Phased array transmitter, transmission method, and program

Abstract

Properly implement compensation for nonlinear distortions. [Solution] A phased array transmitter is provided, which has a k-th distortion compensation unit that outputs a group of k-th coefficients by a k-th neural network model based on the time-series signals of the first to k-th input signals (where k is an integer from 1 to N, and N is an integer of 2 or more), multiplies each value included in the time-series signal of the k-th input signal by each coefficient included in the k-th coefficient group, and outputs the sum of these values ​​as the k-th output signal, wherein the k-th output signal is phase-controlled and amplified in accordance with each of the multiple antennas and radiated wirelessly from the multiple antennas.

Description

Technical Field

[0001] The present disclosure relates to a phased array transmitter, a transmission method, and a program.

Background Art

[0002] Patent Document 1 discloses a technique related to neural network-based digital pre-distortion (DPD). In Patent Document 1, a digital pre-distortion (DPD) actuator is included for receiving an input signal associated with a non-linear component of a radio frequency (RF) transceiver and outputting a pre-distorted signal. The DPD actuator includes a basis function-based actuator for performing a first DPD operation using a set of basis functions associated with a first non-linear characteristic of the non-linear component. The DPD actuator further includes a neural network-based actuator for performing a second DPD operation using a first neural network associated with a second non-linear characteristic of the non-linear component.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the technique described in Patent Document 1, for example, non-linear distortion compensation may not be properly performed.

[0005] An object of the present disclosure is to provide a technique capable of properly performing non-linear distortion compensation in view of the above problems.

Means for Solving the Problems

[0006] In a first aspect of the present disclosure, a phased array transmitter is provided, comprising: an input unit that inputs a time-series signal of a first input signal and a time-series signal of a second input signal; a first distortion compensation unit that outputs a first coefficient group using a first neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, outputs a first output signal obtained by multiplying each value included in the time-series signal of the first input signal with each coefficient included in the first coefficient group and then adding the results; and a second distortion compensation unit that outputs a second coefficient group using a second neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, outputs a second output signal obtained by multiplying each value included in the time-series signal of the second input signal with each coefficient included in the second coefficient group and then adding the results, wherein the first output signal and the second output signal are phase-controlled and amplified in accordance with each of a plurality of antennas and radiated wirelessly from the plurality of antennas.

[0007] Furthermore, in a second aspect of the present disclosure, a transmission method is provided in which a phased array transmitter receives a time-series signal of a first input signal and a time-series signal of a second input signal, outputs a first coefficient group by a first neural network model based on the time-series signal of the first input signal and the time-series signal of the second input signal, outputs a first output signal obtained by multiplying each value in the time-series signal of the first input signal with each coefficient in the first coefficient group and then adding them together, outputs a second output signal obtained by multiplying each value in the time-series signal of the second input signal with each coefficient in the second coefficient group and then adding them together, and the first output signal and the second output signal are phase-controlled and amplified in accordance with each of the plurality of antennas and radiated wirelessly from the plurality of antennas.

[0008] Furthermore, in a third aspect of the present disclosure, a program is provided which causes the computer of a phased array transmitter to perform the following processes: inputting a time-series signal of a first input signal and a time-series signal of a second input signal; outputting a first group of coefficients by a first neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal; outputting a first output signal by multiplying each value in the time-series signal of the first input signal with each coefficient in the first group of coefficients and then adding them together; outputting a second output signal by multiplying each value in the time-series signal of the second input signal with each coefficient in the second group of coefficients and then adding them together; the first output signal and the second output signal are phase-controlled and amplified in correspondence to each of the plurality of antennas and radiated wirelessly from the plurality of antennas. [Effects of the Invention]

[0009] From one perspective, it is possible to properly compensate for nonlinear distortion. [Brief explanation of the drawing]

[0010] [Figure 1] This figure shows an example of the configuration of a phased array transmitter according to the embodiment. [Figure 2] This figure shows an example of the configuration of the beamforming matrix according to the embodiment. [Figure 3] This figure shows an example of the configuration of the beamforming matrix according to the embodiment. [Figure 4] This figure shows an example of the hardware configuration of the distortion compensation section of a phased array transmitter according to the embodiment. [Figure 5] This flowchart shows an example of the processing of the strain compensation unit according to the embodiment. [Figure 6] This figure shows an example of the processing of the strain compensation unit according to the embodiment. [Figure 7] This figure shows an example of the processing of the strain compensation unit according to the embodiment. [Figure 8] This figure shows an example of the processing of the strain compensation unit according to the embodiment. [Modes for carrying out the invention]

[0011] The principles of this disclosure will be described with reference to several exemplary embodiments. These embodiments are described for illustrative purposes only and should be understood as helping those skilled in the art to understand and implement this disclosure without implying any limitation on the scope of this disclosure. The disclosures described herein may be implemented in various ways other than those described below.

[0012] In the following description and claims, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those generally understood by those skilled in the art to which this disclosure belongs.

[0013] Embodiments of the present disclosure will be described below with reference to the drawings. Each drawing is merely illustrative for illustrating one or more embodiments. Each drawing may be associated not only with one specific embodiment but also with one or more other embodiments. As those skilled in the art will understand, various features or steps described with reference to any one drawing can be combined with features or steps shown in one or more other drawings, for example, to create embodiments not explicitly shown or described. Not all features or steps shown in any one drawing to illustrate an exemplary embodiment are necessarily required, and some features or steps may be omitted. The order of steps described in any of the drawings may be changed as appropriate.

[0014] [Embodiment 1] <Structure> Referring to Figure 1, the configuration of the phased array transmitter 10 according to the embodiment will be described. Figure 1 is a diagram showing an example of the configuration of the phased array transmitter 10 according to the embodiment. Although Figures 1 to 3 show an example where N is 4, N in this disclosure is not limited to 4, but can be an integer of 2 or more. The phased array transmitter 10 may communicate in a multiband (and multi-user) manner with different carrier frequencies for each signal, or it may communicate in a single band (and multi-user) manner using the same bandwidth (carrier frequency).

