Delta-sigma modulation device, delta-sigma modulation method, and program

Abstract

To suppress distortion occurring in a signal output through delta-sigma modulation in the process of transmission.SOLUTION: A delta-sigma modulation device 3500 performs delta-sigma modulation on a first signal as an input signal and outputs a second signal. The delta-sigma modulation device 3500 uses a second signal and a third signal generated through a transmission process 3530 of the second signal to output a fourth signal that is an approximation of the signal produced through at least a portion of the transmission process 3530. The delta-sigma modulation device 3500 performs the delta-sigma modulation on the first signal using the fourth signal and outputs the second signal.SELECTED DRAWING: Figure 35

Description

[Technical Field]

[0001] The present disclosure relates to a delta-sigma modulation device, a delta-sigma modulation method, a program, and a recording medium. [Background technology]

[0002] In the field of wireless communications, progress is being made in the development of technologies to achieve high-speed communications in order to cope with increasing traffic. To achieve high-speed communications, wireless communications devices typically need to be equipped with high-speed, high-precision digital-to-analog converters (DACs). However, such devices have the problem of consuming large amounts of power.

[0003] To solve this problem, delta-sigma modulation is sometimes used (see, for example, Patent Document 1, Non-Patent Document 1, and Non-Patent Document 2). Delta-sigma modulation is a process for converting an analog signal as an input signal into a quantized signal (pulse train). Delta-sigma modulation can reduce the required resolution of the DAC (to 1 bit in some cases). This can reduce the power consumption of wireless communication devices.

[0004] Meanwhile, in wireless communication devices, a configuration is known in which multiple functions of a radio access network device are divided into two physically separated devices. For example, a base station is divided into a base band unit (BBU) and a remote unit (RU). The RU may also be called an RRU (remote radio unit), an RRH (remote radio head), or an RAU (remote antenna unit). The BBU and RU are connected via a communication path (e.g., optical fiber). In this configuration, signal transmission between the BBU and RU uses radio over fiber (RoF) technology.

[0005] In recent years, systems using delta-sigma modulation and RoF technology have been studied (see, for example, Non-Patent Document 3). Hereinafter, such systems will be referred to as "RoF systems." In RoF systems, a pulse train output by delta-sigma modulation is transmitted from a BBU to an RU via an optical fiber. In this case, the RU restores the pulse train to the original analog signal before delta-sigma modulation, for example, through an analog band-pass filter (BPF). Since the RU does not need to be equipped with a DAC, the power consumption of the RU can be reduced. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] International Publication No. 2016 / 103981 [Patent Document 2] International Publication No. 2020 / 175581 [Patent Document 3] International Publication No. 2018 / 230112 [Patent Document 4] International Publication No. 2012 / 108421 [Patent Document 5] Japanese Patent Publication No. 2021-164098 [Non-patent literature]

[0007] [Non-Patent Document 1] T. Maehata, S. Kameda, and N. Suematsu, “1- bit feedforward distortion compensation technology for bandpass delta-sigma modulation,” IEICE Trans. Commun., vol.E99-B, no.5, pp.1087-1092, May 2016 [Non-patent document 2] A.Frappe, A.Flament, B.Stefanelli, A.Kaiser, and A.Cathelin, “An all-digital RF signal generator using high-speed SD modulators,” IEEE Journal of Solid-State Circuits, vol.44, no.10, pp.2722-2732, Oct.2009 [Non-patent document 3] H. Li et al., “Real-Time 100-GS / s Sigma-Delta Modulator for All-Digital Radio-over-Fiber Transmission,” J. Lightw. Technol. vol. 38, no. 2, pp. 386-393, Jul. 2019. Summary of the Invention [Problem to be solved by the invention]

[0008] Incidentally, as described in Non-Patent Document 1, it is known that in a configuration using delta-sigma modulation, a phenomenon called spectrum leakage occurs due to waveform distortion. Here, spectrum leakage is a phenomenon in which signal components in bands surrounding a desired frequency band (i.e., bands other than the desired frequency band) are generated.

[0009] Patent Document 1 discloses a technique for compensating for distortion caused by delta-sigma modulation itself. Specifically, the technique of Patent Document 1 compensates for distortion caused by asymmetry between the rising and falling edges of pulses in a pulse train. However, the technique of Patent Document 1 cannot compensate for distortion that occurs during the transmission of a pulse train, for example, in the RoF system described above.

[0010] The present disclosure provides a technique capable of suppressing distortion that occurs in a signal (pulse train) output through delta-sigma modulation during the process of transmitting the signal. [Means for solving the problem]

[0011] In one or more embodiments, a delta-sigma modulation device is provided. The delta-sigma modulation device includes a delta-sigma modulation unit that performs delta-sigma modulation and a modeling processing unit that approximates transmission characteristics of at least a portion of a transmission process. The delta-sigma modulation unit performs the delta-sigma modulation on a first signal as an input signal and outputs a second signal. The modeling processing unit uses the second signal and a third signal generated through a transmission process of the second signal to output a fourth signal that is an approximation of a signal generated through at least a portion of the transmission process. The delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal and outputs the second signal.

[0012] In one or more embodiments, a delta-sigma modulation method is provided, which includes: performing delta-sigma modulation on a first signal as an input signal to output a second signal; using the second signal and a third signal generated through a transmission process of the second signal to output a fourth signal that is an approximation of a signal generated through at least a part of the transmission process; and performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal.

[0013] In one or more embodiments, a computer-readable non-transitory recording medium is provided, having recorded thereon a program that causes a processor to perform the following operations: perform delta-sigma modulation on a first signal as an input signal to output a second signal; use the second signal and a third signal generated through a transmission process of the second signal to output a fourth signal that is an approximation of a signal generated through at least a part of the transmission process; and perform the delta-sigma modulation on the first signal using the fourth signal to output the second signal. [Effects of the Invention]

[0014] According to the above configuration, distortion occurring in the second signal (pulse train) output through delta-sigma modulation can be suppressed during transmission of the second signal. Other problems, configurations, and effects will become clear from the description of the following embodiments. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a diagram illustrating a schematic configuration of a wireless communication device according to a first embodiment. [Figure 2] FIG. 2 is a diagram illustrating a configuration of a delta-sigma modulation section according to the first embodiment. [Figure 3] FIG. 2 is a diagram illustrating a configuration of a neural network processing unit according to the first embodiment. [Figure 4] FIG. 2 is a diagram illustrating a configuration of a neural network learning unit according to the first embodiment. [Figure 5] FIG. 2 is a diagram illustrating the configuration of a modulation unit according to the first embodiment. [Figure 6] FIG. 3 is a diagram illustrating a flow of constructing a transmission model according to the first embodiment. [Figure 7] 3 is a flowchart showing the flow of processing performed by the delta-sigma modulation device according to the first embodiment. [Figure 8] 1 is a diagram illustrating an example of a specific configuration of a wireless communication device according to a first embodiment. [Figure 9] FIG. 2 is a diagram illustrating a hardware configuration of a first device according to the first embodiment. [Figure 10] FIG. 2 is a diagram illustrating a hardware configuration of a second device according to the first embodiment. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the wireless communication device according to the first embodiment. [Figure 12] FIG. 10 is a diagram illustrating a configuration of a delta-sigma modulation unit according to a modified example. [Figure 13] FIG. 10 is a diagram showing the configuration of a first neural network according to a modified example. [Figure 14]FIG. 10 is a diagram showing the configuration of a first neural network according to a modified example. [Figure 15] 10 is a flowchart showing the flow of processing performed by a delta-sigma modulation device according to a modified example. [Figure 16] FIG. 10 is a diagram illustrating a configuration of a neural network learning unit according to a modified example. [Figure 17] FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a second embodiment. [Figure 18] FIG. 10 is a sequence diagram showing a processing flow of the wireless communication device according to the second embodiment. [Figure 19] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 20] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 21] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 22] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 23] FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a modified example. [Figure 24] FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a modified example. [Figure 25] FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a third embodiment. [Figure 26] FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a fourth embodiment. [Figure 27] FIG. 11 is a sequence diagram showing a processing flow of a wireless communication device according to the fourth embodiment. [Figure 28] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 29] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 30] FIG. 10 is a sequence diagram showing a processing flow of a wireless communication device according to a modified example. [Figure 31] FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a modified example. [Figure 32]FIG. 10 is a diagram illustrating a configuration of a wireless communication device according to a modified example. [Figure 33] FIG. 10 is a diagram illustrating a configuration of a delta-sigma modulation device according to a fifth embodiment. [Figure 34] 10 is a flowchart showing the flow of processing performed by a delta-sigma modulation device according to a fifth embodiment. [Figure 35] FIG. 10 is a diagram illustrating a configuration of a delta-sigma modulation device according to a sixth embodiment. [Figure 36] FIG. 2 is a diagram illustrating an example of the configuration of a first model. [Figure 37] FIG. 2 is a diagram illustrating an example of the configuration of a first model. [Figure 38] FIG. 2 is a diagram illustrating an example of the configuration of a first model. [Figure 39] FIG. 2 is a diagram illustrating an example of the configuration of a first model. [Figure 40] FIG. 2 is a diagram illustrating an example of the configuration of a first model. [Figure 41] 13 is a flowchart showing the flow of processing performed by a delta-sigma modulation device according to a sixth embodiment. [Figure 42] FIG. 2 illustrates an example of the configuration of a conversion unit. [Figure 43] FIG. 2 illustrates an example of a configuration of a modeling processing unit. [Figure 44] 10 is a flowchart showing the flow of processing performed by a delta-sigma modulation device according to a modified example. [Figure 45] FIG. 2 illustrates an example of the configuration of a model parameter calculation unit. [Figure 46] FIG. 2 illustrates an example of the configuration of a model parameter calculation unit. [Figure 47] This is an example of a system using delta-sigma modulation and RoF technology. [Figure 48] This is an example of a system using delta-sigma modulation and RoF technology. DETAILED DESCRIPTION OF THE INVENTION

[0016] One or more embodiments will be described below with reference to the accompanying drawings. In this specification and drawings, elements that can be similarly described will be designated by the same reference numerals, and redundant description will be omitted.

[0017] The explanation will be given in the following order: 1. Overview of the embodiment 2. First embodiment 2-1. General configuration of wireless communication device 2-2. Delta-sigma modulation section configuration 2-3.Configuration of neural network processing unit 2-4. Overview of the neural network learning section 2-5.Communication Path Configuration 2-6. Bandpass filter configuration 2-7. Modulation section configuration 2-8.Specific configuration of neural network learning unit 2-9. Processing flow of delta-sigma modulator 2-10.Specific configuration example of wireless communication device 2-11. Configuration of the first device 2-12. Configuration of the second device 2-13.Processing flow of wireless communication device 2-14. Variations 3. Second embodiment 3-1. General configuration of wireless communication device 3-2. Configuration of the first device 3-3. Configuration of the second device 3-4. Overview of processing flow of wireless communication device 3-5. Processing flow of wireless communication device 3-6. Variations 4. Third embodiment 4-1. General configuration of wireless communication device 4-2. Configuration of the first device 4-3. Configuration of the second device 5. Fourth embodiment 5-1. General configuration of wireless communication device 5-2. Configuration of the first device 5-3. Configuration of the second device 5-4. Processing flow of wireless communication device 5-5. Variations 6. Fifth embodiment 6-1.Configuration of Delta-Sigma Modulator 6-2. Processing flow of delta-sigma modulator 7. Sixth embodiment 7-1. Delta-sigma modulator configuration 7-2.Model configuration 7-3. Processing flow of delta-sigma modulator 7-4. Variations

[0018] <<1. Overview of the embodiment>> A summary of one or more embodiments is provided below.

[0019] (1)Technical issues 47 shows an example of a system using delta-sigma modulation and RoF technology. An analog signal 4701 as an input signal is converted into a quantized signal (pulse train) 4702 through a delta-sigma modulation unit 4711. Note that this delta-sigma modulation unit 4711 has a typical one-input, one-output configuration, and the quantized signal (pulse train) is fed back inside the delta-sigma modulation unit 4711. The pulse train 4702 passes through a transmission path 4720 including an optical fiber 4713. The pulse train 4702 is first converted into an optical signal through an E / O conversion unit (Electric-Optic converter) 4712. The optical signal is transmitted on the optical fiber 4713. Thereafter, the optical signal is converted into an electrical signal 4703 through an O / E conversion unit (Optic-Electric converter) 4715. The electrical signal 4703 is converted through a band pass filter (BPF) 4716 into the original analog signal 4704 before delta-sigma modulation. The analog signal 4704 is amplified using an amplifier 4717. The amplified analog signal is output from an antenna 4718.

[0020] 47, ideal delta-sigma modulation is performed. No distortion occurs in pulse train 4702 during the transmission process of pulse train 4702. As a result, no spectrum leakage occurs.

[0021] 48 shows an example of a system using delta-sigma modulation and RoF technology. In this system, distortion occurs in pulse train 4702 during the transmission process of pulse train 4702. For example, distortion occurs when an optical signal passes through optical fiber 4713. This type of distortion is called "optical transmission distortion." Pulse train 4801 represents a state in which optical transmission distortion has occurred.

[0022] Furthermore, distortion also occurs when there is a large amount of power reflection between the O / E conversion unit 4715 and the BPF 4716. This type of distortion is called "electrical transmission distortion." The pulse train 4802 represents a state in which electrical transmission distortion has occurred. Hereinafter, optical transmission distortion and electrical transmission distortion are collectively called "transmission distortion."

[0023] As a result, the transmission distortion affects the frequency components of the analog signal 4803 output via the BPF 4716. Specifically, the above-mentioned spectrum leak occurs, and the signal-to-noise ratio (SNR) of the analog signal 4803 deteriorates. Thus, there is a problem in that the transmission distortion causes degradation of signal quality. In view of the above circumstances, a configuration capable of suppressing the transmission distortion is required.

[0024] (2) Technical Features In one or more embodiments, a delta-sigma modulation device is provided. The delta-sigma modulation device includes a delta-sigma modulation unit that performs delta-sigma modulation and a modeling processing unit that approximates transmission characteristics of at least a portion of a transmission process. The delta-sigma modulation unit performs delta-sigma modulation on a first signal as an input signal and outputs a second signal. The modeling processing unit uses the second signal and a third signal generated through a transmission process of the second signal to output a fourth signal that is an approximation of a signal generated through at least a portion of the transmission process. The delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal and outputs the second signal.

[0025] Generally, the one-input, one-output configuration described in FIG. 47 (i.e., the delta-sigma modulation unit 4711 in which the quantized signal is directly fed back therein) is referred to as delta-sigma modulation. In contrast, in one or more embodiments of the present disclosure, the second signal is not directly fed back as the fourth signal, and the second signal and the fourth signal are separated via a modeling processing unit. The delta-sigma modulation unit has a configuration that receives two input signals, the first signal and the fourth signal, and outputs a second signal, that is, a two-input, one-output configuration. In one or more embodiments of the present disclosure, for convenience, such a two-input, one-output configuration is defined as delta-sigma modulation.

[0026] A delta-sigma modulation device having the above configuration can suppress transmission distortion that occurs in the second signal during the transmission process of the second signal. The delta-sigma modulation device reduces the possibility of spectrum leakage, thereby enabling high-quality signal transmission.

[0027] In another embodiment, a delta-sigma modulation device includes a delta-sigma modulation unit that performs delta-sigma modulation on a first signal as an external input signal and outputs a second signal, a neural network processing unit that includes a first neural network that operates according to a predetermined first parameter, and a neural network learning unit that calculates the first parameter.

[0028] The neural network learning unit calculates the first parameter using the second signal and a third signal generated through a transmission process of the second signal.

[0029] The neural network processing unit updates the first parameter to the first parameter calculated by the neural network learning unit.

[0030] The neural network processing unit uses the second signal to pass through the first neural network and outputs a fourth signal, the fourth signal being an approximation of a signal generated through at least a portion of the transmission process of the second signal.

[0031] The delta-sigma modulation unit performs delta-sigma modulation on the first signal using the fourth signal output from the neural network processing unit, and outputs a second signal.

[0032] In this embodiment, the second signal is not directly fed back as the fourth signal, but the second signal and the fourth signal are separated via a neural network processing unit. The delta-sigma modulation unit is configured to receive the first signal and the fourth signal as input signals and output the second signal, i.e., has a two-input, one-output configuration.

[0033] The neural network learning unit may calculate the first parameter using the third signal and a restored signal of the second signal generated from the third signal.

[0034] A delta-sigma modulation device having the above configuration can suppress transmission distortion that occurs in the second signal during the transmission process of the second signal. The delta-sigma modulation device reduces the possibility of spectrum leakage, thereby enabling high-quality signal transmission.

[0035] <<2. First Embodiment>> Next, the first embodiment and its modified examples will be described with reference to FIGS.

[0036] <2-1. General configuration of wireless communication device> 1 is a diagram showing a schematic configuration of a wireless communication device 100. The wireless communication device 100 is a device that complies with the technical specifications of 3GPP (Third Generation Partnership Project). Specifically, the wireless communication device 100 may be a device that complies with the technical specifications of 5G (5th Generation). Naturally, the wireless communication device 100 is not limited to this example.

