The Lagrange multiplier is dynamically updated in the iteration process, avoiding falling into local optimum.

[0220] In some embodiments, the step S52 includes the following sub-steps:

[0221] The Lagrange multiplier and the decoding matrix of each cell are substituted into the expression of the first precoding matrix to obtain the updated first precoding matrix. It should be noted that the terminal devices of the plurality of cells can calculate the respective Lagrange multipliers in parallel, and update the respective first precoding matrices according to the obtained Lagrange multipliers in parallel.

[0222] In these embodiments, the first precoding matrix is updated according to the Lagrange multiplier and the decoding matrix of each cell, so that the first precoding matrix capable of suppressing the interference signal can be obtained by the iterative solving method.

[0223] Please refer to Figure 6 , Figure 6 The steps of a signal receiving method provided by the embodiments of the present application are shown in the flowchart. In some embodiments of the present application, the method can be applied to the receiving device of each cell for over-the-air computation in a multi-cell, including:

[0224] S601: receiving an uplink signal; the uplink signal includes a precoded signal transmitted by at least one first terminal device in the cell according to a first precoding matrix and a second precoding matrix, an interference signal transmitted by at least one second terminal device in at least one adjacent cell, and a Gaussian white noise signal superimposed in the channel; the source signal is a signal for over-the-air computation.

[0225] Each cell can be provided with a receiving device. The receiving device can receive the precoded signal transmitted by each first terminal device in the cell, and the interference signal transmitted by the plurality of second terminal devices in other cells. In addition, the receiving device also receives the Gaussian white noise signal superimposed in the channel.

[0226] The precoded signal of each terminal device can be transmitted by the terminal device through the steps S201-S203 shown in the above Figure 2 , which will not be described here.

[0227] S602: decoding the uplink signal according to the decoding matrix to obtain the aggregated signal of the source signal of each first terminal device.

[0228] The decoding matrix is ​​determined by minimizing the multi-cell mean square error function; where minimizing the multi-cell mean square error function represents minimizing the sum of the squares of the F-norms of the differences between the decoded signals and the target signals of each of the multi-cell receiving devices; the decoded signal of each receiving device is obtained by processing the received uplink signal using the decoding matrix.

[0229] The decoding matrices of the receiving devices in each of the multiple cells, and the first precoding matrices of the terminal devices in each cell, satisfy the condition of minimizing the multi-cell mean square error function. Specifically, this is obtained by jointly iteratively solving for the minimum mean square error function.

[0230] When processing uplink signals, the aforementioned decoding matrix can decode the converged source signals of each first terminal device in the cell and suppress interference signals sent by second terminal devices in adjacent cells.

[0231] In this embodiment, the terminal device and the receiving device can jointly solve for minimizing the multi-cell mean square error function to obtain the first precoding matrix corresponding to each cell in the multi-cell system and the decoding matrix on the receiving device side. The aforementioned decoding matrix can achieve source signal aggregation within the cell while suppressing cross-cell interference signals and Gaussian white noise signals. When performing over-the-air computation in a multi-cell scenario, each terminal device in each cell processes the source signal used for over-the-air computation using the first and second precoding matrices and then transmits the precoded signal of the source signal uplink. When the receiving device decodes using the aforementioned decoding matrix, it can suppress interference signals transmitted by terminal devices in adjacent cells and channel Gaussian white noise signals, achieving high-precision aggregation of the source signal within the cell and improving the accuracy of over-the-air computation in multi-cell scenarios.

[0232] For each cell's receiving device, when the channel between the terminal device and the receiving device in this cell changes, the decoding matrix of the receiving device in this cell can be iteratively calculated using the expression of the decoding matrix.

[0233] For details, please refer to Figure 7 , Figure 7 The flowchart illustrating the steps for calculating the decoding matrix provided in this application shows that, in some embodiments of this application, the step of calculating the decoding matrix is ​​performed by the receiving device, and the steps include:

[0234] S701: Obtain the first precoding matrix sent by the terminal devices in each cell.

[0235] In the process of receiving device and terminal device interaction iterative calculation of decoding matrix and first precoding matrix, each receiving device can receive the first precoding matrix sent by each terminal device in the cell. Then, the first precoding matrix of the cell is shared with the receiving devices of other cells via the central unit connected by wire. Therefore, the receiving device of each cell can obtain the first precoding matrix sent by the terminal devices of each cell.

[0236] S702: According to the decoding matrix expression and the first precoding matrix of each cell terminal device, the decoding matrix is iteratively calculated; wherein the initial first precoding matrix sent by each cell terminal device satisfies the quasi-unitary constraint; wherein the decoding matrix expression is determined by taking the extreme value of the Lagrange function with respect to the decoding matrix, and the Lagrange function is constructed according to the minimum multi-cell mean square error function and the terminal device transmission power constraint.

[0237] S703: In response to the iteration satisfying the preset condition, the decoding matrix obtained when the preset condition is satisfied is taken as the final decoding matrix of the cell.

[0238] The decoding matrix expression is shown in formula (12).

