Predistortion circuit for a wireless transmitter and method of generating a predistorted baseband signal

Abstract

This application relates to a predistortion circuit for a wireless transmitter and a method for generating a predistorted baseband signal. Specifically, the invention discloses a predistortion circuit for a wireless transmitter, comprising a signal input configured to receive a baseband signal. Furthermore, the predistortion circuit includes a predistorter configured to generate a predistorted baseband signal using the baseband signal and a selection of a first predistorter configuration and a second predistorter configuration.

Description

[0001] This application is a divisional application of Chinese Patent Application No. 201780097294.6, filed on December 29, 2017, entitled "Predistortion Circuit of Wireless Transmitter and Method for Generating Predistortion Baseband Signal". Technical Field

[0002] This disclosure relates to a predistortion circuit for a wireless transmitter and a control circuit for the predistortion circuit. Background Technology

[0003] As a related technology, US2017353163A1 discloses a method for predistorting the input signal of an amplifier device, the method comprising evaluating selection criteria for a computational model of the amplifier device. This computational model provides the output signal of the amplifier device for the input signal. Furthermore, the method includes selecting between a first computational model and a second computational model of the amplifier device based on the evaluated selection criteria. Additionally, the method further includes predistorting the input signal of the amplifier device using the selected computational model. Summary of the Invention

[0004] In one aspect, a predistortion circuit for a wireless transmitter is provided, comprising: a signal input configured to receive a baseband signal; and a predistorter including a plurality of predistortion subcircuits, each predistortion subcircuit configured to generate a corresponding intermediate predistortion signal based on the received baseband signal, wherein the predistorter is configured to generate a predistorted baseband signal based on one or more of the corresponding intermediate predistortion signals according to a predistorter configuration that selectively activates corresponding unique combinations of predistortion subcircuits and a selection from at least a first predistorter configuration and a second predistorter configuration.

[0005] In another aspect, a predistortion circuit for a wireless transmitter is provided, comprising: a signal input configured to receive a baseband signal; a first sub-circuit for calculating a first portion of a predistorted baseband signal based on the received baseband signal and generating a corresponding first intermediate predistortion signal having a first rate; and a second sub-circuit for calculating a second portion of the predistorted baseband signal based on the received baseband signal and generating a corresponding second intermediate predistortion signal having a second rate, wherein the predistortion circuit is configured to generate the predistorted baseband signal based on any one or more of the first intermediate predistortion signal and the second intermediate predistortion signal.

[0006] In another aspect, a method for generating a predistorted baseband signal for a wireless transmitter is provided, comprising: selecting a predistorter configuration from at least a first predistorter configuration or a second predistorter configuration by selectively activating corresponding unique combinations of predistorter sub-circuits in a plurality of predistorter sub-circuits, wherein each predistorter sub-circuit performs a corresponding predistortion calculation; and generating the predistorted baseband signal from the baseband signal according to the selected predistorter configuration. Attached Figure Description

[0007] The following will describe some examples of equipment and / or methods by way of example and with reference to the accompanying drawings, wherein

[0008] Figure 1 An example of a predistortion circuit for a wireless transmitter is shown;

[0009] Figure 2 An example of predistortion circuitry within the transmitter is shown;

[0010] Figure 3 Examples of operational features that can be used to select between at least two predistorter features are shown;

[0011] Figure 4 A flowchart illustrating an example of a method for generating predistorted baseband signals is shown;

[0012] Figure 5 An example of a predistortion circuit with computing nodes operating at different rates is shown;

[0013] Figure 6 Another example of a predistortion circuit with computing nodes operating at different rates is shown;

[0014] Figure 7 An example of how aliasing degradation of a predistorted baseband signal can be mitigated through a predistortion circuit is shown.

[0015] Figure 8 An example of a control circuit is shown;

[0016] Figure 9 An example of the control circuitry within the transmitter is shown;

[0017] Figure 10 An example of a signal spectrum is shown;

[0018] Figure 11 An example of a control circuit configured to compare a reference signal and a feedback signal is shown;

[0019] Figure 12 An overview of the method for comparing reference signal 1210 and feedback signal 1220 is shown;

[0020] Figure 13 It shows according to Figure 11 A graphical representation of the generation of a band-limited signal;

[0021] Figure 14 An example of a specific implementation of millimeter waves is shown;

[0022] Figure 15 The system of equations to be optimized to determine the parameters of the predistorter is shown;

[0023] Figure 16 An example of a device for predistorting baseband signals is shown;

[0024] Figure 17 The signal sampled at the first sampling rate is shown;

[0025] Figure 18 The signal sampled at the second sampling rate is shown;

[0026] Figure 19 The predistorted baseband signal is shown;

[0027] Figures 20 to 22 Another example of a signal spectrum is shown;

[0028] Figure 23 An example of a wireless transceiver is shown;

[0029] Figure 24 A flowchart illustrating an example of a method for predistorting baseband signals is shown; and

[0030] Figure 25 An example of a mobile device using pre-distortion is shown. Detailed Implementation

[0031] Various examples will now be described more fully with reference to the accompanying drawings, some of which illustrate certain aspects. In the drawings, the thickness of lines, layers, and / or regions may be magnified for clarity.

[0032] Therefore, while the other examples are capable of various modifications and alternative forms, some specific examples are shown in the accompanying drawings and will be described in detail thereafter. However, this detailed description does not limit the other examples to the specific forms described. The other examples may cover all modifications, equivalents, and alternative forms falling within the scope of this disclosure. Similar figures throughout the specification refer to similar or analogous elements that may be implemented in the same or modified form when compared with each other, while providing the same or similar functionality.

[0033] It should be understood that when an element is referred to as "connected" or "coupled" to another element, the element may be directly connected or coupled, or connected or coupled via one or more intermediary elements. If "or" is used to combine two elements A and B, it should be understood that all possible combinations are disclosed, i.e., only A, only B, and A and B. An alternative wording for the same combination is "at least one of A and B." The same applies to combinations of more than two elements.

[0034] The terminology used herein is for the purpose of describing particular examples only and is not intended to limit other examples. Whenever the singular forms such as “a,” “an,” and “the” are used and only a single element is used that is not expressly or implicitly defined as mandatory, other examples may use multiple elements to achieve the same functionality. Similarly, when a function is subsequently described as being implemented using multiple elements, other examples may use a single element or processing entity to achieve the same functionality. It should also be understood that the term “comprising”, when used, specifies the presence of the stated features, integers, steps, operations, processes, actions, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, processes, actions, elements, components, and / or any grouping thereof.

[0035] Unless otherwise specified, all terms (including technical and scientific terms) are used herein with their common meaning in the field to which the examples belong.

[0036] Figure 1 An example of a predistortion circuit 100 for a wireless transmitter is schematically shown. The predistortion circuit 100 includes a signal input 110 configured to receive a baseband signal 102. A predistorter 120 is configured to generate a predistorted baseband signal 104 using the baseband signal 102 and a selection of a first predistorter configuration 120a or a second predistorter configuration 120b.

[0037] Predistortion can be used to linearize the output of a power amplifier used to amplify a wireless signal before transmission. The power amplifier of a wireless transmitter introduces nonlinearity based on the baseband signal to be amplified and other operating characteristics of the power amplifier. For example, if the power amplifier operates in saturation (in envelope tracking (ET)), it tends to exhibit more nonlinearity than in an operating mode that does not operate in saturation (using average power tracking (APT)). The predistortion circuit applies a predistortion function to the baseband signal to modify it, thereby anticipating and pre-compensating for the nonlinearity of the power amplifier. If the predistortion function works perfectly, the baseband signal is converted into a predistorted baseband signal such that the output of the power amplifier used to amplify the baseband-based wireless transmission signal (e.g., an RF signal generated by up-mixing the predistorted baseband signal) is not degraded by the power amplifier's nonlinearity. In other words, the nonlinearity of the power amplifier is anticipated and inversely superimposed on the baseband signal to produce an amplified RF signal with the desired signal characteristics and quality. In other words, in order to linearize the signal at the PA output, the predistortion circuit needs to create an IM spectrum at its output that eliminates the IM spectrum of the PA.

[0038] Examples of predistortion circuits allow selection between at least two predistorter configurations to generate a predistorted baseband signal, while the predistorter functions corresponding to the predistorter configurations can have different complexities.

[0039] In the first predistorter configuration 120a, the predistorter circuit 100 executes a first predistortion function, and in the second predistorter configuration 120b, it executes a second predistortion function. The predistortion functions are used to modify the baseband signal and generate a predistorted baseband signal.

[0040] The complexity of a predistorter configuration or the predistortion function executed within that configuration is given, for example, by the number of multipliers and / or adders, or typically by the number of computation nodes used to execute the predistortion function within the currently selected predistorter configuration. Computation nodes within a predistorter can be hardware entities or software components performing single basic computations (e.g., addition or multiplication). Therefore, more complex predistortion functions have more computation nodes than simpler predistortion functions.

[0041] The complexity of the required predistortion function is driven, for example, by the transmission bandwidth of the transmission chain and memory contributions. For instance, if the transmission bandwidth is higher, wideband linearization may require a higher number of multipliers. The same applies to memory contributions. If the transmission chain includes strong memory effects (e.g., due to ET operations, particularly at the band edges where the filter skirt begins), the complexity of the required predistortion function can increase further, leading to a further increase in the number of multipliers / computing nodes.