[0037] The wireless communication device 100 includes a delta-sigma modulation device 110 and a transmission path 120. The wireless communication device 100 converts a signal output by the delta-sigma modulation device 110 into a wireless signal via the transmission path 120, and transmits the wireless signal to the terminal device 190.

[0038] Delta-sigma modulation device 110 includes delta-sigma modulation section 111, neural network processing section 112, and neural network training section 113. Hereinafter, neural network processing section 112 will be referred to as "NN processing section 112." Neural network training section 113 will be referred to as "NN training section 113."

[0039] The transmission path 120 includes a communication path 130 and a transmitter 140. The communication path 130 connects the delta-sigma modulator 111 and the transmitter 140. The transmitter 140 includes an antenna 141, a band-pass filter (BPF) 142, an amplifier 143, and a modulator 144.

[0040] The delta-sigma modulation unit 111 performs delta-sigma modulation on an analog signal as an external input signal, and outputs a quantized signal (1-bit pulse train). The quantized signal is transmitted to the BPF 142 via the communication path 130. The BPF 142 performs bandpass processing (hereinafter referred to as "BPF processing") on the quantized signal, passing only a desired frequency band. The quantized signal is converted into an analog signal through the BPF 142. That is, the quantized signal is restored to the original analog signal before delta-sigma modulation. The amplifier 143 amplifies the analog signal. The antenna 141 outputs the amplified analog signal. In this way, the transmission unit 140 has the function of transmitting a signal to the terminal device 190.

[0041] The amplified analog signal is also output to the modulation unit 144. The modulation unit 144 performs modulation processing on the analog signal and outputs the modulated analog signal to the NN learning unit 113. In this way, the transmission unit 140 also has a function of transmitting (i.e., feeding back) to the delta-sigma modulation device 110 a signal generated through a transmission process of the quantized signal output by the delta-sigma modulation unit 111.

[0042] Hereinafter, the analog signal input to the delta-sigma modulation unit 111 will be referred to as the "first signal." Furthermore, the quantized signal output by the delta-sigma modulation unit 111 will be referred to as the "second signal." Furthermore, the signal generated through the transmission process of the second signal and fed back to the NN learning unit 113 will be referred to as the "third signal." In other words, the third signal is a signal generated in the process of the second signal being transmitted over the transmission path 120. The third signal may include information about transmission distortion occurring in the transmission path 120. Therefore, the third signal is used to feed back information about transmission distortion occurring in the transmission path 120.

[0043] The NN processing unit 112 includes a first neural network 300 that operates according to predetermined parameters (see FIG. 3). Hereinafter, the first neural network 300 will be referred to as the "first NN 300." The parameters of the first NN 300 will be referred to as the "first parameters."

[0044] The NN learning unit 113 receives the second signal and the third signal as input signals. The NN learning unit 113 calculates a first parameter using the second signal and the third signal. The NN learning unit 113 transmits the calculated first parameter to the NN processing unit 112.

[0045] The NN processing unit 112 receives the first parameters from the NN learning unit 113. The NN processing unit 112 updates the first parameters of the first NN 300 to the received first parameters (i.e., the first parameters calculated by the NN learning unit 113).

[0046] The NN processing unit 112 receives the second signal as an input signal. Using the second signal, the NN processing unit 112 outputs an approximation (analog signal) of a signal generated through at least a part of the transmission process of the second signal via the first NN 300. Hereinafter, this approximation is referred to as a "fourth signal." Using the fourth signal, the NN processing unit 112 can feed back to the delta-sigma modulation unit 111 information about transmission distortion occurring in the transmission process of the second signal.

[0047] The delta-sigma modulation unit 111 receives the first signal and the fourth signal as input signals, performs delta-sigma modulation on the first signal using the fourth signal, and outputs a second signal.

[0048] As described above, the delta-sigma modulation unit 111 is configured to receive two input signals, the first signal and the fourth signal, and output the second signal, i.e., a two-input, one-output configuration. The second signal and the fourth signal are separated by interposing the NN processing unit 112 between them. The NN processing unit 112 performs a feedback function. This embodiment is characterized in that the first parameter in the NN processing unit 112 is updated by the NN learning unit 113.

[0049] <2-2. Delta-sigma modulation section configuration> There are two types of devices that implement the delta-sigma modulation unit 111: a band-pass type (see Non-Patent Document 1) and a low-pass type (see Non-Patent Document 2). The delta-sigma modulation unit 111 in this example is a band-pass type device, and is implemented with reference to Non-Patent Document 1.

[0050] 2 is a diagram showing an example of the configuration of the delta-sigma modulation section 111. The delta-sigma modulation section 111 includes an up-converter 210, a loop filter 220, and a quantizer 230.

[0051] The first signal is a baseband signal, and includes an in-phase component signal (hereinafter referred to as "I signal") and a quadrature component signal (hereinafter referred to as "Q signal").

[0052] The upconverter 210 is a two-input, one-output component. The upconverter 210 receives an I signal and a Q signal as input signals. The upconverter 210 upconverts the first signal (I signal and Q signal) to a desired frequency (target frequency) f0.

[0053] The upconverter 210 includes a first multiplier 211a, a second multiplier 211b, and an adder 212. The first multiplier 211a multiplies the I signal by cosωt and outputs the multiplication result to the adder 212. The second multiplier 211b multiplies the Q signal by -sinωt and outputs the multiplication result to the adder 212. Here, "cos()" is a cosine function, and "sin()" is a sine function (the same applies below). Furthermore, ω=2×π×f0.

[0054] The adder 212 adds the multiplication result of the first multiplier 211a and the multiplication result of the second multiplier 211b and outputs the addition result.

[0055] Loop filter 220 is a two-input, one-output element. Loop filter 220 receives the output of upconverter 210 and the fourth signal as input signals. Loop filter 220 performs processing to suppress transmission distortion contained in the feedback component of the input signal (i.e., the fourth signal).

[0056] The loop filter 220 includes a first adder 221 a, a second adder 221 b, and a transfer function processing unit 222.

[0057] First adder 221a adds the output of upconverter 210 (output of adder 212) and the fourth signal, and outputs the addition result to transfer function processing unit 222. Here, the output of first adder 221a is the difference between the output of upconverter 210 and the fourth signal. In other words, the output of first adder 221a may include a component of transmission distortion that occurs in the transmission process of the second signal.

[0058] The transfer function processing unit 222 applies a transfer function to the output of the first adder 221a and outputs a component for suppressing (or canceling) at least a part of the transmission distortion occurring in the transmission process of the second signal (transmission path 120). The transfer function is a function that determines the characteristics of the delta-sigma modulation in this example, and is determined based on a desired signal transfer function, a noise transfer function, etc.

[0059] The second adder 221b adds the output of the upconverter 210 and the output of the transfer function processing unit 222 together, and outputs the addition result to the quantizer 230.

[0060] In this way, loop filter 220 uses the output of upconverter 210 and the fourth signal to output a signal (analog signal) that includes a component for suppressing at least a part of the transmission distortion that occurs in the transmission process of the second signal.

[0061] Quantizer 230 is a 1-bit quantizer that quantizes the output of loop filter 220 (the output of second adder 221b) by 1 bit and outputs a second signal (a 1-bit pulse train).

[0062] In this way, the delta-sigma modulation unit 111 uses the fourth signal to calculate a component for suppressing at least a part of the transmission distortion that occurs in the transmission process of the second signal, and reflects the calculated component in the second signal.

[0063] In another example, the second adder 221b may be omitted from the loop filter 220. In this case, the output of the transfer function processing unit 222 is input to the quantizer 230. The output of the transfer function processing unit 222 includes the output of the upconverter 210 and a component for suppressing at least a part of the transmission distortion that occurs in the transmission process of the second signal.

[0064] <2-3. Configuration of neural network processing unit> 3 is a diagram showing an example of the configuration of the NN processing unit 112. The NN processing unit 112 includes a first NN 300. The first NN 300 receives the second signal as an input signal and outputs a fourth signal.

[0065] In this example, the fourth signal is an approximation (estimate) of a signal generated when the second signal passes through a portion of the transmission path 120. More specifically, the fourth signal is an approximation (estimate) of a signal generated when the second signal passes through the communication path 130, the BPF 142, and the amplifier 143. In other words, the fourth signal is an approximation (estimate) of a radio signal output from the antenna 141.

[0066] Therefore, the fourth signal includes an approximation of at least a portion of the transmission distortion that occurs during the transmission process of the second signal. Specifically, the fourth signal includes an approximation of at least a portion of the transmission distortion that occurs during the process of the second signal passing through communication path 130, BPF 142, and amplifier 143.

[0067] The first NN 300 operates according to first parameters. The first parameters include, for example, weights and biases. For example, if the function f in the following formula (1) is an activation function in a neural network, x is an input, w is a weight, and b is a bias. f(wx+b) (1)

[0068] The first NN 300 includes an input layer 310, an intermediate layer 320, and an output layer 330. The intermediate layer 320 has one layer.

[0069] The input layer 310 includes a node 310a to which the current value of the second signal is input. The input layer 310 further includes a node 310b to which a past value of the second signal is input. Note that "D" represents a delay. The output layer 330 is a linear layer without an activation function. The first parameter is reflected in the hidden layer 320. The output layer 330 outputs the sum of the outputs of the multiple nodes in the hidden layer 320 as a fourth signal.

[0070] The configuration of the first NN 300 is not limited to the configuration shown in FIG. 3. Various commonly used neural networks may be applied to this embodiment. For example, the hidden layer 320 of the first NN 300 may have multiple layers. That is, a multi-layer neural network may be adopted as the first NN 300.

[0071] The NN processing unit 112 receives a first parameter from the NN learning unit 113. The NN processing unit 112 updates the first parameter of the first NN 300 to the received first parameter. That is, the NN processing unit 112 updates the first parameter in accordance with the state of the transmission path 120. Therefore, the NN processing unit 112 can reflect at least a part of the transmission distortion occurring in the current state of the transmission path 120 in the fourth signal.

[0072] <2-4. Overview of the neural network learning section> 4 is a diagram showing an example of the configuration of the NN learning unit 113. The NN learning unit 113 includes a transmission model 410, an error calculation unit 420, and a parameter calculation unit 430.

[0073] The transmission model 410 is a component that models at least a part of the transmission process of the second signal. In this example, the transmission model 410 is a component that models the transmission path 120. The transmission model 410 uses the second signal to output an approximation (estimate) of the third signal.

[0074] The transmission model 410 includes a second neural network 600 that operates according to predetermined parameters (see FIG. 6). Hereinafter, the second neural network 600 will be referred to as the "second NN 600." The parameters of the second NN 600 will be referred to as the "second parameters." The second parameters, like the first parameters, include weights and biases.

[0075] The error calculation unit 420 receives the output of the transmission model 410 (the approximation value of the third signal) and the third signal (in this example, the output of the modulation unit 144) as input signals. The error calculation unit 420 calculates the error (difference) between the output of the transmission model 410 and the third signal. Hereinafter, this error will be referred to as the "approximation error." The error calculation unit 420 outputs the approximation error to the parameter calculation unit 430.

[0076] The parameter calculation unit 430 calculates a first parameter and a second parameter using the approximation error. The parameter calculation unit 430 transmits the calculated first parameter to the NN processing unit 112. Furthermore, the parameter calculation unit 430 updates the second parameter of the second NN 600 to the calculated second parameter.

[0077] In this way, the NN learning unit 113 uses the second signal and the third signal to learn the current state of the transmission path 120. That is, the NN learning unit 113 calculates the first parameter and the second parameter according to the current state of the transmission path 120. The NN learning unit 113 can reflect at least a part of the transmission distortion occurring in the current state of the transmission path 120 in the first NN 300 and the second NN 600.

[0078] <2-5. Communication Path Configuration> The communication path 130 includes one or both of an electrical transmission line (eg, a metal wire) that transmits an electrical signal and an optical transmission line (eg, an optical fiber) that transmits an optical signal.

[0079] In addition, the communication path 130 may include one or a combination of an E / O conversion unit that converts an electrical signal into an optical signal, an O / E conversion unit that converts an optical signal into an electrical signal, a bandpass filter, and a frequency conversion unit.

[0080] <2-6. Bandpass filter configuration> The BPF 142 performs BPF processing on the output of the communication path 130. Here, the desired frequency band is the band from f0-fα to f0+fα, where fα is an arbitrary frequency.

[0081] The BPF 142 may be configured using analog elements. The BPF 142 may be an LC filter or an RLC filter. An LC filter is a filter configured based on a combination of an inductor (L) and a capacitor (C). An RLC filter is a filter configured based on a combination of a resistor (R), an inductor (L), and a capacitor (C).

[0082] <2-7. Modulation section configuration> 5 is a diagram showing an example of the configuration of modulation section 144. Modulation section 144 down-converts the output of amplifier 143 by direct conversion, and outputs the down-converted signal as a third signal.

[0083] The modulation section 144 includes a first modulation section 500a and a second modulation section 500b.

[0084] The first modulation section 500a includes a first multiplier 510a, a first low-pass filter (LPF) 520a, and a first A / D converter (ADC) 530a.

[0085] The first multiplier 510a multiplies the output of the amplifier 143 by cosωt and outputs the multiplication result to the first LPF 520a, where ω=2×π×f0. The first LPF 520a performs low-pass filtering on the output of the first multiplier 510a and outputs the result to the first ADC 530a. The first ADC 530a converts the output of the first LPF 520a into a digital signal, thereby outputting a digital I signal.

[0086] The second modulation section 500b includes a second multiplier 510b, a second low-pass filter (LPF) 520b, and a second A / D converter (ADC) 530b.

[0087] The second multiplier 510b multiplies the output of the amplifier 143 by -sinωt and outputs the multiplication result to the second LPF 520b. The second LPF 520b performs low-pass filtering on the output of the second multiplier 510b and outputs the result to the second ADC 530b. The second ADC 530b converts the output of the second LPF 520b into a digital signal, thereby outputting a digital Q signal.

[0088] Therefore, the third signal includes a digital I signal and a digital Q signal. The third signal is input to the NN learning unit 113.

[0089] <2-8. Specific configuration of neural network learning unit> Next, a specific configuration of the NN learning unit 113 will be described. First, a configuration of the transmission model 410 will be described. FIG.

[0090] As shown in the upper part of FIG. 6, the transmission path 120 includes a communication path 130, a BPF 142, an amplifier 143, and a modulation unit 144. Here, it is assumed that the amplifier 143 does not have strong nonlinearity, that is, the amplifier 143 has strong linearity. Because the BPF 142 also has linearity, the positions of the BPF 142 and the amplifier 143 can be swapped, as shown in the middle part of FIG. 6. That is, the configuration of the transmission path 120 in the upper part can be approximated by the transmission path 120' in the middle part. This allows the communication path 130 and the amplifier 143 to be collectively modeled.

[0091] The communication path 130 and the amplifier 143 are collectively modeled by the second NN 600. The second NN 600 outputs an approximation (estimate) of a signal generated through a portion of the transmission process of the second signal. Specifically, the second NN 600 outputs an approximation of a signal generated during the process of the second signal passing through the communication path 130 and the amplifier 143.

[0092] The BPF 142 is modeled by a digital filter 610. The digital filter 610 may be, for example, a finite impulse response (FIR) filter.

[0093] The modulation unit 144 is modeled by a digital downconverter 620. The digital downconverter 620 has the same functions / properties as the modulation unit 144. Therefore, the digital downconverter 620 downconverts the output of the digital filter 610 and outputs a digital I signal and a digital Q signal.

[0094] Next, we will explain the approximation error calculated by error calculation section 420. The approximation error represents the error between the approximation value of the third signal (the output of transmission model 410) and the actual third signal (the output of modulation section 144).

[0095] When the wireless communication device 100 is started up, information about the transmission distortion occurring in the transmission path 120 is not reflected in the second NN 600, so the transmission model 410 outputs the second signal as is. On the other hand, the actual third signal includes transmission distortion occurring in the transmission process of the second signal. As the magnitude of the transmission distortion (hereinafter referred to as the "distortion amount") increases, the approximation error also increases. If the approximation error is large, this means that information about the transmission distortion currently occurring in the transmission path 120 has not been reflected in the first NN 300 and the second NN 600, and therefore the transmission distortion has not been suppressed. On the other hand, even after the information about the transmission distortion occurring in the transmission path 120 has been reflected in the second NN 600, the distortion amount changes over time. Therefore, if the approximation error is large, this also means that the difference (error) between the distortion amount included in the output of the transmission model 410 and the distortion amount included in the actual third signal is large. Therefore, the approximation error is a value related to the distortion amount occurring in the transmission process of the second signal.

[0096] If the approximation error is small, this means that the output of the transmission model 410 is close to the actual third signal, which means that information about the transmission distortion currently occurring in the transmission path 120 can be reflected in the first NN 300 and the second NN 600, and therefore the transmission distortion can be suppressed.