[0239] In these embodiments, by taking the extreme value of the Lagrange function, the closed-form solution of the decoding matrix is directly obtained, which can accelerate the iterative convergence, and the initial precoding matrix satisfies the quasi-unitary constraint, which can ensure the orthogonality of signal transmission and reduce the error propagation in the initial iteration. When updating the decoding matrix each time, the first precoding matrix reflecting the current interference transformation is automatically adapted, which can realize implicit interference alignment. The above process iteratively optimizes the decoding matrix, and jointly considers the quasi-unitary constraint and power limitation of the first precoding matrix, thereby realizing efficient and robust uplink signal reception and interference management in the multi-cell multiple-input multiple-output system, which is helpful to realize high-precision convergence of the source signal of the cell, and can improve the accuracy of air computing in the multi-cell scenario.

[0240] In some embodiments, as shown in the flow chart of the decoding matrix calculation step, the decoding matrix is calculated based on the following steps: Figure 8

[0241] S81: Obtain the third channel matrix from each second terminal device in each adjacent cell to the receiving device of the cell, calculate the third equivalent channel matrix and the conjugate transpose of the third equivalent channel matrix according to the third channel matrix.

[0242] For each cell, the terminal device in the cell can be regarded as the first terminal device of the cell, and any terminal device of the adjacent cell is called the second terminal device.

[0243] The above third channel matrix can be calculated according to Figure 4 ​Steps S401-S402 of the illustrated embodiment are obtained.

[0244] S82: Calculate the conjugate transpose of the first precoding matrix according to the first precoding matrix of each cell.

[0245] S83: Calculate the second equivalent channel matrix of each first terminal device in the cell to the receiving device in the cell.

[0246] S84: For each neighboring cell of the cell, calculate the third product corresponding to each second terminal device in the cell, calculate the third product corresponding to each first terminal device in the cell; accumulate the third products corresponding to the terminal devices of each cell to obtain a third matrix; wherein the third product corresponding to each second terminal device in each neighboring cell is the product of the third equivalent channel matrix of the second terminal device in the cell, the first precoding matrix of the cell, the conjugate transpose of the first precoding matrix of the cell, and the conjugate transpose of the third equivalent channel matrix; the third product corresponding to each first terminal device in the cell is the product of the second equivalent channel matrix of the first terminal device in the cell, the first precoding matrix of the cell, the conjugate transpose of the first precoding matrix of the cell, and the conjugate transpose of the second equivalent channel matrix.

[0247] S85: Calculate the inverse matrix of the sum of the third matrix and the covariance matrix of the Gaussian white noise.

[0248] S86: Calculate the second equivalent channel matrix of each first terminal device in the cell to the receiving device in the cell, calculate the second product of the second equivalent channel matrix of each first terminal device and the precoding matrix of the cell; sum the second products corresponding to each first terminal device in the cell to obtain a second matrix.

[0249] S87: According to the decoding matrix expression, calculate the product of the inverse matrix and the second matrix to obtain the decoding matrix.

[0250] Specifically, the expression of the decoding matrix is shown in formula (12). is the third matrix; is the second matrix. Substitute the third matrix and the second matrix into formula (12) to obtain the decoding matrix.

[0251] In these embodiments, the decoding matrix is obtained through the above iterative calculation, which further helps to realize the suppression of the interference or noise of the terminal devices of the neighboring cells according to the decoding matrix.

[0252] In the embodiments, a third matrix (a joint covariance matrix of the interference signal and the useful signal) is constructed by accumulating the third product items (interference items) of the terminal devices of the adjacent cells and the third product (useful signal) of the terminal of the cell: the third matrix can explicitly quantify the interference introduced by the terminal devices of the adjacent cells through the third channel matrix; and the effective signal component of the terminal device of the cell through the second equivalent channel matrix is reserved. The respective decoding matrices of the cells are respectively calculated through the distributed computing architecture, so that the efficiency of calculating the decoding matrix can be improved. In addition, the above-mentioned process of solving the decoding matrix can achieve the minimum multi-cell mean square error, and the theoretical optimal receiving performance can be achieved through the decoding matrix. The receiving device of each cell can achieve active suppression of the cross-cell interference signal and the Gaussian white noise signal through the corresponding decoding matrix, which is beneficial to improving the accuracy of the over-the-air computation in the multi-cell scenario.

[0253] The embodiments of the present application further provide a signal processing system, comprising a plurality of terminal devices located in a plurality of cells, and a receiving device located in each cell; wherein,

[0254] The first terminal device of each cell uses a first precoding matrix of the cell to perform first precoding on a source signal used for over-the-air computation, to obtain a first precoded signal, and uses a second precoding matrix to perform second precoding on the first precoded signal, to obtain a precoded signal of the source signal; and the precoded signal is sent uplink.

[0255] The receiving device of each cell receives the uplink signal, and uses a decoding matrix to perform decoding on the uplink signal, to obtain a converged signal of the source signal of each first terminal device of the cell; the uplink signal comprises the precoded signal sent by each first terminal device of the cell, the interference signal sent by at least one second terminal device of an adjacent cell, and a Gaussian white noise signal.

[0256] The first precoding matrix and the decoding matrix are determined based on solving a minimum mean square error function; the minimum multi-cell mean square error function represents that the F-norm square of the difference between the signal obtained by using the decoding matrix to decode the received uplink signal by the receiving device of the cell and the converged signal of the source signal of each first terminal device of the cell is minimum; and the second precoding matrix is the inverse matrix of the channel matrix between the first terminal device and the receiving device of the cell.