[0042] On the other hand, in the case of complex models, predistortion can have a significant impact on current consumption KPIs (key performance indicators). For example, the number of multipliers can determine the current consumption of the predistortion circuit. In conventional approaches, the complexity of the predistorter configuration is driven by the worst-case waveform supported by the transmitter (e.g., maximum transmission bandwidth, constellation density, peak-to-average power ratio), the worst-case memory contribution of the TX chain, the required linearization bandwidth, etc., making a single predistorter configuration suitable for all possible scenarios. However, for less complex waveforms supported by the wireless transmitter, a predistorter configuration with lower complexity may be good enough to achieve, for example, the linear performance or signal quality requirements defined by ACLR and EVM.

[0043] according to Figure 1 The example shown illustrates a predistorter configured to use one of at least two configurations that allow for lower power consumption for a given operating characteristic while meeting all signal quality requirements. Such a simpler configuration can be chosen to conserve energy if a simpler configuration of the predistortion circuitry is sufficient to amplify a given waveform (depending on the current operating characteristics of the wireless transmission circuitry). However, the predistortion function generates a predistorted baseband signal that meets all signal quality requirements.

[0044] According to some examples, a first number of computing nodes are active in a first predistorter configuration, and a second number of computing nodes are active in a second predistorter configuration. Assuming the second predistorter configuration is more complex, the second number of computing nodes can be higher than the first number. If a simpler configuration of the predistorter is sufficient to amplify the waveform of the baseband signal (e.g., the first predistorter configuration can be a simpler configuration than the second predistorter configuration), then the predistorter circuit 120 can switch from the second predistorter configuration to the first predistorter configuration. This flexibility in switching to a simpler configuration of the predistorter (e.g., the first predistorter 120a here) allows for energy savings because the first predistorter configuration requires fewer computing nodes to generate the predistorted baseband signal.

[0045] Existing mobile devices already support many different modulation schemes, ranging from relatively simple 3G voice signals to more complex waveforms such as LTE-60 with 256QAM. The introduction of 5G NR and the upcoming enhancements to the LTE standard will further widen the gap between the “simplest,” least complex waveforms and the most complex waveforms. Therefore, the unnecessary power consumption of conventional predistorters designed to support worst-case or most complex waveforms may increase with future standards. In conventional approaches, the current consumption of the predistortion block or circuitry will be unnecessarily high for a large number of supported waveforms (e.g., LTE signals with low RB allocation), wasting battery current and degrading results in key KPI tests.

[0046] According to some examples, the predistortion circuit 100 includes configuration processing circuitry configured to select a first predistorter configuration or a second predistorter configuration based on the operating characteristics of the wireless transmitter. The operating characteristics available for selecting the predistorter configuration may also depend on the operating modes of other components within the wireless transmitter and / or the characteristics of the baseband signal to be transmitted, as subsequently relative to... Figure 2 The subject of discussion. Figure 2 An example of predistortion circuitry within the transmitter is shown, which allows the predistorter complexity to be adjusted in response to predefined criteria. This reduces current consumption, enables shorter calibration times, and saves memory space used to store digital predistortion (DPD) coefficients.

[0047] like Figure 2 As generally illustrated, the specific implementation of a DPD system includes several tasks, such as predistortion of the baseband signal in the forward path, monitoring and capturing the distorted signal using the observation path, and learning the predistortion function based on a comparison between reference data and data captured by the observation path. In the predistortion system, the coefficients corresponding to the currently selected predistorter configuration (implementing the corresponding predistortion function) can be dynamically updated in this way. For example... Figure 2 As shown, the predistortion circuit 250 according to the example is located in the forward path, and therefore the corresponding techniques described below can be implemented.

[0048] The observation path primarily includes an observation block 210 connected to the RF front-end subsystem 220. The observation block 210 receives a portion of the distortion power amplifier (PA) signal 212 as its input signal 214. The distortion PA signal 212 can be captured via a coupler 222, which is part of the RF front-end subsystem 220. The input signal 214 is down-converted to the baseband (BB) domain and further processed, for example, by performing time alignment, scaling, offset removal, etc., with reference data within the signal conditioning block 230. The processed input signal 216, still containing the distortion of the TX signal, is used as the first input to the DPD update block 240. The corresponding reference signal 218, or the corresponding reference data, is the second input to the DPD update block 240. The reference signal corresponds to the baseband signal used to generate the distortion PA signal 212. In another example, the reference signal may be derived from the output 252 of the pre-distortion circuit 250. Examples of such alternative configurations will follow later. Figure 11 As shown in the image.

[0049] DPD update block 240 updates the predistortion function (e.g., parameters used in a particular configuration of predistortion circuit 250) of the currently used predistorter configuration by comparing the sequence of processed input signals 216 with the time-aligned sequence of reference signals 218.

[0050] The configuration processing circuit 280 determines the appropriate predistorter configuration to be used in the predistortion circuit 250 based on at least one of various operating characteristics of the wireless transmitter. The new predistorter configuration is transmitted to the predistortion block or circuit 250 and to the DPD update block 240. The predistortion circuit 250 may require the new predistorter configuration to select an appropriate or correct predistortion function and apply the correct predistortion function to generate the predistorted baseband signal. The predistorted baseband signal is provided to the RF signal generation circuit 270, which up-converts the predistorted baseband signal to an RF signal and provides the RF signal to the PA 260 for amplification. The DPD update block 240 requires the new configuration to determine and update the coefficients used by the new predistorter configuration within the predistortion circuit 250. In other words, the configuration processing circuit 280 is configured to select either a first predistorter configuration or a second predistorter configuration based on the operating characteristics of the wireless transmitter.

[0051] According to some examples, the operating characteristics considered by the configuration processing circuit 280 include at least one of the following: average power tracking mode, envelope tracking mode, output power range, peak-to-average power ratio of the baseband signal, peak-to-average power ratio of the input signal 214 (e.g., different predistorter functions may be beneficial if the PA is saturated or if significant waveform truncation is introduced due to insufficient power margin for the instantaneous peaks of the waveform), modulation scheme for generating the baseband signal, antenna matching conditions, transmission bandwidth, transmission band, transmission frequency range within the transmission band, number of transmission clusters in the frequency domain, frequency spacing between transmission clusters, bandwidth of each transmission cluster, and acceptable spectral mask. While some examples may use only one of the above operating characteristics, other examples may use any combination of them to infer the predistorter configuration to be used. For example, various standards may receive different weights before combining them to infer an appropriate predistorter configuration.

[0052] Figure 3Examples of how decisions regarding different configurations can be based on bandwidth standards (LTE-40 / -60 or LTE-20 / 15 / 10 / 5 / 3 / 1.4) and mode standards (APT / ET) are shown. Both graphs show the transmit power on the y-axis and the frequency resources used on the x-axis. The upper graph shows an LTE 20 configuration with bandwidth of 20MHz and lower. The predistorter configuration depends on the transmit bandwidth and DC / DC converter mode, which can be APT or ET. At lower transmit power levels below the first threshold 303, the PA can operate in APT mode because the overall power consumption is still modest. In this setting and for the relatively uncomplex signal waveform of the LTE 20 signal, predistortion may be completely unnecessary, considering that the PA in APT mode may not exhibit strong nonlinear effects. For the intermediate power 304 between the first threshold 303 and the second threshold 305, the PA may need to operate in ET mode, resulting in stronger nonlinearity. This may require the use of a moderately complex predistorter model, such as a model with 3 coefficients, like... Figure 3 As shown in column 308 of the upper curve. For high transmission power 306 above the third threshold 305, predistortion with the complexity corresponding to the ET mode at intermediate power may also be required.

[0053] The lower graph illustrates an LTE 40 / 60 configuration with a baseband signal bandwidth of 40MHz and above. Similar to the upper graph, the PA mode changes from APT for the lower power ranges 312, 314, and 316 to ET for the higher power ranges 318 and 320. Due to the complex waveforms of high-bandwidth LTE-40 / 60 signals, the APT mode may already require predistortion with moderate complexity. Figure 3 In the example shown, the corresponding predistorter function has 8 parameters, as shown in column 322. LTE40 / 60 signals at high transmission powers 318 and 320 require the most complex predistorter configurations, where the PA operates in ET mode, and where the predistorter function has 25 parameters. Figure 3 In the examples, the most complex predistorter configurations with transmit powers of 318 and 320 can consume more than three or four times the current or power of the simplest configuration of the intermediate power 304 in LTE 20 and below. Examples using the predistorter circuitry described herein can allow for incremental current savings (e.g., the difference between the 25 parameters of power range 320 or 318 and the 3 parameters of power range 320) because they allow switching between different predistorter configurations, each sufficient for the current setup of the wireless transmitter. The potential savings could increase significantly as the bandwidth of upcoming and future communication standards (e.g., 5G) further increases.

[0054] In another example, configuration processor 280 may further modify the predistorter configuration matching a certain standard based on feedback from DPD update block 240, such as based on optimized residual error, matrix conditions, etc. In these examples, configuration processing circuitry 280 also includes an input interface configured to receive feedback signals based on the output of transmitter power amplifier 260, wherein configuration processing circuitry 280 is further configured to select a first predistorter configuration or a second predistorter configuration based on the feedback signals.

[0055] While the examples of predistortion circuits are not limited to specific implementations of different predistortion functions corresponding to different predistortion configurations, a particular specific implementation is subsequently described. In the following examples, a predistorter configuration based on Volterra series is used, without limiting the scope of other examples.

[0056] The following equation shows the general baseband (BB) representation of a predistorter based on the Volterra series.