[0097] The error calculation unit 420 calculates a first difference, which is the difference between the I signal output by the transmission model 410 and the I signal included in the third signal. Since the I signal is a real component, the first difference is a real component. Furthermore, the error calculation unit 420 calculates a second difference, which is the difference between the Q signal output by the transmission model 410 and the Q signal included in the third signal. Since the Q signal is an imaginary component, the second difference is an imaginary component.

[0098] The error calculation unit 420 calculates an approximation error based on the first difference and the second difference. For example, the approximation error may be a complex number represented by the first difference, which is a real component, and the second difference, which is an imaginary component. In another example, the approximation error may be the larger of the absolute value of the first difference and the absolute value of the second difference. The approximation error may be calculated by one of known calculation methods, as long as it represents the error between the output of the transmission model 410 and the actual third signal.

[0099] The parameter calculation unit 430 calculates a first parameter and a second parameter using the approximation error and the internal parameters of the transmission model 410. The internal parameters of the transmission model 410 may include a second parameter of the second NN 600 and an output value of at least one node in the second NN 600. In another example, the internal parameters of the transmission model 410 may further include a filter coefficient of the digital filter 610 and a sine wave and a cosine wave in the digital downconverter 620. The parameter calculation unit 430 calculates the first parameter and the second parameter so as to reduce the approximation error. For example, the parameter calculation unit 430 calculates the first parameter and the second parameter so as to reduce at least one of the absolute value of the first difference and the absolute value of the second difference.

[0100] <2-9. Processing flow of delta-sigma modulator> Next, the flow of processing by the delta-sigma modulation device 110 will be described with reference to Fig. 7. Fig. 7 is a flowchart showing an example of the flow of processing by the delta-sigma modulation device 110.

[0101] When the wireless communication device 100 is started up (or the wireless communication device 100 is reset), the delta-sigma modulation unit 111 performs delta-sigma modulation on a first signal as an external input signal, and outputs a second signal (701).

[0102] The NN learning unit 113 calculates the first parameter and the second parameter using the second signal and the third signal (702). The NN learning unit 113 transmits the calculated first parameter to the NN processing unit 112.

[0103] The NN processing unit 112 updates the first parameters of the first NN 300 to the first parameters calculated by the NN learning unit 113. Furthermore, the NN learning unit 113 updates the second parameters of the second NN 600 in the transmission model 410 to the calculated second parameters (703).

[0104] The NN processing unit 112 uses the second signal to pass through the first NN 300 and output a fourth signal (704).

[0105] The delta-sigma modulation unit 111 performs delta-sigma modulation on the first signal using the fourth signal, and outputs a second signal (705).

[0106] Thereafter, the processes of steps 702 to 705 are repeatedly executed.

[0107] The above configuration has the following advantages. The delta-sigma modulation device 110 receives, as a fed-back third signal, information about transmission distortion occurring in the transmission process (transmission path 120) of the second signal. The delta-sigma modulation device 110 calculates an approximation error using the second signal and the third signal, and calculates a first parameter and a second parameter using the approximation error. The delta-sigma modulation device 110 updates the first parameter of the first NN 300 to the calculated first parameter, and updates the second parameter of the second NN 600 to the calculated second parameter. This causes information about transmission distortion occurring in the transmission path 120 to be reflected in the first NN 300 and the second NN 600. The delta-sigma modulation device 110 performs delta-sigma modulation on the first signal using the fourth signal (the output of the first NN 300). This allows the delta-sigma modulation device 110 to perform noise shaping, passing the signal component of the target frequency f0 and shifting noise near the frequency f0 out of the band.

[0108] Delta-sigma modulation device 110 can suppress transmission distortion that occurs in the transmission process of the second signal by using the feedback processing described above. Therefore, delta-sigma modulation device 110 can suppress spectral leakage. Delta-sigma modulation device 110 can transmit high-quality signals.

[0109] <2-10. Specific configuration examples of wireless communication devices> Next, a specific configuration of a wireless communication device will be described with reference to Figures 8 to 10. Figure 8 shows an example of a specific configuration of a wireless communication device 800. Components that have already been described are assigned the same reference numerals, and detailed description thereof will be omitted.

[0110] The wireless communication device 800 is a node of a radio access network (RAN). In this example, the wireless communication device 800 is a base station. The wireless communication device 800 performs wireless communication with terminal devices 190 located within its coverage area.

[0111] The wireless communication device 800 includes multiple devices (or multiple nodes). Specifically, the wireless communication device 800 includes a first device 810 and a second device 820. The first device 810 is a node of a wireless access network and may be referred to as a "first wireless access network device." The second device 820 is a node of a wireless access network and may be referred to as a "second wireless access network device."

[0112] The first device 810 performs processing related to the first physical layer. In this example, the processing related to the first physical layer includes BPF processing and amplification processing.

[0113] The second device 820 performs processing related to a second physical layer that is higher than the first physical layer. In this example, the processing related to the second physical layer includes delta-sigma modulation. Note that the first and second physical layers may be considered to be included in the first layer (lowest layer) of multiple layers that make up the communication protocol implemented in the first device 810 and the second device 820.

[0114] The first device 810 and the second device 820 are connected via a first communication path 131. The second device 820 transmits a second signal to the first device 810 via the first communication path 131.

[0115] The first device 810 and the second device 820 are connected via a second communication path 132. The first device 810 transmits information related to the transmission between the first device 810 and the second device 820 to the second device 820 via the second communication path 132. Hereinafter, the information related to the transmission will be referred to as "transmission-related information."

[0116] In this example, the transmission-related information is information related to the transmission of a second signal transmitted from second device 820 to first device 810. Specifically, the transmission-related information is a third signal (i.e., the output of modulation unit 144) generated in the transmission process of the second signal. The third signal includes information related to transmission distortion that occurred in the transmission process of the second signal.

[0117] <2-11. Configuration of the first device> The first device 810 includes an antenna 141, a BPF 142, an amplifier 143, a modulator 144, and an information transmitter 145. The information transmitter 145 controls communication with the second device 820 via the second communication path 132. Specifically, the information transmitter 145 controls transmission of transmission-related information.

[0118] 9 is a diagram showing an example of the hardware configuration of the first device 810. The first device 810 includes a communication interface 910, a storage unit 920, and a processing unit 930.

[0119] The communication interface 910 is an interface for communicating with other devices. The communication interface 910 includes an antenna 141 for wireless communication. Furthermore, the communication interface 910 includes a connection terminal, a connection circuit, etc. for communicating with the second device 820 via the first communication path 131. The communication interface 910 also includes a connection terminal, a connection circuit, etc. for communicating with the second device 820 via the second communication path 132.

[0120] The storage unit 920 includes volatile memory and non-volatile memory. The volatile memory may include, for example, a random access memory (RAM). The non-volatile memory may include, for example, one or more of a read only memory (ROM), a hard disk drive (HDD), and a solid state drive (SSD). The non-volatile memory stores program code (instructions) for implementing one or more functions of the first device 810.

[0121] The processing unit 930 includes one or more processors. The one or more processors may include, for example, one or more of a central processing unit (CPU), a micro processing unit (MPU), and a microcontroller. The processing unit 930 executes program code (instructions) stored in the memory unit 920 to realize one or more functions of the first device 810.

[0122] Processing unit 930 may include one or more analog elements and / or analog circuits. Processing unit 930 may use one or more analog elements and / or analog circuits to implement one or more functions of first device 810. Note that in this specification, the expression "A and / or B" should be interpreted as "A or B" or "A and B."

[0123] <2-12. Configuration of the second device> The second device 820 includes a delta-sigma modulation unit 111 , an NN processing unit 112 , and an NN learning unit 113 .

[0124] 10 is a diagram showing an example of the hardware configuration of the second device 820. The second device 820 includes a communication interface 1010, a storage unit 1020, and a processing unit 1030.

[0125] The communication interface 1010 is an interface for communicating with other devices. The communication interface 1010 includes a connection terminal, a connection circuit, etc. for communicating with the first device 810 via the first communication path 131. The communication interface 1010 also includes a connection terminal, a connection circuit, etc. for communicating with the first device 810 via the second communication path 132.

[0126] The communication interface 1010 may include a connection terminal, a connection circuit, etc. for communicating with a node (not shown) of a core network. Furthermore, the communication interface 1010 may include an antenna for wireless communication with the first device 810.

[0127] The storage unit 1020 includes volatile memory and non-volatile memory. The volatile memory may include, for example, RAM. The non-volatile memory may include, for example, one or more of ROM, HDD, and SSD. The non-volatile memory stores program code (instructions) for implementing one or more functions of the second device 820.

[0128] The processing unit 1030 includes one or more processors. The one or more processors may include, for example, one or more of a CPU, an MPU, and a microcontroller. The processing unit 1030 executes program code (instructions) stored in the memory unit 1020 to realize one or more functions of the second device 820.

[0129] The processing unit 1030 may include one or more analog elements and / or analog circuits. The processing unit 1030 may implement one or more functions of the second device 820 using one or more analog elements and / or analog circuits.

[0130] <2-13. Processing flow of wireless communication device> Next, the flow of processing by wireless communication device 800 will be described with reference to Fig. 11. Fig. 11 is a sequence diagram showing an example of the flow of processing by wireless communication device 800.

[0131] The second device 820 executes the process of step 701 in Figure 7 (1101). Therefore, the second device 820 transmits a second signal to the first device 810.

[0132] The first device 810 receives the second signal, and then the first device 810 (in this example, the information sending unit 145) sends the transmission-related information (in this example, the third signal) to the second device 820 (1102).

[0133] The second device 820 receives the transmission-related information and executes the processes of steps 702 to 704 in FIG. 7 (1103).

[0134] The second device 820 executes (1104) the process of step 705 in Fig. 7. Therefore, the second device 820 transmits the second signal to the first device 810. Thereafter, the processes of steps 1102 to 1104 are repeatedly executed.

[0135] The information transmitting unit 145 of the first device 810 may transmit the transmission-related information to the second device 820 according to a predetermined rule. The information transmitting unit 145 may periodically (i.e., at predetermined time intervals) transmit the transmission-related information to the second device 820. The information transmitting unit 145 may also non-periodically transmit the transmission-related information to the second device 820.

[0136] The above configuration has the following advantages. The first device 810 can feed back transmission-related information to the second device 820. The second device 820 can receive the transmission-related information and perform appropriate processing for transmission distortion occurring in the transmission process of the second signal. For example, the second device 820 can update the first parameter of the first NN 300 using the transmission-related information. This allows the second device 820 to transmit to the first device 810 a second signal including a component that suppresses transmission distortion. Therefore, transmission between the two devices (the first device 810 and the second device 820) can be improved.

[0137] <2-14. Variations> The technology according to the present disclosure is not limited to the above-described embodiments.

[0138] (1) First Modification The delta-sigma modulation unit 111 may be a low-pass type device. Fig. 12 is a diagram showing an example of the configuration of the delta-sigma modulation unit 111. Hereinafter, the I signal included in the first signal will be referred to as the "first I signal." The Q signal included in the first signal will be referred to as the "first Q signal."

[0139] The delta-sigma modulation section 111 includes a downconverter 1210 , a first loop filter 1220 , a second loop filter 1230 , a first quantizer 1240 , a second quantizer 1250 , and an upconverter 1260 .

[0140] The downconverter 1210 includes a first multiplier 1211a and a second multiplier 1211b.

[0141] First multiplier 1211a multiplies the fourth signal by cosωt to generate an I component of the fourth signal, where ω=2×π×f0. Hereinafter, the I component generated from the fourth signal will be referred to as the "second I signal." First multiplier 1211a outputs the second I signal to first loop filter 1220.

[0142] Second multiplier 1211b multiplies the fourth signal by −sinωt to generate a Q component of the fourth signal. Hereinafter, the Q component generated from the fourth signal will be referred to as a “second Q signal.” Second multiplier 1211b outputs the second Q signal to second loop filter 1230.

[0143] Thus, the downconverter 1210 downconverts the fourth signal into a second I signal and a second Q signal.

[0144] If the signal sampling rate is set to 1 / 4 of the target frequency f0, i.e., f0 / 4, an example of a cosωt signal sequence will be [1, 0, -1, 0, 1, ...], and an example of a -sinωt signal sequence will be [0, -1, 0, 1, 0, ...]. The symbol " / " here represents division. This simplifies the multiplication process described above. This modification may include such processing.

[0145] The first loop filter 1220 performs processing to suppress transmission distortion included in the feedback component (i.e., the second I signal). Specifically, the first loop filter 1220 uses the first I signal and the second I signal to output a signal including a first component for suppressing at least a part of the transmission distortion occurring in the transmission process of the second signal (i.e., the transmission path 120).

[0146] The first loop filter 1220 includes a first adder 1221 a, a second adder 1221 b, and a transfer function processing unit 1222.

[0147] First adder 1221a adds the first I signal and the second I signal. First adder 1221a outputs the addition result to transfer function processing unit 1222. Here, the output of first adder 1221a is the difference between the first I signal and the second I signal. In other words, the output of first adder 1221a includes some of the transmission distortion components that occur in the transmission process of the second signal.

[0148] The transfer function processing unit 1222 applies a transfer function to the output of the first adder 1221a and outputs a first component for suppressing at least a part of the transmission distortion occurring in the transmission path 120. The transfer function is a function that determines the characteristics of the delta-sigma modulation in this example, and is determined based on a desired signal transfer function, a noise transfer function, etc.

[0149] The second adder 1221b adds the first I signal and the output of the transfer function processing unit 1222 together, and outputs the addition result to the first quantizer 1240.

[0150] The first quantizer 1240 is a 1-bit quantizer. The first quantizer 1240 quantizes the output of the second adder 1221b with 1 bit and outputs the first quantized signal to the upconverter 1260.

[0151] The second loop filter 1230 performs processing to suppress transmission distortion included in the feedback component (i.e., the second Q signal). Specifically, the second loop filter 1230 uses the first Q signal and the second Q signal to output a signal including a second component for suppressing at least a part of the transmission distortion occurring in the transmission process of the second signal (i.e., the transmission path 120).

[0152] The second loop filter 1230 includes a first adder 1231 a, a second adder 1231 b, and a transfer function processing unit 1232.

[0153] First adder 1231a adds the first Q signal and the second Q signal. First adder 1231a outputs the addition result to transfer function processing unit 1232. Here, the output of first adder 1231a is the difference between the first Q signal and the second Q signal. In other words, the output of first adder 1231a includes some of the transmission distortion components that occur in the transmission process of the second signal.

[0154] The transfer function processing unit 1232 applies a transfer function to the output of the first adder 1231a and outputs a second component for suppressing at least a part of the transmission distortion occurring in the transmission path 120. The transfer function is a function that determines the characteristics of the delta-sigma modulation in this example, and is determined based on a desired signal transfer function and a noise transfer function.

[0155] The second adder 1231 b adds the first Q signal and the output of the transfer function processing unit 1232 together, and outputs the addition result to the second quantizer 1250 .

[0156] The second quantizer 1250 is a 1-bit quantizer. The second quantizer 1250 quantizes the output of the second adder 1231b with 1 bit and outputs the second quantized signal to the upconverter 1260.

[0157] The upconverter 1260 is a two-input, one-output component and includes a first multiplier 1261a, a second multiplier 1261b, and an adder 1262.

[0158] The first multiplier 1261 a multiplies the first quantized signal by cosωt and outputs the multiplication result to the adder 1262 .

[0159] The second multiplier 1261b multiplies the second quantized signal by −sinωt and outputs the multiplication result to the adder 1262.

[0160] Adder 1262 adds the output of first multiplier 1261a and the output of second multiplier 1261b and outputs a second signal. In this way, upconverter 1260 upconverts the first quantized signal and the second quantized signal and outputs the second signal.

[0161] Note that if the signal sampling rate is set to ¼ of the desired frequency f0, i.e., f0 / 4, then the signal sequence for cosωt will be [1, 0, −1, 0, 1, …], and the signal sequence for −sinωt will be [0, −1, 0, 1, 0, …]. The symbol “ / ” here represents division. This simplifies the above multiplication process. Furthermore, since the outputs of the first quantizer 1240 and the second quantizer 1250 are binarized signal examples (i.e., either 1 or −1), the output of the adder 1262 is also a binarized signal example. The quantization of the signal is not impaired by the processing of the upconverter 1260. This modification may include such processing.

[0162] The above configuration has the following advantages: The fourth signal (the output of the first NN 300) is fed back to the first loop filter 1220 and the second loop filter 1230 via the downconverter 1210. This enables the delta-sigma modulation unit 111 to perform noise shaping, which passes the signal component of the target frequency f0 and shifts noise near the frequency f0 out of band.

[0163] Note that the second adder 1221b may be omitted from the first loop filter 1220. In this case, the output of the transfer function processing unit 1222 is input to the first quantizer 1240. The transfer function in the transfer function processing unit 1222 is set so that the output of the transfer function processing unit 1222 includes the first I signal and the above-described first component.

