Apparatus and method for compensating for non-linearity of power amplifier

Abstract

This disclosure provides an apparatus and method for compensating for nonlinearity in a power amplifier. An apparatus configured to perform wireless communication to accelerate a neural network includes: a predistortion circuit configured to generate a predistorted input signal by performing predistortion on an input signal based on a parameter set including multiple coefficients; a power amplifier configured to generate an output signal by amplifying an RF signal based on the predistorted input signal; and a parameter acquisition circuit configured to obtain second memory polynomial modeling information corresponding to an operating frequency band based on first memory polynomial modeling information corresponding to each of a plurality of frequency bands, and to use the second memory polynomial modeling information to obtain a parameter set according to an indirect learning structure.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit of Korean Patent Application No. 10-2019-0083441, filed on July 10, 2019 with the Korean Intellectual Property Office, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] This disclosure generally relates to an apparatus and method for compensating for nonlinearity in a power amplifier (such as a power amplifier for a wireless communication device). Background Technology

[0004] Apparatus for wireless communication may include a transmitter that provides a radio frequency (RF) signal output to an antenna. The transmitter may include components for generating the RF signal from a baseband signal, such as a mixer for up-converting the baseband signal to an RF signal, one or more filters, and an RF power amplifier. When the baseband signal is processed by the transmitter's components, the RF signal may become distorted due to the characteristics of the components. For example, the power amplifier may exhibit, in particular, nonlinear gain and phase variations with the input signal power, and this nonlinearity may degrade communication quality by distorting the RF output signal. In the case of digital baseband signals, distortion due to operation in the gain compression region of the power amplifier can lead to excessive bit errors, especially for bits represented by relatively higher power signals. To reduce distortion, digital predistortion techniques or RF predistortion techniques can be used to predistort the input RF signal of the power amplifier in a manner complementary to the characteristics of the power amplifier. However, at higher RF frequencies and in the presence of multiple antenna elements in an antenna array (where mutual coupling may affect the power amplifier), RF signal distortion may be exacerbated and may be more difficult to compensate for using predistortion methods. Summary of the Invention

[0005] Embodiments of the present invention provide an apparatus and method for compensating for nonlinearity in a power amplifier, and more specifically, an apparatus and method for adaptively performing predistortion across various frequency bands while reducing the storage space required for predistortion memory.

[0006] According to one aspect of the present invention, an apparatus configured to perform wireless communication to accelerate a neural network is provided, the apparatus comprising: a predistortion circuit configured to generate a predistorted input signal by performing predistortion on an input signal based on a parameter set including a plurality of coefficients; a power amplifier configured to generate an output signal by amplifying an RF signal based on the predistorted input signal; and a parameter acquisition circuit configured to obtain second memory polynomial modeling information corresponding to an operating frequency band based on first memory polynomial modeling information corresponding to each of a plurality of frequency bands, and to obtain a parameter set according to an indirect learning structure by using the second memory polynomial modeling information.

[0007] According to another aspect of the present invention, a method for processing a signal of a processing device is provided, the method comprising: determining at least one frequency band comprising at least a portion of the operating frequency band of the device from a plurality of frequency bands divided from an entire frequency band on which the device is configured to operate; obtaining an autocorrelation matrix corresponding to the operating frequency band and a cross-correlation vector corresponding to the operating frequency band based on frequency band information corresponding to the determined at least one frequency band; and generating an output signal by performing predistortion on an input signal using a coefficient matrix obtained based on the autocorrelation matrix corresponding to the operating frequency band and the cross-correlation vector corresponding to the operating frequency band.

[0008] According to another aspect of the present invention, an apparatus configured to perform wireless communication is provided, the apparatus comprising: a memory configured to store multiple frequency band information and instructions for operation of the apparatus; a processor configured to perform predistortion on an input signal in a given frequency band and generate a predistorted input signal by executing at least one instruction among the instructions stored in the memory; and a power amplifier configured to generate an output signal by amplifying the predistorted input signal, wherein the processor is configured to determine at least one frequency band among the multiple frequency bands that includes at least a portion of the given frequency band, and to perform predistortion calculation on the input signal using a coefficient matrix obtained by using frequency band information corresponding to the determined at least one frequency band among the multiple frequency band information.

[0009] According to another aspect of the present invention, a method is provided for performing predistortion calculation on an input signal spanning a frequency band spanning a first frequency band and a second frequency band, performed by an apparatus. The method includes: obtaining an autocorrelation matrix corresponding to the frequency band based on the center frequency of the frequency band, a first autocorrelation matrix corresponding to the first frequency band, and a second autocorrelation matrix corresponding to the second frequency band; obtaining a cross-correlation vector corresponding to the frequency band based on the center frequency of the frequency band, a first cross-correlation vector corresponding to the first frequency band, and a second cross-correlation vector corresponding to the second frequency band; and performing predistortion on the input signal using a coefficient matrix obtained based on the obtained autocorrelation matrix and the obtained cross-correlation vector.

[0010] According to another aspect of the present invention, an apparatus configured to perform wireless communication is provided, the apparatus comprising: a memory configured to store multiple band information of each of a plurality of frequency bands divided in an entire frequency band on which the apparatus is configured to operate; a processor configured to generate a predistorted input signal by performing predistortion on an input signal in a given frequency band; and an amplifier configured to generate an output signal based on the predistorted input signal provided by the processor, wherein the processor is configured to perform predistortion on the input signal by using a coefficient matrix obtained based on first band information corresponding to the first band and second band information corresponding to the second band when the given frequency band spans a first band and a second band among the plurality of frequency bands, and to perform predistortion on the input signal by using a coefficient matrix obtained based on third band information corresponding to the third band and fourth band information corresponding to the fourth band when the given frequency band spans a third band and a fourth band among the plurality of frequency bands.

[0011] According to another aspect of the present invention, an apparatus configured to perform wireless communication is provided, the apparatus comprising: a memory configured to store multiple memory polynomial modeling information corresponding to each of a plurality of frequency bands in a frequency band to which the apparatus is configured to operate; and a processor configured to obtain second memory polynomial modeling information corresponding to the operating frequency band using at least one first memory polynomial modeling information corresponding to at least one frequency band in the plurality of frequency bands that includes at least a portion of the operating frequency band, and to generate a predistorted input signal by performing predistortion on an input signal using a parameter set obtained using the second memory polynomial modeling information. Attached Figure Description

[0012] Embodiments of the inventive concept will become clearer from the following detailed description taken in conjunction with the accompanying drawings, in which:

[0013] Figure 1 An apparatus illustrating an example embodiment of a concept according to the present invention;

[0014] Figure 2 This illustrates an example of the input-output characteristics of a power amplifier;

[0015] Figure 3 An apparatus illustrating an example embodiment of a concept according to the present invention;

[0016] Figure 4 This illustrates multiple frequency bands included in the entire frequency band according to an example embodiment of the inventive concept;

[0017] Figure 5 The entire frequency band and a given frequency band are shown in an example embodiment of the present invention;

[0018] Figure 6 This is a diagram of a parameter acquisition circuit according to an exemplary embodiment of the present invention;

[0019] Figure 7 Frequency band information is shown in an example embodiment of the concept according to the present invention;

[0020] Figure 8 This is a flowchart of a signal processing method of an apparatus according to an exemplary embodiment of the present invention;

[0021] Figure 9 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention;

[0022] Figure 10 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention;

[0023] Figure 11 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention;

[0024] Figure 12 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention;

[0025] Figure 13 This is a parameter acquisition circuit based on an example embodiment of the present invention;

[0026] Figure 14 This is a parameter acquisition circuit based on an example embodiment of the present invention;

[0027] Figure 15 The entire frequency band and multiple frequency zones are shown in an example embodiment according to the inventive concept;

[0028] Figure 16 The entire frequency band and multiple frequency zones are shown in an example embodiment according to the inventive concept;

[0029] Figure 17 A communication device is shown as an example embodiment of a concept according to the present invention. Detailed Implementation

[0030] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0031] Figure 1An apparatus 10 is illustrated according to an example embodiment of the invention. Apparatus 10 may include predistortion circuitry 100, transmitter 200, and parameter acquisition circuitry 300. Apparatus 10 may be a communication device (typically a wireless communication device). Apparatus 10 may be a base station or user equipment included in a wireless communication system. As a non-limiting example, the wireless communication system may be a wireless communication system using a cellular network, such as a 5G wireless system, a Long Term Evolution (LTE) system, LTE Advanced, Code Division Multiple Access (CDMA) system, a Global System for Mobile Communications (GSM) system, or any other wireless communication system, such as a Wireless Local Area Network (WLAN), WiFi, and Bluetooth. A base station may be referred to as a Node B, Evolved Node B (eNB), segment, site, Base Transceiver System (BTS), Access Point (AP), Relay Node and Remote Radio Header End (RRH), Radio Unit (RU), small cell, etc. User equipment may be referred to as a terminal device, Mobile Station (MS), Mobile Terminal (MT), User Terminal (UT), Subscriber Station (SS), wireless device, handheld device, etc. Figure 1 In addition to the components shown, the device 10 may also include various other components.