[0057] and

[0058] N is the memory depth, P is the kernel order, and K is the maximum order. Again, note that the examples are not limited to a specific polynomial representation. The polynomial representations above are merely examples to better understand the use of different predistorter configurations.

[0059] In this particular example, a single configuration corresponds to a set of Volterra kernels actually used. The Volterra kernels are thus characterized by the order of the polynomial p, the memory depth N, and the time index.

[0060] Example :

[0061] x(n-k1)*|x(n-k1)| p-1

[0062] x(n-k1)*|x(n-k2)| p-1

[0063] x(n-k1)*x(n-k2)*x(n-k3) *

[0064] Each core is weighted by coefficients h(p)k1..kp. The sum of the cores generates a predistorted baseband signal 252, which is used to linearize the transmission chain by pre-compensating the nonlinearity of the power amplifier 260, which receives the RF signal after it has been generated in the RF generation block 270.

[0065] However, changing the coefficients does not generate a new configuration because the coefficients are continuously updated by DPD update block 240 to achieve optimal predistortion with the currently selected predistorter configuration. Therefore, updating the coefficients from a first non-zero value to a second non-zero value changes the predistortion function, but is not considered a new configuration in the context of this example.

[0066] Changing a coefficient from a non-zero value to zero, or vice versa, can be interpreted as changing the configuration, since a coefficient with a zero value means that a kernel is not being used. In the context of the example, the first configuration differs from the second configuration if the first set of active kernels (with a first number of compute nodes) associated with the first configuration differs from the second set of active kernels (with a second number of compute nodes) associated with the second configuration. Therefore, an active kernel means that a kernel is being used and contributing to the signal generated by the predistortion circuitry.

[0067] While examples of predistortion circuitry can be used in all specific implementations where predistortion of signals is desired, the following considerations detail applications in the field of wireless communications. Some examples of implementing digital predistortion (DPD) to support LTE upgrades or 5G modulation schemes. 5G NR (New Radio) is a new communications standard expected to begin commercial deployment in 2020. 5G NR will be characterized by higher bandwidth and more complex modulation schemes in the uplink compared to, for example, 4G. For example, mobile stations may need to support aggregated bandwidths of up to 200MHz (2x100MHz) below 6GHz and above 1GHz in the mmW range (>24GHz). Baseline uplink modulation schemes that can be used in 5G NR are DFT-s-OFDM (which is similar to SC-FDMA used in the LTE uplink) or CP-OFDM.

[0068] As already mentioned, the examples are not limited to modulation schemes based on LTE or 5G NR standards. Other examples also apply to any modulation scheme that imposes strict linearity requirements on the TX chain.

[0069] In summary, at least two challenges have emerged for mobile terminals that support 5G NR:

[0070] Higher channel bandwidth (up to 200MHz) in the sub-6GHz range, and expected to be greater than 1GHz in the mmW range.

[0071] - Signals with a higher peak-to-average power ratio (PAPR) can be used by employing methods similar to SC-FDMA.

[0072] High-efficiency modulation schemes using 256QAM and OFDM are employed to increase throughput.

[0073] More complex signals with higher PAPR and higher constellation density (e.g., 256QAM, 1024QAM) place more stringent demands on the linearity of the TX chain. Higher linearity requirements mean fewer AMAM and AMPM transitions. This is typically achieved by providing sufficient PA margin to prevent modulation peak truncation and by flattening the PA's AMPM response through increased PA quiescent current. However, both of these measures significantly increase the PA's current consumption. Digital predistortion (e.g., in the baseband domain), according to one example, can be used to achieve the linearity requirement while mitigating the increase in PA current consumption. Appropriate predistortion of the PA input signal requires less PA margin and PA quiescent current while maintaining sufficient linearity to meet, for example, ACLR (Adjacent Channel Leakage Ratio) and EVM (Error Vector Magnitude) targets.

[0074] Figure 4 A flowchart illustrating an example of a method for generating a predistorted baseband signal for a wireless transmitter is shown schematically. The method includes selection 410 for selecting a first predistorter configuration or a second predistorter configuration, and predistortion 420 for generating a predistorted baseband signal using the baseband signal and the selected predistorter configuration.

[0075] The above considerations rely on the following principles, and can be used to determine additional decision criteria for selecting predistorter configurations. Increasing transmission bandwidth also increases the complexity of the predistorter because the linearity of the TX chain depends on the instantaneous RF frequency and the rate of change of the envelope of the modulated RF signal. This is because the linearity depends on the IQ data stream of the modulated RF carrier. At higher bandwidths, the AMAM and AMPM responses are no longer constant and depend on the sequence of modulation symbols. This is also known as the memory effect, and becomes more pronounced at higher transmission bandwidths. Several potential effects introduce the memory effect into the TX chain. The effects are partially different for ET (Envelope Tracking) and APT (Average Power Tracking) systems. Compared to APT operation, ET introduces significant nonlinear effects in the transmission chain, and these nonlinear effects are partially dependent on the RF frequency. Therefore, the nonlinearity can vary over several MHz, resulting in a dispersed TX channel. In a dispersed TX channel, the predistortion characteristics are strongly mapped to the absolute RF frequency. If the channel is distributed, and if the second frequency range is offset by several MHz, the DPD coefficients of a given predistortion function optimized for the first frequency range will show a smaller linear improvement in the second frequency range. To overcome the frequency dependence of AMAM and AMPM responses, memory predistortion is required.

[0076] Some of the related effects of distributed channels in ET mode are as follows:

[0077] - ET delay dispersion in frequency

[0078] - Insufficient tracker bandwidth cuts through the envelope bandwidth and introduces latency variations.

[0079] -VCC bandwidth in PA module is too low

[0080] - This effect is attributed to the PA load pull of the TX filter: in ET systems with PAs operating in compression (where the ET system exhibits higher load sensitivity), this effect is more pronounced. The TX filter is characterized by its input impedance (=PA load impedance), which has a high frequency dependence due to the resonator used to shape the filter characteristics.

[0081] ET delay dispersion and load pull attributable to the TX filter are the dominant effects of channel dispersion. Although both effects also occur in the center of the TX band, they become more pronounced at the band edges as we transition from the passband to the stopband.

[0082] At high transmission bandwidths (e.g., >400MHz), if it may be difficult to implement an ET mode that supports >400MHz with reasonable efficiency and system complexity, then APT operation can be a PA provisioning scheme. In APT, at least two main effects introduce memory into the TX chain:

[0083] This effect is attributed to the PA load pull of the TX filter. It also occurs in ET, but is less noticeable in APT operation where the PA operates linearly.

[0084] - Limited PA bias bandwidth: This effect also occurs in ET systems, but is usually masked by the VCC bandwidth limitation in ET systems. In APT systems, this effect is more pronounced. A well-designed PA bias network provides low impedance from DC up to the BB modulation bandwidth. Low impedance is required to avoid remodulation effects, which introduce intermodulation errors into the spectrum of the PA output signal.

[0085] Signal degradation caused by finite PA bias bandwidth depends on the envelope of the RF signal and the rate of envelope change.

[0086] The PA load pulling effect depends on the instantaneous RF frequency and the impedance response of the TX filter across the modulation bandwidth. When the instantaneous modulation frequency is at the first frequency, the PA is loaded by the first impedance; when the instantaneous modulation frequency is at the second frequency, the PA is loaded by the second impedance. These two impedances can be quite different, especially when the interval between the instantaneous frequencies is large, which can occur for signals with high transmission bandwidths. Therefore, the AMAM and AMPM responses obtained at the first frequency differ from those at the second frequency.

[0087] To linearize the PA output in such cases, the predistortion circuit needs to create an IM spectrum that eliminates the intermodulation (IM) spectrum of the PA at its output. Assuming a transmission bandwidth of 1 GHz, the requirements for the predistortion circuit are illustrated exemplarily. The bandwidth (→x^3) of the IM3 spectrum is therefore 3 GHz, and the bandwidth (→x^5) of the IM5 spectrum is 5 GHz. The predistorter needs to generate the IM3 and IM5 spectra in the BB domain to achieve the desired linearization of the PA output. For example, if both the IM3 and IM5 spectra should be eliminated, the required bandwidth in the BB domain is + / - 2.5 GHz. Therefore, the computational nodes (e.g., multipliers) used to generate the predistorted signal need to operate at a sampling rate of 5 GHz to prevent the IM5 spectrum from overlapping (aliasing) with its frequency domain counterpart.

[0088] However, compute nodes or multipliers operating at 5 GHz may draw a large amount of current.

[0089] Figure 5 The example of the predistortion circuit 500 shown allows for the deployment of memory predistortion in transmission systems characterized by high transmission bandwidth, while reducing the current consumption and design complexity of the predistortion circuit 500. Figure 5 An example is shown where different predistorter configurations can be distinguished by the operating rates of the compute node groups within the different configurations.

[0090] exist Figure 5 In the example, the predistortion circuit 500 is divided into three parts or sub-circuits 510, 520, and 530 (also referred to as blocks). This can facilitate a single predistorter configuration or different predistorter configurations in any combination of sub-circuits 510, 520, and 530. The computation nodes of each sub-circuit 510, 520, and 530 operate at different sampling rates. Specifically, a first number of computation nodes are configured to operate at a first rate, and a second number of computation nodes are configured to operate at a second rate, which is higher than the first rate. Another example may have two sub-circuits that are selectively or jointly activated for the predistorter configuration. More than three sub-circuits may also be appropriate depending on the order of the predistortion function to be implemented and the IM spectrum to be considered. Another example is not limited to a specific number of blocks or specific mathematical operations performed by the blocks. The total predistortion function performed by the predistortion circuit 500 is decomposed into partial functions with different bandwidth requirements, and the partial functions are mapped to sub-circuits operating at different appropriate rates.