[0257] In this embodiment, the minimum multi-cell mean square error function is solved by combining the terminal devices of different cells and the receiving devices of different cells, and the first precoding matrix corresponding to each cell in the multi-cell and the decoding matrix on the receiving device side are obtained. The decoding matrix can realize the convergence of the source signals of the cell while suppressing the cross-cell interference signals and Gaussian white noise signals. When the over-the-air calculation is performed in the multi-cell scenario, after each terminal device in each cell processes the source signals used for over-the-air calculation using the first precoding matrix and the second precoding matrix, the receiving device can suppress the interference signals sent by the terminal devices of the adjacent cells and the channel Gaussian white noise signals when decoding the received uplink signals using the decoding matrix, and can realize high-precision convergence of the source signals of the cell and improve the accuracy of over-the-air calculation in the multi-cell scenario.

[0258] In some embodiments, the first precoding matrix and the decoding matrix are obtained by the terminal devices and the receiving devices through the following operations:

[0259] First, the terminal devices of each cell generate an initial first precoding matrix and send the initial first precoding matrix to the receiving devices of the cell.

[0260] Second, the terminal devices of each cell and the receiving devices of each cell perform the following joint interference alignment iteration operation.

[0261] First, the receiving devices of each cell receive the first precoding matrix sent by the terminal devices of the cell, and then share the first precoding matrix of the cell with the receiving devices of other cells via the central unit connected to them by wire. The first precoding matrix of the terminal devices of the cell received by each receiving device of each cell for the first time is the initial first precoding matrix.

[0262] Second, each receiving device calculates the decoding matrix according to the decoding matrix expression and the first precoding matrix sent by the terminal devices of each cell to obtain the decoding matrix used for iterative calculation. Each receiving device can share the decoding matrix of each cell via the central unit connected to it by wire, and then the receiving devices of each cell send all the decoding matrices to each terminal device of the cell. Then, each terminal device calculates the Lagrange multiplier based on the decoding matrix of each cell; updates the first precoding matrix according to the Lagrange multiplier to obtain the first precoding matrix used for iterative calculation; and sends the first precoding matrix to the receiving devices of the cell for the next iteration operation.

[0263] Finally, each terminal device stores the first precoding matrix obtained when the iteration operation stops satisfying the preset condition as the final first precoding of the cell, and each receiving device stores the decoding matrix obtained when the iteration operation stops satisfying the preset condition as the final decoding matrix.

[0264] wherein the expression of the first precoding matrix is obtained by solving a Lagrange function for the first precoding matrix, and the expression of the decoding matrix is obtained by solving the Lagrange function for the decoding matrix; the Lagrange function is constructed by the minimum mean square error function and the terminal device transmit power constraint.

[0265] In the embodiments, the expression of the first precoding matrix and the expression of the decoding matrix are obtained by using a Lagrange function constructed by using a preset constraint and a minimum multi-cell mean square error function, and the first precoding matrix and the decoding matrix of each cell are calculated by jointly iterating the terminal devices and the receiving device of each cell by randomly generating an initial first precoding matrix. Since the first precoding matrix and the decoding matrix are obtained by solving the above Lagrange function, the source signal of the terminal device is precoded by using the first precoding matrix, the precoded signal is decoded by using the decoding matrix, the decoded signal of each cell satisfies the minimum multi-cell mean square error function, the receiving device of each cell can effectively detect the aggregated signal of the source signal of the terminal device in the cell and suppress the interference signal and the noise signal of the adjacent cell, and the accuracy of the over-the-air computation is improved.

[0266] The method in the embodiments of the present application has been described above, and the device for executing the above method provided by the embodiments of the present application is described below. It can be understood by those skilled in the art that the method and the device can be combined and referenced with each other, the related device provided by the embodiments of the present application can execute the steps in the above method, and can also achieve the beneficial effects possessed by the above method embodiments.

[0267] In some embodiments, the present application provides a signal sending device, which can be arranged in a terminal device, and the device comprises:

[0268] a first precoding module, configured to perform first precoding on the source signal of the first terminal device based on the first precoding matrix to obtain a first precoded signal; wherein the source signal is used for over-the-air computation;

[0269] a second precoding module, configured to perform second precoding on the first precoded signal based on the second precoding matrix to obtain a target precoded signal of the source signal;

[0270] a sending module, configured to send the precoded signal obtained by the second precoding in uplink;

[0271] wherein the first precoding matrix is determined based on solving a minimum multi-cell mean square error function, and the minimum multi-cell mean square error function represents the minimum of the sum of squares of the F-norm of the difference between the decoding signal of the receiving device of each cell and the target signal; and the second precoding matrix is the inverse matrix of the channel matrix between the first terminal device and the receiving device of the cell.

[0272] In the embodiment, the minimum multi-cell mean square error function is solved by the joint terminal device and receiving device to obtain the first precoding matrix corresponding to each cell in the multi-cell and the decoding matrix on the receiving device side; when the over-the-air computation is performed in the multi-cell scenario, each terminal device in each cell uses the first precoding matrix and the second precoding matrix to process the source signal used for over-the-air computation, so that the source signals of multiple different terminal devices in the same cell are aligned to participate in the over-the-air computation, thereby realizing the over-the-air computation, and meanwhile, the receiving device can suppress the interference signals and noise signals transmitted by the terminal devices in the adjacent cells when decoding by using the corresponding decoding matrix, and the accuracy of the over-the-air computation in the multi-cell scenario can be improved.