[0032] Transmitter 200 can process the baseband predistorted signal PDS and generate an RF output signal OS. For example, transmitter 200 may include an up-converter 202 for up-converting the predistorted signal PDS to an input RF signal RFin and a power amplifier 204 for amplifying the signal RFin to generate the output signal OS. Herein, the input RF signal RFin may be referred to as the predistorted RF signal and may be referred to as the RF signal based on the predistorted signal PDS. Transmitter 200 and / or predistortion circuitry 100 may also include various filters (not shown), such as a low-pass filter, to smooth inter-symbol transitions.

[0033] Typically, the output characteristics of power amplifier 204 may require linearity, but due to the inherent characteristics of power amplifier 204 or various external factors, power amplifier 204 may exhibit nonlinear characteristics. In other words, such as Figure 2 As shown, the output characteristics of power amplifier 204 can exhibit nonlinear characteristics.

[0034] Figure 2 Example input-output characteristics of power amplifier 204 are shown.

[0035] The solid line represents the desired gain characteristic of power amplifier 204. For example... Figure 2 As shown, the characteristic of desired gain can be represented by the characteristic that the input voltage is proportional to the output voltage.

[0036] However, the actual gain of power amplifier 204 can exhibit the same characteristics as... Figure 2The same characteristics as the dashed lines in the diagram. In other words, the actual gain characteristic of the power amplifier 204 can be represented by a nonlinear characteristic in a specific region where the input voltage and output voltage are not proportional to each other.

[0037] Further reference Figure 1 Predistortion can be used to compensate for the nonlinearity of the power amplifier 204. Predistortion can be described as a technique that predistorts the input signal IS based on characteristics that are complementary to the nonlinearity of the power amplifier 204.

[0038] For example, predistortion circuit 100 can generate a baseband predistorted signal PDS by performing predistortion on the input signal IS. Predistortion circuit 100 can provide the predistorted signal PDS to transmitter 200. Power amplifier 204 can generate an output signal OS by amplifying the predistorted RF signal RFI. As predistortion circuit 100 performs predistortion on the input signal IS, the nonlinearity of power amplifier 204 can be compensated. In an embodiment, predistortion circuit 100 can perform digital predistortion on the input signal IS.

[0039] In one embodiment, the predistortion circuit 100 may perform predistortion on the input signal IS based on a parameter set PS. The parameter set PS may be provided by the parameter acquisition circuit 300. The parameter set PS may include multiple coefficients for predistortion.

[0040] Here, the input signal IS and the predistortion signal PDS can be digital signals, such as phase shift keying (PSK) signals, quadrature amplitude modulation (QAM) signals, etc. The output signal OS can be referred to as the modulated digital signal.

[0041] The predistortion circuit 100 can be modeled using a polynomial as shown in Equation 1. Here, polynomial modeling can be referred to as memory polynomial modeling. In Equation 1, x(n) represents a sample of the input signal IS, z(n) represents a sample of the predistortion signal PDS, and a q This represents the coefficients used for predistortion, and Q represents the nonlinear order.

[0042] [Formula 1]

[0043]

[0044] The predistortion circuit 100 can be implemented in various forms. According to embodiments, the predistortion circuit 100 can be implemented in hardware or software. When the predistortion circuit 100 is implemented in hardware, it may include circuitry for performing predistortion on the input signal IS. When the predistortion circuit 100 is implemented in software, it may be implemented by a processor (…). Figure 3The predistortion is performed by executing a program and / or instructions loaded into memory 400 (or any processor in device 10) to perform the predistortion. In other examples, the predistortion circuitry 100 may be implemented by a combination of software and hardware (such as firmware).

[0045] The parameter acquisition circuit 300 can obtain the parameter set PS based on the predistortion signal PDS and the output signal OS, and can provide the generated parameter set PS to the predistortion circuit 100.

[0046] In an embodiment, the parameter acquisition circuit 300 can obtain a parameter set PS comprising multiple coefficients based on an indirect learning structure. The indirect learning structure can represent a learning structure in which the difference between the intermediate signal obtained from the output signal OS and the predistorted signal PDS (rather than the difference between the input signal IS and the output signal OS) is minimized. Where y(n) represents a sample of the output signal OS, w(n) represents a sample of the intermediate signal, and a... kq When Q represents the nonlinear order and K represents the memory depth, the intermediate signal can be a signal obtained using a polynomial such as Equation 2 below.

[0047] [Equation 2]

[0048]

[0049] The parameter acquisition circuit 300 can obtain a parameter set PS that reduces the difference between the intermediate signal and the predistortion signal PDS obtained based on the output signal OS as described above, by using an indirect learning structure. In an embodiment, the parameter acquisition circuit 300 can obtain a parameter set PS that minimizes the mean square error (MSE) between the intermediate signal and the predistortion signal PDS. When z(n) represents a sample of the predistortion signal PDS, w(n) represents a sample of the intermediate signal, and e(n) represents a sample of the error signal, the MSE can be obtained using Equation 3 below.

[0050] [Formula 3]

[0051] MSE = E[e 2 [(n)]=E[|z(n)-w(n)| 2 ]

[0052] The factor that minimizes the MSE can be obtained by using a Wiener filter. In other words, the parameter acquisition circuit 300 can obtain the coefficient 'a' that minimizes the MSE by applying a Wiener filter. kqThe parameter set PS. In this case, when applying the Wiener filter, the process of obtaining the coefficients that minimize MSE may involve solving the matrix equation in Equation 4 below. In Equation 4, y(n) represents the sample of the output signal OS, z(n) represents the sample of the predistorted signal PDS, Q represents the nonlinear order, and K represents the memory depth.

[0053] [Formula 4]

[0054] Ax = b

[0055] A = E[Y(n)Y] H (n)],x=[a 00 a 01 a 02 …a K-1Q-1 ] T b = E[Y(n)z H (n)],

[0056] Y(n)=[y(n),y(n)|y(n)|,,…,y(n)|y(n)| Q-1 ,y(n-1),…,y(n-K+1)|y(n-K+1)| Q-1 ] T

[0057] In Equation 4, the matrix with T as a superscript denotes the transpose matrix. The matrix with H as a superscript is the Hermitian matrix, which is obtained by transposing the matrix after applying conjugation to all its elements. In other words, the matrix with H as a superscript is the conjugate transpose matrix. E[] denotes the expectation of the values ​​included therein. Y(n) can be a KQ×1 matrix with elements corresponding to the i*j-th row and the first column having the values ​​shown in Equation 5 below.

[0058] [Formula 5]

[0059] Y(n)[i*j,1]=y(n-i+1)|y(n-i+1)| j-1

[0060] Here, x can be a KQ×1 matrix whose elements corresponding to the i*jth row and the first column have the values ​​in Equation 6.

[0061] [Formula 6]

[0062] x[i*j,1]=a (i-1)(j-1)

[0063] As a result, the matrix A used in the method of applying the Wiener filter can be a KQ×KQ matrix, and the matrix b can be a KQ×1 matrix. In other words, the matrix b can be a vector with KQ elements. Hereinafter, for ease of description, in the method of applying the Wiener filter to Equation 4, the matrix A used in the equivalent matrix equation can be referred to as the autocorrelation matrix, and the matrix b can be referred to as the cross-correlation vector. In addition, the matrix x including the coefficients can be referred to as the coefficient matrix. In other words, obtaining the parameter set PS can include obtaining the coefficient matrix by obtaining the solution of the matrix equation of Equation 4. In this way, the autocorrelation matrix and cross-correlation vector required to obtain the coefficient matrix can be referred to as "memory polynomial modeling information" (interchangeably, only "polynomial modeling information"), which can be information based on signal measurement and subsequent calculations using the measurement results, and this information can be stored in the memory 400 within the device 10.