[0091] If all sub-circuits are used within a predistorter configuration, each sub-circuit generates partial signals 512, 514, and 516 as part of the predistortion signal 518. After rate matching is applied to equalize the different sampling rates, partial signals 512, 514, and 516 are combined at the outputs of sub-circuits 510, 520, and 530. For example, a sampling rate converter can be configured to match the outputs of a first number of computing nodes with a second rate to generate a rate-matched output for the first number of computing nodes. Furthermore, a combining circuit can be configured to combine the outputs of a second number of computing nodes with the rate-matched outputs of the first number of computing nodes. Figure 5 In this context, it is assumed that subcircuit 510 (#1) includes all linear terms that do not increase the bandwidth of the input signal, subcircuit 520 (#2) includes all 3rd-order kernels that generate a signal with 3 times the bandwidth of the input signal, and subcircuit 530 (#3) includes all 5th-order kernels that generate a signal with 5 times the bandwidth of the input signal. Therefore, subcircuit 520 operates at 3 times the sampling rate of subcircuit 510, and subcircuit 530 operates at 5 times the sampling rate of subcircuit 510 to avoid aliasing.

[0092] Therefore, the highest sampling rate and highest current consumption are only used by fifth-order cores that require the highest sampling rate to avoid aliasing. The same applies to third-order cores. The example of predistortion circuit 500 ensures that the compute nodes of predistortion circuit 500 operate only at the sampling rate required to prevent aliasing. This example of predistortion circuit 500 helps save energy compared to a single-rate implementation where all compute nodes (multipliers) would run at the highest rate. This example of predistortion circuit 500 also saves energy compared to a multiphase implementation that would increase the number of multipliers operating at lower sampling rates (which reduces some design challenges at high sampling rates but does not help reduce current consumption).

[0093] Compared to alternative methods, the current consumption of the predistorter circuit 500 is significantly reduced by running the computing nodes (e.g., multipliers) at different rates based on, for example, the signal bandwidth.

[0094] Figure 7 This illustrates how the spectrum of the predistorted baseband signal can be degraded due to aliasing if an insufficient rate is selected for the operation of the predistortion circuit. Figure 7 In this context, it is assumed that the third-order intermodulation distortion will be considered by the predistortion function. The main spectrum 710 of the predistorted baseband signal overlaps with the first copy 712 of the predistorted baseband signal, thereby degrading the quality of the predistorted baseband signal if the sampling rate of the selected computation node is too low. When using the predistorter configuration according to the example of the predistortion circuit, a sufficiently high sampling rate can be selected for each IM order to be considered without increasing the current consumption beyond what is required to achieve the goal. In other words, Figure 7This illustrates harmful interference in the IM spectrum caused by excessively low sampling rates. Figure 3 In this process, the overlap of the IM3 spectrum will degrade the linearization results because the overlap adds an undesirable IM contribution to the spectrum of the predistorter circuit.

[0095] Figure 6 Another example of the predistortion circuit 600 is shown. The first-order computation is performed at the lowest rate in the first sub-circuit 610 (#1), the second-order computation at a medium rate in the second sub-circuit 620 (#2), and the third-order computation at the highest rate in the third sub-circuit 630 (#3). (Except in...) Figure 5 In addition to the example, sub-circuits 610, 620, and 630 are connected in series and include sampling rate converters 612, 622, and 632 to adjust the sampling rate between subsequent sub-circuits in the series-connected sub-circuit chain.

[0096] In another example, the total number of multipliers is reduced by using intermediate results generated in the first block in at least the second block after proper rate matching. This is achieved through... Figure 5 The vertical connection between sub-circuits 510, 520 and 530 is shown.

[0097] In summary, digital predistortion of ultra-high bandwidth signals in mobile devices can be severely constrained by battery current consumption. In mobile devices, current consumption is critical for a good customer experience and for addressing the critical heat dissipation issues that occur in small form factor devices. For example, for very high bandwidth signals, a predistorter configuration with moderate complexity may already require 30 to 60 multipliers, which, depending on the specific technology and digital implementation, can draw hundreds of mA from the predistortion circuitry (in addition to the PA current).

[0098] The increased transmission bandwidth introduced by 5G NR will bring new challenges to mobile terminals. The increase in transmission bandwidth may outpace the deployment of new process nodes (e.g., 28nm, 16nm, 10nm) that artificially improve baseline current each year. As a negative consequence, the current consumption of conventional digital predistorters will also increase significantly despite the new process nodes. There is a strong need to introduce new methods and evolve digital circuits so that current consumption increases less with increasing transmission bandwidth. Otherwise, excessive current consumption of the digital components at ultra-high bandwidth will hinder or delay their deployment. The examples described in this article can mitigate the increase in current consumption.

[0099] In summary, the aforementioned methods and examples of predistortion circuits perform the computations required to generate a predistorted signal at at least two sampling rates, where the sampling rate depends on the bandwidth of the signal after applying specific mathematical operations for predistortion. Examples of predistortion circuits include at least two sub-circuits, each generating a portion of the predistorted signal, with the sampling rate of the first sub-circuit differing from that of the second sub-circuit.

[0100] Examples of predistortion circuits can facilitate the deployment of predistorted signals with ultra-high bandwidth, which will occur, for example, in 5G mmW applications. They reduce the increased current consumption and design complexity resulting from the upcoming increase in transmission bandwidth. Examples of predistortion circuits allow for the provision of mobile phones with small form factors and leading-edge battery current consumption.

[0101] For example, in 5G NR mobile terminals, predistortion circuitry is expected to have a significant impact on battery current (→ operating time) and power dissipation / heat generation (→ device size and form factor).

[0102] Suitable for implementation according to Figure 5 or Figure 6 A specific example of the predistortion function for a predistortion circuit is a function based on the Volterra series. The following equation shows the general BB representation of a Volterra-based predistorter.

[0103] and

[0104] N is the memory depth, P is the kernel order, and K is the maximum order. However, other examples are not limited to this particular polynomial representation. The polynomial representation is given merely to better illustrate the meaning of the x^3, x^5, and x^7 terms in a broader context.

[0105] While the preceding examples discussed the forward path of a system employing predistortion, the other examples discussed the observation path and control circuitry used to update the predistortion parameters in the predistortion circuitry for generating the predistorted baseband signal.

[0106] Figure 8An example of a control circuit 800 for a predistortion circuit is shown. The control circuit 800 for the predistortion circuit includes a feedback signal input 810 configured to receive a feedback signal 812 associated with the output of a power amplifier, and a bandwidth limiting circuit 820 configured to limit the bandwidth of the feedback signal 812 to derive a band-limited feedback signal 822. A parameter processing circuit 830 is configured to update the predistortion parameters used in the predistortion circuit based on the band-limited feedback signal 822. Optionally, another example may include an output interface 840 to output the updated predistortion parameters. The bandwidth limitation can be performed by any filter circuit in the digital or analog domain. For example, an FIR filter can be used in the digital domain.

[0107] Using the band-limited feedback signal 822 to update the predistortion parameters used in the predistortion circuit can significantly reduce the power consumption of the control circuit 800 and other components in the feedback or observation path, because power-consuming components that can operate at high bandwidth can be avoided.

[0108] However, conventional implementations use a continuous and broad spectrum for predistortion learning and compute updated predistortion parameters. Predistortion learning describes the process of generating a predistortion function designed to compensate for nonlinear effects introduced by the analog transmission system. For predistortion learning, an observation path is needed that captures a portion of the distorted RF output signal of the transmission system (e.g., generated by a power amplifier) ​​and down-converts it to the BB domain for further processing, such as... Figure 9 As shown. For example, for the future communication standards already described, compared to conventional methods, Figure 8 The energy savings in the example shown can be significant. Assuming a signal bandwidth of 1 GHz, the IM spectrum can exhibit a bandwidth of 5 GHz (covering the product of IM3 and IM5). Therefore, the observation path of a conventional solution must be flat in terms of delay and amplitude over a 5 GHz bandwidth in the RF domain, and, for example, a 5 GHz ADC sampling rate would be required to meet the Nyquist criterion. However, a 5 GHz ADC is complex and draws a large amount of current.

[0109] Conversely, this example of control circuitry provides a good solution for deploying memory predistortion in transmission systems characterized by high transmission bandwidth (e.g., 5G NR) while reducing the current consumption and design complexity of the predistortion implementation. The example shown subsequently discusses the observation of the signal output from the power amplifier and introduces the concept of reducing the complexity of the observation path.

[0110] Figure 9An example of control circuitry 940 within the transmitter circuitry is shown. The transmitter circuitry includes a predistortion circuitry 910 configured to receive a baseband signal 912 and generate a predistorted baseband signal 922. To generate the predistorted baseband signal 922, the predistortion circuitry uses a selected predistorter model. The predistortion function of the predistorter model is optimized for predistortion using dynamically updated predistortion parameters. The predistortion parameters are continuously updated by the control circuitry 940 based on feedback from the amplified radio frequency signal 932. An upmixer 920 uses the predistorted baseband signal 922 and a local oscillator signal to generate a radio frequency signal. A power amplifier 930 is coupled to the output of the upmixer 920 and generates the amplified radio frequency signal 932.