[0015] The phased array transmitter 10 has input sections 11-1 to 11-N (in this disclosure, N is an integer of 2 or more) (hereinafter, unless otherwise necessary, it will also be simply referred to as "input section 11"). The phased array transmitter 10 also has distortion compensation sections (DPD: Digital Predistortion) 12-1 to 12-N (hereinafter, unless otherwise necessary, it will also be simply referred to as "distortion compensation section 12").

[0016] Furthermore, the phased array transmitter 10 has DACs (Digital-to-Analog Converters, D / A converters) 13-1 to 13-N (hereinafter, unless otherwise necessary, they will also be simply referred to as "DAC13"). The phased array transmitter 10 also has a beamforming matrix (phased array antenna) 14.

[0017] Each of the input sections 11-1 to 11-N is an input signal x1 to x2, which is data transmitted wirelessly from the phased array transmitter 10. N Each of these digital data is input (acquired). For example, input unit 11-1 acquires input signal x1, input unit 11-2 acquires input signal x2, and input unit 11-N acquires input signal x N Obtain it.

[0018] The distortion compensation unit 12-k (wherein k is an integer from 1 to N) receives input signals x1 to x acquired by input units 11-1 to 11-N, respectively. NBased on this, an output signal y is generated in which signal interference due to non-linear distortion of the amplifier arranged for each antenna is cancelled. k This is output to DAC13-k. More specifically, the distortion compensation unit 12-k causes the k-th neural network model (the k-th learned model) to output the k-th coefficient group based on the respective time-series signals of the input signals x1 to x N . Then, the distortion compensation unit 12-k multiplies each value included in the time-series signal of the input signal x k (the k-th input signal) and each coefficient included in the k-th coefficient group, and outputs, as the output signal y k (the k-th output signal), the value obtained by adding them after multiplication. Therefore, each of the distortion compensation units 12-1 to 12-N generates output signals y1 to y N .

[0019] DAC13-k converts the data from the distortion compensation unit 12-k into an analog signal and outputs it to the beamforming matrix 14. The beamforming matrix 14 performs phase control and amplification corresponding to each of the output signals y1 to y N for each of the plurality of antennas, and radiates them wirelessly from the plurality of antennas. As a result, non-linear distortion occurs in the output signals y1 to y N to which the inverse characteristics of signal interference due to non-linear distortion of the amplifier arranged for each antenna are added, so that each of the original input signals x1 to x N is transmitted to a receiver (not shown) by radio waves. Note that a signal having an IF frequency in a frequency band lower than the target RF frequency may be output to DAC13-k. In that case, a frequency mixer and a bandpass filter may be arranged after DAC13-k, and the frequency mixer may perform conversion from IF to RF. Then, the bandpass filter may remove signal components other than the desired RF component generated by the frequency mixer and output the desired RF signal.

[0020] (Regarding the beamforming matrix 14) Figure 2 shows an example of the configuration of a beamforming matrix 14 according to an embodiment. In the example in Figure 2, the beamforming matrix 14 includes array banks 141-1 to 141-M, adders 142-1 to 142-M, amplifiers (PA, power amplifiers) 143-1 to 143-M, and antennas 144-1 to 144-M. In this disclosure, M is an integer of 2 or more. For example, M may be an integer greater than N.

[0021] In the example in Figure 2, array bank 141-m (wherein m is an integer from 1 to M) is used for each output signal y1~y after analog conversion. N The phases of each are controlled by a phase shifter and output to adder 142-m. Adder 142-m outputs y1~y with controlled phases. N The signals are added together and output to amplifier 143-m. Amplifier 143-m amplifies the amplitude (intensity) of the signal from adder 142-m and outputs it to antenna 144-m. Antenna 144-m transmits the signal from amplifier 143-m to a receiver (not shown) via radio waves. The beamforming matrix 14 shown in Figure 2 may also be called a full array configuration. In addition to the phase, the amplitude value may be controlled in the above phase transformer to improve the accuracy of beam control.

[0022] Figure 3 shows another example of the configuration of the beamforming matrix 14 according to the embodiment. In the example of Figure 3, the beamforming matrix 14 has submatrices 14-1 to 14-L (wherein L is an integer of 2 or more). Submatrices 14-1 to 14-L are configured to handle input signals x1 to x N The signal is transmitted wirelessly to a receiver (not shown). In the example in Figure 3, only the detailed configuration of submatrix 14-1 is shown, but each of the submatrixes 14-2 to 14-L has a similar configuration to submatrix 14-1. Submatrix 14-k includes phasers 21-k-1 to 21-kL, amplifiers 22-k-1 to 22-kL, and antennas 23-k-1 to 23-kL.

[0023] In the example in Figure 3, submatrix 14-k represents the output signal y after analog conversion.k The phase of the signal is controlled by phase shifters 21-k-1 to 21-kL, and output to amplifiers 22-k-1 to 22-kL, respectively. Each of the amplifiers 22-k-1 to 22-kL amplifies the amplitude (intensity) of the signals from each of the phase shifters 21-k-1 to 21-kL and outputs it to each of the antennas 23-k-1 to 23-kL. Each of the antennas 23-k-1 to 23-kL transmits the signals from each of the amplifiers 22-k-1 to 22-kL via radio waves to a receiver (not shown). The beamforming matrix 14 shown in Figure 3 may also be referred to as a subarray configuration.

[0024] <Hardware configuration of the distortion compensation unit 12> Figure 4 shows an example of the hardware configuration of the strain compensation unit 12 of the phased array transmitter 10 according to the embodiment. In the example in Figure 4, the strain compensation unit 12 (computer 100) includes a processor 101, memory 102, and a communication interface 103. These units may be connected by a bus or the like. The memory 102 stores at least a portion of the program 104. The communication interface 103 includes an interface necessary for communication with other network elements.

[0025] When program 104 is executed in cooperation with the processor 101 and memory 102, etc., the computer 100 performs at least some of the processing of embodiments of this disclosure. Memory 102 may be of any type. Memory 102 may, in non-limiting examples, be a non-temporary computer-readable storage medium. Memory 102 may also be implemented using any suitable data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. Although only one memory 102 is shown for computer 100, computer 100 may have several physically different memory modules. Processor 101 may be of any type. Processor 101 may include one or more general-purpose computers, dedicated computers, microprocessors, digital signal processors (DSPs), and, in non-limiting examples, processors based on multicore processor architectures. Computer 100 may have multiple processors, such as application-specific integrated circuit chips that are time-dependent to a clock that synchronizes the main processor.