[0064] Regarding the operating frequency, the operating frequency band can vary depending on the type or configuration of device 10. For example, the center frequency and bandwidth of the operating frequency band set for the operation of device 10 can vary. When the operating frequency band changes, the memory polynomial modeling information required to obtain the coefficient matrix x from parameter acquisition circuit 300 can vary depending on the characteristics of power amplifier 204. In other words, without implementing the inventive concept taught herein, device 10 may need to store memory polynomial modeling information for every possible frequency band to provide highly reliable predistortion. However, storing all memory polynomial modeling information for every possible frequency band may result in a large memory storage space requirement. On the other hand, using the embodiments described herein, a method for performing predistortion for various frequency bands can be implemented using a smaller memory storage space.

[0065] The parameter acquisition circuit 300 of an exemplary embodiment of the present invention can use frequency band information (FSI) of multiple frequency bands pre-divided throughout the entire frequency band to obtain memory polynomial modeling information corresponding to a given (operating) frequency band, wherein the device 10 is configured to operate over the entire frequency band. The frequency band information (FSI) may be stored in memory 400. Hereinafter, the term "given frequency band" may refer to the operating frequency band currently allocated for the operation of device 10. "Given frequency band" may be interchangeably referred to as a sub-band or channel of the entire frequency band. Many given frequency bands may exist within the entire frequency band of operation of device 10.

[0066] To this end, device 10 can divide the entire frequency band into multiple frequency bands, and after obtaining polynomial modeling information corresponding to each of the multiple frequency bands, it can generate polynomial modeling information corresponding to each of the multiple frequency bands and frequency band information FSI including the center frequency of each of the multiple frequency bands, and memory 400 can store the frequency band information FSI. (Refer to...) Figure 4 , Figure 15 and Figure 16 For a more detailed description of multiple frequency bands, please refer to [reference needed]. Figure 7 The frequency band information (FSI) is described in detail. The number of frequency bands "N" may be less than the number of allowed given frequency bands "M" within the entire frequency band on which device 10 is configured to operate. Instead of measuring and calculating polynomial modeling information separately for each given frequency band, this information can be obtained by calculation based on the polynomial modeling information of one or more frequency bands overlapping with the given frequency band. Therefore, by obtaining polynomial modeling information for a smaller number of frequency bands N, the amount of measurements and calculations performed in advance to cover all allowed given frequency bands and / or the amount of memory space used to store the information can be reduced.

[0067] According to an embodiment, the parameter acquisition circuit 300 can determine at least one frequency band among a plurality of frequency bands, including at least a portion of a given frequency band, and can obtain polynomial modeling information corresponding to the frequency band provided by the frequency band information FSI corresponding to the determined at least one frequency band.

[0068] For example, the parameter acquisition circuit 300 can obtain an autocorrelation matrix corresponding to at least one determined frequency band, a center frequency corresponding to at least one determined frequency band, and an autocorrelation matrix corresponding to a frequency band given based on the center frequency of a given frequency band. Additionally, the parameter acquisition circuit 300 can obtain a cross-correlation vector corresponding to at least one determined frequency band, a center frequency corresponding to at least one determined frequency band, and a cross-correlation vector corresponding to a frequency band given based on the center frequency of a given frequency band. The parameter acquisition circuit 300 can obtain a coefficient matrix based on the obtained autocorrelation matrix and the obtained cross-correlation vector, and can output the coefficient matrix, or the coefficients included in the coefficient matrix, as a parameter set PS. For example, the parameter acquisition circuit 300 can obtain the coefficient matrix by performing a calculation that multiplies the cross-correlation vector by the inverse of the obtained autocorrelation matrix. Alternatively, in an embodiment, the parameter acquisition circuit 300 can obtain the coefficient matrix by performing an iterative approximation calculation using the obtained autocorrelation matrix and the obtained cross-correlation vector, and in this case, an embodiment using the conjugate gradient method is also applicable.

[0069] The following figures describe in detail a method for obtaining polynomial modeling information corresponding to a given frequency band based on band information FSI for multiple frequency bands, for example, a method for obtaining the autocorrelation matrix and cross-correlation vector corresponding to a given frequency band by using parameter obtaining circuit 300.

[0070] The parameter acquisition circuit 300 can be implemented in various forms, and according to embodiments, it can be implemented in hardware or software. When the parameter acquisition circuit 300 is implemented in hardware, it may include circuitry for generating a parameter set PS based on the output signal OS and the predistortion signal PDS. Alternatively, for example, when the parameter acquisition circuit 300 is implemented in software, such as... Figure 3 As shown, this can be achieved by using a processor ( Figure 3 The parameter acquisition circuit 300 can be generated by any processor in the device 10 (500) executing a program and / or instructions loaded into the memory 400 to generate the parameter set PS. However, the embodiments are not limited thereto, and the parameter acquisition circuit 300 can be implemented by a combination of software and hardware (such as firmware).

[0071] Memory 400 may be a storage area for storing data, and may store, for example, an operating system (OS), various programs, and various data. Memory 400 may include at least one of volatile memory and non-volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable PROM (EEPROM), flash memory, phase-change random access memory (RAM) (PRAM), magnetic RAM (MRAM), resistive RAM (RRAM), ferroelectric RAM (FRAM), etc. Volatile memory may include dynamic RAM (DRAM), static RAM (SRAM), synchronous DRAM (SDRAM), phase-change RAM (PRAM), magnetic RAM (MRAM), resistive RAM (RRAM), ferroelectric RAM (FeRAM), etc. In addition, in embodiments, memory 400 may include at least one of hard disk drive (HDD), solid-state drive (SSD), compact flash memory (CF), secure digital storage (SD), miniature secure digital storage, extreme digital storage (xD), or memory stick. In embodiments, memory 400 may semi-permanently or temporarily store data generated by a processor ( Figure 3 The 500 in the memory stores the program and multiple instructions executed by the processor. Additionally, the memory 400 can store data for the processor (…). Figure 3 The various information or data used in the calculation or operation of 500 (in the context of the calculation or operation).

[0072] According to the apparatus 10 of the exemplary embodiment based on the inventive concept, when the parameter acquisition circuit 300 obtains memory polynomial modeling information, such as an autocorrelation matrix or a cross-correlation vector, for obtaining the coefficient matrix based on band information FSI for multiple frequency bands, the amount of data to be stored in the memory 400 can be reduced. In other words, the storage space of the memory 400 used or required to perform predistortion in the apparatus 10 can be reduced.

[0073] Furthermore, since the device 10 of the exemplary embodiment of the present invention is adaptively capable of performing predistortion in a wide range of given frequency bands by using only a small amount of storage space in the memory 400, the reliability of wireless communication of the device 10 can also be improved.

[0074] Figure 3 An apparatus 20 is shown as an example embodiment of a concept according to the present invention. Specifically, Figure 3 yes Figure 1 The diagram illustrates an implementation example of the predistortion circuit 100, transmitter 200, parameter acquisition circuit 300, and memory 400. (Details omitted.) Figure 3 China has already referred to Figure 1 Redundancy descriptions of the predistortion circuit 100, transmitter 200, parameter acquisition circuit 300, and memory 400 are given.

[0075] The device 20 may include a transmitter 200, a memory 400, and a processor 500, and the processor 500 may include a predistortion circuit 100 and a parameter acquisition circuit 300.

[0076] Processor 500 can control all operations of device 20; for example, processor 500 may be a central processing unit (CPU). Processor 500 may include only a single processor core, or alternatively, may include multiple processor cores (“multi-core”). Processor 500 can process or execute programs and / or data stored in memory 400. In embodiments, processor 500 can control various functions or perform various calculations of device 20 by executing programs stored in memory 400.

[0077] According to an exemplary embodiment of the present invention, the processor 500 can generate a predistorted signal PDS by performing predistortion on the input signal IS. In the embodiment, the processor 500 can obtain memory polynomial modeling information corresponding to a given frequency band based on frequency band information FSI of multiple frequency bands stored in the memory 400, obtain a parameter set PS based on the obtained polynomial modeling information, and perform predistortion on the input signal IS based on the obtained parameter set PS.

[0078] In the apparatus 20 of the exemplary embodiment of the present invention, since the processor 500 obtains memory polynomial modeling information, such as an autocorrelation matrix or a cross-correlation vector, for obtaining the coefficient matrix based on band information FSI about multiple frequency bands, the amount of data to be stored in the memory 400 can be reduced. In other words, the storage space of the memory 400 used or required to perform predistortion in the apparatus 20 can be reduced.

[0079] Furthermore, since the device 20 is adaptively able to perform predistortion in various given frequency bands over a wide bandwidth using only a small amount of storage space in the memory 400, the reliability of the wireless communication of the device 20 can also be improved.