[0111] Control circuitry 940 for predistortion circuitry 910 includes a feedback signal input 941 configured to receive a feedback signal 934 associated with the output of power amplifier 930. According to some examples, feedback signal 934 is a copy of a low-power amplified radio frequency signal 932. Bandwidth limiting circuitry 942 is configured to limit the bandwidth of feedback signal 934 to derive band-limited feedback signal 936. Control circuitry 940 also includes a baseband signal input 943 configured to receive a baseband signal or reference signal 912. Additional bandwidth limiting circuitry 944 is configured to limit the bandwidth of baseband signal 912 to derive band-limited baseband signal 914. Parameter processing circuitry 946 is configured to update predistortion parameters used within predistortion circuitry 910 based on band-limited feedback signal 936 and band-limited baseband signal 914. Figure 9 In a specific example, the band-limited feedback signal 936 is further digitized by an ADC 945. Further signal shaping is performed using a signal conditioning circuit 947 and a sampling circuit 949 to enable direct comparison of the band-limited feedback signal 936 (as an output from the sampling circuit 949) and the band-limited baseband signal 914. For example, the signal conditioning circuit 947 and the sampling circuit 949 may perform sampling rate conversion and / or time alignment, allowing corresponding samples of the band-limited feedback signal 936 and the band-limited baseband signal 914 to be directly compared via a comparator circuit 946a (which is part of the parameter processing circuit 940). Based on this comparison, an update circuit 946b within the parameter processing circuit 946 calculates and updates the predistortion parameters, which are then passed to the predistortion circuit 910 to close the control loop and optimize the linearization implemented by the selected predistorter model.

[0112] In other words, Figure 9The transmitter circuit / communication system includes: a forward path, wherein digital predistortion is used to improve the linearity of the transmitted signal; an observation path coupled to the transmission path such that the coupled signal (feedback signal 934) includes nonlinear distortion of the transmission path; a comparator block that compares the coupled signal with a reference signal; and an update block that modifies the predistortion function in the forward path based on the result of the comparator block.

[0113] In summary, predistortion parameters are updated based on comparisons between feedback and reference signals at discrete frequency points within the transmission spectrum. The observation bandwidths of the feedback and reference signals can be selected to ensure that adjacent observation channels do not overlap. The reference signal can be derived from the baseband signal before predistortion, such as... Figure 9 As shown. Another example also shows how the reference signal can be derived from the predistorted baseband signal at the output of the predistortion circuit.

[0114] Figure 10 This demonstrates the concept of using discrete sample intervals in the frequency domain instead of the full spectrum to determine predistortion parameters. Two signals (a reference signal and a feedback signal) are sampled at the same frequency offset with the same measurement bandwidth. Figure 10 The spectrum within the transmission bandwidth 1010 and adjacent frequency portions 1020a and 1020b is schematically shown. The widths of the adjacent frequency portions 1020a and 1020b to be monitored depend on the order of the IM distortion to be considered.

[0115] Figure 10 A specific example of 13 observation channels 1030a to 1030m with limited bandwidth is shown. The predistortion parameters can be updated independently for each of the observation channels 1030a to 1030m. According to some examples, the observation channels 1030a to 1030m are then processed in a predetermined order to account for contributions from the entire spectrum used to update the predistortion parameters. The spacing between adjacent observation channels does not need to be equidistant. Depending on the response of the intermodulation spectrum, for example, if the change in the IM spectrum with frequency is more pronounced than the change in the IM spectrum with a second frequency offset, it may also be beneficial to choose a closer spacing at the first frequency offset. Figure 10 As shown, the reference spectrum of the reference signal and the coupled spectrum of the feedback signal are divided into blocks 1030a to 1030m, and the predistortion function (predistortion parameters) is updated based on the comparison of the blocks. According to some examples, the comparison blocks of the reference signal and the feedback signal have the same bandwidth. However, adjacent blocks may have different bandwidths.

[0116] like Figure 10As shown, the examples use narrowband signals to determine predistortion parameters. According to some examples, the measurement bandwidth is several megahertz, such as 1 MHz, 3 MHz, 5 MHz, or 10 MHz. Using band-limited signals significantly reduces complexity and current consumption. Furthermore, in some examples, the determination of predistortion parameters (DPD coefficients) can be performed at a lower rate, depending on the measurement bandwidth and, for example, the Nyquist formula. However, the predistortion circuitry or block that creates the predistorted baseband signal can itself operate at full rate to avoid aliasing of the IM spectrum generated by the predistorter circuitry.

[0117] Figure 11 An example is shown on how the corresponding frequency blocks can be generated. Figure 9 In contrast to the example, the reference signal is generated from the predistorted baseband signal at the output of the predistortion circuit 1110. Figure 9 Similar to the previous approach, it is desirable to reduce the bandwidth of the observation path by narrow-band sampling of the feedback and reference signals in the frequency domain. Therefore, the bandwidth of the components within the observation path does not need to be high enough to capture the complete feedback signal, resulting in reduced complexity and power consumption. (Similar to...) Figure 9 , Figure 11 An upmixer 1120 is shown that generates an RF signal using a local oscillator signal 1121. A power amplifier 1130 amplifies the RF signal. Control circuitry 1140 includes bandwidth limiting circuitry 1142 to limit the bandwidth of the feedback signal, thereby deriving a band-limited feedback signal 1146. Figure 11 In the example, the bandwidth limiting circuit 1142 includes an adjustable mixer 1143 configured to down-mix the feedback signal 1141 using a feedback oscillator signal 1144 to generate a down-mixed feedback signal 1145. The feedback oscillator signal 1144 has the frequency of the local oscillator signal 1121 plus the observed frequency offset Δf.

[0118] A bandpass filter circuit 1147 is configured to bandpass filter the downmixed feedback signal 1145 to generate a band-limited feedback signal 1146. Prior to filtering, the downmixed feedback signal 1145 is digitized by an ADC 1149. In summary, the bandwidth-limited feedback signal is generated by first downmixing the RF signal with the frequency of the feedback oscillator signal, and then bandpass filtering the signal to obtain an observation frequency block centered on the frequency of the feedback oscillator signal and having the bandwidth given by the bandpass filter.

[0119] The control circuit 1140 also includes a reference signal generation circuit 1150, which includes a reference signal input configured to receive a predistorted baseband signal 1111 from the output of the predistortion circuit 1110. The reference signal generation circuit 1150 also includes a bandwidth limiting circuit configured to limit the bandwidth of the reference signal to derive a band-limited reference signal. The additional bandwidth limiting circuit includes a frequency shifter 1151 configured to shift the predistorted baseband signal by the frequency offset; and an additional bandpass filter circuit 1153 configured to bandpass filter the shifted predistorted baseband signal 1111 to generate a band-limited predistorted signal 1114. The band-limited predistorted signal 1114 and the band-limited reference signal are created as corresponding frequency blocks within the spectrum such that the optimization circuit 1155 can directly compare the two signals to derive updated predistortion parameters. In other words, the additional bandwidth limiting circuit 1150 includes a frequency shifter 1151, which is configured to shift the reference signal by the reciprocal of the observed frequency offset.

[0120] In summary, to achieve proper frequency domain sampling, the LO frequency of the mixer in the observation path is scanned, and this LO frequency is typically set to the intermediate frequency of the frequency block. The LO frequency is scanned from block to block, the signal is down-converted, filtered (so that the reference signal / predistortion baseband signal and the coupled feedback signal undergo the same filtering), and fed to the optimization circuit 1155 (optimizer block).

[0121] When scanning the LO frequency, the LO phase at the new frequency may be unknown in some cases. An unknown LO phase will prevent the learning of predistortion coefficients because the optimization circuit 1155 cannot distinguish whether the observed phase shift is caused by the PA or by a change in the LO frequency. Some examples optionally include a switch 1170, which can be configured to connect an input signal from before PA 1130 to mixer 1143. In a first measurement, the input signal can then be connected to mixer 1143. Since the input phase then does not include PA phase distortion, the signal can be used to determine the LO phase without the contribution from PA 1130. In a second measurement, the output signal of PA feedback signal 1145 can then be measured as an input to optimizer circuit 1155. In other words, control circuit 1140 may also include a phase control input coupled to the output of the upmixer to receive a phase control signal, wherein control circuit 1140 is further configured to adjust the phase of the feedback signal based on the phase control signal.

[0122] Figure 12 Again, schematically showing in Figure 11 The example executes a process to construct a set of equations designed to determine the predistortion coefficients of the selected predistorter configuration.

[0123] During the optimization process, linear equations are solved, for example, using the least squares (LS) algorithm. Figure 15 An example of such a system of linear equations is shown for a predistortion function based on Volterra series.

[0124] The rows of matrix A are derived from different blocks or observation channels, with at least one row for each observation channel. For observation channels, data can be accumulated, such as... Figure 12 As shown. Reference signal 1210 and feedback signal 1220 can be collected and bandwidth-limited by means of filters 1230 and 1240, respectively. In order to be able to use samples of the two signals to calculate predistortion parameters, further matching of signal samples can be performed within optimization circuit 1255. For example, rate matching circuits 1262 and 1264 can be used to align the sampling rate. Optionally, phase calibration can be performed on the feedback signal, as shown above. For this purpose, an optional phase correction circuit 1266 for the local oscillator signal 1144 and a subsequent time alignment circuit 1268 can be present. After ambiguity compensation and correct time alignment, samples of reference signal 1210 and feedback signal 1220 can be used as Figure 15 The input to the linear equation shown.