[0026] Embodiments of the present disclosure may be implemented in hardware or in dedicated circuitry, software, logic, or any combination thereof. Some embodiments may be implemented in hardware, while others may be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device.

[0027] This disclosure also provides at least one computer program product tangibly stored on a non-temporary computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions contained in a program module, and is executed on a device on a target real or virtual processor to perform the processes or methods of this disclosure. The program module includes routines, programs, libraries, objects, classes, components, data structures, etc., that perform a specific task or implement a specific abstract data type. The functionality of the program module may be combined or divided among the program module as desired in various embodiments. The machine-executable instructions of the program module can be executed on a local or distributed device. On a distributed device, the program module can reside on both local and remote storage media.

[0028] Program code for performing the methods of this disclosure may be written in any combination of one or more programming languages. These program codes are provided to a processor or controller of a general-purpose computer, a dedicated computer, or other programmable data processing device. When the program code is executed by the processor or controller, the functions / operations in the flowchart and / or block diagrams it implements are performed. The program code may run entirely on a machine, partially on a machine, partially as a standalone software package, partially on a machine, partially on a remote machine, or entirely on a remote machine or server.

[0029] Programs can be stored and supplied to a computer using various types of non-temporary computer-readable media. Non-temporary computer-readable media include various types of tangible recording media. Examples of non-temporary computer-readable media include magnetic recording media, magneto-optical recording media, optical disc media, and semiconductor memory. Magnetic recording media include, for example, flexible disks, magnetic tapes, and hard disk drives. Magneto-optical recording media include, for example, magneto-optical disks. Optical disc media include, for example, Blu-ray discs, CD (Compact Disc)-ROM (Read Only Memory), CD-R (Recordable), and CD-RW (ReWritable). Semiconductor memory includes, for example, solid-state drives, mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (random access memory). Programs may also be supplied to a computer using various types of temporary computer-readable media. Examples of temporary computer-readable media include electrical signals, optical signals, and electromagnetic waves. Temporary computer-readable media can supply programs to a computer via wired communication channels such as electric wires and optical fibers, or via wireless communication channels.

[0030] <Processing> Next, an example of the processing of the strain compensation unit 12 according to the embodiment will be described with reference to Figures 5 to 7. Figure 5 is a flowchart showing an example of the processing of the strain compensation unit 12 according to the embodiment. Figures 6 and 7 show an example of the processing of the strain compensation unit 12 according to the embodiment. Note that the processing in Figure 5 may be performed at each point in time when transmitting data wirelessly.

[0031] In step S101, each of the input units 11-1 to 11-N receives input signals x1 to x wirelessly transmitted from the phased array transmitter 10. N Input (acquire) the respective digital data.

[0032] Next, each distortion compensation section 12-k processes the input signal x1~x N Each data based on this is input to a trained model (the trained model of the kth) for generating the kth transmission signal (step S102). Here, the distortion compensation unit 12-k processes the input signals x1~x as shown in Figures 6 to 7. N The signals (time-series signals) at each point in time up to the present may be used as the input data (explanatory variables) for the k-th trained model. Each trained model may be pre-configured in the phased array transmitter 10.

[0033] In the example shown in Figure 6, the distortion compensation unit 12-1 uses the value 611, which is the result of a specific operation on the current time t value x1(t) of the input signal x1, as the first input data for the first trained model 601. Details of the specific operation will be described later.

[0034] Furthermore, the distortion compensation unit 12-1 uses a value 612, which is the result of a specific calculation on the value x1(t-1) of the input signal x1 at a specific period before the current time t, as the second input data of the first trained model 601. Also, the distortion compensation unit 12-1 uses a value 61j, which is the result of a specific calculation on the value x1(tj) of the input signal x1 at a specific period j (where j is an integer greater than or equal to 2) before the current time t, as the j-th input data of the first trained model 601.

[0035] Furthermore, for the input signal x2, the distortion compensation unit 12-1 uses the value 621, which is the result of a specific calculation on the value x2(t), as the input data for the (j+1)th, similar to the case of the input signal x1. Also, the distortion compensation unit 12-1 uses the value 622, which is the result of a specific calculation on the value x2(t-1), as the input data for the (j+2)th, and the distortion compensation unit 12-1 uses the value 62j, which is the result of a specific calculation on the value x2(tj), as the input data for the (j+j)th,.

[0036] Furthermore, the distortion compensation unit 12-1, for input signal x3, uses value 631, which is the result of a specific calculation on value x3(t), as the input data for the j+j+1, similar to the case of input signal x1. The distortion compensation unit 12-1 also uses value 632, which is the result of a specific calculation on value x3(t-1), as the input data for the j+j+2. Finally, the distortion compensation unit 12-1 uses value 63j, which is the result of a specific calculation on value x3(tj), as the input data for the j+j+j.

[0037] Furthermore, for input signal x4, the distortion compensation unit 12-1 uses the value 641, which is the result of a specific calculation on the value x4(t), as the input data for the j+j+j+1, similar to the case of input signals x1 to x3. Also, the distortion compensation unit 12-1 uses the value 642, which is the result of a specific calculation on the value x4(t-1), as the input data for the j+j+j+2. Finally, the distortion compensation unit 12-1 uses the value 64j, which is the result of a specific calculation on the value x4(tj), as the input data for the j+j+j+j.

[0038] In the example shown in Figure 7, the distortion compensation unit 12-2 uses a value 711, which is the result of a specific calculation on the current time t value x1(t) of the input signal x1, as the first input data of the second trained model 701. The distortion compensation unit 12-2 also uses a value 712, which is the result of a specific calculation on the value x1(t-1) of the input signal x1 at a specific period before the current time t, as the second input data of the second trained model 701. Furthermore, the distortion compensation unit 12-2 uses a value 71j, which is the result of a specific calculation on the value x1(tj) of the input signal x1 at a specific period j times before the current time t, as the j-th input data of the second trained model 701.