[0164] The second adder 1231b may be omitted from the second loop filter 1230. In this case, the output of the transfer function processing unit 1232 is input to the second quantizer 1250. The transfer function in the transfer function processing unit 1232 is set so that the output of the transfer function processing unit 1232 includes the first Q signal and the above-described second component.

[0165] (2) Second Modification The first NN 300 may be configured to output an approximation of a signal generated through a portion of the transmission path 120. As an example, the first NN 300 outputs an approximation of a signal generated through a process in which the second signal passes through the communication path 130. With this configuration, it is possible to suppress at least a portion of the transmission distortion occurring in the transmission path 120.

[0166] (3) Third Modification The first NN 300 is not limited to the above example. Fig. 13 is a diagram showing an example of the configuration of the first NN 300.

[0167] The input layer 310 may include nodes 310c and 310d to which the result of a logical operation of two nodes in the input layer 310 is input. For example, the logical operation may include a logical product (AND) and an exclusive OR (XOR), etc. The input layer 310 may also include nodes to which a logical operation between a first result of the logical operation of two nodes and a second result of the logical operation of two nodes is input. That is, the input layer 310 may include nodes to which the result of a logical operation of two or more nodes in the input layer 310 is input.

[0168] The above configuration has the following advantages: The first NN 300 can efficiently generate nonlinear distortion such as intersymbol interference that occurs when the second signal passes through the transmission path 120. The accuracy of the fourth signal output by the first NN 300 is improved.

[0169] 14 is a diagram illustrating an example of the configuration of the first NN 300. The node 310a may have an output line (connection line) through which the current value of the second signal is output to the output layer 330 without passing through the hidden layer 320.

[0170] The above configuration has the following advantages: The current value of the second signal does not pass through the hidden layer 320, so no weight is applied to the current value of the second signal; The first NN 300 can efficiently reflect the residual from the current input value in the fourth signal; The accuracy of the fourth signal output by the first NN 300 is improved.

[0171] The first NN 300 may be a combination of the above-described configurations. The input layer 310 includes at least a node to which a current value of the second signal is input. The input layer 310 may further include at least one of a node to which a past value of the second signal is input, a node to which a result of a logical operation of two or more nodes in the input layer 310 is input, and a node having an output line through which the current value of the second signal is output to the output layer 330 without passing through the hidden layer 320. Furthermore, the first NN 300 may include various known structures.

[0172] (4) Fourth Modification The processing flow of delta-sigma modulation device 110 is not limited to the above example. Fig. 15 is a flowchart showing an example of the processing flow of delta-sigma modulation device 110. The flowchart of Fig. 15 is a flowchart in which step 1501 is added to the flowchart of Fig. 7.

[0173] After step 701, the NN learning unit 113 determines whether a predetermined first condition is met (1501). If the first condition is met, the delta-sigma modulation device 110 executes the process of step 702 and the process of step 703. Therefore, the first parameter of the first NN 300 and the second parameter of the second NN 600 are updated.

[0174] If the first condition is not met, the delta-sigma modulation device 110 does not execute the process of step 702 and the process of step 703. That is, the first parameter of the first NN 300 and the second parameter of the second NN 600 are not updated.

[0175] The first condition is a condition that the approximation error is greater than a predetermined first magnitude. Specifically, if the approximation error is the greater of the absolute value of the first difference and the absolute value of the second difference, the first condition may be a condition that the approximation error is greater than a predetermined first threshold Th1. If the approximation error is expressed by a complex number, the first condition may be a condition related to the absolute value of the complex number (i.e., the distance from the origin on the complex plane). The first condition may be a condition related to the absolute value of the approximation error (the distance from the origin) being greater than a predetermined first distance threshold. In another example, the first condition may include a condition related to the first difference and a condition related to the second difference. For example, if the absolute value of the first difference exceeds a predetermined first real component threshold and / or the absolute value of the second difference exceeds a predetermined first imaginary component threshold, the NN learning unit 113 may determine that the first condition is met.

[0176] The state of the transmission path 120 changes from moment to moment. By using the first condition as described above, the delta-sigma modulation device 110 can update the first parameter of the first NN 300 and the second parameter of the second NN 600 in accordance with the current state of the transmission path 120.

[0177] As described above, if the approximation error is large, this means that the output of the transmission model 410 (the approximation value of the third signal) is far from the actual third signal. In other words, this means that the accuracy of the output of the transmission model 410 is low, and the accuracy of the fourth signal output by the first NN 300 is also low. Therefore, there is a strong need to reflect the current state of the transmission path 120 (especially information related to transmission distortion) in the first NN 300 and the second NN 600. Therefore, if the first condition is met, the NN learning unit 113 determines "Yes" in step 1501. Then, the delta-sigma modulation device 110 executes the processing of step 702 and the processing of step 703.

[0178] On the other hand, if the approximation error is small, this means that the output of the transmission model 410 is close to the actual third signal. Therefore, there is little need to update the first parameter of the first NN 300 and the second parameter of the second NN 600. Therefore, the NN learning unit 113 determines "No" in step 1501. The delta-sigma modulation device 110 then proceeds to step 704.

[0179] The above configuration has the following advantages: The delta-sigma modulation device 110 executes the processes of steps 702 and 703 only when there is a high need to reflect the current state of the transmission path 120 in the first NN 300 and the second NN 600. Because the delta-sigma modulation device 110 does not execute the processes of steps 702 and 703 when they are not necessary, the processing load on the delta-sigma modulation device 110 can be reduced.

[0180] After the first condition is met, the delta-sigma modulation device 110 may repeatedly execute the processes of steps 702 and 703 until a predetermined second condition is met. Specifically, the parameter calculation unit 430 continues the process of calculating the first parameter and the second parameter until the second condition is met. The parameter calculation unit 430 continues the process of updating the second parameter of the second NN 600 until the second condition is met. The NN processing unit 112 continues the process of updating the first parameter of the first NN 300 until the second condition is met.

[0181] The second condition is that the approximation error is smaller than a predetermined second magnitude. The second magnitude is smaller than the first magnitude. Specifically, when the approximation error is the larger of the absolute value of the first difference and the absolute value of the second difference, the second condition may be that the approximation error is smaller than a predetermined second threshold Th2. The second threshold Th2 is smaller than the first threshold Th1. When the approximation error is expressed by a complex number, the second condition may be a condition related to the absolute value of the complex number (i.e., the distance from the origin on the complex plane). The second condition may be a condition related to the absolute value of the approximation error (the distance from the origin) being smaller than a predetermined second distance threshold. The second distance threshold is smaller than the first distance threshold. In another example, the second condition may include a condition related to the first difference and a condition related to the second difference. For example, if the absolute value of the first difference is less than a predetermined second real component threshold and / or the absolute value of the second difference is less than a predetermined second imaginary component threshold, the NN learning unit 113 may determine that the second condition is met. For example, the second real component threshold is less than the first real component threshold. The second imaginary component threshold is less than the first imaginary component threshold.

[0182] (5) Fifth Modification The configuration of the NN learning unit 113 is not limited to the above example.

[0183] The NN learning unit 113 further includes a delay adjustment unit 1610 and a gain adjustment unit 1620.

[0184] The delay adjustment unit 1610 receives the second signal as an input signal and delays the second signal so that the timing at which the output of the transmission model 410 is input to the error calculation unit 420 is synchronized with the timing at which the third signal is input to the error calculation unit 420.

[0185] The gain adjustment unit 1620 receives the third signal as an input signal. The gain adjustment unit 1620 multiplies the third signal by a gain. Specifically, the gain adjustment unit 1620 regards the third signal as a complex number and multiplies each of the I signal and Q signal included in the third signal by a complex gain.

[0186] The delay value in the delay adjustment unit 1610 and the complex gain in the gain adjustment unit 1620 are set so that the approximation error output from the error calculation unit 420 is minimized.

[0187] (6) Sixth Modification The modulation section 144 may down-convert the output of the amplifier 143 using a superheterodyne system and output an IF signal.

[0188] (7) Seventh Modification The modulator 144 in the transmission path 120 may be omitted. In this configuration, the digital downconverter 620 in the transmission model 410 is omitted. The error calculator 420 calculates an approximation error between the output of the transmission model 410 and a third signal (in this example, the output of the amplifier 143).

[0189] (8) Eighth Modification The configuration of the transmission model 410 is not limited to the above example. The configuration of the transmission model 410 may be changed as appropriate according to the configuration of the delta-sigma modulation device 110 and the configuration of the transmission path 120. For example, a commonly used digital distortion compensation technique may be applied to the first signal (I signal and Q signal). The configuration of the transmission model 410 may be changed accordingly.

[0190] The second NN 600 may be a neural network that models the entire transmission path 120 .

[0191] If the modulation unit 144 has a configuration that downconverts the output of the amplifier 143 using a superheterodyne system, the transmission model 410 may include a corresponding component. The transmission model 410 may also include a component that outputs an IF signal from the output of the digital filter 610 using a superheterodyne system.

[0192] The amplifier 143 may be modeled separately rather than being modeled as the second NN 600. In another example, the modeling of the amplifier 143 may be omitted.

[0193] The second NN 600 may be a component that models a part of the transmission path 120. For example, the second NN 600 may be a neural network that outputs the same signal (fourth signal) as the first NN 300. That is, the second NN 600 may output an approximation of a signal generated when the second signal passes through the communication path 130, the BPF 142, and the amplifier 143.

[0194] The second NN 600 may have the same structure as the first NN 300. The second NN 600 includes an input layer, at least one hidden layer, and an output layer. The input layer includes at least a node to which a current value of the second signal is input. The input layer may further include at least one of a node to which a past value of the second signal is input, a node to which a result of a logical operation of two or more nodes in the input layer is input, and a node having an output line through which the current value of the second signal is output to the output layer without passing through a hidden layer.

[0195] <<3. Second Embodiment>> Next, a second embodiment and its modified examples will be described with reference to Figures 17 to 24. In the second embodiment, the same components as those in the first embodiment are denoted by the same reference numerals, and detailed description of these components will be omitted.

[0196] <3-1. General configuration of wireless communication device> 17 is a diagram showing the configuration of a wireless communication device 1700. The wireless communication device 1700 includes a first device 810 and a second device 820.

[0197] The first communication path 131 includes an optical fiber 131a. The first communication path 131 includes an E / O conversion unit 131b at an end of the optical fiber 131a on the side of the second device 820. The E / O conversion unit 131b converts an electrical signal into an optical signal. Furthermore, the first communication path 131 includes an O / E conversion unit 131c at an end of the optical fiber 131a on the side of the first device 810. The O / E conversion unit 131c converts the optical signal into an electrical signal.

[0198] The second communication path 132 includes an optical fiber 132a. The second communication path 132 includes an E / O conversion unit 132b at an end of the optical fiber 132a that faces the second device 820. Furthermore, the second communication path 132 includes an O / E conversion unit 132c at an end of the optical fiber 132a that faces the first device 810.

[0199] <3-2. Configuration of the first device> The first device 810 includes an antenna 141, a BPF 142, an amplifier 143, a modulator 144, and an information transmitter 145. The information transmitter 145 transmits transmission-related information to the second device 820 via the second communication path 132. The transmission-related information in this example is a third signal (output of the modulator 144), as in the first embodiment.

[0200] <3-3. Configuration of the second device> The second device 820 includes a delta-sigma modulation unit 111, an NN processing unit 112, an NN learning unit 113, and an instruction transmission unit 114. The delta-sigma modulation unit 111 transmits a second signal to the first device 810 via a first communication path 131. The instruction transmission unit 114 transmits instruction signals (an initial instruction signal, a first instruction signal, and a second instruction signal) described below to the first device 810 via the first communication path 131.

[0201] <3-4. Overview of processing flow of wireless communication device> Next, an outline of the processing flow of wireless communication device 1700 will be described. Specifically, the processing flow of first device 810 transmitting transmission-related information to second device 820 will be described.

[0202] The instruction transmitting unit 114 determines whether a condition regarding the amount of distortion occurring in the transmission process of the second signal is satisfied. For example, the condition is the first condition regarding the approximation error. If the first condition is satisfied, the instruction transmitting unit 114 transmits a first instruction signal to the first device 810.

[0203] In this example, the first instruction signal is a signal instructing the first device 810 to transmit transmission-related information to the second device 820. The first instruction signal includes information about the number of times Nk that the first device 810 will transmit the transmission-related information to the second device 820, where Nk is an integer equal to or greater than 1. The information transmitting unit 145 transmits the transmission-related information to the second device 820 in response to the first instruction signal.

[0204] The instruction transmitting unit 114 determines whether the second condition regarding the approximation error is satisfied. If the second condition is satisfied, the instruction transmitting unit 114 transmits a second instruction signal to the first device 810.

[0205] In this example, the second instruction signal is a signal that instructs the first device 810 to stop transmitting the transmission-related information to the second device 820. The information transmitting unit 145 stops transmitting the transmission-related information in response to the second instruction signal.

[0206] <3-5. Processing flow of wireless communication device> Next, the processing flow of wireless communication device 1700 will be described with reference to Fig. 18. Fig. 18 is a sequence diagram showing an example of the processing flow of wireless communication device 1700.

[0207] The instruction sending unit 114 sends (1801) an initial instruction signal to the first device 810 via the first communication path 131. The initial instruction signal is an instruction signal for causing the first device 810 to send transmission-related information to the second device 820 only once.

[0208] The delta-sigma modulation unit 111 transmits (1802) the second signal to the first device 810 via the first communication path 131. The information transmission unit 145 transmits (1803) the transmission-related information (in this example, the third signal) to the second device 820 via the second communication path 132.

[0209] The second device 820 receives the transmission-related information. The second device 820 (i.e., the NN processing unit 112 and the NN learning unit 113) executes the processes of steps 702 and 703 in FIG. 7 (1804). Hereinafter, the series of processes of steps 702 and 703 will be collectively referred to as the "learning process." The NN learning unit 113 transmits the approximation error calculated in the process of step 702 to the instruction sending unit 114.

[0210] The instruction transmitting unit 114 determines whether the first condition is met using the approximation error. In this example, the instruction transmitting unit 114 determines that the first condition is met (1805). Therefore, the instruction transmitting unit 114 transmits a first instruction signal to the first device 810 via the first communication path 131 (1806).

[0211] The delta-sigma modulator 111 transmits the second signal to the first device 810 via the first communication path 131 (1807).

[0212] In response to the first instruction signal, the information transmitting unit 145 transmits (1808) the transmission-related information to the second device 820 via the second communication path 132. This is the first transmission of the transmission-related information from the time when the first device 810 received the first instruction signal.

[0213] The second device 820 executes the learning process (1809). Every time the second device 820 executes the learning process, the instruction sending unit 114 determines whether the second condition is met.

[0214] The transmission of the second signal, the transmission-related information, and the learning process are repeatedly performed, thereby gradually reducing the approximation error.

[0215] The delta-sigma modulator 111 then transmits 1810 the second signal to the first device 810 via the first communication path 131 .

[0216] The information transmitting unit 145 transmits (1811) the transmission-related information to the second device 820 via the second communication path 132. This is the Nj-th transmission of the transmission-related information, based on the time when the first device 810 received the first instruction signal. <Nkである。

[0217] The second device 820 executes a learning process (1812). The instruction sending unit 114 determines that the second condition is met (1813). Therefore, the instruction sending unit 114 sends a second instruction signal to the first device 810 via the first communication path 131 (1814). In response to the second instruction signal, the information sending unit 145 stops sending the transmission-related information.

[0218] Thereafter, the instruction transmitting unit 114 transmits an initial instruction signal to the first device 810 via the first communication path 131 every time a predetermined waiting time Tw has elapsed (1815).

[0219] Note that when the information transmitting unit 145 transmits the transmission-related information Nk times after the first device 810 receives the first instruction signal, the information transmitting unit 145 stops transmitting the transmission-related information. When a predetermined waiting time Tw has elapsed since the first device 810 received the transmission-related information Nk times, the instruction transmitting unit 114 transmits an initial instruction signal to the first device 810.

[0220] In another example, the second instruction signal may further include information about the waiting time Tw. In this configuration, the information transmitting unit 145 transmits the transmission-related information to the second device 820 every time the waiting time Tw elapses. The instruction transmitting unit 114 determines whether the first condition is met every time the information transmitting unit 145 receives the transmission-related information. If the first condition is met, the instruction transmitting unit 114 transmits the first instruction signal to the first device 810.

[0221] The above configuration has the following advantages. If the first condition is met (i.e., if the approximation error is large), this means that the current state of the transmission path 120 (particularly, information about transmission distortion) cannot be reflected in the first NN 300 and the second NN 600, and therefore transmission distortion cannot be suppressed. In this situation, the first device 810 feeds back transmission-related information to the second device 820. The second device 820 performs a learning process using the transmission-related information to update the first parameter of the first NN 300 and the second parameter of the second NN 600. This makes it possible to suppress transmission distortion that occurs during the transmission process of the second signal. Furthermore, it is possible to realize a distributed MIMO (Multiple Input Multiple Output) system that increases the number of simultaneous connections and transmission capacity.

[0222] <3-6. Variations> The technology according to the present disclosure is not limited to the above-described embodiments.