[0125] Each row of matrix A includes reference data points predistorted by a predistortion function (from a band-limited predistortion baseband signal). Each predistorted data point in vector b is compared to a point in the coupled data / band-limited feedback signal of vector h. This is done for a large number (e.g., thousands) of data points, and the predistortion coefficients h... k They are determined to satisfy a certain metric, such as least squares error.

[0126] According to some examples, if low-pass filtered data is used, the optimization circuit 1155 can operate at a low rate.

[0127] Figure 13 It was shown again Figure 11 How can the bandwidth limiting circuit and other bandwidth limiting circuits work together to ensure that the corresponding frequency blocks of the predistorted baseband signal and feedback signal are used through the optimization circuit 1155?

[0128] Figure 14 This is an example of how it can be used in mmW applications. In mmW designs, an RF headend 1410 with key analog functions including transmission, reception, and antenna beamforming can be attractive. Since the communication device will have multiple headends for MIMO purposes, the RF and LO signals are provided by special coaxial cables. Therefore, the observation path can be multiplexed within the transmitter circuitry 1430.

[0129] In summary, the previous examples enable the practical use of predistortion for signals with extremely high bandwidth, which will occur, for example, in 5G mmW applications. This will reduce the increased current consumption and design complexity resulting from the upcoming increase in transmission bandwidth. These examples illustrate observation paths and introduce methods that reduce the complexity of observation paths by using multiple narrowband signals instead of a single wideband signal, which provides at least the following effects:

[0130] • Lower sampling rate of the ADC (100 times or more, depending on the signal bandwidth).

[0131] • The requirements for amplitude and frequency response of the observation path caused by narrowband signals are relaxed → less compensation effort is required in the digital domain to equalize the observation path and for less critical RF analog designs.

[0132] The requirements for delay error between the coupled signal and the reference signal are relaxed. Currently, a delay error of 1...3ns is permissible for a 60MHz bandwidth signal. For a 1GHz bandwidth, the permissible error will be <<1ns, making it difficult to find a feasible implementation for full-bandwidth monitoring.

[0133] Furthermore, cellular data transmission standards not only increase bandwidth but also output power. In typical implementations, mobile terminals can transmit at a maximum output power of 23 dBm. The upcoming Power Category 2 defines 26 dBm of output power for all TDD bands, and Power Category 1 defines 31 dBm of output power for the low-frequency Band 14. Due to high bandwidth, high channel frequency, power-saving envelope tracking technology, and higher output power, the time-dependent nonlinearity (i.e., memory effect) of analog circuits becomes increasingly pronounced. Because nonlinearity generates higher frequencies (harmonics), calculations related to the predistortion model must be performed at high sampling rates. Complex mathematical models (Volterra series), high word lengths (due to the exponentiation of harmonics), and high sampling frequencies (due to the frequency spread of harmonics) can lead to numerous multiplication operations, accompanied by significant current consumption.

[0134] Figure 16 An example of a device 1600 for predistorting a baseband signal 1601 is shown, which allows for a (significant) reduction in current consumption.

[0135] The apparatus 1600 includes a predistortion circuit 1610 configured to determine samples of a baseband signal 1601 at a first sampling rate. For example, the first sampling rate may be equal to or higher than the sampling rate of the baseband signal 1601. Furthermore, the predistortion circuit 1610 is configured to calculate and output samples of a predistorted baseband signal 1602 at a second sampling rate based on predistortion parameters (e.g., based on Volterra series) and the samples of the baseband signal 1601. The second sampling rate is lower than the first sampling rate. That is, the predistortion circuit 1610 uses an output sampling rate lower than the input sampling rate of the baseband signal 1601 to calculate the predistorted baseband signal 1602.

[0136] The device 1600 also includes an upsampling filter 1620, which is configured to calculate samples of the predistorted baseband signal 1602 at a first sampling rate based on samples of the predistorted baseband signal 1602 at a second sampling rate.

[0137] A higher first sampling rate allows tracking of all higher-order exponential harmonics in the baseband signal 1601, enabling the processing of the baseband signal 1601 to be known with sufficient accuracy. On the other hand, a lower second sampling rate allows for current savings, as predistortion is performed at this lower sampling rate. The upsampling filter then restores the input sampling rate of the predistortion circuit 1610. However, since the upsampling filter uses fixed coefficients and since no exponential order occurs, upsampling alone is much simpler to compute than predistortion at a higher sampling rate. Therefore, current is saved and reasonable harmonic attenuation is achieved.

[0138] This can be combined from the following Figures 17 to 20 This becomes even clearer with the given non-restrictive numerical examples.

[0139] Figure 17 It shows the expression |∑ n＝1...5 The analog signal 1710 has a period from –π to π, where exp(It) / n| is the length of one cycle. Signal 1710 consists of a carrier wave plus its attenuated 2nd, 3rd, 4th, and 5th harmonics. Furthermore, Figure 17 Digital samples of the signal, 1700, ..., 1709, are shown. Figure 17 In the example, signal 1710 was oversampled ten times.

[0140] Figure 18 The same signal 1710 is shown. However, in Figure 18 In the example, signal 1710 is oversampled only five times, as shown in digital samples 1801, ..., 1805.

[0141] Assuming signal 1710 represents the process of pre-distorting the baseband signal, then from Figure 17 and Figure 18It is evident that predistortion circuits / algorithms (e.g., for memory digital predistortion, MDPD) can hardly estimate / follow the fifth exponential harmonic at low sampling rates. Figure 18 In the example, digital sample 1803 has an amplitude value of 1.8, while the subsequent power amplifier actually receives an analog input signal with a maximum amplitude value of 2.3. Therefore, the predistortion circuit / algorithm for MDPD will attempt to predistort the digital sample value to 1.8 using several weighted exponential functions, while the correct digital sample should be 2.3. In such cases... Figure 17 In the case of ten oversampling operations shown, the error is much smaller than that of five oversampling operations (digital sample 1705 has an amplitude value of approximately 2.3).

[0142] Examples of baseband signal 1910 and the corresponding predistortion baseband signal 1920 are in Figure 19 As shown in the diagram. Compared to the baseband signal 1910, the predistorted baseband signal 1920 includes additional signal components associated with the 3rd and 5th harmonics (3rd and 5th intermodulation distortion). The higher-order exponential harmonics are attenuated compared to the desired signal components around the carrier frequency (approximately 0 MHz frequency deviation).

[0143] The predistortion circuit 1610 of device 1600 allows tracking of higher-order exponential harmonics because the first sampling rate (i.e., the input sampling rate) is high enough. Furthermore, the second sampling rate (i.e., the output sampling rate) is low enough to save current and achieve reasonable attenuation of higher-order exponential harmonics. In other words, device 1600 can be understood as a novel downsampling MDPD method. In the second step, the (higher-order) upsampling filter 1620 allows recovery of the original MDPD input sampling rate. Since the upsampling filter 1620 uses fixed coefficients and because no exponential order occurs, it is much simpler than MDPD calculations (e.g., Volterra series). Therefore, device 1600 allows for current savings.

[0144] Although downsampling by the predistortion circuit 1610 can cause aliasing of the fifth harmonic in the predistortion baseband signal, the adaptive algorithm of the upsampling filter 1620, which receives the downsampled signal as input, will inherently optimize the addition of the aliased signal to the original signal.

[0145] The above text combined Figures 16 to 19 The effect of the predistortion mentioned above is Figure 20 As shown in the image. Figure 20 The spectra of two LTE 20 signals, 2010 and 2020, are shown. Signal 2010 is generated based on a baseband signal a without pre-distortion, while signal 2020 is generated based on the same baseband signal by using device 1600 for MDPD.

[0146] The input signal to the MDPD is an LTE carrier aggregation signal with a bandwidth of ±19MHz. A fifth-order MDPD based on Volterra is used for predistortion. The input sampling rate (i.e., the first sampling rate) of the predistortion circuit is 184MHz, while the output sampling rate of the predistortion circuit is 92MHz. This sampling frequency produces an MDPD edge of 92 / 2MHz = 46MHz. By comparing signals 2010 and 2020, it is evident that the MDPD based on the proposed concept provides up to the Nyquist rate (i.e., half the sampling rate at...). Figure 20 The signal attenuation in the example is 46MHz. Therefore, ACLR can be reduced.

[0147] However, in some examples, it may be beneficial not to reduce the sampling rate. This is in Figure 21 and Figure 22 As exemplarily shown in the figure. Figure 21 The spectra of two LTE 20 signals, 2110 and 2120, are shown. Signal 2110 is generated based on a baseband signal a without predistortion, while signal 2120 is generated based on the same baseband signal using a downsampled MDPD. The input signal to the MDPD is also an LTE carrier aggregation signal with a bandwidth of ±19MHz. A fifth-order MDPD based on Volterra is used for predistortion. (See from...) Figure 21 As can be seen, the LTE carrier aggregation signal comprises two narrow-band spectrums (i.e., horn spectrums) that are widely separated from each other. Due to downsampling MDPD, a second aliasing is generated in the LTE 20 signal 2120 for the 3rd and 5th harmonics.

[0148] By performing MDPD without downsampling, aliasing can be avoided, and 3rd and 5th harmonics can be reduced. This is in Figure 22 It is shown in the middle, Figure 22 The spectrum of two LTE 20 signals, 2210 and 2220, is shown. Signal 2210 essentially corresponds to... Figure 21 Signal 2110 is shown in the figure. Signal 2220 is generated based on the same baseband signal as signal 2210 using MDPD without downsampling (i.e., normal MDPD).