[0039] Furthermore, the distortion compensation unit 12-2, similar to the case of input signal x1, uses value 721, which is the result of a specific calculation on value x2(t), as the input data for the (j+1)th input signal x2. The distortion compensation unit 12-2 also uses value 722, which is the result of a specific calculation on value x2(t-1), as the input data for the (j+2)th input signal. Finally, the distortion compensation unit 12-2 uses value 72j, which is the result of a specific calculation on value x2(tj), as the input data for the (j+j)th input signal.

[0040] Furthermore, the distortion compensation unit 12-2, for input signal x3, uses value 731, which is the result of a specific calculation on value x3(t), as the input data for the j+j+1, similar to the case of input signal x1. The distortion compensation unit 12-2 also uses value 732, which is the result of a specific calculation on value x3(t-1), as the input data for the j+j+2. Finally, the distortion compensation unit 12-2 uses value 73j, which is the result of a specific calculation on value x3(tj), as the input data for the j+j+j.

[0041] Furthermore, for input signals x1 to x3, the distortion compensation unit 12-2 uses the value 741, which is the result of a specific calculation on the value x4(t), as the input data for the j+j+j+1 position, similar to the case of input signals x1 to x3. The distortion compensation unit 12-2 also uses the value 742, which is the result of a specific calculation on the value x4(t-1), as the input data for the j+j+j+2 position. Finally, the distortion compensation unit 12-2 uses the value 74j, which is the result of a specific calculation on the value x4(tj), as the input data for the j+j+j+j position. The distortion compensation units 12-3 to 12-N are the same as those 12-1 to 12-2.

[0042] (Regarding specific operations) The distortion compensation unit 12-k may calculate, for example, the square of the signal amplitude as the specific operation described above. In this case, for example, the value that is the first input data of the kth trained model (for example, value 611 in Figure 6, value 711 in Figure 7) is |x1(t)| 2 Therefore, the value considered as the second input data (for example, the value 612 in Figure 6 and the value 712 in Figure 7) is |x1(t-1)| 2 This is the result. In addition, the distortion compensation unit 12-k may calculate the amplitude of the signal as a specific operation as described above. In this case, for example, the value that is the first input data of the k-th trained model (for example, value 611 in Figure 6 and value 711 in Figure 7) becomes |x1(t)|, and the value that is the second input data (for example, value 612 in Figure 6 and value 712 in Figure 7) becomes |x1(t-1)|.

[0043] Furthermore, the distortion compensation unit 12-k may calculate, as a specific operation described above, at least one of the following: the real part of the signal, the imaginary part of the signal, the square of the real part of the signal, the square of the imaginary part of the signal, and the product of the real and imaginary parts of the signal.

[0044] Furthermore, the distortion compensation unit 12-k performs the above-mentioned specific calculation, for example, input signals x1~x N Alternatively, the square of the amplitude of the linear sum of can be calculated. In this case, for example, the values ​​that are the first input data of the k-th trained model (for example, the value 611 in Figure 6 and the value 711 in Figure 7) are |x1(t)+w2x2(t)+···+w N x N (t) 2 Therefore, the values ​​considered as the second input data (for example, the value 612 in Figure 6 and the value 712 in Figure 7) are |x1(t-1)+w2x2(t-1)+···+w N x N (t-1)| 2 This is the result. Also, w2~w N This is a specific weighting coefficient. Furthermore, the distortion compensation unit 12-k performs the aforementioned specific calculation, for example, input signals x1~x N The amplitude of the linear sum of can also be calculated. In this case, for example, the values ​​that are the first input data of the k-th trained model (for example, the value 611 in Figure 6 and the value 711 in Figure 7) are |x1(t)+w2x2(t)+···+w N x N (t)|, and the values ​​considered as the second input data (for example, the value 612 in Figure 6 and the value 712 in Figure 7) are |x1(t-1)+w2x2(t-1)+···+w N x N (t-1)|

[0045] Furthermore, the distortion compensation unit 12-k may, as a specific operation as described above, calculate the dot product or cross product of the first input signal and the second input signal, for example, by considering the I signal and Q signal of each input signal as separate components of the vector. In this case, for example, the value that is the first input data of the kth trained model (for example, value 611 in Figure 6 and value 711 in Figure 7) becomes Real(x1(t))×Real(x2(t))+Imag(x1(t))×Imag(x2(t)), and the value that is the second input data (for example, value 612 in Figure 6 and value 712 in Figure 7) becomes Real(x1(t-1))×Real(x2(t-1))+Imag(x1(t-1))×Imag(x2(t-1)). Note that Real(x) represents the real part of x, and Imag(x) represents the imaginary part of x.

[0046] Furthermore, the distortion compensation unit 12-k may, as a specific operation as described above, calculate, for example, the amplitude value of a linear sum of different signals. The examples of specific operations described above can be used in combination as appropriate.

[0047] Next, each distortion compensation unit 12-k in Figures 6 and 7 applies the input signal x to each coefficient included in the kth coefficient group, which is the output of H (in this disclosure, H is an integer of 2 or more) of the kth trained model. k The output signal y is the sum of the values ​​obtained by multiplying the time-series signals at H points in time. k This is output to DAC13-k (step S103). As a result, each output signal y has the inverse characteristic of signal interference due to the nonlinear distortion of the amplifiers placed for each antenna added to it. k This is output. Therefore, it is possible to effectively add inverse characteristics that are in line with the physical model of signal interference due to nonlinear distortion of the amplifiers placed for each antenna, thereby achieving higher distortion compensation performance with less computation. Note that the value of H may be pre-set in the phased array transmitter 10, for example.

[0048] Here, the distortion compensation unit 12-k, as shown in Figures 6 and 7, uses the input signal x of each output data (each target variable) of the k-th trained model. k Alternatively, the values ​​obtained by multiplying each of the time-series signals may be added together.

[0049] In the example shown in Figure 6, the distortion compensation unit 12-1 multiplies the first output data of the first trained model 601 by the value x1(t) 651, and then multiplies the second output data by the value x1(t-1) 652. Similarly, the distortion compensation unit 12-1 multiplies the h-th output data by the value x1(th) for each integer h less than or equal to H (an integer greater than or equal to 2) 65h. Finally, the distortion compensation unit 12-1 outputs the sum of the results of each multiplication 651 to 65h 661 as the output signal y1.