[0080] Figure 4 The illustration shows the first frequency band FS_1 to the fifth frequency band FS_5 included in the entire frequency band according to an exemplary embodiment of the present invention. (Refer to...) Figure 1 right Figure 4 Describe it.

[0081] The device 10 can operate in a specific frequency band depending on the type of device 10 or the settings applied thereto. Figure 4 The diagram shows the entire frequency band, including all frequencies available in the band. For example, the entire frequency band may include the band between a first frequency f_1 and a sixth frequency f_6.

[0082] To obtain polynomial modeling information, device 10 can divide the entire frequency band into multiple frequency bands (FS_1, FS_2, FS_3, FS_4, and FS_5). For ease of description, the number of frequency bands and their corresponding bandwidths are merely exemplary and not limited to specific frequency bands. Figure 4 The width is shown. For example, the first frequency band FS_1 can represent the frequency band between the first frequency f_1 and the second frequency f_2, and the center frequency of the first frequency band FS_1 can be the first center frequency fc_1. Similarly, the second frequency band FS_2 can represent the frequency band between the second frequency f_2 and the third frequency f_3, and the center frequency of the second frequency band FS_2 can be the second center frequency fc_2. In the same way, the third frequency band FS_3, the fourth frequency band FS_4, and the fifth frequency band FS_5 can be understood.

[0083] In this embodiment, the first frequency band FS_1 to the fifth frequency band FS_5 may have the same bandwidth. However, the embodiment is not limited to this; the first frequency band FS_1 to the fifth frequency band FS_5 may have different bandwidths from each other. For example, a frequency band near the center of the entire frequency band may have a narrower bandwidth than a frequency band located near the edge of the entire frequency band.

[0084] In this embodiment, device 10 can obtain memory polynomial modeling information for each of multiple frequency bands (FS_1, FS_2, FS_3, FS_4, and FS_5), and the memory 400 in device 10 can store the memory polynomial modeling information obtained for each of the multiple frequency bands (FS_1, FS_2, FS_3, FS_4, and FS_5) and the center frequency of each of the multiple frequency bands (FS_1, FS_2, FS_3, FS_4, and FS_5) as frequency band information FSI. (Refer to...) Figure 7 The frequency band information (FSI) is described in more detail.

[0085] Figure 5 The entire frequency band and a given frequency band are shown in an example embodiment of the concept according to the present invention. Specifically, Figure 5 The given frequency band is shown as follows Figure 4 The diagram shows the situation of a specific frequency band within the entire frequency band. See below for reference. Figure 1 right Figure 5 Describe it.

[0086] Device 10 may operate in a specific frequency band depending on various factors such as the type of device 10 or its settings, and this specific frequency band may be referred to as a given frequency band. The given frequency band may have a center frequency fc.

[0087] Figure 5 The illustration shows a given frequency band spanning both the third frequency band FS_3 and the fourth frequency band FS_4, but the embodiment is merely illustrative and not limited thereto. For example, a given frequency band may span at least one other frequency band.

[0088] In an embodiment, the parameter acquisition circuit 300 in device 10 can determine at least one frequency band (FS_1, FS_2, FS_3, FS_4, and FS_5) that includes at least a portion of a given frequency band. For example, the parameter acquisition circuit 300 can determine that the third frequency band FS_3 and the fourth frequency band FS_4 include at least a portion of a given frequency band.

[0089] In an embodiment, the parameter acquisition circuit 300 may obtain memory polynomial modeling information corresponding to a given frequency band based on memory polynomial modeling information corresponding to at least one determined frequency band. For example, the parameter acquisition circuit 300 may obtain memory polynomial modeling information corresponding to a given frequency band based on memory polynomial modeling information corresponding to a third frequency band FS_3 and memory polynomial modeling information corresponding to a fourth frequency band FS_4. Each piece of memory polynomial modeling information may include an autocorrelation matrix and a cross-correlation vector used in a matrix equation used to minimize the difference between the intermediate signal based on the output signal OS and the predistortion signal PDS according to the application of the Wiener filter. For example, each piece of memory polynomial modeling information may include the autocorrelation matrix A and the cross-correlation vector b in Equation 4.

[0090] Reference Figure 6 The method by which the parameter acquisition circuit 300 obtains memory polynomial modeling information corresponding to a given frequency band is described in more detail.

[0091] Figure 6 This is a diagram illustrating a parameter acquisition circuit 300 according to an exemplary embodiment of the present invention. The parameter acquisition circuit 300 may correspond to... Figure 1 and Figure 3 The parameters are obtained from circuit 300. (Also refer to...) Figure 1 and Figure 5 right Figure 6 Describe it.

[0092] The parameter acquisition circuit 300 may include an autocorrelation matrix acquisition circuit 310, a cross-correlation vector acquisition circuit 320, and a coefficient matrix acquisition circuit 330.

[0093] The autocorrelation matrix acquisition circuit 310 can obtain the autocorrelation matrix ACM corresponding to a given frequency band based on the frequency band information FSI of multiple frequency bands and provide the autocorrelation matrix ACM to the coefficient matrix acquisition circuit 330.

[0094] The cross-correlation vector acquisition circuit 320 can obtain the cross-correlation vector CCV corresponding to a given frequency band based on the frequency band information FSI of multiple frequency bands and provide the cross-correlation vector CCV to the coefficient matrix acquisition circuit 330.

[0095] The coefficient matrix acquisition circuit 330 can obtain the coefficient matrix CM corresponding to a given frequency band based on the autocorrelation matrix ACM and the cross-correlation vector CCV corresponding to the given frequency band. For example, the coefficient matrix acquisition circuit 330 can obtain the coefficient matrix CM by performing a calculation that multiplies the cross-correlation vector CCV by the inverse of the autocorrelation matrix ACM. The coefficient matrix acquisition circuit 330 can output the coefficient matrix CM as a parameter set PS.

[0096] Figure 7 Frequency band information (FSI) is shown according to an example embodiment of the concept of the present invention. Also refer to... Figure 1 and Figure 5 right Figure 7 Describe it.

[0097] The Frequency Band Information (FSI) can include the first FSI_1 to the Nth FSI_N (where N is a natural number of 2 or greater) corresponding to each of the multiple frequency bands. For example, when the entire frequency band is divided into N frequency bands, the FSI can include the frequency band information corresponding to each of the N frequency bands.

[0098] The first frequency band information FSI_1 is described as a representative of the first to Nth frequency band information FSI_N. The first frequency band information FSI_1 may include the memory polynomial modeling information corresponding to the first frequency band FS_1, and may include the first center frequency fc_1 of the first frequency band FS_1. The memory polynomial modeling information corresponding to the first frequency band FS_1 may include the first autocorrelation matrix ACM_1 and the first cross-correlation vector CCV_1. In other words, the first frequency band information FSI_1 may include the first autocorrelation matrix ACM_1, the first cross-correlation vector CCV_1, and the first center frequency fc_1. Combined with... Figure 6 and Figure 7The following figures illustrate a method for obtaining memory polynomial modeling information corresponding to a given frequency band.

[0099] Figure 8 This is a flowchart of a signal processing method of apparatus 10 according to an exemplary embodiment of the present invention. (Refer to...) Figure 1 right Figure 8 Describe it.

[0100] Device 10 can determine at least one frequency band among a plurality of frequency bands that includes at least a portion of a given frequency band (S110). For example, such as Figure 4 As shown, multiple frequency bands can represent frequency bands divided within an entire frequency band. In an embodiment, device 10 can determine at least one frequency band comprising at least a portion of a given frequency band based on the start and end frequencies of that given frequency band. For example, while referring to... Figure 5 When given as Figure 5 When the frequency band shown is used, device 10 can determine that the given frequency band spans the third frequency band FS_3 and the fourth frequency band FS_4.

[0101] Apparatus 10 can obtain the autocorrelation matrix and cross-correlation vector based on the band information FSI corresponding to the at least one determined frequency band (S120). In other words, apparatus 10 can obtain memory polynomial modeling information corresponding to a given frequency band based on the band information FSI corresponding to the at least one determined frequency band. For example, the parameter acquisition circuit 300 in apparatus 10 can obtain memory polynomial modeling information corresponding to a given frequency band based on the memory polynomial modeling information corresponding to each of the at least one determined frequency bands, the center frequency corresponding to each of the at least one determined frequency bands, and the center frequency fc of the given frequency band. For example, as described in more detail below, the parameter acquisition circuit 300 can obtain the autocorrelation matrix corresponding to a given frequency band based on the autocorrelation matrix corresponding to each of the at least one determined frequency bands, the center frequency corresponding to each of the at least one determined frequency bands, and the center frequency fc of the given frequency band. Similarly, the parameter acquisition circuit 300 can obtain the cross-correlation vector corresponding to a given frequency band based on the cross-correlation vector corresponding to each of the at least one determined frequency band, the center frequency corresponding to each of the at least one determined frequency band, and the center frequency fc of the given frequency band.