[0223] (1) First Modification The first instruction signal is not limited to the above example. The first instruction signal may include information about a period Tp at which the first device 810 transmits the transmission-related information to the second device 820. In response to the first instruction signal, the information transmitting unit 145 transmits the transmission-related information to the second device 820 every time the period Tp elapses.

[0224] The period Tp may be k1 times the learning period, where k1 is an integer equal to or greater than 1. The learning period is the time required from the time the second device 820 receives the transmission-related information until the second device 820 updates the first parameter of the first NN 300 and the second parameter of the second NN 600.

[0225] The period Tp may be defined by a frame, subframe, or slot defined in 3GPP. A frame is defined by multiple subframes. A subframe is defined by one or more slots. The period Tp may be defined as "k2 times a frame," where k2 is an integer equal to or greater than 1. As an example, one frame is 10 milliseconds (ms). The period Tp may be defined as "k3 times a subframe," where k3 is an integer equal to or greater than 1. As an example, one subframe is 1 ms. The period Tp may be defined as "k4 times a slot," where k4 is an integer equal to or greater than 1.

[0226] FIG. 19 is a sequence diagram showing an example of the processing flow of wireless communication device 1700.

[0227] The instruction transmitting unit 114 transmits a first instruction signal to the first device 810 (1901). The delta-sigma modulating unit 111 transmits a second signal to the first device 810 (1902). The information transmitting unit 145 transmits transmission-related information (in this example, a third signal) to the second device 820 (1903). The second device 820 executes a learning process (1904). Thereafter, the delta-sigma modulating unit 111 repeatedly transmits the second signal to the first device 810 (1905).

[0228] A period Tp has elapsed since the information transmitting unit 145 executed the process of step 1903. The information transmitting unit 145 transmits the transmission-related information to the second device 820 (1906). The second device 820 executes a learning process (1907). Thereafter, the information transmitting unit 145 transmits the transmission-related information to the second device 820 every time the period Tp elapses (1908).

[0229] The above configuration has the following advantages. The first device 810 periodically feeds back transmission-related information to the second device 820. The second device 820 periodically performs a learning process using the transmission-related information to update the first parameter of the first NN 300 and the second parameter of the second NN 600. This makes it possible to suppress transmission distortion that occurs during the transmission process of the second signal.

[0230] (2) Second Modification The information transmitting unit 145 may transmit the transmission-related information to the second device 820 when a condition related to transmission between the first device 810 and the second device 820 is met. For example, the information transmitting unit 145 may estimate the amount of distortion that occurred in the transmission process of the second signal based on the second signal. The information transmitting unit 145 determines whether a third condition related to the amount of distortion is met. The third condition is that the amount of distortion is greater than a predetermined magnitude. When the third condition is met, the information transmitting unit 145 transmits the transmission-related information to the second device 820.

[0231] The amount of distortion may be estimated as follows. The information transmitting unit 145 performs Fourier transform processing or the like on the second signal to analyze the frequency components of the second signal. If the amount of distortion generated during the transmission of the second signal is large, the frequency components of the second signal outside the desired frequency band (the band from f0-fα to f0+fα) described above will be large. Taking this into consideration, the information transmitting unit 145 calculates the signal power of the band outside the desired frequency band.

[0232] The third condition may be that the signal power of the band outside the desired frequency band is greater than a third threshold value Th3. If the signal power of the band outside the desired frequency band is greater, the information transmitting unit 145 may determine that the amount of distortion is greater than a predetermined value.

[0233] The signal to be subjected to the Fourier transform process may be any of the signals 1701 to 1703 shown in FIG. A signal 1701 after passing through the first communication path 131 and before passing through the BPF 142 Signal 1702 after passing through BPF 142 and before passing through amplifier 143 Signal 1703 after passing through amplifier 143

[0234] FIG. 20 is a sequence diagram showing an example of the processing flow of wireless communication device 1700.

[0235] The delta-sigma modulation unit 111 transmits the second signal to the first device 810 (2001). The information transmission unit 145 estimates the amount of distortion as described above and determines whether the third condition is met (2002). In this example, the information transmission unit 135 determines that the third condition is met (2003). Therefore, the information transmission unit 145 transmits transmission-related information to the second device 820 (2004). The second device 820 executes a learning process (2005). Thereafter, the delta-sigma modulation unit 111 repeatedly transmits the second signal to the first device 810 (2006).

[0236] Thereafter, a first time Ta elapses from the time point when the information transmitting unit 145 executes the process of step 2002. The information transmitting unit 145 estimates the amount of distortion as described above, and determines whether or not the third condition is met (2007). In this manner, the information transmitting unit 145 estimates the amount of distortion every time the first time Ta elapses. If the third condition is met, the information transmitting unit 145 transmits the transmission-related information to the second device 820. On the other hand, if the third condition is not met, the information transmitting unit 145 does not transmit the transmission-related information to the second device 820.

[0237] The above configuration has the following advantages: When the third condition is met (i.e., when the amount of distortion is relatively large), the first device 810 feeds back the transmission-related information to the second device 820. The first device 810 can feed back the transmission-related information to the second device 820 in an appropriate situation without a first instruction signal from the second device 820.

[0238] In another example, the information transmitting unit 145 may estimate the amount of distortion every time the first device 810 receives the second signal.

[0239] (3) Third Modification The process shown in Fig. 18 may be combined with the process shown in Fig. 19. Fig. 21 is a sequence diagram showing an example of the flow of processing by wireless communication device 1700.

[0240] Steps 2101 to 2107 in Fig. 21 are the same as steps 1901 to 1907 in Fig. 19. Therefore, detailed descriptions of these steps will be omitted. The information transmitting unit 145 transmits transmission-related information to the second device 820 every time the period Tp elapses in response to a first instruction signal (2101) including information about the period Tp.

[0241] Every time the second device 820 executes the learning process, the instruction sending unit 114 determines whether the first condition is met.

[0242] After the second device 820 performs the learning process in step 2107, the instruction sending unit 114 determines that the first condition is met (2108). The instruction sending unit 114 sends a first instruction signal including information about the number of times Nk to the first device 810 (2109).

[0243] Steps 2110 to 2117 in Fig. 21 are the same as steps 1807 to 1814 in Fig. 18. Therefore, detailed descriptions of these steps will be omitted. Note that in steps 2110 to 2117, the time interval at which the first device 810 transmits transmission-related information to the second device 820 is shorter than the period Tp.

[0244] After the instruction sending unit 114 sends the second instruction signal in step 2117, the instruction sending unit 114 sends a first instruction signal including information about the period Tp to the first device 810. In response, the information sending unit 145 sends transmission-related information to the second device 820 every time the period Tp elapses.

[0245] The above configuration has the following advantages. When the first condition is met (i.e., when the approximation error is large), the first device 810 feeds back transmission-related information to the second device 820 at a relatively short period. The second device 820 can quickly reflect the current state of the transmission path 120 in the first NN 300 and the second NN 600. On the other hand, when the second condition is met (i.e., when the approximation error is small), there is little need to update the first parameter of the first NN 300 and the second parameter of the second NN 600. In this situation, the first device 810 feeds back transmission-related information to the second device 820 at a relatively long period Tp. The processing load on the second device 820 can be reduced.

[0246] (4) Fourth Modification The process shown in Fig. 18 may be combined with the process shown in Fig. 20. Fig. 22 is a sequence diagram showing an example of the flow of processing by wireless communication device 1700.

[0247] Steps 2201 to 2205 in Fig. 22 are the same processing as steps 2001 to 2005 in Fig. 20. Therefore, detailed explanation of these steps will be omitted.

[0248] Every time the second device 820 executes the learning process, the instruction sending unit 114 determines whether the first condition is met.

[0249] After the second device 820 performs the learning process in step 2205, the instruction sending unit 114 determines that the first condition is met (2206). The instruction sending unit 114 sends a first instruction signal including information about the number of times Nk to the first device 810 (2207).

[0250] Steps 2208 to 2215 in Figure 22 are the same processing as steps 1807 to 1814 in Figure 18. Therefore, detailed descriptions of these steps will be omitted. In steps 2208 to 2215, every time the second device 820 transmits a second signal to the first device 810, the first device 810 transmits transmission-related information to the second device 820. Therefore, the time interval at which the first device 810 transmits the transmission-related information to the second device 820 is relatively short.

[0251] After instruction sending unit 114 sends the second instruction signal in step 2215, delta-sigma modulation unit 111 sends (2216) the second signal to first device 810. Information sending unit 145 estimates the amount of distortion as described above and determines whether the third condition is met (2217).

[0252] Each time the first device 810 receives the second signal, the information transmitting unit 145 estimates the amount of distortion and determines whether the third condition is met. If the third condition is met, the information transmitting unit 145 transmits the transmission-related information to the second device 820. In this case, as described above, the instruction transmitting unit 114 determines whether the first condition is met.

[0253] On the other hand, if the third condition is not met, the information transmitting unit 145 does not transmit the transmission-related information to the second device 820.

[0254] The above configuration has the following advantages. In the first stage, the first device 810 feeds back the transmission-related information to the second device 820 only when the third condition is met (i.e., when the amount of distortion is relatively large). The second device 820 executes the learning process and determines whether the first condition is met only when the first device 810 has fed back the transmission-related information. This reduces the processing load on the second device 820.

[0255] Next, in the second stage, when the first condition is met (i.e., when the approximation error is large), the second device 820 instructs the first device 810 to feed back transmission-related information to the second device 820 at relatively short intervals. In the second stage, the second device 820 can reflect the current state of the transmission path 120 (particularly, information related to transmission distortion) to the first NN 300 and the second NN 600 at relatively short intervals.

[0256] (5) Fifth Modification The first device 810 and the second device 820 may be connected by a communication path (e.g., an optical fiber). The second signal may be transmitted via the optical fiber, and the transmission-related information may be transmitted via the same optical fiber.

[0257] For example, first device 810 and second device 820 may be configured to communicate with each other using wavelength division multiplexing, a well-known technique for transmitting optical signals of different wavelengths over a single optical fiber.

[0258] The first device 810 and the second device 820 may be configured to communicate with each other using polarization multiplexing, a well-known technique in which horizontally polarized waves and vertically polarized waves are transmitted over a single optical fiber.

[0259] The first device 810 and the second device 820 may be configured to communicate with each other using time-division multiplexing, a well-known technique in which multiple signals are transmitted sequentially over a single optical fiber so that the signals do not overlap in time.

[0260] The optical fiber may be a multi-core fiber having multiple cores. In this case, the first device 810 and the second device 820 may be configured to communicate with each other using a spatial multiplexing technique. Spatial multiplexing is a well-known technique in which different signals are transmitted for each core.

[0261] (6) Sixth Modification 23 is a diagram showing the configuration of a wireless communication device 2300. Wireless communication device 2300 includes a first device 810 and a second device 820. Wireless communication device 2300 has a configuration for outputting N types of signals, where N is an integer greater than or equal to 2.

[0262] Specifically, first device 810 includes a plurality of antennas 141-1 to 141-N, a plurality of BPFs 142-1 to 142-N, and a plurality of amplifiers 143-1 to 143-N.

[0263] The second device 820 includes a plurality of delta-sigma modulation sections 111-1 to 111-N. Furthermore, the first device 810 and the second device 820 are connected via a plurality of first communication paths 131-1 to 131-N. The plurality of delta-sigma modulation sections 111-1 to 111-N are connected to a plurality of antennas 141-1 to 141-N via a plurality of first communication paths 131-1 to 131-N, respectively. For example, the delta-sigma modulation section 111-N is connected to the antenna 141-N via the first communication path 131-N, a BPF 142-N, and an amplifier 143-N.

[0264] First device 810 includes switch 146. Switch 146 can switch the signal to be output to modulation section 144. With this configuration, first device 810 can feed back transmission-related information (in this example, the third signal) for each of the N types of second signals to second device 820. This makes it possible to suppress transmission distortion that occurs in the transmission process of each of the N types of second signals.

[0265] (7) Seventh Modification 24 is a diagram showing the configuration of a wireless communication device 2400. The wireless communication device 2400 includes a first device 810 and a second device 820. The wireless communication device 2400 includes a transmitting function and a receiving function.

[0266] Hereinafter, the link through which a signal is transmitted from the second device 820 via the first device 810 to the terminal device 190 may be referred to as the "downlink." A signal transmitted on the downlink may be referred to as a "downlink signal."

[0267] On the other hand, the link through which a signal is transmitted from the terminal device 190 to the second device 820 via the first device 810 may be referred to as the "uplink." A signal transmitted on the uplink may be referred to as an "uplink signal."

[0268] The second device 820 transmits a downlink signal to the first device 810 via the first communication path 131. The first device 810 transmits an uplink signal to the second device 820 via the second communication path 132.

[0269] The first device 810 includes an antenna switch 147. The antenna switch 147 includes a first terminal 147a connected to the amplifier 143, a second terminal 147b connected to the antenna 141, and a third terminal 147c connected to a low noise amplifier (LNA) 148.

[0270] The antenna switch 147 can operate in three modes: a transmit mode, a receive mode, and a feedback mode.

[0271] In the transmission mode, the first terminal 147a and the second terminal 147b are connected to each other, and the downlink signal is output from the antenna 141.

[0272] In the reception mode, the second terminal 147b and the third terminal 147c are connected. Therefore, the uplink signal is transmitted to the second device 820 via the LNA 148, the modulation unit 144, the information transmission unit 145, and the second communication path 132. The second device 820 performs predetermined processing on the uplink signal using a reception function (not shown).

[0273] In the feedback mode, the first terminal 147a and the third terminal 147c are connected. Therefore, the downlink signal is sent to the information transmitting unit 145 via the LNA 148 and the modulating unit 144. The information transmitting unit 145 transmits transmission-related information to the second device 820 via the second communication path 132 at a predetermined timing. The transmission-related information in this example is information related to the downlink signal to be transmitted from the first device 810 to the terminal device 190. Specifically, the transmission-related information is a signal (i.e., a third signal) obtained by frequency-converting (down-converting) the downlink signal through the modulating unit 144.

[0274] In such a configuration, the wireless communication device 2400 may transmit downlink signals and receive uplink signals using a Time Division Duplex (TDD) scheme. TDD is a scheme in which uplink signals and downlink signals are transmitted alternately in different slots. Note that these slots may also be referred to as "time slots."

[0275] When the wireless communication device 2400 operates in the TDD system, the instruction transmitter 114 operates as follows. In the case of an uplink slot, the second communication path 132 is used for transmitting an uplink signal, but the first communication path 131 is not used. Therefore, the instruction transmitter 114 transmits a first instruction signal to the first device 810 via the first communication path 131 in the uplink slot. The instruction transmitter 114 transmits a second instruction signal to the first device 810 via the first communication path 131 in the uplink slot.

[0276] Furthermore, the information transmitting unit 145 operates as follows: In the case of a downlink slot, the first communication path 131 is used for transmitting a downlink signal, but the second communication path 132 is not used. Therefore, the information transmitting unit 145 transmits transmission-related information to the second device 820 via the second communication path 132 in the downlink slot.

[0277] <<4. Third Embodiment>> Next, a third embodiment will be described with reference to Fig. 25. In the third embodiment, the same components as those in the first and second embodiments are denoted by the same reference numerals, and detailed description of these components will be omitted.

[0278] <4-1. General configuration of wireless communication device> 25 is a diagram showing the configuration of a wireless communication device 2500. The wireless communication device 2500 includes a first device 810 and a second device 820. The first device 810 and the second device 820 are connected via a first communication path 131.

[0279] <4-2. Configuration of the first device> The first device 810 includes an antenna 141, a BPF 142, an amplifier 143, an antenna 150, a frequency conversion unit 151, a second modulation unit 152, and a third modulation unit 153. The third modulation unit 153 has the same configuration as the modulation unit 144. The third modulation unit 153 processes the analog signal (corresponding to the downlink signal) amplified by the amplifier 143 as described above, and outputs a third signal (including an I signal and a Q signal).

[0280] The second modulation unit 152 receives the output of the third modulation unit 153. The second modulation unit 152 performs signal modulation processing on the output of the third modulation unit 153. A common modulation method may be used as the signal modulation processing. For example, single carrier modulation or OFDM (Orthogonal Frequency Division Multiplexing) may be used as the signal modulation processing. The frequency conversion unit 151 receives the output of the second modulation unit 152. The frequency conversion unit 151 upconverts the output of the second modulation unit 152. In this case, the frequency conversion unit 151 upconverts the output of the second modulation unit 152 to a frequency f1 different from the frequency f0. The output of the frequency conversion unit 151 is transmitted via the antenna 150.

[0281] <4-3. Configuration of the second device> The second device 820 includes a delta-sigma modulation unit 111, an NN processing unit 112, an NN learning unit 113, an antenna 115, a demodulation unit 117, and a modulation unit 144. The second device 820 receives a signal output from an antenna 150 of the first device 810 via the antenna 115. The modulation unit 144 processes the received signal as described above. The demodulation unit 117 demodulates the output of the modulation unit 144. The demodulation method of the demodulation unit 117 corresponds to the modulation method of the second modulation unit 152. The demodulation unit 117 outputs the demodulated third signal (including an I signal and a Q signal) to the NN learning unit 113.