[0149] Since downsampling MDPD may be disadvantageous for certain rare spectrums, the device 1600 can be adjusted accordingly. Specifically, the predistortion circuit 1610 can be further configured to receive information about the spectral allocation of the baseband signal data in the spectrum (i.e., the shape of the resulting signal spectrum; for example, indicated by the allocated resource blocks). If the spectral allocation meets a first decision criterion (the minimum / maximum bandwidth of the allocated spectrum or the minimum / maximum distance between the allocated spectrums; for example, such as...), Figure 20The predistortion circuit 1610 can be further configured to calculate samples of the predistorted baseband signal at a second sampling rate (i.e., the input sampling rate). If the spectrum allocation meets the second decision criterion (e.g., such as...), Figure 22 (As shown in the spectrum), the predistortion circuit 1610 can be further configured to calculate samples of the predistorted baseband signal at a first sampling rate (i.e., a lower output sampling rate). Therefore, if the spectrum allocation meets the second decision criterion, the upsampling filter 1620 can be deactivated.

[0150] In other words, the device 1600 supports multi-rate DPD. Depending on the known transmitted signal, the circuitry or algorithm used for predistortion can vary. For example, it can operate in decimation mode using N multipliers, or it can operate at N times the sampling rate using N / n multipliers per multiphaser. Alternatively, an improved upsampling filter with better aliasing suppression can be used. Furthermore, alternatively, the MDPD can be turned off, and the PA can operate in a more linear average power point tracking mode.

[0151] Figure 23 An example of a wireless transceiver 2300 using downsampling predistortion is shown. Baseband circuitry 2330 provides baseband signal 2301 to predistortion circuitry 2310. Predistortion circuitry 2310 samples baseband signal 2301 at its sampling rate and outputs predistorted baseband signal 2302 at a lower sampling rate. Upsampling filter 2320 upsamples predistorted baseband signal 2302 to the original sampling rate of baseband signal 2301.

[0152] The predistorted baseband signal 2302 is then up-mixed to radio frequency using upmixer 2340, and further converted to an analog representation by ADC 2350. The analog predistorted baseband signal is amplified by PA 2360 and provided to antenna 2390 via duplexer 2380. Figure 23 As shown, the supply voltage V of PA 2360 cc The envelope tracking circuit 2370 includes an envelope tracking path for determining the envelope of the predistortion baseband signal 2302, an ADC for digitization, and a method for providing V based on the digitized envelope information. cc (DC to DC converters are provided.)

[0153] The downsampling MPDP used in the wireless transceiver 2300 allows the generation of RF signals for radiation into the environment, which have reduced signal distortion and reduced power consumption.

[0154] The predistortion circuit 2310 can also support different / multiple configurations as described above. Therefore, the wireless transceiver 2300 includes a feedback path from the antenna 2390 to the adaptive circuit 2315. The feedback path (e.g., via a coupler) receives a portion of the distorted PA output signal. The feedback signal is down-converted to the baseband domain and subsequently digitized by the ADC 2325. The feedback receiver 2335 further processes the feedback signal. The adaptive circuit 2315 updates the predistortion parameters (e.g., the predistortion function) of the currently used predistorter configuration by comparing the sequence of the processed feedback signal with the time-aligned sequence of the baseband signal 2301 as described above. Alternatively or additionally, the feedback signal may be provided by the conventional receive path 2395 of the wireless transceiver 2300.

[0155] To summarize the above aspects regarding downsampling MDPD, Figure 24 A flowchart of a method 2400 for predistorting a baseband signal is also shown. Method 2400 includes determining samples of the baseband signal 2402 at a first sampling rate. Furthermore, method 2400 includes calculating samples of the predistorted baseband signal 2404 at a second sampling rate based on predistortion parameters and the samples of the baseband signal. The second sampling rate is lower than the first sampling rate. Method 2400 also includes calculating samples of the predistorted baseband signal 2406 at the first sampling rate based on the samples of the predistorted baseband signal at the second sampling rate.

[0156] Because of its high input sampling rate, the MDPD can track the amplitude of the baseband signal with high precision. Furthermore, due to its low output sampling rate, the MDPD only calculates the required output samples, resulting in low current consumption. Therefore, the MDPD achieves both high precision and low current consumption.

[0157] Combined with the proposed technology or one or more examples described above (e.g. Figures 16 to 23 The text mentions further details and aspects of the method. The method may include one or more aspects corresponding to the proposed technique or one or more additional optional features of one or more examples described above.

[0158] Figure 25 Examples of specific implementations of predistortion using one or more aspects of the proposed technology or one or more of the examples above are shown. Figure 25 An example of a mobile device 2500 (e.g., a mobile phone, smartphone, tablet, or laptop) is illustrated.

[0159] The mobile device 2500 includes a wireless transmitter or transmitter circuitry 2510 according to one or more aspects of the proposed technology or one or more examples above. At least one antenna 2260 of the mobile device 2500 is coupled to the wireless transmitter or transmitter circuitry 2510.

[0160] According to one or more aspects of the proposed technology or one or more of the examples described above, the wireless transmitter or transmitter circuit 2510 may include a predistortion circuit 2520, a control circuit 2530 for the predistortion circuit 2520, and / or means for predistorting the baseband signal 2540. The output of the predistortion circuit 2520 and / or the means for predistorting the baseband signal 2540 may be coupled to PA 2550 for amplifying the predistorted signal.

[0161] Therefore, mobile devices capable of generating high-bandwidth transmission signals with reduced current consumption can be provided.

[0162] The aforementioned pre-distorted wireless communication circuit, using the proposed technology or one or more of the examples above, can be configured to operate according to a mobile communication network or system standardized under the 3rd Generation Partnership Project (3GPP). The mobile or wireless communication system may correspond to, for example, 5G New Radio (5G NR), Long Term Evolution (LTE), LTE-A, High Speed ​​Packet Access (HSPA), Universal Mobile Telecommunications System (UMTS) or UMTS Terrestrial Radio Access Network (UTRAN), Evolved UTRAN (e-UTRAN), Global System for Mobile Communications (GSM) or Enhanced Data Rate GSM Evolution (EDGE) network, or GSM / EDGE Radio Access Network (GERAN). Alternatively, the wireless communication circuitry can be configured to operate according to mobile communication networks employing different standards, such as the Global Microwave Access Interoperability (WIMAX) network IEEE 802.16 or the Wireless Local Area Network (WLAN) IEEE 802.11, typically Orthogonal Frequency Division Multiple Access (OFDMA) networks, Time Division Multiple Access (TDMA) networks, Code Division Multiple Access (CDMA) networks, Wideband CDMA (WCDMA) networks, Frequency Division Multiple Access (FDMA) networks, Space Division Multiple Access (SDMA) networks, etc.

[0163] The embodiments described in this article can be summarized as follows:

[0164] Example 1 is a predistortion circuit for a wireless transmitter, comprising: a signal input configured to receive a baseband signal; and a predistorter configured to generate a predistorted baseband signal using the baseband signal and a selection of a first predistorter configuration and a second predistorter configuration.

[0165] In Embodiment 2, the first predistorter in the predistortion circuit described in Embodiment 1 is configured to execute a first predistortion function, and the second predistorter is configured to execute a second predistortion function.

[0166] In Embodiment 3, the predistortion circuit described in Embodiment 1 or 2 further includes: a first number of computing nodes active in a first predistorter configuration; and a second number of computing nodes active in a second predistorter configuration.

[0167] In Example 4, the second quantity in the predistortion circuit described in Example 3 is higher than the first quantity.

[0168] In Embodiment 5, a first number of computing nodes in the predistortion circuit described in Embodiment 3 are configured to operate at a first rate, wherein a second number of computing nodes are configured to operate at a second rate, and wherein the second rate is higher than the first rate.

[0169] In Embodiment 6, the predistortion circuit described in Embodiment 5 further includes a sampling rate converter configured to match the output of a first number of computing nodes with a second rate to generate a rate-matched output for the first number of computing nodes.

[0170] In Embodiment 7, the predistortion circuit described in Embodiment 6 further includes a combination circuit configured to combine the outputs of a second number of computing nodes with the rate-matched outputs of a first number of computing nodes.

[0171] In embodiment 8, the predistortion circuit in any of the preceding embodiments further includes a configuration processing circuit, which is configured to select a first predistorter configuration or a second predistorter configuration based on the operating characteristics of the wireless transmitter.

[0172] In Example 9, the operating characteristics of the predistortion circuit described in Example 8 include at least one of average power tracking mode, envelope tracking mode, output power range, peak-to-average power ratio of the baseband signal, modulation scheme for generating the baseband signal, and antenna matching conditions.

[0173] In Example 10, the operating characteristics of the predistortion circuit described in Example 8 or 9 include at least one of the following: transmission bandwidth, transmission frequency band, transmission frequency range within the transmission frequency band, number of transmission clusters in the frequency domain, frequency spacing between transmission clusters, and bandwidth of each transmission cluster.

[0174] In Example 11, the operating characteristics of the predistortion circuit in any one of Examples 8 to 10 include an acceptable spectral mask.

[0175] In Embodiment 12, the configuration processing circuit in the predistortion circuit described in Embodiment 8 further includes an input interface configured to receive a feedback signal based on the output of the transmitter's power amplifier, wherein the configuration processing circuit is further configured to select a first predistorter configuration or a second predistorter configuration based on the feedback signal.