[0102] Apparatus 10 can perform predistortion (S130) based on a coefficient matrix obtained from the autocorrelation matrix and cross-correlation vector already obtained in operation S120. For example, parameter acquisition circuit 300 can obtain the coefficient matrix by performing a calculation that multiplies the cross-correlation vector corresponding to a given frequency band by the inverse of the autocorrelation matrix corresponding to the given frequency band, and can provide the obtained coefficient matrix to predistortion circuit 100 as a parameter set PS. Predistortion circuit 100 can generate a predistorted signal PDS by performing predistortion on the input signal IS based on parameter set PS.

[0103] Figure 9 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention. Specifically, Figure 9 yes Figure 8 The flowchart below details the example operation of S120. Refer to the following example. Figure 1 right Figure 9 Describe it.

[0104] The device 10 can obtain the autocorrelation matrix corresponding to the given frequency band based on the center frequency fc of the given frequency band, the center frequency corresponding to the determined at least one frequency band, and the autocorrelation matrix corresponding to the determined at least one frequency band (S220). These operations (S220) can be performed by the parameter acquisition circuit 300.

[0105] The device 10 can obtain the cross-correlation vector corresponding to the given frequency band based on the center frequency fc of the given frequency band, the center frequency corresponding to the determined at least one frequency band, and the cross-correlation vector corresponding to the determined at least one frequency band (S240). These operations (S240) can also be performed by the parameter acquisition circuit 300.

[0106] Figure 10 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention. Specifically, Figure 10 yes Figure 9 A detailed flowchart illustrating operation S220 is provided. (Refer to...) Figure 1 right Figure 10 Describe it.

[0107] To facilitate understanding of the concept of this invention, Figure 10 A flowchart is shown in the case where at least one frequency band is defined, including a first frequency band FS_1 and a second frequency band FS_2, and a given frequency band spans both the first frequency band FS_1 and the second frequency band FS_2. Each of the following operations S322, S324, S342, S344, and S360 can be performed by the parameter acquisition circuit 300 of device 10.

[0108] The device 10 can obtain a first frequency shift vector (S322) based on the center frequency fc corresponding to a given frequency band and a first center frequency fc_1, which is the center frequency of the first frequency band FS_1. In an embodiment, the first frequency shift vector can be obtained by the following equation 7. In equation 7, FSV1 represents the first frequency shift vector, fc represents the center frequency of the given frequency band, fc1 represents the first center frequency, T1 represents the sampling time, and K represents the memory depth.

[0109] [Formula 7]

[0110] FSV1 = ee(fc - fc1),

[0111]

[0112] In Equation 7, ee(x) can be used as an intermediate function, and when Q is a nonlinear order, ee(x) can be a vector comprising K×Q elements. For example, ee(x) can represent 1 repeated Q times, and then e 2πxT1 Repeated Q times, then e 2π2xT1 Repeated Q, finally e 2π(K-1)xT1 A vector that is repeated Q times.

[0113] The device 10 can obtain a first temporary autocorrelation matrix based on a first frequency shift vector and a first autocorrelation matrix corresponding to the first frequency band FS_1 (S324). In an embodiment, the first temporary autocorrelation matrix can be obtained by the following equation 8. In equation 8, TACM1 represents the first temporary autocorrelation matrix, ACM1 represents the first autocorrelation matrix, and FSV1 represents the first frequency shift vector.

[0114] [Formula 8]

[0115]

[0116] In Equation 8, the matrix with T as the superscript represents the transpose matrix. In the calculation, "·" denotes matrix multiplication. This represents the Hadamard product calculation. The Hadamard product calculation can be represented as the multiplication of elements at the same position in the two matrices to be multiplied, and can be called element-wise multiplication.

[0117] Similarly, device 10 can obtain a second frequency shift vector (S342) based on the center frequency fc corresponding to a given frequency band and a second center frequency fc_2, which is the center frequency of the second frequency band FS_2. In an embodiment, the second frequency shift vector can be obtained by the following equation 9. In equation 9, FSV2 can represent the second frequency shift vector, fc can represent the center frequency of the given frequency band, fc2 can represent the second center frequency, T1 can represent the sampling time, and K can represent the memory depth.

[0118] [Formula 9]

[0119] FSV2 = ee(fc - fc2)

[0120]

[0121] In Equation 9, ee(x) can be used as an intermediate function, and when Q is a nonlinear order, ee(x) can be a vector containing K×Q elements. For example, ee(x) can represent 1 repeated Q times, and then e 2πxT1 Repeated Q times, then e 2π2xT1 Repeated Q times, finally e 2π(K-1)T1 A vector that is repeated Q times.

[0122] The device 10 can obtain the second temporary autocorrelation matrix (S344) based on the second frequency shift vector and the second autocorrelation matrix corresponding to the second frequency band FS_2. In an embodiment, the second temporary autocorrelation matrix can be obtained by the following equation 10. In equation 10, TACM2 can represent the second temporary autocorrelation matrix, ACM2 can represent the second autocorrelation matrix, and FSV2 can represent the second frequency shift vector.

[0123] [Formula 10]

[0124]

[0125] In Equation 10, a matrix with T as a superscript can represent a transpose matrix. In the calculation, "·" represents matrix multiplication. The Hadamard product can be represented as a multiplication operation between elements at the same position in the two matrices to be multiplied, and can be called element-wise multiplication.

[0126] In other words, operation S320, which includes operations S322 and S324 and is used to obtain the first temporary autocorrelation matrix, can be substantially similar to operation S340, which includes operations S342 and S344 and is used to obtain the second temporary autocorrelation matrix.

[0127] The device 10 can obtain the autocorrelation matrix corresponding to a given frequency band based on the first temporary autocorrelation matrix and the second temporary autocorrelation matrix (S360), for example, by summing the first temporary autocorrelation matrix and the second temporary autocorrelation matrix. The autocorrelation matrix corresponding to the given frequency band can be obtained by the following equation 11, where TACM1 represents the first temporary autocorrelation matrix, TACM2 represents the second temporary autocorrelation matrix, and ACM represents the autocorrelation matrix corresponding to the given frequency band.

[0128] [Equation 11]

[0129] ACM = TACM1 + TACM2

[0130] Figure 11 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention. Specifically, Figure 11 It can be Figure 9 The detailed operation flowchart of operation S220 is shown below. Figure 1 right Figure 11 Describe it.

[0131] To facilitate understanding of the concept of this invention, Figure 11 A flowchart is shown in the case where at least one frequency band is included, namely, a first frequency band FS_1, a second frequency band FS_2, and a third frequency band FS_3.

[0132] Figure 11 The operation of S320 can be combined with Figure 10 The operation is basically the same as that of S320, and Figure 11 Operation S340 can be combined with Figure 10 Operation S340 is essentially the same. In the following description, operations S352, S354 and S370 can be executed by the parameter acquisition circuit 300 of device 10.

[0133] The device 10 can obtain the third frequency shift vector (S352) based on the center frequency fc corresponding to the given frequency band and the third center frequency fc_3, which is the center frequency of the third frequency band FS_3. The detailed method for obtaining the third frequency shift adjacent can be similar to the methods in Equations 7 and 9.

[0134] Device 10 can obtain the third temporary autocorrelation matrix (S354) based on the third frequency shift vector and the third autocorrelation matrix corresponding to the third frequency band FS_3. The detailed method for obtaining the third temporary autocorrelation matrix can be similar to the methods in Equations 8 and 10.

[0135] In other words, operation S350, which includes operations S352 and S354, can be substantially similar to operations S320 and S340.

[0136] Device 10 can obtain the autocorrelation matrix corresponding to a given frequency band based on the first temporary autocorrelation matrix, the second temporary autocorrelation matrix, and the third temporary autocorrelation matrix (S370), for example, by summing the first temporary autocorrelation matrix, the second temporary autocorrelation matrix, and the third temporary autocorrelation matrix. The autocorrelation matrix corresponding to the given frequency band can be obtained by the following equation 12, where TACM1, TACM2, and TACM3 represent the first temporary autocorrelation matrix, the second temporary autocorrelation matrix, and the third temporary autocorrelation matrix, respectively, and ACM represents the autocorrelation matrix corresponding to the given frequency band.