[0273] In one embodiment, the decoding signal of each receiving device is obtained by processing the received uplink signal by using the decoding matrix of the receiving device, the uplink signal of each receiving device includes the interference signal transmitted by at least one terminal device in the adjacent cell of the receiving device, the precoding signal transmitted by multiple first terminal devices in the cell of the receiving device, and the Gaussian white noise signal superimposed in the channel; and the target signal corresponding to each receiving device is the converged signal of the source signals of multiple terminal devices in the cell of the receiving device.

[0274] In one embodiment, the signal transmission device further includes a first encoding matrix determination module configured to obtain the first precoding matrix based on the following first operation:

[0275] The decoding matrix transmitted by the receiving device of each cell is obtained, wherein the decoding matrix is calculated by the receiving device based on the initial first precoding matrix transmitted by the terminal device of each cell; the initial first precoding matrix transmitted by each terminal device to the corresponding access point for the first time is generated by the terminal device of each cell and satisfies the quasi-unitary constraint;

[0276] The first precoding matrix is iteratively calculated based on the expression of the first precoding matrix and the decoding matrix of each cell.

[0277] In the embodiment, the receiving device of each cell feeds back the interference information (implicit in the decoding matrix) by using the local decoding matrix, without the centralized global channel state information, so that the overhead of the backhaul link can be reduced, the decoding matrix is calculated by the terminal device initial precoding matrix, and the actual interference coupling relationship among the multi-cells is actually reflected. The interference information is transmitted by the decoding matrix in the iteration process, so that the final precoding matrix has the interference alignment characteristic, and the system efficiency can be improved; in addition, the optimal solution is gradually approached by the iteration algorithm, and an acceptable solution result can be obtained after a limited number of iterations. Compared with directly solving the high-dimensional non-convex optimization problem, the calculation complexity is significantly reduced.

[0278] In one embodiment, the first precoding matrix determining module is further configured to determine the expression of the first precoding matrix based on the following steps:

[0279] The expression of the first precoding matrix is obtained by taking the extreme value of the Lagrange function with respect to the first precoding matrix, wherein,

[0280] The Lagrange function is constructed according to the minimum mean square error function and the terminal device transmission power constraint.

[0281] In the embodiment, by constructing the Lagrange function according to the minimum multi-cell mean square error function and the transmission power constraint, the optimization problem with equality constraints is converted into an unconstrained optimization problem using the Lagrange multiplier method. By taking the partial derivative of the Lagrange function with respect to the first precoding matrix, the explicit expression of the first precoding can be derived, which is convenient for subsequent iterative implementation. In addition, the Lagrange function can balance the signal quality and power efficiency control, improve the signal-to-noise ratio by minimizing the multi-cell mean square error, and avoid terminal device overload through power constraint, which conforms to the actual terminal device limit and avoids terminal overload through power constraint. It can effectively suppress the interference signals of adjacent cells and further improve the accuracy of air computing. It conforms to the actual device limit.

[0282] In one embodiment, the first precoding matrix determining module is further configured to calculate the Lagrange multiplier according to the decoding matrix corresponding to each cell,

[0283] update the first precoding matrix according to the Lagrange multiplier and the decoding matrix corresponding to each cell;

[0284] replace the initial first precoding matrix with the updated first precoding matrix to iteratively calculate the first precoding matrix;

[0285] In response to the iterative calculation satisfying the preset condition, the first precoding matrix obtained when the preset condition is satisfied is taken as the final first precoding matrix.

[0286] In the embodiment, in the process of iteratively calculating the first precoding matrix, the Lagrange multiplier is used to update the first precoding matrix. Since the Lagrange multiplier dynamically adjusts the weight of the minimum multi-cell mean square error and the power constraint, the first precoding matrix can adapt to different channel conditions.

[0287] In one embodiment, the preset condition includes:

[0288] The number of iterations reaches a predetermined value.

[0289] In one embodiment, the first encoding matrix determination module is further configured to: in the plurality of adjacent cells, for the current cell, calculate a first product corresponding to each first terminal device in the current cell, and accumulate the first products corresponding to each first terminal device in the current cell to obtain a first matrix; wherein the first product corresponding to each first terminal device in the current cell is a product of a conjugate transpose of a first equivalent channel matrix between the first terminal device and a receiving device in a neighboring cell of the current cell, a decoding matrix of the neighboring cell, a conjugate transpose of the decoding matrix of the neighboring cell, and a first equivalent channel matrix of the neighboring cell, and a product of a conjugate transpose of a second equivalent channel matrix between the first terminal device and a receiving device in the current cell, a decoding matrix of the current cell, a conjugate transpose of the decoding matrix of the current cell, and the second equivalent channel matrix;

[0290] According to the second channel matrix between each first terminal device in the current cell and a receiving device in the current cell, a second equivalent channel matrix is determined, a second product of a conjugate transpose of the second equivalent channel matrix and a decoding matrix corresponding to each first terminal device in the current cell is calculated, and the second products corresponding to each first terminal device are accumulated to obtain a second matrix;

[0291] Based on the first matrix and the second matrix, a Lagrange multiplier equation is solved to obtain the Lagrange multiplier.