[0050] In the example shown in Figure 7, the distortion compensation unit 12-2 multiplies the first output data of the second trained model 701 by the value x2(t) 751, and then multiplies the second output data by the value x2(t-1) 752. Similarly, for each integer h less than or equal to H, the distortion compensation unit 12-2 multiplies the h-th output data by the value x2(th) 75h. Finally, the distortion compensation unit 12-2 outputs the sum of the results of each multiplication 751 to 75h 761 as the output signal y2.

[0051] Similarly, the distortion compensation unit 12-k uses the first output data of the k-th trained model and the value x k Multiply (t) by the second output data and the value x k Multiply by (t-1). Similarly, the distortion compensation unit 12-k multiplies the h-th output data and the value x for each integer h less than or equal to H which is an integer greater than or equal to 2. k (th) is multiplied by (th). Then, the distortion compensation unit 12-k outputs the sum of the results of each multiplication to the output signal y k Output as follows.

[0052] Next, DAC13-k receives the output signals y1~y from each distortion compensation unit 12. N The signal is converted to an analog signal and output to the beamforming matrix 14 (step S104).

[0053] Next, the beamforming matrix 14 outputs the signal y1~y NThe signal is phase-controlled and amplified for each of the multiple antennas, and then transmitted wirelessly from the multiple antennas to a receiver (not shown) (step S105). Note that the output signals y1~y N The result of mixing due to the nonlinear distortion of the amplifiers placed for each antenna is the original input signal x1~x N The radio waves are emitted. As a result, the receiver receives the original input signal x1~x N It is received.

[0054] (Example of pruning a pre-trained model) Each distortion compensation unit 12-k may remove (prune) connections of weights that have become smaller during the training process of the kth trained model. This reduces, for example, the number of connections (number of parameters) of the kth trained model, further reducing the computational load.

[0055] (Example of updating a trained model) Each distortion compensation unit 12-k may feed back the signal for distortion compensation and update the kth learned model. This allows for better compensation of nonlinear distortion, for example, in response to the nonlinear distortion of the actual amplifier. In this case, each distortion compensation unit 12-k receives, for example, the input signal x k and the received signal z k The parameters of the k-th trained model may be updated to reduce the error. In this case, each distortion compensation unit 12-k is, for example, based on the input signal x k and the received signal z k To reduce the error, the connection of the k-th trained model network may be removed.

[0056] Each distortion compensation section 12-k controls the output signal y k A receiver that receives radio waves based on this signal through space receives the received signal z k You may obtain it.

[0057] Furthermore, each distortion compensation section 12-k receives the output signal y amplified by amplifiers 143-1 to 143-M. k Based on the channel matrix between the phased array transmitter 10 and the receiver, the received signal zk The channel matrix may be estimated. For estimation, circuits such as array banks and submatrices used in Figures 2 and 3 may be used. The channel matrix may also be a matrix composed of the amplitude and phase variations of the propagation path (channel) between each transmitting antenna and receiving antenna. In this case, each distortion compensation unit 12-k may estimate the channel matrix using, for example, a signal known to the phased array transmitter 10 and receiver (e.g., a pilot signal).

[0058] Furthermore, each distortion compensation unit 12-k may predict the radio wave propagation characteristics based, for example, the spatial positional relationship between the phased array transmitter 10 and the receiver, and the arrangement of the antennas of the phased array transmitter 10, and estimate the channel matrix based on the predicted radio wave propagation characteristics.

[0059] <Other> For example, in Beyond 5G and similar technologies, phased array transmitters with multiple antennas and beam control are considered essential for simultaneously transmitting high-quality data to numerous user terminals without interference. Similarly, in satellite communications and similar applications, phased array transmitters with multiple antennas and beam control are considered essential for simultaneously transmitting high-quality data to numerous ground stations without interference.

[0060] In phased array transmitters, nonlinear distortion in the amplifiers located at each antenna causes signals that were originally designed not to interfere with each other to mix, resulting in signal interference. While conventional multi-beam digital predistortion (DPD) can be used to cancel this nonlinear distortion, the high computational load makes practical application difficult.

[0061] According to this disclosure, it is possible to appropriately compensate for the nonlinear distortion of the amplifiers arranged for each antenna. For example, by generating an output signal in which the inverse characteristics of the amplifiers arranged for each antenna are effectively added to the input signal, higher distortion compensation performance can be obtained with less computation.

[0062] <Variation> The phased array transmitter 10 may be a device contained in a single housing, but the phased array transmitter 10 of this disclosure is not limited to this. Each part of the phased array transmitter 10 (for example, the strain compensation unit 12) may be implemented by cloud computing, which consists of, for example, one or more computers. Alternatively, the strain compensation unit 12 and the beamforming matrix 14 may be configured as separate devices. Such phased array transmitters 10 are also included as examples of the "phased array transmitter" of this disclosure.

[0063] [Embodiment 2] Figure 8 shows an example of the processing of the strain compensation unit according to the embodiment.

[0064] In Embodiment 2, the strain compensation unit 12-k shown in Figures 6 and 7 above can be replaced with a strain compensation unit as shown in Figure 8.

[0065] In the example shown in Figure 8, the distortion compensation unit 12-1 uses the value 811, which is the result of a specific calculation performed on the current time t value x1(t) of the input signal x1, as the first input data for the first trained model 801. Details of the specific calculation will be described later.

[0066] Furthermore, the distortion compensation unit 12-1 uses a value 812, which is the result of a specific calculation on the value x1(t-1) of the input signal x1 at a specific period before the current time t, as the second input data of the first trained model 801. Also, the distortion compensation unit 12-1 uses a value 81j, which is the result of a specific calculation on the value x1(tj) of the input signal x1 at a specific period j (where j is an integer greater than or equal to 2) before the current time t, as the j-th input data of the first trained model 801.

[0067] Furthermore, the distortion compensation unit 12-1 also uses the value 821, which is the result of a specific calculation on the value x2(t), as the input data for the input signal x2, similar to the case of input signal x1. The distortion compensation unit 12-1 also uses the value 822, which is the result of a specific calculation on the value x2(t-1), as the input data for the j+2nd position. Finally, the distortion compensation unit 12-1 uses the value 82j, which is the result of a specific calculation on the value x2(tj), as the input data for the j+j position.