[0137] [Equation 12]

[0138] ACM = TACM1 + TACM2 + TACM3

[0139] Reference Figure 10 To describe an implementation where a given frequency band spans two frequency bands, refer to Figure 11 This describes an embodiment where a given frequency band spans three frequency bands. In the case of embodiments where a given frequency band spans four or more frequency bands, see reference... Figure 10 and Figure 11 The method described for obtaining the autocorrelation matrix can be similarly extrapolated to obtain the autocorrelation matrix using four or more temporary autocorrelation matrices.

[0140] Figure 12 This is a flowchart of a signal processing method according to an exemplary embodiment of the present invention. Specifically, Figure 12 It can be Figure 9 The detailed operation flowchart of operation S240 is provided. (Refer to...) Figure 1 right Figure 12 Describe it.

[0141] To facilitate understanding of the concept of this invention, Figure 12 A flowchart is shown in the case where at least one determined frequency band includes a first frequency band FS_1 and a second frequency band FS_2. In the following description, operations S422, S424, S442, S444, and S460 can be performed by the parameter acquisition circuit 300 of device 10.

[0142] The device 10 can obtain a first frequency shift vector (S422) based on the center frequency fc corresponding to a given frequency band and a first center frequency fc_1, which is the center frequency of the first frequency band FS_1. In an embodiment, the first frequency shift vector can be obtained by referencing... Figure 10 Equation 7 describes the results.

[0143] The device 10 can obtain a first temporary cross-correlation vector based on a first frequency shift vector and a first cross-correlation vector corresponding to the first frequency band FS_1 (S424). In an embodiment, the first temporary cross-correlation vector can be obtained by the following equation 13. In equation 13, TCCV1 represents the first temporary cross-correlation vector, CCV1 represents the first cross-correlation vector, and FSV1 represents the first frequency shift vector.

[0144] [Equation 13]

[0145]

[0146] In Equation 13, a matrix with T as a superscript can represent a transpose matrix. In the calculation, "·" represents matrix multiplication. The Hadamard product can be represented as the multiplication of elements at the same position in the two matrices to be multiplied, and can be called element-wise multiplication.

[0147] Similarly, device 10 can obtain a second frequency shift vector (S442) based on the center frequency fc corresponding to a given frequency band and a second center frequency fc_2, which is the center frequency of the second frequency band FS_2. In an embodiment, the second frequency shift vector can be obtained by referring to... Figure 10 Equation 9 describes the results.

[0148] Device 10 can obtain a second temporary cross-correlation vector (S444) based on a second frequency shift vector and a second cross-correlation vector corresponding to the second frequency band FS_2. In an embodiment, the second temporary cross-correlation vector can be obtained by the following equation 14. In equation 14, TCCV2 represents the second temporary cross-correlation vector, CCV2 represents the second cross-correlation vector, and FSV2 represents the second frequency shift vector.

[0149] [Formula 14]

[0150]

[0151] In Equation 14, the matrix with T as the superscript represents the transpose matrix. In the calculation, "·" can represent matrix multiplication. The Hadamard product can be represented as the multiplication of elements at the same position in the two matrices to be multiplied, and can be called element-wise multiplication.

[0152] In other words, operation S420, which includes operations S422 and S424 and is an operation to obtain the first temporary cross-correlation vector, can be substantially similar to operation S440, which includes operations S442 and S444 and is an operation to obtain the second temporary cross-correlation vector.

[0153] Device 10 can obtain the cross-correlation vector corresponding to a given frequency band based on a first temporary cross-correlation vector and a second temporary cross-correlation vector (S460), for example, by summing the first temporary cross-correlation vector and the second temporary cross-correlation vector. The cross-correlation vector corresponding to the given frequency band can be obtained by the following equation 15. Here, TCCV1 represents the first temporary cross-correlation vector, TCCV2 represents the second temporary cross-correlation vector, and CCV represents the cross-correlation vector corresponding to the given frequency band.

[0154] [Formula 15]

[0155] CCV = TCCV1 + TCCV2

[0156] Reference Figure 12 The given description suggests that the method for obtaining the cross-correlation vector can be implemented using an extrapolation approach similar to that used in embodiments where a given frequency band spans three or more frequency bands.

[0157] Figure 13This is a diagram of a parameter acquisition circuit 300 according to an exemplary embodiment of the present invention.

[0158] The parameter acquisition circuit 300 may include an autocorrelation matrix acquisition circuit 310, a cross-correlation vector acquisition circuit 320, and a coefficient matrix acquisition circuit 330. Descriptions of the autocorrelation matrix acquisition circuit 310, the cross-correlation vector acquisition circuit 320, and the coefficient matrix acquisition circuit 330 have been referenced. Figure 6 Given, its redundant description is omitted.

[0159] The autocorrelation matrix acquisition circuit 310 may include a first segment determination circuit 312, a first frequency shift vector acquisition circuit 314, and an autocorrelation matrix calculation circuit 316.

[0160] The first segment determination circuit 312 can determine at least one frequency band that includes at least a portion of a given frequency band from a plurality of frequency bands. The first segment determination circuit 312 can provide the determined segment information DS to the first frequency shift vector acquisition circuit 314.

[0161] The first frequency shift vector acquisition circuit 314 can select a frequency band information FSI corresponding to at least one frequency band among multiple frequency band information FSIs based on the determined segment information DS, and can generate a frequency shift vector FSV based on the frequency band information FSI corresponding to the determined at least one frequency band. For example, the first frequency shift vector acquisition circuit 314 can, according to a reference Figure 10 Equation 7 describes the generation of the frequency shift vector FSV.

[0162] The autocorrelation matrix calculation circuit 316 can obtain the autocorrelation matrix ACM corresponding to a given frequency band based on the frequency shift vector FSV. For example, when the given frequency band spans a first frequency band and a second frequency band, the autocorrelation matrix calculation circuit 316 can obtain the autocorrelation matrix ACM corresponding to a given frequency band based on the frequency shift vector FSV by using... Figure 10 The same method as operations S324, S344, and S360 is used to obtain the autocorrelation matrix ACM corresponding to a given frequency band.

[0163] The cross-correlation vector acquisition circuit 320 may include a second segment determination circuit 322, a second frequency shift vector acquisition circuit 324, and a cross-correlation vector calculation circuit 326.

[0164] The second segment determination circuit 322 can determine at least one frequency band that includes at least a portion of a given frequency band from a plurality of frequency bands. The second segment determination circuit 322 can provide the determined segment information DS to the second frequency shift vector acquisition circuit 324.

[0165] The second frequency shift vector acquisition circuit 324 can select a frequency band information FSI corresponding to at least one frequency band determined in the frequency band information FSI based on the determined segment information DS, and can generate a frequency shift vector FSV based on the frequency band information FSI corresponding to the determined at least one frequency band. For example, the second frequency shift vector acquisition circuit 324 can, according to a reference Figure 10 Equation 7 describes the generation of the frequency shift vector FSV.

[0166] The cross-correlation vector calculation circuit 326 can obtain the cross-correlation vector CCV corresponding to a given frequency band based on the frequency shift vector FSV. For example, when the given frequency band spans a first frequency band and a second frequency band, the cross-correlation vector calculation circuit 326 can obtain the cross-correlation vector CCV based on the frequency shift vector FSV by using... Figure 12 The same method as operations S424, S444, and S460 is used to obtain the cross-correlation vector CCV corresponding to a given frequency band.

[0167] Figure 14 This is a diagram of a parameter acquisition circuit 300 according to an exemplary embodiment of the present invention. Specifically, Figure 14 This illustrates another implementation of the parameter acquisition circuit 300. Figure 14 An embodiment showing the segment determination circuit 340 and the frequency shift vector acquisition circuit 350 being shared is shown as follows. Figure 13 Implementation example.

[0168] The parameter acquisition circuit 300 may include an autocorrelation matrix acquisition circuit 310, a cross-correlation vector acquisition circuit 320, a coefficient matrix acquisition circuit 330, a segment determination circuit 340, and a frequency shift vector acquisition circuit 350. (The remaining text is omitted as it is already referenced.) Figure 6 Descriptions of the autocorrelation matrix acquisition circuit 310, the cross-correlation vector acquisition circuit 320, and the coefficient matrix acquisition circuit 330 are given.