[0292] In the embodiment, the first matrix and the second matrix are calculated, the Lagrange multiplier is calculated based on the first matrix and the second matrix, and then the first precoding matrix is updated according to the Lagrange multiplier and the decoding matrix of each cell, so that the first precoding matrix capable of suppressing the interference signal can be obtained by an iterative solution method. In the above process, a closed-form solution is realized instead of a numerical search, and convergence can be accelerated. The Lagrange multiplier is dynamically updated in the iteration process, and local optimization is avoided.

[0293] In one embodiment, the first precoding matrix determination module is further configured to: bring the Lagrange multiplier and the decoding matrix of each cell into an expression of the first precoding matrix to obtain an updated first precoding matrix.

[0294] In some embodiments, the present application provides a signal receiving device, which can be arranged in a receiving device of a cell, and the device comprises:

[0295] The receiving module is configured to receive an uplink signal; the uplink signal comprises a precoding signal obtained by precoding a source signal of each first terminal device in the current cell according to a first precoding matrix and a second precoding matrix, an interference signal transmitted by at least one second terminal in at least one adjacent cell, and a Gaussian white noise signal superimposed in a channel; the source signal is a signal for over-the-air computation.

[0296] a decoding module configured to decode the uplink signals according to a decoding matrix to obtain a converged signal of the source signals of the at least one first terminal device;

[0297] The decoding matrix is determined based on solving a minimum multi-cell mean square error function, the minimum multi-cell mean square error function represents a minimum of a sum of squares of F-norms of differences between decoded signals of the receiving devices of the cells and target signals, and the decoded signal of each receiving device is obtained by processing the received uplink signals using the decoding matrix.

[0298] In the embodiment, the minimum multi-cell mean square error function is solved in combination with the terminal devices and the receiving devices to obtain the first precoding matrix corresponding to each cell in the multi-cell and the decoding matrix on the receiving device side; the decoding matrix can converge the source signals of the cell and suppress the cross-cell interference signals and the Gaussian white noise signals. In the multi-cell scenario, after the terminal devices in each cell process the source signals for the over-the-air computation using the first precoding matrix and the second precoding matrix to send the precoded signals of the uplink source signals, the receiving device can suppress the interference signals sent by the terminal devices of the adjacent cells and the channel Gaussian white noise signals when decoding the received uplink signals using the decoding matrix, can converge the source signals of the cell with high precision, and can improve the accuracy of the over-the-air computation in the multi-cell scenario.

[0299] In some embodiments, the signal receiving apparatus further comprises a decoding matrix determination module configured to determine the decoding matrix based on a first operation as follows:

[0300] obtaining the first precoding matrix sent by the terminal devices of the cells;

[0301] updating the decoding matrix according to the decoding matrix expression and the first precoding matrix of the terminal devices of the cells to iteratively calculate the decoding matrix; the initial first precoding matrix sent by the terminal devices of the cells satisfies the quasi-unitary constraint; the decoding matrix expression is determined by taking the extreme value of a Lagrange function with respect to the decoding matrix, and the Lagrange function is constructed according to the minimum mean square error function and the terminal device transmission power constraint;

[0302] in response to the iterative calculation satisfying a preset condition, taking the decoding matrix obtained when the preset condition is satisfied as the final decoding matrix of the cell.

[0303] In the embodiment, a closed-form solution of the decoding matrix is directly obtained by taking the extreme value of the Lagrange function, the iterative convergence can be accelerated, the initial precoding matrix satisfies the quasi-unitary constraint, the orthogonality of signal transmission can be ensured, and the error propagation in the initial iteration stage is reduced. When the decoding matrix is updated each time, the first precoding matrix reflecting the current interference transformation is automatically adapted, and implicit interference alignment can be realized. The above process iteratively optimizes the decoding matrix, and realizes efficient and robust uplink signal reception and interference management in the multi-cell MIMO system in combination with the quasi-unitary constraint and the power limitation of the first precoding matrix, which is helpful to realize high-precision convergence of the source signal in the cell, and can improve the accuracy of the over-the-air calculation in the multi-cell scenario.

[0304] In some embodiments, the decoding matrix determination module is further configured to:

[0305] obtain a third channel matrix from each second terminal device in each adjacent cell to the receiving device of the cell, calculate a third equivalent channel matrix according to the third channel matrix, and a conjugate transpose of the third equivalent channel matrix;

[0306] calculate a conjugate transpose of the precoding matrix according to the precoding matrix of each cell;

[0307] In the plurality of adjacent cells, for an adjacent cell of the cell, calculate a third product corresponding to each second terminal device in the cell, for the cell, calculate a third product corresponding to each first terminal device in the cell, and accumulate the third products corresponding to the terminal devices of the plurality of cells respectively to obtain a third matrix; wherein the third product corresponding to each second terminal device in each adjacent cell is a product of the third equivalent channel matrix of the second terminal device in the cell, the first precoding matrix of the cell, the conjugate transpose of the first precoding matrix of the cell, and the conjugate transpose of the third equivalent channel matrix; wherein the third product corresponding to each first terminal device in the cell is a product of the second equivalent channel matrix of the first terminal device in the cell, the first precoding matrix of the cell, the conjugate transpose of the first precoding matrix of the cell, and the conjugate transpose of the second equivalent channel matrix;

[0308] calculate an inverse matrix of a sum of the third matrix and a covariance matrix of the Gaussian white noise;

[0309] calculate a second equivalent channel matrix from each first terminal device in the cell to the receiving device of the cell,

[0310] calculate a second product of the second equivalent channel matrix of each first terminal device and the precoding matrix of the cell, and sum the second products corresponding to the first terminal devices in the cell to obtain a second matrix;

[0311] calculate a product of the inverse matrix and the second matrix according to the decoding matrix expression to obtain the decoding matrix.