[0282] The above configuration has the following advantages. The first device 810 can feed back transmission-related information to the second device 820 via spatial transmission. The transmission-related information in this example is information related to a downlink signal to be transmitted from the first device 810 to the terminal device 190. Therefore, the second device 820 can update the first parameter of the first NN 300 and the second parameter of the second NN 600 using the transmission-related information. This makes it possible to suppress transmission distortion that occurs in the transmission process of the second signal.

[0283] <<5. Fourth Embodiment>> Next, a fourth embodiment and its modified examples will be described with reference to Figures 26 to 32. In the fourth embodiment, the same components as those in the first to third embodiments are denoted by the same reference numerals, and detailed description of these components will be omitted.

[0284] <5-1. General configuration of wireless communication device> 26 is a diagram showing the configuration of a wireless communication device 2600. The wireless communication device 2600 includes a first device 810 and a second device 820. The first device 810 and the second device 820 are connected via a first communication path 131. Furthermore, the first device 810 and the second device 820 are connected via a second communication path 132.

[0285] <5-2. Configuration of the first device> The first device 810 includes an antenna 141 , a BPF 142 , an amplifier 143 , a modulation unit 144 , an information transmission unit 145 , a digital restoration unit 149 , and an NN learning unit 113 .

[0286] The digital restoration unit 149 receives the output of the first communication path 131. The digital restoration unit 149 converts the output of the first communication path 131 into a 1-bit digital signal. At this time, the transmission distortion contained in the output of the first communication path 131 is shaped and the digital signal is output. This restores the original second signal output by the delta-sigma modulation unit 111. In this way, the digital restoration unit 149 generates a restored signal of the second signal from the third signal (in this example, the signal generated by the second signal passing through the first communication path 131). Hereinafter, such a restored signal of the second signal may be referred to as a "fifth signal."

[0287] The NN learning unit 113 calculates a first parameter of the first NN 300 and a second parameter of the second NN 600 using the fifth signal (the restored second signal) and the third signal output by the modulation unit 144 (i.e., the third signal generated when the second signal passes through the first communication path 131, the BPF 142, the amplifier 143, and the modulation unit 144). The NN learning unit 113 updates the second parameter of the second NN 600 to the calculated second parameter. Furthermore, the NN learning unit 113 outputs the calculated first parameter to the information transmission unit 145. The information transmission unit 145 transmits the first parameter to the second device 820 via the second communication path 132. Therefore, the transmission-related information in this example is the first parameter. The first parameter is information used to output an approximation of a signal generated in the transmission process of the second signal in the first NN 300. Therefore, the first parameter is information related to the transmission between the first device 810 and the second device 820. In other words, since the first parameter is also used to output an approximation of transmission distortion occurring in the transmission process of the second signal, the first parameter can also be said to be information related to transmission distortion.

[0288] <5-3. Configuration of the second device> The second device 820 includes a delta-sigma modulation unit 111, an NN processing unit 112, and an instruction transmission unit 114. The delta-sigma modulation unit 111 transmits a second signal to the first device 810 via a first communication path 131. The instruction transmission unit 114 transmits a first instruction signal to the first device 810 via the first communication path 131. The instruction transmission unit 114 transmits the second instruction signal to the first device 810 via the first communication path 131. The NN processing unit 112 receives a first parameter via the second communication path 132. The NN processing unit 112 updates the first parameter of the first NN 300 to the received first parameter.

[0289] <5-4. Processing flow of wireless communication device> Next, the flow of processing by wireless communication device 2600 will be described with reference to Fig. 27. Fig. 27 is a sequence diagram showing an example of the flow of processing by wireless communication device 2600.

[0290] When the wireless communication device 2600 is started up (or the wireless communication device 2600 is reset), the instruction sending unit 114 sends 2701 a first instruction signal to the first device 810. In this example, the first instruction signal includes information about the number of times Nk.

[0291] The delta-sigma modulation unit 111 transmits the second signal to the first device 810 (2702). The NN learning unit 113 calculates a first parameter and a second parameter (2703). The NN learning unit 113 updates the second parameter of the second NN 600 to the calculated second parameter.

[0292] The information transmitting unit 145 transmits the transmission-related information (in this example, the first parameter) to the second device 820 (2704). This is the first transmission of the transmission-related information, based on the time when the first device 810 received the first instruction signal. The second device 820 receives the transmission-related information. The NN processing unit 112 updates the first parameter of the first NN 300 to the received first parameter.

[0293] Each time the NN learning unit 113 calculates the first parameter and the second parameter, the information transmitting unit 145 determines whether the second condition is met.

[0294] Thereafter, the transmission of the second signal, the calculation of the first and second parameters, and the transmission of the transmission-related information are repeatedly performed, thereby gradually reducing the approximation error.

[0295] The delta-sigma modulation unit 111 transmits (2705) the second signal to the first device 810. The NN learning unit 113 calculates (2706) the first parameter and the second parameter.

[0296] The information transmitting unit 145 transmits the transmission-related information to the second device 820 (2707). This is the Nj-th transmission of the transmission-related information, based on the time when the first device 810 received the first instruction signal. <Nkである。

[0297] At this point, the information transmitting unit 145 determines that the second condition is met (2708). The information transmitting unit 145 transmits a notification to the second device 820 (2709). This notification is for informing the second device 820 that the second condition is met. In response to this notification, the instruction transmitting unit 114 transmits a second instruction signal to the first device 810 (2710). This causes the information transmitting unit 145 to stop transmitting the transmission-related information.

[0298] After that, when the waiting time Tw has elapsed, the instruction sending unit 114 sends a first instruction signal including information about the number of times Nk to the first device 810 (2711).

[0299] The above configuration has the following advantages. The first device 810 repeatedly feeds back transmission-related information (first parameters) to the second device 820 until the second condition is met. The second device 820 can update the first parameters of the first NN 300 using the transmission-related information. This makes it possible to suppress transmission distortion that occurs in the transmission process of the second signal. Furthermore, it is possible to realize a distributed MIMO system that increases the number of simultaneous connections and transmission capacity.

[0300] <5-5. Variations> The technology according to the present disclosure is not limited to the above-described embodiments.

[0301] (1) First Modification FIG. 28 is a sequence diagram showing an example of the processing flow of wireless communication device 2600.

[0302] When the wireless communication device 2600 is powered up (or reset), the instruction transmitter 114 transmits a first instruction signal to the first device 810 (2801). The first signal includes information about the period Tp. The period Tp may be defined by the learning period, as described above. Furthermore, the period Tp may be defined by a frame, a subframe, or a slot, as described above.

[0303] The delta-sigma modulation unit 111 transmits the second signal to the first device 810 (2802). The NN learning unit 113 calculates the first parameter and the second parameter (2803). The NN learning unit 113 updates the second parameter of the second NN 600 to the calculated second parameter.

[0304] The information transmitting unit 145 transmits (2804) the transmission-related information (in this example, the first parameter) to the second device 820. The second device 820 receives the transmission-related information. The NN processing unit 112 updates the first parameter of the first NN 300 to the received first parameter.

[0305] The delta-sigma modulator 111 then repeatedly transmits the second signal to the first device 810 (2805).

[0306] A period Tp has elapsed since the NN learning unit 113 executed the process of step 2803. The NN learning unit 113 calculates the first parameter and the second parameter (2806). The information transmitting unit 145 transmits the transmission-related information to the second device 820 (2807).

[0307] The above configuration has the following advantages: The first device 810 periodically feeds back transmission-related information (first parameters) to the second device 820. The second device 820 can periodically update the first parameters of the first NN 300 using the transmission-related information.

[0308] The NN learning unit 113 may calculate the first parameter and the second parameter each time the first device 810 receives the second signal. The information transmitting unit 145 may transmit the transmission-related information to the second device 820 each time a period Tp elapses.

[0309] (2) Second Modification FIG. 29 is a sequence diagram showing an example of the processing flow of wireless communication device 2600.

[0310] When the wireless communication device 2600 is started up (or the wireless communication device 2600 is reset), the delta-sigma modulation unit 111 transmits a second signal to the first device 810 (2901). The NN learning unit 113 calculates a first parameter and a second parameter (2902). The NN learning unit 113 outputs an approximation error to the information transmission unit 145. The information transmission unit 145 uses the approximation error to determine whether or not a first condition is met. In this example, the information transmission unit 145 determines that the first condition is met (2903). In this case, the information transmission unit 145 transmits transmission-related information (in this example, the first parameter) to the second device 820 (2904).

[0311] Thereafter, the delta-sigma modulation unit 111 repeatedly transmits the second signal to the first device 810 (2905). The NN learning unit 113 calculates the first parameter and the second parameter (2906). Then, every time the NN learning unit 113 calculates the first parameter and the second parameter, the information transmission unit 145 determines whether or not the first condition is met. If the first condition is met, the information transmission unit 145 transmits the transmission-related information to the second device 820. On the other hand, if the first condition is not met, the information transmission unit 145 does not transmit the transmission-related information to the second device 820.

[0312] The above configuration has the following advantages: When the first condition is met (that is, when the approximation error is large), the first device 810 can transmit the transmission-related information (first parameter) to the second device 820.

[0313] (3) Third Modification The process shown in Fig. 27 may be combined with the process shown in Fig. 28. Fig. 30 is a sequence diagram showing an example of the flow of processing by wireless communication device 2600.

[0314] Steps 3001 to 3009 in Fig. 30 are the same as steps 2701 to 2709 in Fig. 27. Therefore, detailed descriptions of these steps will be omitted. Note that in steps 3001 to 3009, the time interval at which the first device 810 transmits transmission-related information to the second device 820 is shorter than the period Tp.

[0315] When the instruction transmitting unit 114 receives the notification in step 3009, the instruction transmitting unit 114 transmits (3010) a first instruction signal including the period Tp to the first device 810. Steps 3011 to 3016 in Fig. 30 are the same process as steps 2802 to 2807 in Fig. 28. Therefore, detailed description of these steps will be omitted.

[0316] The above configuration has the following advantages. First, the first device 810 transmits transmission-related information (first parameters) to the second device 820 at a relatively short period. The second device 820 can quickly reflect the state of the transmission path 120 in the first NN 300. On the other hand, when the second condition is met (i.e., when the approximation error is small), there is little need to update the first parameters of the first NN 300. In such a situation, the first device 810 feeds back the transmission-related information to the second device 820 at a relatively long period Tp.

[0317] 29 may be combined with the sequence of FIG. 30. For example, the information transmitting unit 145 may determine whether or not a first condition is met during the period Tp (i.e., between step 3012 and step 3015). If the first condition is met, the information transmitting unit 145 transmits the transmission-related information to the second device 820. On the other hand, if the first condition is not met, the information transmitting unit 145 does not transmit the transmission-related information to the second device 820. According to this configuration, even during the period Tp, if the approximation error becomes large, the first device 810 can feed back the transmission-related information to the second device 820.

[0318] (4) Fourth Modification The wireless communication device 2600 may have the configuration of the seventh modified example of the second embodiment. That is, the wireless communication device 2600 may operate in the TDD system. In this configuration, the instruction transmitter 114 operates as follows. The instruction transmitter 114 transmits a first instruction signal to the first device 810 via the first communication path 131 in an uplink slot. The instruction transmitter 114 transmits a second instruction signal to the first device 810 via the first communication path 131 in an uplink slot.

[0319] Furthermore, the information transmitting unit 145 operates as follows: The information transmitting unit 145 transmits transmission-related information to the second device 820 via the second communication path 132 in a downlink slot.

[0320] (5) Fifth Modification 31 is a diagram showing the configuration of a wireless communication device 2601. The components of the wireless communication device 2601 are the same as the components of the wireless communication device 2600 in FIG.

[0321] The output of the first communication path 131 is directly input to the digital restoration unit 149. The fifth signal (the restored second signal) output by the digital restoration unit 149 is input to the BPF 142. The fifth signal passes through the BPF 142, the amplifier 143, and the modulation unit 144, and is input to the NN learning unit 113. Therefore, the NN learning unit 113 calculates the first parameter and the second parameter using the "fifth signal" and the "third signal generated after the fifth signal has passed through the BPF 142, the amplifier 143, and the modulation unit 144."

[0322] The first device 810 and the second device 820 of this example may operate according to any of the sequences shown in FIGS.

[0323] (6) Sixth Modification 32 is a diagram showing the configuration of a wireless communication device 3100. The wireless communication device 3100 includes a first device 810 and a second device 820. The first device 810 and the second device 820 are connected via a first communication path 131.

[0324] The first device 810 includes an antenna 141, a BPF 142, an amplifier 143, a modulation unit 144, an information transmission unit 145, a digital restoration unit 149, an NN learning unit 113, an antenna 150, a frequency conversion unit 151, and a second modulation unit 152.

[0325] The second device 820 includes a delta-sigma modulation unit 111 , an NN processing unit 112 , an instruction transmission unit 114 , an antenna 115 , a modulation unit 116 , and a demodulation unit 117 .

[0326] In this example, the first device 810 transmits transmission-related information (first parameters) to the second device 820 via the antenna 150. The second device 820 receives the transmission-related information via the antenna 115. The operation of the first device 810 and the operation of the second device 820 will be described below.

[0327] The second modulation unit 152 receives the output (first parameter) of the NN learning unit 113. The second modulation unit 152 performs signal modulation processing on the output of the NN learning unit 113. A common modulation method may be used as the signal modulation processing. For example, single carrier modulation, OFDM, or the like may be used as the signal modulation processing. The frequency conversion unit 151 receives the output of the second modulation unit 152. The frequency conversion unit 151 upconverts the output of the second modulation unit 152. In this case, the frequency conversion unit 151 upconverts the output of the second modulation unit 152 to a frequency f1 different from the frequency f0. The information transmission unit 145 receives the output of the frequency conversion unit 151. The information transmission unit 145 transmits transmission-related information via the antenna 150. The transmission-related information in this example is the output of the frequency conversion unit 151, which is the first parameter.

[0328] The second device 820 receives the signal output from the antenna 150 via the antenna 115. The modulation unit 116 down-converts the received signal and outputs an I signal and a Q signal. The demodulation unit 117 demodulates the I signal and the Q signal back to their original digital values ​​(first parameters). The demodulation method of the demodulation unit 117 corresponds to the modulation method of the second modulation unit 152.

[0329] The demodulation unit 117 outputs the first parameters to the NN processing unit 112. The NN processing unit 112 receives the first parameters. The NN processing unit 112 updates the first parameters of the first NN 300 to the received first parameters.

[0330] The first device 810 and the second device 820 may operate according to any of the sequences shown in FIGS.

[0331] <<6. Fifth Embodiment>> Next, a fifth embodiment will be described with reference to Figures 33 to 34. The first to fourth embodiments described above are specific embodiments, but the fifth embodiment is a more generalized embodiment.

[0332] <6-1. Configuration of Delta-Sigma Modulator> 33 is a diagram showing the configuration of a delta-sigma modulation device 3200. The delta-sigma modulation device 3200 includes a delta-sigma modulation unit 3201, a neural network processing unit 3202, and a neural network learning unit 3203. Hereinafter, the neural network processing unit 3202 will be referred to as the "NN processing unit 3202," and the neural network learning unit 3203 will be referred to as the "NN learning unit 3203."

[0333] The components 3201 to 3203 of the delta-sigma modulation device 3200 may be implemented by one or more processors and a memory. The one or more processors may include, for example, one or more of a CPU, an MPU, and a microcontroller. The memory may include a volatile memory and a non-volatile memory. The memory may store program code (instructions). The one or more processors may implement the functions of the delta-sigma modulation device 3200 by executing the program code stored in the memory.

[0334] The delta-sigma modulation unit 3201 may operate in the same manner as the above-described delta-sigma modulation unit 111. As an example, the delta-sigma modulation unit 3201 performs delta-sigma modulation.

[0335] The NN processing unit 3202 may operate in the same manner as the above-described NN processing unit 112. As an example, the NN processing unit 3202 includes a first neural network 3202a that operates according to predetermined first parameters. Hereinafter, the first neural network 3202a will be referred to as the "first NN 3202a." The first NN 3202a may have the same configuration as the above-described first NN 300. The first parameters may include weights and biases.

[0336] The NN learning unit 3203 may operate in the same manner as the above-described NN learning unit 113. As an example, the NN learning unit 3203 calculates a first parameter of the first NN 3202a.

[0337] <6-2. Processing flow of delta-sigma modulator> FIG. 34 is a flowchart illustrating an example of the processing flow of delta-sigma modulation device 3200.

[0338] The delta-sigma modulation unit 3201 performs delta-sigma modulation on a first signal as an external input signal, and outputs a second signal (3301).

[0339] The NN learning unit 3203 calculates a first parameter of the first NN 3202a using the second signal and a third signal generated through a transmission process of the second signal (3302).