[0068] Furthermore, the distortion compensation unit 12-1, for input signal x3, uses value 831, which is the result of a specific calculation on value x3(t), as the input data for the j+j+1, similar to the case of input signal x1. The distortion compensation unit 12-1 also uses value 832, which is the result of a specific calculation on value x3(t-1), as the input data for the j+j+2. Finally, the distortion compensation unit 12-1 uses value 83j, which is the result of a specific calculation on value x3(tj), as the input data for the j+j+j.

[0069] Furthermore, for input signal x4, the distortion compensation unit 12-1 uses the value 841, which is the result of a specific calculation on the value x4(t), as the input data for the j+j+j+1, similar to the case of input signals x1 to x3. Also, the distortion compensation unit 12-1 uses the value 842, which is the result of a specific calculation on the value x4(t-1), as the input data for the j+j+j+2. Finally, the distortion compensation unit 12-1 uses the value 84j, which is the result of a specific calculation on the value x4(tj), as the input data for the j+j+j+j.

[0070] Furthermore, the distortion compensation units 12-2 to 12-N are the same as those for the distortion compensation unit 12-1.

[0071] (Regarding specific operations) The distortion compensation unit 12-k may calculate, for example, the square of the signal amplitude as the specific operation described above. In this case, for example, the value that is the first input data of the k-th trained model (for example, the value 811 in Figure 8) is |x1(t)| 2 Therefore, the value considered as the second input data (for example, the value 812 in Figure 8) is |x1(t-1)| 2This is the result. In addition, the distortion compensation unit 12-k may calculate the amplitude of the signal as a specific operation as described above. In this case, for example, the value that is the first input data of the k-th trained model (for example, value 811 in Figure 8) becomes |x1(t)|, and the value that is the second input data (for example, value 812 in Figure 8) becomes |x1(t-1)|.

[0072] Furthermore, the distortion compensation unit 12-k may calculate, as a specific operation described above, at least one of the following: the real part of the signal, the imaginary part of the signal, the square of the real part of the signal, the square of the imaginary part of the signal, and the product of the real and imaginary parts of the signal.

[0073] Furthermore, the distortion compensation unit 12-k performs the above-mentioned specific calculation, for example, input signals x1~x N Alternatively, the square of the amplitude of the linear sum of can be calculated. In this case, for example, the value that is the first input data of the k-th trained model (for example, the value 811 in Figure 8) is |x1(t)+w2x2(t)+···+w N x N (t) 2 Therefore, the value considered as the second input data (for example, the value 812 in Figure 8) is |x1(t-1)+w2x2(t-1)+···+w N x N (t-1)| 2 This is the result. Also, w2~w N This is a specific weighting coefficient. Furthermore, the distortion compensation unit 12-k performs the aforementioned specific calculation, for example, input signals x1~x N The amplitude of the linear sum of can also be calculated. In this case, for example, the value that is the first input data of the kth trained model (for example, the value 811 in Figure 8) is |x1(t)+w2x2(t)+···+w N x N (t)| is obtained, and the value considered as the second input data (for example, the value 812 in Figure 8) is |x1(t-1)+w2x2(t-1)+···+w N x N (t-1)|

[0074] Furthermore, the distortion compensation unit 12-k may, as a specific operation as described above, calculate the dot product or cross product of the first input signal and the second input signal, for example, by considering the I signal and Q signal of each input signal as separate components of the vector. In this case, for example, the value that is considered the first input data of the k-th trained model (for example, value 811 in Figure 8) becomes Real(x1(t))×Real(x2(t))+Imag(x1(t))×Imag(x2(t)), and the value that is considered the second input data (for example, value 812 in Figure 8) becomes Real(x1(t-1))×Real(x2(t-1))+Imag(x1(t-1))×Imag(x2(t-1)). Note that Real(x) represents the real part of x, and Imag(x) represents the imaginary part of x.

[0075] Furthermore, the distortion compensation unit 12-k may, as a specific operation as described above, calculate, for example, the amplitude value of a linear sum of different signals. The examples of specific operations described above can be used in combination as appropriate.

[0076] Next, each distortion compensation unit 12-k in Figure 8 applies the input signal x1 to x to each coefficient included in the kth coefficient group, which is the output of H (in this disclosure, H is an integer of 2 or more) of the kth trained model. N The values ​​obtained by multiplying the time-series signals at H points in time are added together (summed) and output as the output signal yk to DAC13-k (step S103). As a result, each output signal yk has the inverse characteristics of signal interference due to nonlinear distortion of the amplifiers arranged for each antenna added to it. Therefore, since the inverse characteristics that are in line with the physical model of signal interference due to nonlinear distortion of the amplifiers arranged for each antenna can be effectively added, higher distortion compensation performance can be obtained with less computation. Note that the value of H may be set in advance in the phased array transmitter 10, for example.

[0077] Here, the distortion compensation unit 12-k, as shown in Figure 8, inputs each output data (each target variable) of the k-th trained model to the input signal x k Alternatively, the values ​​obtained by multiplying each of the time-series signals may be added together.

[0078] In the example in Figure 8, the distortion compensation unit 12-k multiplies the first output data of the second trained model 801 with the value x1(t) 851-1, and multiplies the second output data with the value x1(t-1) 851-2. It multiplies the h+1th output data with the value x2(t) 852-1. It multiplies the (N-1)h+1th output data with the value x N (t) is multiplied by 85N-1. The compensation unit 12-k outputs the value obtained by adding the results of each multiplication 851-1 to 85N-h to form the output signal y k It is outputting as follows.

[0079] Steps S101 to S102 and steps S104 to S105 are the same as in Embodiment 1.

[0080] Although the present disclosure has been described above with reference to embodiments (Embodiment 1 and Embodiment 2), the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure can be made as understandable to those skilled in the art within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate.