[0312] In the embodiment, the third matrix (joint covariance matrix of interference signal and useful signal) is constructed by accumulating the third product term (interference term) of the terminal device of the adjacent cell and the third product (useful signal) of the terminal device of the cell. The third matrix can explicitly quantify the interference introduced by the terminal device of the adjacent cell through the third channel matrix; and the effective signal component of the terminal device of the cell through the second equivalent channel matrix is reserved. The distributed computing architecture is used to realize that each cell respectively calculates the decoding matrix of itself, so that the efficiency of calculating the decoding matrix can be improved. In addition, the above-mentioned process of solving the decoding matrix can realize the minimization of the multi-cell mean square error, and the theoretical optimal receiving performance can be achieved through the decoding matrix. The receiving device of each cell can realize the active suppression of the cross-cell interference signal and the Gaussian white noise signal through the corresponding decoding matrix, which is beneficial to improve the accuracy of the over-the-air computation in the multi-cell scenario.

[0313] It should be noted that the module names involved in the embodiments of the present application can be defined as other names, as long as the functions of the modules can be realized, and the names of the modules are not limited specifically.

[0314] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the embodiments of the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation portal for user to select authorization or refusal.

[0315] The signal sending and signal receiving methods of the embodiments of the present application have been described above, and the device provided by the embodiments of the present application for executing the above-mentioned methods will be described below. Those skilled in the art can understand that the methods and devices can be combined and referenced with each other, and the related device provided by the embodiments of the present application can execute the steps in the method of the above-mentioned list sorting.

[0316] The signal sending method and the signal receiving method provided by the embodiments of the present application can be applied in electronic devices with communication function. The electronic device includes a terminal device, and the specific device form of the terminal device can refer to the above-mentioned related description, which will not be described here.

[0317] The embodiments of the present application provide a terminal device, which comprises a processor and a memory; the memory stores computer execution instructions; and the processor executes the computer execution instructions stored in the memory, so that the terminal device executes the above-mentioned method.

[0318] An embodiment of the present application provides a chip. The chip comprises a processor configured to invoke a computer program in a memory to execute the technical solutions in the above embodiments. The implementation principle and technical effects are similar to those of the above related embodiments, and will not be repeated here.

[0319] An embodiment of the present application further provides a computer readable storage medium. The computer readable storage medium stores a computer program. The computer program is executed by a processor to implement the above method. The method described in the above embodiments can be implemented by software, hardware, firmware or any combination thereof, in whole or in part. If implemented in software, the functions can be stored in or transmitted as one or more instructions or codes on a computer readable medium. The computer readable medium can include computer storage medium and communication medium, and can also include any medium that can carry computer programs from one place to another. The storage medium can be any target medium that can be accessed by a computer.

[0320] In a possible implementation, the computer readable medium can include RAM, ROM, compact disc read-only memory (CD-ROM) or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that is targeted to carry desired program codes in the form of instructions or data structures and can be accessed by a computer. Moreover, any connection is appropriately referred to as a computer readable medium. For example, if software is transmitted from a website, server or other remote source using a coaxial cable, optical fiber cable, twisted pair, digital subscriber line (DSL) or wireless technology (such as infrared, radio and microwave), the coaxial cable, optical fiber cable, twisted pair, DSL or wireless technology (such as infrared, radio and microwave) is included in the definition of the medium. As used herein, magnetic disks and optical disks include compact disks, laser disks, optical disks, digital versatile disks (DVD), floppy disks and Blu-ray disks, in which magnetic disks usually reproduce data magnetically, and optical disks reproduce data optically with a laser. Combinations of the above should also be included in the scope of the computer readable medium.

[0321] An embodiment of the present application provides a computer program product, which comprises a computer program, when the computer program is executed, causes a computer to execute the above method.

[0322] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as a combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions of one or more flows and / or blocks Figure 1 The functions of one or more flows and / or blocks

[0323] The above detailed description of the embodiments of the present application further describes the purposes, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above is only a specific implementation of the embodiments of the present application, and is not used to limit the protection scope of the embodiments of the present application. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A signal transmission method, applied to a terminal device, characterized in that, The method comprises: performing first precoding on a source signal of a first terminal device based on a first precoding matrix to obtain a first precoded signal, wherein the source signal is used for over-the-air computation; performing second precoding on the first precoded signal based on a second precoding matrix to obtain a precoded signal of the source signal; uplink transmitting the precoded signal; wherein the first precoding matrix is determined based on solving a minimum multi-cell mean square error function, and the minimum multi-cell mean square error function represents a minimum sum of squares of F-norms of differences between decoded signals of receiving devices of each cell and target signals; and the second precoding matrix is an inverse matrix of a channel matrix between the first terminal device and the receiving devices of the cell.