[0340] The NN processing unit 3202 updates the first parameters of the first NN 3202a to the first parameters calculated by the NN learning unit 3203 (3303).

[0341] The NN processing unit 3202 uses the second signal to output a fourth signal through the first NN 3202a (3304). The fourth signal is an approximation of a signal generated through at least a part of the transmission process of the second signal.

[0342] The delta-sigma modulation unit 3201 performs delta-sigma modulation on the first signal using the fourth signal output from the NN processing unit 3202, and outputs a second signal (3305).

[0343] The NN learning unit 3203 may calculate the first parameter of the first NN 3202a using the third signal and a restored signal of the second signal generated from the third signal.

[0344] The above configuration has the following advantages: Delta-sigma modulation device 3200 can suppress transmission distortion occurring in the transmission process of the second signal by the above-described feedback processing.

[0345] After the process of step 3305, the processes of steps 3302 to 3305 may be repeatedly executed.

[0346] <<7. Sixth Embodiment>> Next, a sixth embodiment and its modified example will be described with reference to FIGS.

[0347] <7-1. Configuration of Delta-Sigma Modulator> 35 is a diagram showing the configuration of a delta-sigma modulation device 3500. The delta-sigma modulation device 3500 includes a delta-sigma modulation section 3510 and a modeling processing section 3520.

[0348] The delta-sigma modulation unit 3510 performs delta-sigma modulation. The modeling processing unit 3520 performs processing to approximate the transmission characteristics of at least a part of the transmission process 3530.

[0349] The components 3510 and 3520 of the delta-sigma modulator 3500 may be implemented by one or more processors and memories, similar to the hardware configuration exemplarily described with reference to FIG. 9 . The one or more processors may include, for example, one or more of a CPU, an MPU, and a microcontroller. The memory may include a volatile memory and a non-volatile memory. The memory may store program code (instructions). The one or more processors may implement the functionality of the delta-sigma modulator 3500 by executing the program code stored in the memory.

[0350] As in the above-described embodiment, the analog signal input to the delta-sigma modulation unit 3510 is referred to as the "first signal." Furthermore, the quantized signal output by the delta-sigma modulation unit 3510 is referred to as the "second signal." The transmission process 3530 is a transmission process for the second signal and may be referred to as a transmission path for the second signal. The signal generated via the second signal transmission process 3530 and fed back to the modeling processing unit 3520 is referred to as the "third signal." The third signal may include information about transmission distortion occurring in the transmission process 3530. Therefore, the third signal is used to feed back information about the transmission distortion occurring in the transmission process 3530.

[0351] The delta-sigma modulation unit 3510 performs delta-sigma modulation on the first signal and outputs a second signal. The modeling processing unit 3520 uses the second signal and a third signal generated through a transmission process 3530 of the second signal to output a fourth signal that is an approximation of the signal generated through at least a part of the transmission process 3530. The delta-sigma modulation unit 3510 performs delta-sigma modulation on the first signal using the fourth signal and outputs the second signal.

[0352] <7-2. Model configuration> Next, a description will be given of the model used in modeling processing unit 3520. Hereinafter, this model will be referred to as the "first model." Modeling processing unit 3520 uses first model 3600 to output the fourth signal.

[0353] FIG. 36 is a diagram showing an example of the configuration of a first model 3600. The input signal to the first model 3600 is a second signal, represented by x(n). In FIG. 36, "D" represents a delay, as in the above-described embodiment. "α" is a parameter. The output signal (i.e., the fourth signal) of the first model 3600 is obtained by the linear sum of x(ni)×x(nj), which is a mixing of the current signal x(ni) and the past signal x(nj) of the second signal. If the fourth signal is y(n), y(n) is expressed by the following equation:

number

[0354] It should be noted that the mixing included in the output of the fourth signal is not limited to the mixing of the two signals described above. For example, the mixing included in the output of the fourth signal may include a linear sum of three or more mixings x(n i )×x(n j )×x(n k ). Hereinafter, such a model will be referred to as a "polynomial model."

[0355] The above configuration has the following advantages: First model 3600 can efficiently generate distortion due to mixing of current and voltage that occurs when the second signal passes through transmission process 3530. Therefore, the accuracy of the fourth signal output by first model 3600 is improved.

[0356] The first model may be a nonlinear model such as a neural network (NN), a logical operation, or a polynomial. For example, the first model may include the NN described in the first to fifth embodiments and its modified examples (FIG. 37). The first model may also be a combination of one or more of the NN, the logical operation, and the polynomial.

[0357] 37 is a diagram showing an example of the configuration of a first model 3700. The first model 3700 includes a neural network (NN). The fourth signal y(n) is obtained by adding the product of the output of the NN and the second signal x(n) and the product of the second signal x(n) and a constant (β).

[0358] In this way, the first model 3700 may be a model in which the output of the nonlinear model is the gain for the current signal (second signal x(n)). The above configuration has the following advantages. The first model 3700 can efficiently generate pattern-dependent gain changes that occur as the second signal passes through the transmission process 3530. The first model 3700 can efficiently generate distortion by multiplying the gain by the current signal. Therefore, the accuracy of the fourth signal output by the first model 3700 is improved.

[0359] 38 is a diagram showing an example of the configuration of a first model 3800. The first model 3800 includes a logical operation. In this example, for ease of explanation, the logical operation is AND. In this example, the fourth signal y(n) is expressed by the following equation:

number

[0360] Note that the logical operator included in the first model 3800 is not limited to AND and may include other logical operators (such as OR and XOR). The first model 3800 may include a combination of multiple logical operators. For example, the first model 3800 may include a combination of an OR operator and an AND operator, OR(AND(x(ni), x(nj)), x(nk)). In this case, if x(n) is a binary value of -1 or 1, the first model 3800 may include scaling such that the input of the logical operator becomes a binary value of 0 or 1. Specifically, the input of the logical operator is scaled from a binary value of -1, 1 to a binary value of 0, 1 by adding 1 to the conventional input and then multiplying it by 0.5.

[0361] The above configuration has the following advantages. The first model 3800 can efficiently generate a change in gain that accompanies a change in the pattern that occurs when the second signal passes through the transmission process 3530. The first model 3800 can efficiently generate distortion by multiplying the gain by the current signal. Therefore, the accuracy of the fourth signal output by the first model 3800 is improved.

[0362] FIG. 39 is a diagram showing an example of the configuration of a first model 3900. The polynomial used in the first model 3900 is the same as the polynomial described using FIG. 36. In this example, the fourth signal y(n) is obtained by adding the product of the output of the polynomial in FIG. 36 and the second signal x(n) and the product of the second signal x(n) and a constant (β). Specifically, the fourth signal y(n) is expressed by the following equation.

number

[0363] The above configuration has the following advantages. The first model 3900 can efficiently generate a change in gain due to the mixing of current and voltage that occurs when the second signal passes through the transmission process 3530. The first model 3900 can efficiently generate distortion by multiplying the gain by the current signal. Therefore, the accuracy of the fourth signal output by the first model 3900 is improved.

[0364] The first model may include one or a combination of two or more models selected from the models in Figures 13, 14, and 36 to 39. Figure 40 is a diagram showing an example of the configuration of a first model 4000. The first model 4000 is a combination of the polynomial model in Figure 39 and the logical operator model in Figure 38. The first model 4000 can generate the fourth signal with higher accuracy.

[0365] <7-3. Processing flow of delta-sigma modulator> FIG. 41 is a flowchart illustrating an example of the processing flow of delta-sigma modulation device 3500.

[0366] The delta-sigma modulation unit 3510 performs delta-sigma modulation on the first signal and outputs a second signal (4101). The modeling processing unit 3520 uses the second signal and a third signal generated through a transmission process 3530 of the second signal to output a fourth signal that is an approximation of the signal generated through at least a part of the transmission process 3530 (4102). The delta-sigma modulation unit 3510 performs delta-sigma modulation on the first signal using the fourth signal and outputs a second signal (4103).

[0367] The above configuration has the following advantages: Delta-sigma modulation device 3500 can suppress transmission distortion occurring in transmission process 3530 by the above-described feedback processing.

[0368] <7-4. Variations> (1) First Modification The output of the first model may be a complex number. That is, the output of the first model may include a signal representing a real component and a signal representing an imaginary component. A first model that outputs a signal representing such a complex number is realized by extending the coefficients in FIGS. 13, 14, and 36 to 40 to complex numbers. In another example, a first model that outputs a signal representing a complex number is realized by extending the weights and biases of a neural network (NN) to complex numbers.

[0369] The modeling processing unit 3520 outputs a fourth signal that is a signal representing a complex number. For example, there are cases where the feedback value to the delta-sigma modulation unit 3510 is limited to a real number. In this case, the delta-sigma modulation device 3500 may further include a conversion unit 4200. The conversion unit 4200 converts the fourth signal that is a signal representing a complex number into a fourth signal that is a signal representing a real number. With this configuration, the fourth signal that is a real number is fed back to the delta-sigma modulation unit 3510.

[0370] 42 is a diagram showing an example of the configuration of the conversion unit 4200. The conversion unit 4200 converts a complex fourth signal into a real fourth signal through frequency conversion and inverse frequency conversion operations. The frequency conversion and inverse frequency conversion operations are operations that shift the frequency of the fourth signal from a desired frequency band and return the shifted frequency to the original frequency band, thereby outputting a real signal.

[0371] For the above calculation, the conversion unit 4200 includes a downconverter 4210 and an upconverter 4220. The downconverter 4210 downconverts the complex fourth signal and outputs the downconverted signal to the upconverter 4220. The upconverter 4220 upconverts the signal and outputs the upconverted signal as a real fourth signal.

[0372] Note that the order of down-conversion and up-conversion may be reversed in the processing within the conversion unit 4200. That is, the real fourth signal may be obtained by up-converting the complex fourth signal and then down-converting the up-converted fourth signal.

[0373] The configuration of Fig. 42 can be applied to one or a combination of two or more selected from the models of Fig. 13, Fig. 14, and Figs. 36 to 40. According to this configuration, the expressive power of the first model can be expanded, and as a result, the accuracy of the fourth signal is improved.

[0374] (2) Second Modification 43 is a diagram showing an example of the configuration of the modeling processing unit 3520. The modeling processing unit 3520 includes a model output calculation unit 3521 and a model parameter calculation unit 3522. The model output calculation unit 3521 includes a first model that operates in accordance with a predetermined first parameter. The first model may have any of the configurations described above. The model parameter calculation unit 3522 calculates the first parameter.

[0375] FIG. 44 is a flowchart illustrating an example of the processing flow of delta-sigma modulation device 3500.

[0376] The delta-sigma modulation unit 3510 performs delta-sigma modulation on the first signal and outputs a second signal (4401).

[0377] The model parameter calculation unit 3522 calculates the first parameter using the third signal generated through the second signal transmission process 3530 and the second signal (4402).

[0378] In addition, the model parameter calculation unit 3522 may calculate the first parameter using a third signal generated through a second signal transmission process 3530 and a restored signal of the second signal generated from the third signal.

[0379] The model output calculation unit 3521 updates the first parameters to the first parameters calculated by the model parameter calculation unit 3522 (4403).

[0380] The model output calculation unit 3521 uses the second signal to output (4404) a fourth signal that is an approximation of a signal that would be generated through at least a portion of the transmission process 3530 through the first model.

[0381] The delta-sigma modulation unit 3510 performs delta-sigma modulation on the first signal using the fourth signal output from the model output calculation unit 3521, and outputs a second signal (4405).

[0382] (3) Third Modification FIG. 45 illustrates an exemplary configuration of model parameter calculation unit 3522. Model parameter calculation unit 3522 may include a second model 4510, an error calculation unit 4520, and a parameter calculation unit 4530. Second model 4510 is a second model that models at least a part of transmission process 3530. Second model 4510 may have the same configuration as the first model described above. Second model 4510 uses a second signal to generate an approximation of a signal generated through at least a part of transmission process 3530, and outputs the approximation to error calculation unit 4520. For example, second model 4510 outputs an approximation of a third signal. Error calculation unit 4520 calculates an error between the approximation of the third signal and the third signal, and outputs the error to parameter calculation unit 4530. Parameter calculation unit 4530 calculates a first parameter using the error. The parameter calculation unit 4530 outputs the calculated first parameter to the model output calculation unit 3521.

[0383] The parameter calculation section 4530 may calculate a first parameter when the error is greater than a predetermined first magnitude.

[0384] The parameter calculation unit 4530 may continue to calculate the first parameter until the error is smaller than a predetermined second magnitude, which is smaller than the first magnitude.

[0385] Second model 4510 may have the same configuration as transmission model 410. Error calculation section 4520 may execute the same processing as error calculation section 420 described above. Parameter calculation section 4530 may execute the same processing as parameter calculation section 430 described above.

[0386] The second model 4510 may operate according to predetermined second parameters. The parameter calculation unit 4530 may calculate the second parameters using the error and update the second parameters of the second model 4510 to the second parameters calculated by the parameter calculation unit 4530.

[0387] (4) Fourth Modification 46 is a diagram showing an example of the configuration of model parameter calculation section 3522. Model parameter calculation section 3522 further includes a delay adjustment section 4610 and a gain adjustment section 4620. Gain adjustment section 4620 multiplies the third signal by a gain. Delay adjustment section 4610 delays the second signal so that the timing at which the approximation value of the third signal is input to error calculation section 4520 and the timing at which the third signal is input to error calculation section 4520 are synchronized.

[0388] The delay adjustment unit 4610 may perform the same processing as the delay adjustment unit 1610 described above. The gain adjustment unit 4620 may perform the same processing as the gain adjustment unit 1620 described above.

[0389] The parameter calculation unit 4530 may calculate a first parameter and a second parameter using the error and internal parameters of the second model 4510. The internal parameters may include the second parameter and an output value of at least one node in the second model 4510.

[0390] (5) Fifth Modification The transmission process 3530 may have the configuration of the transmission path 120 described in the first to fifth embodiments. For example, the transmission process 3530 may include a communication path, a bandpass filter, an amplifier, and a modulator.

[0391] The modeling processing unit 3520 may include the digital restoration unit 149 described above. The digital restoration unit 149 generates a restored signal of the second signal from a third signal (in this example, a signal generated by the second signal passing through the communication path). This restored signal of the second signal is referred to as a "fifth signal," as described above. The modeling processing unit 3520 may calculate the first parameter using the fifth signal and the third signal.

[0392] The model parameter calculation unit may calculate the first parameter using the fifth signal and the third signal generated through the transmission process 3530.

[0393] The model parameter calculation unit may calculate the first parameter using a fifth signal and the third signal generated by the fifth signal passing through the band-pass filter, the amplifier, and the modulator.

[0394] The modeling processing unit 3520 may calculate a second parameter of the second model 4510 using the fifth signal and the third signal.

[0395] (6) Sixth Modification The delta-sigma modulation unit 3510 may have a configuration similar to that of the delta-sigma modulation unit 111 described in the first to fifth embodiments. The modeling processing unit 3520 may have the NN processing unit 112 and the NN learning unit 113 described in the first to fifth embodiments.

[0396] It should be noted that the above-described embodiment and modified examples are merely examples, and the scope of the technical idea of ​​the present disclosure is not limited to the above-described configurations. Other aspects conceivable within the scope of the technical idea of ​​the present disclosure are also included in the scope of the present disclosure.

[0397] The processing steps shown in the flowcharts or sequence diagrams do not necessarily have to be performed in the order shown. The processing steps may be performed in an order different from that shown, or two or more processing steps may be performed in parallel. Furthermore, some processing steps may be deleted, and additional processing steps may be added.

[0398] The arrows in the drawings are merely examples showing the direction of signal (data) flow from one component to another, and do not exclude bidirectional communication between the two components.

[0399] In this specification, "transmitting X to Y" is not limited to directly transmitting X to Y, but also includes indirectly transmitting X to Y (i.e., X is transmitted to another node, and X is transmitted from the other node to Y). Similarly, "receiving X from Y" is not limited to directly receiving X from Y, but also includes indirectly receiving X from Y (i.e., X is transmitted from Y to another node, and X is received from the other node). Therefore, a repeater may be disposed between the first device 810 and the second device 820. The repeater may repeat the transmission of the second signal and / or the transmission-related information.

[0400] The functions of the devices described herein may be realized by software, hardware, or a combination of software and hardware. Program code (instructions) constituting the software may be stored, for example, in a computer-readable recording medium inside or outside each device, and may be loaded into memory and executed by a processor at runtime. Alternatively, a non-transitory computer-readable recording medium having the program code recorded thereon may be provided.

[0401] Some or all of the above-described embodiments and modified examples can be described as, but are not limited to, the following supplementary notes.

[0402] (Appendix A1) a delta-sigma modulation unit that performs delta-sigma modulation on a first signal as an external input signal and outputs a second signal; a neural network processing unit including a first neural network that operates according to predetermined first parameters; a neural network learning unit that calculates the first parameter; Equipped with the neural network learning unit calculates the first parameter using a third signal generated through a transmission process of the second signal and the second signal or a restored signal of the second signal generated from the third signal; the neural network processing unit updates the first parameter to the first parameter calculated by the neural network learning unit; the neural network processing unit uses the second signal to output a fourth signal that is an approximation of a signal generated through at least a part of the transmission process through the first neural network; the delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal output from the neural network processing unit, and outputs the second signal. Delta-sigma modulator.