[0169] The segment determination circuit 340 may have the same characteristics as... Figure 13 The first segment determination circuit 312 and the second segment determination circuit 322 have essentially the same function. In other words, the segment determination circuit 340 can determine at least one frequency band that includes at least a portion of a given frequency band from a plurality of frequency bands. The segment determination circuit 340 can provide the determined segment information DS to the frequency shift vector acquisition circuit 350.

[0170] The frequency shift vector acquisition circuit 350 can have the same characteristics as... Figure 13 The first frequency shift vector acquisition circuit 314 and the second frequency shift vector acquisition circuit 324 in the circuit have essentially the same function.

[0171] Figure 14 The autocorrelation matrix acquisition circuit 310 in the middle can have the same characteristics as... Figure 13 The autocorrelation matrix calculation circuit 316 in the middle has basically the same function. Figure 14 The cross-correlation vector acquisition circuit 320 in the middle can have the same characteristics as... Figure 13 The cross-correlation vector calculation circuit 326 in the middle has basically the same function.

[0172] Figure 15 The entire frequency band and multiple frequency bands (FS_1, FS_2, FS_3, FS_4, and FS_5) are shown in an exemplary embodiment of the present invention. (See also...) Figure 15 ,and Figure 4 Different frequency bands (FS_1, FS_2, FS_3, FS_4 and FS_5) may have different bandwidths.

[0173] For example, the bandwidth of a frequency band near the center of the entire frequency band can be smaller than the bandwidth of a frequency band near the edge of the entire frequency band.

[0174] For example, the bandwidth of the second frequency band FS_2, the third frequency band FS_3, and the fourth frequency band FS_4 can be represented as the first bandwidth Δf1, and the bandwidth of the first frequency band FS_1 and the fifth frequency band FS_5 near the edge of the entire frequency band can be represented as the second bandwidth Δf2. In an embodiment, the first bandwidth Δf1 may be smaller than the second bandwidth Δf2.

[0175] Figure 16 The entire frequency band and multiple frequency bands (FS_1, FS_2, FS_3, FS_4, and FS_5) are shown in an exemplary embodiment of the present invention. (See also...) Figure 16 ,and Figure 4 Different frequency bands (FS_1, FS_2, FS_3, FS_4 and FS_5) may have different bandwidths.

[0176] For example, the bandwidth of the third frequency band FS_3 can be the first bandwidth Δf1, the bandwidth of the second frequency band FS_2 and the fourth frequency band FS_4 can be the second bandwidth Δf2, and the bandwidth of the first frequency band FS_1 and the fifth frequency band FS_5 can be the third bandwidth Δf3. The first bandwidth Δf1 can be smaller than the second bandwidth Δf2, and the second bandwidth Δf2 can be smaller than the third bandwidth Δf3.

[0177] Figure 17 A communication device 1000 according to an example embodiment of the concept of the present invention is shown. For example... Figure 17As shown, the communication device 1000 may include an application-specific integrated circuit (ASIC) 1100, an application-specific instruction set processor (ASIP) 1300, a memory 1500, a main processor 1700, and a main memory 1900. Two or more of the ASIC 1100, ASIP 1300, and main processor 1700 can communicate with each other. Furthermore, at least two or more of the ASIC 1100, ASIP 1300, memory 1500, main processor 1700, and main memory 1900 may be embedded in a single chip.

[0178] ASIP 1300 may include application-specific integrated circuits that support application-specific instruction sets and execute instructions contained within those instruction sets. Memory 1500 may communicate with ASIP 1300 and may serve as non-volatile memory to store multiple instructions executed by ASIP 1300. For example, memory 1500 may include any type of memory accessible by ASIP 1300, such as, as non-limiting examples, random access memory (RAM), read-only memory (ROM), magnetic tape, magnetic disk, optical disk, volatile memory, non-volatile memory, and combinations thereof.

[0179] The main processor 1700 can control the communication device 1000 by executing multiple instructions. For example, the main processor 1700 can control the ASIC 1100 and ASIP 1300, and process data received via the MIMO channel, or process user input to the communication device 1000. The main memory 1900 can communicate with the main processor 1700 and can store multiple instructions executed by the main processor 1700 as non-volatile memory. For example, the main memory 1900 can include any type of memory accessible by the main processor 1700, such as, as non-limiting examples, RAM, ROM, magnetic tape, magnetic disk, optical disk, volatile memory, non-volatile memory, and combinations thereof.

[0180] The above-described method for compensating the nonlinearity of the transmitter according to an example embodiment of the present invention can be achieved by... Figure 17 The communication device 1000 may be executed by at least one of the components included in the communication device 1000. For example, the processor 500 described above may be included in... Figure 17The method of compensating for the nonlinearity of the transmitter 200 can be implemented as at least one of ASIC 1100, ASIP 1300, memory 1500, main processor 1700, and main memory 1900. In some embodiments, at least one operation of the above-described method for compensating for the nonlinearity of the transmitter 200 can be implemented as a plurality of instructions stored in memory 1500. In some embodiments, ASIP 1300 can perform at least one operation of the method for compensating for the nonlinearity of the transmitter 200 by executing the plurality of instructions stored in memory 1500. In some embodiments, at least one operation of the method for compensating for the nonlinearity of the transmitter 200 can be implemented in a hardware block designed and included in ASIC 1100 using logic synthesis or the like. In some embodiments, at least one operation of the method for compensating for the nonlinearity of the power amplifier 200 can be implemented as a plurality of instructions stored in main memory 1900, and main processor 1700 can perform at least one operation of the method for compensating for the nonlinearity of the power amplifier 200 by executing the plurality of instructions stored in main memory 1900.

[0181] Although the inventive concept has been specifically shown and described with reference to embodiments thereof, it will be understood that various changes in form and detail may be made therein without departing from the spirit and scope of the appended claims. For example, while the inventive concept has been specifically shown and described in conjunction with wireless communication applications, in other examples, the inventive concept can be applied to correcting nonlinearities in wired communication systems.

Claims

1. An apparatus configured to perform wireless communication, the apparatus comprising: a pre-distortion circuit configured to generate a pre-distorted signal by performing pre-distortion on an input signal based on a parameter set including a plurality of coefficients; a power amplifier configured to generate an output signal by amplifying a radio frequency (RF) signal based on the pre-distorted signal; and a parameter obtaining circuit configured to obtain second memory polynomial modeling information corresponding to an operating frequency band based on first memory polynomial modeling information corresponding to at least one frequency segment of a plurality of frequency segments of an entire frequency band on which the apparatus is configured to operate overlapping at least a portion of the operating frequency band, the parameter obtaining circuit configured to obtain the parameter set according to an indirect learning structure by using the second memory polynomial modeling information, and wherein the obtaining of the second memory polynomial modeling information corresponding to the operating frequency band based on the first memory polynomial modeling information includes generating at least one frequency shift vector based on a center frequency corresponding to each of the at least one frequency segment and a center frequency of the operating frequency band. The parameter obtaining circuit is configured to determine the at least one frequency segment and obtain the second memory polynomial modeling information based on the first memory polynomial modeling information corresponding to the determined at least one frequency segment.

2. The apparatus of claim 1, wherein, Each of the first memory polynomial modeling information and the second memory polynomial modeling information includes:

3. The apparatus of claim 2, wherein, an autocorrelation matrix and a cross-correlation vector, the autocorrelation matrix and the cross-correlation vector being used for a matrix equation that reduces a size of a difference between an intermediate signal based on the output signal and the pre-distorted signal according to an application of a Wiener filter. The parameter obtaining circuit is configured to:

4. The apparatus of claim 3, wherein, generate the at least one frequency shift vector based on a center frequency corresponding to each of the determined at least one frequency segment and a center frequency of the operating frequency band, and obtain an autocorrelation matrix corresponding to the operating frequency band based on the at least one frequency shift vector and an autocorrelation matrix corresponding to each of the determined at least one frequency segment. The parameter obtaining circuit is configured to:

5. The apparatus of claim 3, wherein, when the determined at least one frequency segment includes a first frequency segment and a second frequency segment, generate a first frequency shift vector based on a first center frequency corresponding to the first frequency segment and a center frequency of the operating frequency band, and obtain a first temporary autocorrelation matrix based on the first frequency shift vector and a first autocorrelation matrix corresponding to the first frequency segment; generate a second frequency shift vector based on a second center frequency corresponding to the second frequency segment and the center frequency of the operating frequency band, and obtain a second temporary autocorrelation matrix based on the second frequency shift vector and a second autocorrelation matrix corresponding to the second frequency segment; and obtain the autocorrelation matrix corresponding to the operating frequency band by performing a calculation of summing the first temporary autocorrelation matrix and the second temporary autocorrelation matrix. 6.The apparatus of claim 3, the parameter obtaining circuit is configured to: ​ wherein ​ generating the at least one frequency shift vector based on a center frequency of the operating frequency band and a center frequency corresponding to each of the determined at least one frequency segment, and obtaining a cross-correlation vector corresponding to the operating frequency band based on the at least one frequency shift vector and a cross-correlation vector corresponding to the determined at least one frequency segment.