2. The method of claim 1, wherein, The decoded signal of each receiving device is obtained by processing a received uplink signal using a decoding matrix, and the uplink signal corresponding to each receiving device comprises interference signals transmitted by terminal devices of neighboring cells of the receiving device, precoded signals transmitted by a plurality of first terminal devices of the cell to which the receiving device belongs, and a Gaussian white noise signal superimposed in the channel; and the target signal corresponding to each receiving device is a converged signal of the source signals of the plurality of first terminal devices of the cell to which the receiving device belongs.

3. The method of claim 1, wherein, The first precoding matrix is obtained based on a first operation as follows: obtaining decoding matrices transmitted by receiving devices of each cell, wherein for each cell, the decoding matrix of the cell is calculated by the receiving device based on initial first precoding matrices transmitted by terminal devices of each cell; and the initial first precoding matrix transmitted by each terminal device to the corresponding receiving device for the first time is generated by the terminal devices of each cell and satisfies a quasi-unitary constraint; iteratively calculating the first precoding matrix based on an expression of the first precoding matrix and the decoding matrices of each cell.

4. The method of claim 3, wherein, The expression of the first precoding matrix is determined based on the following steps: obtaining the expression of the first precoding matrix by taking the extreme value of the first precoding matrix based on a Lagrange function; wherein, the Lagrange function is constructed based on the minimum multi-cell mean square error function and a terminal device transmit power constraint.

5. The method of claim 4, wherein, The iteratively calculating the first precoding matrix based on the expression of the first precoding matrix and the decoding matrices of each cell comprises: calculating a Lagrange multiplier according to the decoding matrices respectively corresponding to each cell; updating the first precoding matrix according to the Lagrange multiplier and the decoding matrices respectively corresponding to each cell; replacing the initial first precoding matrix with the updated first precoding matrix to iteratively calculate the first precoding matrix; in response to the iterative calculation satisfying a preset condition, taking the first precoding matrix obtained when the preset condition is satisfied as a final first precoding matrix.

6. The method of claim 5, wherein, The preset condition comprises: the number of iterations reaches a predetermined value.

7. The method of claim 5, wherein, The calculating the Lagrange multiplier according to the decoding matrix comprises: For the cell, a first product corresponding to each first terminal device in the cell is calculated, and each first product corresponding to each first terminal device in the cell is accumulated to obtain a first matrix; wherein the first product corresponding to each first terminal device in the cell is a product of a conjugate transpose of a first equivalent channel matrix between the first terminal device and a receiving device of a neighboring cell of the first terminal device, a decoding matrix of the neighboring cell, a conjugate transpose of the decoding matrix of the neighboring cell, and a first equivalent channel matrix of the neighboring cell; and a product of a conjugate transpose of a second equivalent channel matrix between the first terminal device and a receiving device of the cell, a decoding matrix of the cell, a conjugate transpose of the decoding matrix of the cell, and the second equivalent channel matrix; A second equivalent channel matrix is determined according to a second channel matrix between each first terminal device in the cell and the receiving device of the cell, a second product of a conjugate transpose of the second equivalent channel matrix and a decoding matrix corresponding to each first terminal device in the cell is calculated, and each second product corresponding to each first terminal device is accumulated to obtain a second matrix; A Lagrange multiplier equation is solved based on the first matrix and the second matrix to obtain the Lagrange multiplier.

8. The method of claim 5, wherein, The updating of the first precoding matrix according to the Lagrange multiplier comprises: The Lagrange multiplier and the decoding matrix of each cell are substituted into an expression of the first precoding matrix to obtain the updated first precoding matrix.

9. A signal receiving method applied to a receiving device in a cell, the method comprising: The method comprises: Receiving an uplink signal; the uplink signal comprises a precoded signal obtained by precoding a respective source signal of at least one first terminal device in the cell according to a first precoding matrix and a second precoding matrix, an interference signal transmitted by at least one second terminal of at least one neighboring cell, and a Gaussian white noise signal superimposed in the channel; the source signal is a signal for over-the-air computation; Decoding the uplink signal according to a decoding matrix to obtain a converged signal of the respective source signal of the at least one first terminal device; The decoding matrix is determined based on solving a minimum multi-cell mean square error function; the minimum multi-cell mean square error function represents that a sum of squares of F-norms of differences between decoding signals of receiving devices of each cell and target signals is minimized; the decoding signal of each receiving device is obtained by processing a received uplink signal using a decoding matrix.

10. The method of claim 9, wherein, The decoding matrix is obtained based on a first operation as follows: Obtaining a first precoding matrix transmitted by each terminal device of each cell; Updating the decoding matrix according to a decoding matrix expression and the first precoding matrix of each terminal device of each cell to iteratively calculate the decoding matrix; wherein an initial first precoding matrix transmitted by each terminal device satisfies a quasi-unitary constraint; wherein the decoding matrix expression is determined by taking an extreme value of a Lagrange function with respect to the decoding matrix, and the Lagrange function is constructed according to the minimum multi-cell mean square error function and a terminal device transmit power constraint; In response to the iterative calculation satisfying a preset condition, the decoding matrix obtained when the preset condition is satisfied is taken as a final decoding matrix of the cell.