[0403] (Appendix A2) the delta-sigma modulation unit calculates a component for suppressing at least a part of the distortion occurring in the transmission process using the fourth signal, and reflects the calculated component in the second signal. 1. A delta-sigma modulator as described in Appendix A1.

[0404] (Appendix A3) The delta-sigma modulation unit an upconverter that upconverts the first signal; a loop filter that uses the output of the upconverter and the fourth signal to output a signal including a component for suppressing at least a part of distortion occurring in the transmission process; a quantizer that quantizes the signal output by the loop filter and outputs the second signal; Equipped with 1. A delta-sigma modulator as described in Appendix A1.

[0405] (Appendix A4) the first signal includes a first in-phase component signal and a first quadrature component signal; The delta-sigma modulation unit a downconverter that downconverts the fourth signal into a second in-phase component signal and a second quadrature component signal; a first loop filter that uses the first in-phase component signal and the second in-phase component signal to output a signal including a first component for suppressing at least a part of distortion occurring in the transmission process; a first quantizer that quantizes the signal output by the first loop filter and outputs a first quantized signal; a second loop filter that uses the first quadrature component signal and the second quadrature component signal to output a signal including a second component for suppressing at least a part of the distortion; a second quantizer that quantizes the signal output by the second loop filter and outputs a second quantized signal; an up-converter that up-converts the first quantized signal and the second quantized signal and outputs the second signal; Equipped with 1. A delta-sigma modulator as described in Appendix A1.

[0406] (Appendix A5) The neural network learning unit a transmission model that models at least a part of the transmission process, the transmission model using the second signal to output an approximation value of the third signal; an error calculation unit that calculates an error between the approximation of the third signal and the third signal; a parameter calculation unit that calculates the first parameter using the error; Equipped with A delta-sigma modulation device according to any one of appendices A1 to A4.

[0407] (Appendix A6) the parameter calculation unit calculates the first parameter when the error is greater than a predetermined first magnitude; 1. A delta-sigma modulation device as described in Appendix A5.

[0408] (Appendix A7) the parameter calculation unit continues calculating the first parameter until the error becomes smaller than a predetermined second magnitude; The second magnitude is smaller than the first magnitude. 1. A delta-sigma modulation device as described in Appendix A6.

[0409] (Appendix A8) the transmission model includes a second neural network that operates according to predetermined second parameters and outputs an approximation of a signal generated through at least a portion of the transmission process; The parameter calculation unit calculating the second parameter using the error; updating the second parameters of the second neural network to the second parameters calculated by the parameter calculation unit; The delta-sigma modulation device according to any one of appendices A5 to A7.

[0410] (Appendix A9) The parameter calculation unit calculating the first parameter and the second parameter using the error and an internal parameter of the transmission model; the internal parameters include the second parameters and an output value of at least one node in the second neural network; 10. A delta-sigma modulator as described in Appendix A8.

[0411] (Appendix A10) The neural network learning unit a gain adjustment unit that multiplies the third signal by a gain; a delay adjustment unit that delays the second signal so that a timing at which the approximation value of the third signal is input to the error calculation unit and a timing at which the third signal is input to the error calculation unit are synchronized; Further comprising: The delta-sigma modulation device according to any one of appendices A5 to A9.

[0412] (Appendix A11) the transmission process includes a communication path, a bandpass filter, an amplifier, and a modulator; a restoration unit that generates a fifth signal as the restored signal of the second signal from the third signal generated via the communication path, The neural network learning unit the fifth signal; and the third signal generated through the transmission process; Calculating the first parameter using The delta-sigma modulation device according to any one of appendices A1 to A10.

[0413] (Appendix A12) the transmission process includes a communication path, a bandpass filter, an amplifier, and a modulator; a restoration unit that generates a fifth signal as the restored signal of the second signal from the third signal generated via the communication path, The neural network learning unit the fifth signal; and The fifth signal is the third signal generated through a process of passing through the bandpass filter, the amplifier, and the modulator. Calculating the first parameter using The delta-sigma modulation device according to any one of appendices A1 to A10.

[0414] (Appendix A13) the first neural network comprises an input layer, at least one hidden layer, and an output layer; the input layer includes a node to which a current value of the second signal is input; The input layer further comprises: a node to which a past value of the second signal is input; a node to which a result of a logical operation of two or more nodes in the input layer is input; a node having an output line through which the current value of the second signal is output to the output layer without passing through the hidden layer; including at least one of The delta-sigma modulation device according to any one of appendices A1 to A12.

[0415] (Appendix A14) the fourth signal includes an approximation of at least a portion of the distortion occurring during at least a portion of the transmission process. The delta-sigma modulation device according to any one of appendices A1 to A13.

[0416] (Appendix A15) performing delta-sigma modulation on a first signal as an external input signal to output a second signal; Calculating a first parameter of a first neural network using a third signal generated through a transmission process of the second signal and the second signal or a restored signal of the second signal generated from the third signal; updating the first parameters of the first neural network to the calculated first parameters; outputting a fourth signal that is an approximation of a signal generated through at least a portion of the transmission process through the first neural network using the second signal; performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal; A delta-sigma modulation method including:

[0417] (Appendix A16) performing delta-sigma modulation on a first signal as an external input signal to output a second signal; Calculating a first parameter of a first neural network using a third signal generated through a transmission process of the second signal and the second signal or a restored signal of the second signal generated from the third signal; updating the first parameters of the first neural network to the calculated first parameters; outputting a fourth signal that is an approximation of a signal generated through at least a portion of the transmission process through the first neural network using the second signal; performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal; A non-transitory computer-readable recording medium that records a program that causes a processor to execute the above.

[0418] (Appendix B1) a delta-sigma modulation unit that performs delta-sigma modulation; a modeling processor that approximates at least a part of the transmission characteristics of the transmission process; Equipped with the delta-sigma modulation unit performs the delta-sigma modulation on a first signal as an input signal and outputs a second signal; the modeling processing unit uses the second signal and a third signal generated through a transmission process of the second signal to output a fourth signal that is an approximation of a signal generated through at least a part of the transmission process; the delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal, and outputs the second signal. Delta-sigma modulator.

[0419] (Appendix B2) The modeling processing unit a model output calculation unit including a first model that operates according to predetermined first parameters; a model parameter calculation unit that calculates the first parameter; Equipped with the model parameter calculation unit calculates the first parameters using the third signal and the second signal or a restored signal of the second signal generated from the third signal; the model output calculation unit updates the first parameters to the first parameters calculated by the model parameter calculation unit; the model output calculation unit uses the second signal to pass through the first model and output the fourth signal; 10. The delta-sigma modulator of claim 1.

[0420] (Appendix B3) the delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal output from the model output calculation unit, and outputs the second signal. 1. A delta-sigma modulator as described in Appendix B2.

[0421] (Appendix B4) the delta-sigma modulation unit calculates a component for suppressing at least a part of the distortion occurring in the transmission process using the fourth signal, and reflects the calculated component in the second signal. 10. The delta-sigma modulator of claim 1.

[0422] (Appendix B5) The delta-sigma modulation unit an upconverter that upconverts the first signal; a loop filter that uses the output of the upconverter and the fourth signal to output a signal including a component for suppressing at least a part of distortion occurring in the transmission process; a quantizer that quantizes the signal output by the loop filter and outputs the second signal; Equipped with 10. The delta-sigma modulator of claim 1.

[0423] (Appendix B6) the first signal includes a first in-phase component signal and a first quadrature component signal; The delta-sigma modulation unit a downconverter that downconverts the fourth signal into a second in-phase component signal and a second quadrature component signal; a first loop filter that uses the first in-phase component signal and the second in-phase component signal to output a signal including a first component for suppressing at least a part of distortion occurring in the transmission process; a first quantizer that quantizes the signal output by the first loop filter and outputs a first quantized signal; a second loop filter that uses the first quadrature component signal and the second quadrature component signal to output a signal including a second component for suppressing at least a part of the distortion; a second quantizer that quantizes the signal output by the second loop filter and outputs a second quantized signal; an up-converter that up-converts the first quantized signal and the second quantized signal and outputs the second signal; Equipped with 10. The delta-sigma modulator of claim 1.

[0424] (Appendix B7) The model parameter calculation unit a second model that models at least a part of the transmission process, the second model using the second signal to output an approximation of the third signal; an error calculation unit that calculates an error between the approximation of the third signal and the third signal; a first parameter calculation unit that calculates the first parameter using the error; Equipped with 1. A delta-sigma modulator as described in Appendix B2.

[0425] (Appendix B8) the first parameter calculation unit calculates the first parameter when the error is greater than a predetermined first magnitude; 10. The delta-sigma modulator of claim 7.

[0426] (Appendix B9) the first parameter calculation unit continues to calculate the first parameter until the error becomes smaller than a predetermined second magnitude; The second magnitude is smaller than the first magnitude. 10. The delta-sigma modulator of claim 8.

[0427] (Appendix B10) the second model operates according to predetermined second parameters; The first parameter calculation unit calculating the second parameter using the error; updating the second parameters of the second model to the second parameters calculated by the first parameter calculation unit; 10. The delta-sigma modulator of claim 7.

[0428] (Appendix B11) the first parameter calculation unit calculates the first parameter and the second parameter using the error and an internal parameter of the second model; the internal parameters include the second parameters and an output value of at least one node in the second model; 10. The delta-sigma modulator of claim 8, wherein the delta-sigma modulator is a delta-sigma modulator.

[0429] (Appendix B12) The model parameter calculation unit a gain adjustment unit that multiplies the third signal by a gain; a delay adjustment unit that delays the second signal so that a timing at which the approximation value of the third signal is input to the error calculation unit and a timing at which the third signal is input to the error calculation unit are synchronized; Further comprising: 10. The delta-sigma modulator of claim 7.

[0430] (Appendix B13) the transmission process includes a communication path, a bandpass filter, an amplifier, and a modulator; a restoration unit that generates a fifth signal as the restored signal of the second signal from the third signal generated via the communication path, The model parameter calculation unit the fifth signal; and the third signal generated through the transmission process; Calculating the first parameter using 1. A delta-sigma modulator as described in Appendix B2.

[0431] (Appendix B14) the transmission process includes a communication path, a bandpass filter, an amplifier, and a modulator; a restoration unit that generates a fifth signal as the restored signal of the second signal from the third signal generated via the communication path, The model parameter calculation unit the fifth signal; and The fifth signal is the third signal generated through a process of passing through the bandpass filter, the amplifier, and the modulator. Calculating the first parameter using 1. A delta-sigma modulator as described in Appendix B2.

[0432] (Appendix B15) the fourth signal includes an approximation of at least a portion of the distortion occurring during at least a portion of the transmission process. 10. The delta-sigma modulator of claim 1.

[0433] (Appendix B16) the fourth signal is a signal representing a complex number, a conversion unit that converts the fourth signal representing the complex number into the fourth signal representing a real number, The delta-sigma modulation device according to any one of appendices B1 to B15.

[0434] (Appendix B17) The conversion unit includes an up-converter and a down-converter. 10. The delta-sigma modulator of claim 8, wherein the delta-sigma modulator is a delta-sigma modulator.

[0435] (Appendix B18) performing delta-sigma modulation on a first signal as an input signal to output a second signal; outputting a fourth signal that is an approximation of a signal generated through at least a part of the transmission process using the second signal and a third signal generated through a transmission process of the second signal; performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal; A delta-sigma modulation method including:

[0436] (Appendix B19) performing delta-sigma modulation on a first signal as an input signal to output a second signal; outputting a fourth signal that is an approximation of a signal generated through at least a part of the transmission process using the second signal and a third signal generated through a transmission process of the second signal; performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal; A non-transitory computer-readable recording medium that records a program that causes a processor to execute the above. [Industrial Applicability]

[0437] Distortion occurring in a signal output through delta-sigma modulation can be suppressed during transmission of the signal. [Explanation of symbols]

[0438] 110: Delta-sigma modulator 111: Delta Sigma Modulation Section 112: Neural network processing unit 113: Neural network learning section 120: Transmission path 810: First device 820: Second device 3200: Delta-Sigma Modulator 3201: Delta Sigma Modulation Section 3202: Neural network processing unit 3203: Neural Network Learning Department 3500: Delta-Sigma Modulator 3510: Delta-Sigma Modulation Section 3520: Modeling processing section 3530: Transmission process

Claims

1. a delta-sigma modulation unit that performs delta-sigma modulation; a modeling processor that approximates at least a part of the transmission characteristics of the transmission process; Equipped with the delta-sigma modulation unit performs the delta-sigma modulation on a first signal as an input signal and outputs a second signal; the modeling processing unit uses the second signal and a third signal generated through a transmission process of the second signal to output a fourth signal that is an approximation of a signal generated through at least a part of the transmission process; the delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal, and outputs the second signal. Delta-sigma modulator.

2. The modeling processing unit a model output calculation unit including a first model that operates according to predetermined first parameters; a model parameter calculation unit that calculates the first parameter; Equipped with the model parameter calculation unit calculates the first parameters using the third signal and the second signal or a restored signal of the second signal generated from the third signal; the model output calculation unit updates the first parameters to the first parameters calculated by the model parameter calculation unit; the model output calculation unit uses the second signal to pass through the first model and output the fourth signal; 2. The delta-sigma modulation device according to claim 1.

3. the delta-sigma modulation unit performs the delta-sigma modulation on the first signal using the fourth signal output from the model output calculation unit, and outputs the second signal.

3. The delta-sigma modulation device according to claim 2.

4. the delta-sigma modulation unit calculates a component for suppressing at least a part of distortion occurring in the transmission process using the fourth signal, and reflects the calculated component in the second signal.

2. The delta-sigma modulation device according to claim 1.

5. The delta-sigma modulation unit an upconverter that upconverts the first signal; a loop filter that uses the output of the upconverter and the fourth signal to output a signal including a component for suppressing at least a part of distortion occurring in the transmission process; a quantizer that quantizes the signal output by the loop filter and outputs the second signal; Equipped with 2. The delta-sigma modulation device according to claim 1.

6. the first signal includes a first in-phase component signal and a first quadrature component signal; The delta-sigma modulation unit a downconverter that downconverts the fourth signal to a second in-phase component signal and a second quadrature component signal; a first loop filter that uses the first in-phase component signal and the second in-phase component signal to output a signal including a first component for suppressing at least a part of distortion occurring in the transmission process; a first quantizer that quantizes the signal output by the first loop filter and outputs a first quantized signal; a second loop filter that uses the first quadrature component signal and the second quadrature component signal to output a signal including a second component for suppressing at least a part of the distortion; a second quantizer that quantizes the signal output by the second loop filter and outputs a second quantized signal; an up-converter that up-converts the first quantized signal and the second quantized signal and outputs the second signal; Equipped with 2. The delta-sigma modulation device according to claim 1.

7. The model parameter calculation unit a second model that models at least a part of the transmission process, the second model using the second signal to output an approximation of the third signal; an error calculation unit that calculates an error between the approximation of the third signal and the third signal; a first parameter calculation unit that calculates the first parameter using the error; Equipped with 3. The delta-sigma modulation device according to claim 2.

8. the second model operates according to predetermined second parameters; The first parameter calculation unit calculating the second parameter using the error; updating the second parameters of the second model to the second parameters calculated by the first parameter calculation unit; 8. The delta-sigma modulation device according to claim 7.

9. The first parameter calculation unit calculating the first parameters and the second parameters using the error and the internal parameters of the second model; the internal parameters include the second parameters and an output value of at least one node in the second model; 9. The delta-sigma modulation device according to claim 8.

10. The model parameter calculation unit a gain adjustment unit that multiplies the third signal by a gain; a delay adjustment unit that delays the second signal so that a timing at which the approximation value of the third signal is input to the error calculation unit and a timing at which the third signal is input to the error calculation unit are synchronized; Further comprising:

8. The delta-sigma modulation device according to claim 7.

11. the fourth signal is a signal representing a complex number, a conversion unit that converts the fourth signal representing the complex number into the fourth signal representing a real number, The delta-sigma modulation device according to any one of claims 1 to 10.

12. The conversion unit includes an up-converter and a down-converter.

12. The delta-sigma modulation device according to claim 11.

13. performing delta-sigma modulation on a first signal as an input signal to output a second signal; outputting a fourth signal that is an approximation of a signal generated through at least a part of the transmission process using the second signal and a third signal generated through a transmission process of the second signal; performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal; A delta-sigma modulation method including:

14. performing delta-sigma modulation on a first signal as an input signal to output a second signal; outputting a fourth signal that is an approximation of a signal generated through at least a part of the transmission process using the second signal and a third signal generated through a transmission process of the second signal; performing the delta-sigma modulation on the first signal using the fourth signal to output the second signal; A program that causes a processor to execute the following.

Images