[0081] Some or all of the embodiments described above may also be described as follows, but are not limited to the following. Some or all of the elements (e.g., configuration and function) described in each appendix dependent on Appendix 1 may also be dependent on other independent appendices of other categories in a similar manner. Some or all of the elements described in any appendix may be applicable to various hardware, software, recording means, systems, and methods for recording software. (Note 1) An input section that receives the time-series signal of the first input signal and the time-series signal of the second input signal, A first distortion compensation unit outputs a first set of coefficients using a first neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, and outputs a first output signal obtained by multiplying each value included in the time-series signal of the first input signal with each coefficient included in the first set of coefficients and then adding them together. The system includes a second distortion compensation unit that outputs a second set of coefficients using a second neural network model based on the time-series signals of the first input signal and the second input signal, and outputs a second output signal obtained by multiplying each value included in the time-series signal of the second input signal by each coefficient included in the second set of coefficients and then adding them together, The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are wirelessly radiated from the multiple antennas. Phased array transmitter. (Note 2) The input unit receives the time-series signal of the third input signal, The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal, and outputs the first output signal by multiplying each value included in the time-series signal of the first input signal with each coefficient included in the first set of coefficients and then adding the results together. The second distortion compensation unit outputs the second set of coefficients using the second neural network model based on the time-series signal of the first input signal, the time-series signal of the second input signal, and the time-series signal of the third input signal. It then outputs the second output signal by multiplying each value included in the time-series signal of the second input signal by each coefficient included in the second set of coefficients and adding them together. The phased array transmitter has a third distortion compensation unit that outputs a third set of coefficients using a third neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal, and outputs a third output signal which is the sum of the values ​​obtained by multiplying each value included in the time-series signal of the third input signal with each coefficient included in the third set of coefficients. The first output signal, the second output signal, and the third output signal are phase-controlled and amplified in accordance with each of the plurality of antennas, and are radiated wirelessly from the plurality of antennas. Phased array transmitter as described in Appendix 1. (Note 3) The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on at least one of the amplitude of each signal included in the time-series signal of the first input signal and the square of the amplitude. Phased array transmitter as described in Appendix 1 or 2. (Note 4) The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on at least one of the real part, imaginary part, square of the real part, square of the imaginary part, and the product of the real and imaginary parts of each signal included in the time series signal of the first input signal. Phased array transmitter as described in Appendix 1 or 2. (Note 5) The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on the amplitude of the linear sum of the first input signal and the second input signal, the square of the amplitude, and at least one of the dot product or cross product of the first input signal and the second input signal, where the I signal and Q signal of each input signal are considered as separate components of a vector, respectively. Phased array transmitter as described in Appendix 1 or 2. (Note 6) The first distortion compensation unit updates the first neural network model based on a first received signal obtained from a receiver that has received radio waves based on the first output signal. Phased array transmitter as described in Appendix 1 or 2. (Note 7) The first distortion compensation unit updates the first neural network model based on the first output signal amplified by the amplifiers corresponding to each of the plurality of antennas and the first received signal estimated based on the channel matrix between the phased array transmitter and the receiver. Phased array transmitter as described in Appendix 1 or 2. (Note 8) The first distortion compensation unit removes network connections in the first neural network model based on the first output signal and the first received signal when a radio wave based on the first output signal is received by a receiver. Phased array transmitter as described in Appendix 1 or 2. (Note 9) Phased array transmitters The time-series signal of the first input signal and the time-series signal of the second input signal are input. Based on the time-series signals of the first input signal and the time-series signals of the second input signal, a first neural network model outputs a first set of coefficients, and the first output signal is output by multiplying each value in the time-series signal of the first input signal with each coefficient in the first set of coefficients, and then adding the results together. Based on the time-series signals of the first input signal and the time-series signals of the second input signal, a second neural network model outputs a second set of coefficients, and the second output signal is obtained by multiplying each value in the time-series signal of the second input signal with each coefficient in the second set of coefficients, and then adding the results together. The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are wirelessly radiated from the multiple antennas. Sending method. (Note 10) The time-series signal of the first input signal and the time-series signal of the second input signal are input. Based on the time-series signals of the first input signal and the time-series signals of the second input signal, a first neural network model outputs a first set of coefficients, and the first output signal is output by multiplying each value in the time-series signal of the first input signal with each coefficient in the first set of coefficients, and then adding the results together. Based on the time-series signals of the first input signal and the second input signal, a second neural network model outputs a second set of coefficients. The second output signal is obtained by multiplying each value in the time-series signal of the second input signal by each coefficient in the second set of coefficients, and then adding the results together. The processing is performed by the computer in the phased array transmitter. The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are wirelessly radiated from the multiple antennas. program. (Note 11) An input section that receives the time-series signal of the first input signal and the time-series signal of the second input signal, A first distortion compensation unit outputs a first set of coefficients using a first neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, and outputs a first output signal obtained by multiplying each value included in the time-series signals of the first input signal and the time-series signals of the second input signal with each coefficient included in the first set of coefficients, and then adding them together. The system includes a second distortion compensation unit that outputs a second set of coefficients using a second neural network model based on the time-series signals of the first input signal and the second input signal, and outputs a second output signal obtained by multiplying each value included in the time-series signals of the first input signal and the time-series signals of the second input signal with each coefficient included in the second set of coefficients, and then adding them together. The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are wirelessly radiated from the multiple antennas. Phased array transmitter. (Note 12) The input unit receives the time-series signal of the third input signal, The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal. It then outputs the first output signal by multiplying each value contained in the time-series signals of the first input signal, the second input signal, and the third input signal by each coefficient contained in the first set of coefficients and adding them together. The second distortion compensation unit outputs the second set of coefficients using the second neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal. It then outputs the second output signal by multiplying each value contained in the time-series signals of the first input signal, the second input signal, and the third input signal, and each coefficient contained in the second set of coefficients, and then adding them together. The phased array transmitter has a third distortion compensation unit that outputs a third set of coefficients using a third neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal, and outputs a third output signal which is the sum of the values ​​included in the time-series signals of the first input signal, the second input signal, and the third input signal, and the coefficients included in the third set of coefficients, respectively, after multiplying them by each other. The first output signal, the second output signal, and the third output signal are phase-controlled and amplified in accordance with each of the plurality of antennas, and are radiated wirelessly from the plurality of antennas. Phased array transmitter as described in Appendix 11. [Explanation of Symbols]

[0082] 10 Phased Array Transmitter 11 Input section 12 Distortion compensation section 13 DAC 14 Beamforming Matrix

Claims

1. An input section that receives the time-series signal of the first input signal and the time-series signal of the second input signal, A first distortion compensation unit outputs a first set of coefficients using a first neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, and outputs a first output signal obtained by multiplying each value included in the time-series signal of the first input signal with each coefficient included in the first set of coefficients and then adding them together. The system includes a second distortion compensation unit that outputs a second set of coefficients using a second neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, and outputs a second output signal obtained by multiplying each value included in the time-series signal of the second input signal with each coefficient included in the second set of coefficients and then adding them together, The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are radiated wirelessly from the multiple antennas. Phased array transmitter.