7. The apparatus of claim 3, wherein, the parameter obtaining circuit is configured to: when the determined at least one frequency segment includes a first frequency segment and a second frequency segment, generate a first frequency shift vector based on a first center frequency corresponding to the first frequency segment and a center frequency of the operating frequency band, and obtain a first temporary cross-correlation vector based on the first frequency shift vector and a first cross-correlation vector corresponding to the first frequency segment; generate a second frequency shift vector based on a second center frequency corresponding to the second frequency segment and the center frequency of the operating frequency band, and obtain a second temporary cross-correlation vector based on the second frequency shift vector and a second cross-correlation vector corresponding to the second frequency segment; and obtain the cross-correlation vector corresponding to the operating frequency band by performing a calculation of summing the first temporary cross-correlation vector and the second temporary cross-correlation vector.

8. The apparatus of claim 1, further comprising a memory configured to store frequency segment information, the frequency segment information including the first memory polynomial modeling information and a center frequency corresponding to each of the plurality of frequency segments.

9. The apparatus of claim 1, wherein, the plurality of frequency segments each have a same frequency width.

10. The apparatus of claim 1, wherein, a center frequency segment among the plurality of frequency segments has a frequency width smaller than a frequency width of an edge frequency segment, the center frequency segment including a center frequency of the entire frequency band, and the edge frequency segment including at least one of a maximum frequency and a minimum frequency of the entire frequency band.

11. A method of processing a signal within an apparatus, the method comprising: determining at least one frequency segment including at least a portion of an operating frequency band of the apparatus from among a plurality of frequency segments divided from an entire frequency band over which the apparatus is configured to operate; obtaining an autocorrelation matrix corresponding to the operating frequency band and a cross-correlation vector corresponding to the operating frequency band based on frequency segment information corresponding to the determined at least one frequency segment; and generating an output signal by performing pre-distortion on an input signal using a coefficient matrix obtained based on the autocorrelation matrix corresponding to the operating frequency band and the cross-correlation vector corresponding to the operating frequency band, wherein the step of obtaining the autocorrelation matrix corresponding to the operating frequency band and the cross-correlation vector corresponding to the operating frequency band includes generating at least one frequency shift vector based on a center frequency of the operating frequency band and a center frequency corresponding to each of the determined at least one frequency segment.

12. The method of claim 11, Also included are: the plurality of frequency segments are determined by dividing the entire frequency band at equal intervals; and storing in a memory in the apparatus a center frequency corresponding to each of the plurality of frequency segments and an autocorrelation matrix and a cross-correlation vector corresponding to each of the plurality of frequency segments.

13. The method of claim 11, Also included are: dividing the plurality of frequency bands such that a frequency band closer to a center of the entire frequency band has a smaller frequency width than a frequency band farther from the center of the entire frequency band; and storing, in a memory of the apparatus, a center frequency corresponding to each of the plurality of frequency bands and an autocorrelation matrix and a cross-correlation vector corresponding to each of the plurality of frequency bands.

14. The method of claim 11, wherein the step of obtaining the autocorrelation matrix corresponding to the operating frequency band and the cross-correlation vector corresponding to the operating frequency band includes: obtaining the autocorrelation matrix corresponding to the operating frequency band based on the at least one frequency shift vector and the autocorrelation matrix corresponding to each of the determined at least one frequency band.

15. The method of claim 14, wherein the determined at least one frequency band includes a first frequency band and a second frequency band, wherein the step of generating the at least one frequency shift vector includes: generating a first frequency shift vector based on a first center frequency corresponding to the first frequency band and a center frequency of the operating frequency band; and generating a second frequency shift vector based on a second center frequency corresponding to the second frequency band and the center frequency of the operating frequency band, wherein the step of obtaining the autocorrelation matrix corresponding to the operating frequency band includes: obtaining a first temporary autocorrelation matrix based on the first frequency shift vector and a first autocorrelation matrix corresponding to the first frequency band; obtaining a second temporary autocorrelation matrix based on the second frequency shift vector and a second autocorrelation matrix corresponding to the second frequency band; and obtaining the autocorrelation matrix corresponding to the operating frequency band by performing a calculation of summing the first temporary autocorrelation matrix and the second temporary autocorrelation matrix.

16. The method of claim 15, wherein, the step of obtaining the first temporary autocorrelation matrix includes: performing a vector multiplication of the first frequency shift vector and a transposed version of the first frequency shift vector; and obtaining the first temporary autocorrelation matrix by performing a Hadamard product calculation using the first autocorrelation matrix and a result matrix of the vector multiplication.

17. The method of claim 11, wherein the step of obtaining the autocorrelation matrix corresponding to the operating frequency band and the cross-correlation vector corresponding to the operating frequency band includes: obtaining the cross-correlation vector corresponding to the operating frequency band based on the at least one frequency shift vector and the cross-correlation vector corresponding to each of the determined at least one frequency band.

18. The method of claim 17, wherein, the determined at least one frequency band includes a first frequency band and a second frequency band, wherein the step of generating the at least one frequency shift vector includes: generating a first frequency shift vector based on a first center frequency corresponding to the first frequency band and a center frequency of the operating frequency band; and generating a second frequency shift vector based on a second center frequency corresponding to the second frequency band and the center frequency of the operating frequency band, wherein the step of obtaining the cross-correlation vector corresponding to the operating frequency band includes: obtaining a first temporary cross-correlation vector based on the first frequency shift vector and a first cross-correlation vector corresponding to the first frequency band; obtaining a second temporary cross-correlation vector based on the second frequency shift vector and a second cross-correlation vector corresponding to the second frequency band; and obtaining a cross-correlation vector corresponding to the operating band by performing a calculation of summing the first temporary cross-correlation vector and the second temporary cross-correlation vector.

19. The method of claim 18, wherein, the step of obtaining the first temporary cross-correlation vector comprises: performing vector multiplication of the first frequency shift vector and the transposed first frequency shift vector; and obtaining the first temporary cross-correlation vector by performing Hadamard product calculation using the first cross-correlation vector and a result matrix of the vector multiplication.

20. An apparatus configured to perform wireless communication, the apparatus comprising: a memory configured to store a plurality of pieces of frequency band information of a plurality of frequency bands and instructions of operations of the apparatus; a processor configured to perform pre-distortion on an input signal in a given frequency band and generate a pre-distorted input signal by executing at least one instruction among the instructions stored in the memory; and a power amplifier configured to generate an output signal by amplifying the pre-distorted input signal, wherein the processor is configured to: determine at least one frequency band including at least a portion of the given frequency band among the plurality of frequency bands, and perform pre-distortion calculation on the input signal by using a coefficient matrix obtained by using frequency band information corresponding to the determined at least one frequency band among the plurality of pieces of frequency band information, wherein the processor is configured to obtain an autocorrelation matrix corresponding to the given frequency band based on a center frequency corresponding to the determined at least one frequency band, an autocorrelation matrix corresponding to the determined at least one frequency band, and a center frequency corresponding to the given frequency band, and wherein obtaining the autocorrelation matrix corresponding to the given frequency band comprises generating at least one frequency shift vector based on the center frequency corresponding to the determined at least one frequency band and a center frequency of the given frequency band.

21. The apparatus of claim 20, wherein each of the plurality of pieces of frequency band information comprises: a center frequency corresponding to one of the plurality of frequency bands, an autocorrelation matrix corresponding to the one of the plurality of frequency bands, and a cross-correlation vector corresponding to the one of the plurality of frequency bands.

22. The apparatus of claim 21, wherein, the processor is configured to: obtain a cross-correlation vector corresponding to the given frequency band based on a center frequency corresponding to the determined at least one frequency band, a cross-correlation vector corresponding to the determined at least one frequency band, and a center frequency corresponding to the given frequency band, and obtain the coefficient matrix based on the autocorrelation matrix corresponding to the given frequency band and the cross-correlation vector corresponding to the given frequency band.

Images