11. The method of claim 10, wherein, The updating of the decoding matrix according to the decoding matrix expression and the first precoding matrix of each terminal device in each cell comprises: obtaining a third channel matrix from each second terminal device in each adjacent cell to the receiving device in the cell, calculating a third equivalent channel matrix according to the third channel matrix, and a conjugate transpose of the third equivalent channel matrix; calculating a conjugate transpose of the first precoding matrix according to the first precoding matrix of each cell; calculating a second equivalent channel matrix from each first terminal device in the cell to the receiving device in the cell; for each adjacent cell of the cell, calculating a third product corresponding to each second terminal device in the cell, calculating a third product corresponding to each first terminal device in the cell, and accumulating the third products corresponding to the terminal devices in the cells to obtain a third matrix; wherein the third product corresponding to each second terminal device in each adjacent cell is a product of the third equivalent channel matrix of the second terminal device in the cell, the first precoding matrix of the cell, the conjugate transpose of the first precoding matrix of the cell, and the conjugate transpose of the third equivalent channel matrix; the third product corresponding to each first terminal device in the cell is a product of the second equivalent channel matrix of the first terminal device in the cell, the first precoding matrix of the cell, the conjugate transpose of the first precoding matrix of the cell, and the conjugate transpose of the second equivalent channel matrix; calculating an inverse matrix of a sum of the third matrix and a covariance matrix of Gaussian white noise; calculating a second product of the second equivalent channel matrix of each first terminal device and the first precoding matrix of the cell, summing the second products corresponding to each first terminal device in the cell to obtain a second matrix; calculating a product of the inverse matrix and the second matrix according to the decoding matrix expression to obtain the decoding matrix.

12. A signal processing system comprising a plurality of terminal devices located in a plurality of cells and a receiving device located in each of the cells; wherein each first terminal device in each cell uses a first precoding matrix of the cell to perform first precoding on a source signal used for over-the-air calculation to obtain a first precoded signal, and uses a second precoding matrix to perform second precoding on the first precoded signal to obtain a precoded signal of the source signal; and the precoded signal is sent uplink; the receiving device in each cell receives an uplink signal, uses a decoding matrix to decode the uplink signal, and obtains a converged signal of the source signal of each first terminal device in the cell; the uplink signal comprises the precoded signal sent by each first terminal device in the cell, an interference signal sent by at least one second terminal device in an adjacent cell, and a Gaussian white noise signal. The first precoding matrix and the decoding matrix are determined based on solving a minimum mean square error function; the minimum mean square error function represents that a difference between a signal obtained by decoding, by a receiving device of the cell, the uplink signal after decoding the uplink signal using the decoding matrix and a converged signal of source signals of a plurality of first terminal devices of the cell is a square of an F-norm minimum; and the second precoding matrix is an inverse matrix of a channel matrix between the first terminal device and the receiving device of the cell.

13. The system of claim 12, wherein, The first precoding matrix and the decoding matrix are obtained by the terminal device and the receiving device through the following operations: The terminal device of each cell generates an initial first precoding matrix and sends the initial first precoding matrix to the receiving device of the cell; The terminal device of each cell and the receiving device of each cell perform the following joint interference alignment iteration operation: The receiving device of each cell receives the first precoding matrix sent by the terminal device of the cell, and the receiving devices of all the cells share the first precoding matrix of each cell through a central unit connected to the receiving devices by wire; the receiving device of each cell sends the shared first precoding matrix of each cell to the terminal device of the cell; and the first precoding matrix of the terminal device of the cell received by the receiving device of each cell for the first time is the initial first precoding matrix; The decoding matrix is calculated according to the decoding matrix expression and the first precoding matrix, and the decoding matrix for iteration calculation is obtained; The receiving devices of all the cells share the decoding matrix of each cell through the central unit connected to the receiving devices by wire; and send the decoding matrix of each cell obtained to the terminal device of the cell; The terminal device calculates a Lagrange multiplier based on the decoding matrix of each cell, updates the first precoding matrix according to the Lagrange multiplier, and obtains the first precoding matrix for iteration calculation; and sends the first precoding matrix to the receiving device of the cell; The first precoding matrix and the decoding matrix obtained when the iteration calculation is stopped to meet a preset condition are taken as the final first precoding matrix and the decoding matrix. The expression of the first precoding matrix is obtained by solving an extreme value of a Lagrange function for the first precoding matrix, and the expression of the decoding matrix is obtained by solving an extreme value of the Lagrange function for the decoding matrix; and the Lagrange function is constructed based on the minimum mean square error function and a terminal device transmission power constraint.

14. An electronic device, comprising: The electronic device comprises: a processor and a memory; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory, so that the electronic device executes the method in any one of claims 1-11.

15. A computer-readable storage medium, the computer-readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to implement the method in any one of claims 1-11.

16. A chip system, characterized by The electronic device comprises at least one processor and a communication interface, the communication interface and the at least one processor are interconnected by wire, and the at least one processor is used to run a computer program or instructions to execute the method in any one of claims 1-11.

17. A computer program product, characterised in that, The computer program, when executed, causes a computer to execute the method in any one of claims 1-11.

Images