2. The input unit receives the time-series signal of the third input signal, The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on the time-series signal of the first input signal, the time-series signal of the second input signal, and the time-series signal of the third input signal, and outputs the first output signal by multiplying each value included in the time-series signal of the first input signal with each coefficient included in the first set of coefficients and then adding the results together. The second distortion compensation unit outputs the second set of coefficients using the second neural network model based on the time-series signal of the first input signal, the time-series signal of the second input signal, and the time-series signal of the third input signal. It then outputs the second output signal by multiplying each value included in the time-series signal of the second input signal by each coefficient included in the second set of coefficients and adding them together. The phased array transmitter has a third distortion compensation unit that outputs a third set of coefficients using a third neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal, and outputs a third output signal which is the sum of the values ​​obtained by multiplying each value included in the time-series signal of the third input signal with each coefficient included in the third set of coefficients. The first output signal, the second output signal, and the third output signal are phase-controlled and amplified in accordance with each of the plurality of antennas, and are radiated wirelessly from the plurality of antennas. The phased array transmitter according to claim 1.

3. The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on at least one of the amplitude of each signal included in the time-series signal of the first input signal and the square of the amplitude. A phased array transmitter according to claim 1 or 2.

4. The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on at least one of the real part, imaginary part, square of the real part, square of the imaginary part, and the product of the real and imaginary parts of each signal included in the time series signal of the first input signal. A phased array transmitter according to claim 1 or 2.

5. The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on the amplitude of the linear sum of the first input signal and the second input signal, the square of the amplitude, and at least one of the dot product or cross product of the first input signal and the second input signal, where the I signal and Q signal of each input signal are considered as separate components of a vector. A phased array transmitter according to claim 1 or 2.

6. The first distortion compensation unit updates the first neural network model based on the first received signal obtained from a receiver that has received radio waves based on the first output signal. A phased array transmitter according to claim 1 or 2.

7. The first distortion compensation unit updates the first neural network model based on the first output signal amplified by the amplifiers corresponding to each of the plurality of antennas and the first received signal estimated based on the channel matrix between the phased array transmitter and the receiver. A phased array transmitter according to claim 1 or 2.

8. The first distortion compensation unit removes network connections in the first neural network model based on the first output signal and the first received signal when a radio wave based on the first output signal is received by a receiver. A phased array transmitter according to claim 1 or 2.

9. Phased array transmitters The time-series signal of the first input signal and the time-series signal of the second input signal are input. Based on the time-series signals of the first input signal and the time-series signals of the second input signal, a first neural network model outputs a first set of coefficients, and the first output signal is output by multiplying each value in the time-series signal of the first input signal with each coefficient in the first set of coefficients, and then adding the results together. Based on the time-series signals of the first input signal and the second input signal, a second neural network model outputs a second set of coefficients, and the second output signal is obtained by multiplying each value in the time-series signal of the second input signal with each coefficient in the second set of coefficients, and then adding the results together. The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are radiated wirelessly from the multiple antennas. Sending method.

10. The time-series signal of the first input signal and the time-series signal of the second input signal are input. Based on the time-series signals of the first input signal and the time-series signals of the second input signal, a first neural network model outputs a first set of coefficients, and the first output signal is output by multiplying each value in the time-series signal of the first input signal with each coefficient in the first set of coefficients, and then adding the results together. Based on the time-series signals of the first input signal and the second input signal, a second neural network model outputs a second set of coefficients. The second output signal is obtained by multiplying each value in the time-series signal of the second input signal by each coefficient in the second set of coefficients, and then adding the results together. The processing is performed by the computer in the phased array transmitter. The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are radiated wirelessly from the multiple antennas. program.

11. An input section that receives the time-series signal of the first input signal and the time-series signal of the second input signal, A first distortion compensation unit outputs a first set of coefficients using a first neural network model based on the time-series signals of the first input signal and the time-series signals of the second input signal, and outputs a first output signal obtained by multiplying each value included in the time-series signals of the first input signal and the time-series signals of the second input signal with each coefficient included in the first set of coefficients, and then adding them together. The system includes a second distortion compensation unit that outputs a second set of coefficients using a second neural network model based on the time-series signals of the first input signal and the second input signal, and outputs a second output signal obtained by multiplying each value included in the time-series signals of the first input signal and the time-series signals of the second input signal with each coefficient included in the second set of coefficients, and then adding them together. The first output signal and the second output signal are phase-controlled and amplified in accordance with each of the multiple antennas, and are radiated wirelessly from the multiple antennas. Phased array transmitter.

12. The input unit receives the time-series signal of the third input signal, The first distortion compensation unit outputs the first set of coefficients using the first neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal. It then outputs the first output signal by multiplying each value contained in the time-series signals of the first input signal, the second input signal, and the third input signal with each coefficient contained in the first set of coefficients, and then adding the results together. The second distortion compensation unit outputs the second coefficient group using the second neural network model based on the time-series signal of the first input signal, the time-series signal of the second input signal, and the time-series signal of the third input signal. It then outputs the second output signal by multiplying each value contained in the time-series signal of the first input signal, the time-series signal of the second input signal, and the time-series signal of the third input signal with each coefficient contained in the second coefficient group, and then adding the results together. The phased array transmitter has a third distortion compensation unit that outputs a third set of coefficients using a third neural network model based on the time-series signals of the first input signal, the second input signal, and the third input signal, and outputs a third output signal which is the sum of the values ​​included in the time-series signals of the first input signal, the second input signal, and the third input signal, and the coefficients included in the third set of coefficients, respectively. The first output signal, the second output signal, and the third output signal are phase-controlled and amplified in accordance with each of the plurality of antennas, and are radiated wirelessly from the plurality of antennas. The phased array transmitter according to claim 11.

Images