[0176] Example 13 is a predistortion circuit for a wireless transmitter, comprising: a signal input configured to receive a baseband signal; and a predistorter configured to use a first sub-circuit to calculate a first portion of a predistorted baseband signal at a first rate, and use a second sub-circuit to calculate the first portion of the predistorted baseband signal at a second rate, thereby generating a predistorted baseband signal.

[0177] In Embodiment 14, the predistortion circuit described in Embodiment 13 further includes a rate matching circuit configured to adjust the first rate and the second rate to the sampling rate of the predistortion baseband signal.

[0178] In Embodiment 15, the predistortion circuit described in Embodiment 13 or 14 further includes a combination circuit configured to combine a first portion of the predistortion baseband signal and a second portion of the predistortion baseband signal to generate a predistortion baseband signal.

[0179] Example 16 is a method for generating a predistorted baseband signal for a wireless transmitter, comprising: selecting a first predistorter configuration or a second predistorter configuration; and using the baseband signal and the selected predistorter configuration to generate the predistorted baseband signal.

[0180] In Example 17, the method described in Example 16 further includes: executing a first predistortion function in a first predistortion configuration; and executing a second predistortion function in a second predistortion configuration.

[0181] In Example 18, the method described in Example 16 or 17 further includes: using a first number of calculations in a first predistorter configuration; and using a second number of calculations in a second predistorter configuration.

[0182] In Example 19, in the method described in Example 18, the second quantity is higher than the first quantity.

[0183] In Example 20, the method described in Example 18 further includes: performing a first number of calculations at a first rate; and performing a second number of calculations at a second rate, wherein the first rate is higher than the second rate.

[0184] In Example 21, the method described in Example 18 further includes matching the output of a first number of computations with a second rate to generate a rate-matched output of a first number of computation nodes.

[0185] In Example 22, the method described in Example 21 further includes combining the output of the second number of calculations with the rate-matched output of the first number of calculations.

[0186] In embodiment 23, the method of any one of embodiments 16 to 22 further includes selecting a first predistorter configuration or a second predistorter configuration based on the operating characteristics of the wireless transmitter.

[0187] In Example 24, the operating characteristics of the method described in Example 23 include at least one of average power tracking mode, envelope tracking mode, output power range, peak-to-average power ratio of baseband signal, modulation scheme for generating baseband signal, and antenna matching conditions.

[0188] In Example 25, the operational characteristics of the method described in Example 23 or 24 include at least one of the following: transmission bandwidth, transmission frequency band, transmission frequency range within the transmission frequency band, number of transmission clusters in the frequency domain, frequency spacing between transmission clusters, and bandwidth of each transmission cluster.

[0189] In Example 26, the operational characteristics of the method in any one of Examples 23 to 25 include an acceptable spectral mask.

[0190] In embodiment 27, the method of any one of embodiments 16 to 26 further includes: receiving a feedback signal based on the output of the power amplifier of the transmitter; and selecting a first predistorter configuration or a second predistorter configuration based on the feedback signal.

[0191] Example 28 is a wireless transmitter that includes a predistortion circuit according to any one of Examples 1 to 15.

[0192] In embodiment 29, the wireless transmitter described in embodiment 28 further includes a power amplifier coupled to the output of the predistortion circuit.

[0193] Example 30 is a mobile device that includes a wireless transmitter according to one of Examples 28 or 29.

[0194] In embodiment 31, the mobile device described in embodiment 30 further includes at least one antenna coupled to the wireless transmitter.

[0195] Example 32 is a non-transitory computer-readable medium having a program stored thereon, the program having program code for performing the method described in any one of Examples 16 to 27 when the program is executed on a computer or processor.

[0196] Example 33 is a computer program having program code configured to perform the method described in any one of Examples 16 to 27 when the computer program is executed on a computer or processor.

[0197] Example 34 is a control circuit for a predistortion circuit, the control circuit comprising: a feedback signal input configured to receive a feedback signal associated with the output of a power amplifier; a bandwidth limiting circuit configured to limit the bandwidth of the feedback signal to derive a band-limited feedback signal; and a parameter processing circuit configured to update the predistortion parameters used in the predistortion circuit based on the band-limited feedback signal.

[0198] In embodiment 35, the control circuit described in embodiment 34 further includes an output interface configured to output updated predistortion parameters.

[0199] In embodiment 36, the bandwidth limiting circuit in the control circuit described in embodiment 34 or 35 includes: an adjustable mixer configured to downmix a feedback signal using a feedback oscillator signal to generate a downmixed feedback signal, the feedback oscillator signal having the frequency of a local oscillator signal plus an observed frequency offset; and a bandpass filter circuit configured to bandpass filter the downmixed feedback signal to generate a band-limited feedback signal.

[0200] In embodiment 37, the control circuit according to any one of embodiments 34 to 36 further includes a reference signal generation circuit, which includes: a reference signal input terminal configured to receive a reference signal related to the output of the predistortion circuit or related to the input of the predistortion circuit; and an additional bandwidth limiting circuit configured to limit the bandwidth of the reference signal to derive a band-limited reference signal, wherein the configuration processing circuit is configured to update the predistortion parameters based on the band-limited observation signal and the band-limited reference signal.

[0201] In embodiment 38, the additional bandwidth limiting circuit in the control circuit of embodiment 37 includes: a frequency shifter configured to shift a reference signal by a frequency offset; and an additional bandpass filter circuit configured to bandpass filter the shifted reference signal to generate a band-limited feedback signal.

[0202] In Example 39, the frequency offset in the control circuit described in Example 38 corresponds to the observed frequency offset of the feedback signal.

[0203] Example 40 is a transmitter circuit comprising: a predistortion circuit configured to receive a baseband signal and generate a predistorted baseband signal based on predistortion parameters; an upmixer configured to generate a radio frequency (RF) signal using the predistorted baseband signal and a local oscillator signal; a power amplifier coupled to the output of the upmixer and configured to generate an amplified RF signal; and a control circuit for the predistortion circuit, the control circuit comprising: a feedback signal input configured to receive a signal related to the output of the power amplifier. The circuit includes: a feedback signal; a bandwidth limiting circuit configured to limit the bandwidth of the feedback signal to derive a band-limited feedback signal; a reference signal input configured to receive a reference signal related to the output of the predistortion circuit or related to the input of the predistortion circuit; a further bandwidth limiting circuit configured to limit the bandwidth of the reference signal to derive a band-limited reference signal; and a parameter processing circuit configured to update the predistortion parameters used within the predistortion circuit based on the band-limited feedback signal and the band-limited reference signal.

[0204] In Example 41, the bandwidth of the band-limited feedback signal and the bandwidth of the band-limited reference signal are equal in the transmitter circuit described in Example 40.

[0205] In embodiment 42, the control circuit in the transmitter circuit described in embodiment 40 or 41 further includes a phase control input terminal coupled to the output terminal of the upmixer to receive a phase control signal, wherein the control circuit is further configured to adjust the phase of the feedback signal based on the phase control signal.

[0206] In embodiment 43, the bandwidth limiting circuit in the transmitter circuit of any one of embodiments 40 to 42 further includes an adjustable mixer configured to downmix a feedback signal using a feedback oscillator signal having an observed frequency offset relative to the frequency of a local oscillator signal; and the additional bandwidth limiting circuit includes a frequency shifter configured to shift a reference signal by the reciprocal of the observed frequency offset.

[0207] Example 44 is a mobile device including a transmitter circuit according to one of Examples 40 to 43.

[0208] In Embodiment 40, the mobile device of Embodiment 40 further includes at least one antenna coupled to the transmitter circuitry.

[0209] Example 46 is a method for determining parameters of a predistortion circuit, the method comprising: receiving a feedback signal associated with the output of a power amplifier; limiting the bandwidth of the feedback signal to derive a band-limited feedback signal; and updating the predistortion parameters used in the predistortion circuit based on the band-limited feedback signal.

[0210] In Example 47, the method described in Example 46 further includes: downmixing the feedback signal using a feedback oscillator signal to generate a downmixed feedback signal, thereby generating a downmixed feedback signal having the frequency of the local oscillator signal plus the observed frequency offset; and bandpass filtering the downmixed feedback signal.

[0211] In embodiment 48, the method described in embodiment 46 or 47 further includes: receiving a reference signal related to the output of the predistortion circuit or related to the input of the predistortion circuit; limiting the bandwidth of the reference signal to derive a band-limited reference signal; and updating the predistortion parameters based on the band-limited observation signal and the band-limited reference signal.

[0212] In Example 49, the method of any one of Examples 46 to 48 further includes shifting the reference signal by a frequency offset.

[0213] In Example 50, the frequency offset in the method described in Example 49 corresponds to the observed frequency offset of the feedback signal.

[0214] Example 51 is a non-transitory computer-readable medium having a program stored thereon, the program having program code for performing the method described in any one of Examples 46 to 50 when the program is executed on a computer or processor.

[0215] Example 52 is a computer program having program code configured to perform the method described in any one of Examples 46 to 50 when the computer program is executed on a computer or processor.

[0216] Example 53 is an apparatus for predistorting a baseband signal, comprising: a predistortion circuit configured to determine samples of the baseband signal at a first sampling rate and to calculate samples of the predistorted baseband signal at a second sampling rate based on predistortion parameters and the samples of the baseband signal, wherein the second sampling rate is lower than the first sampling rate; and an upsampling filter configured to calculate samples of the predistorted baseband signal at the first sampling rate based on samples of the predistorted baseband signal at the second sampling rate.

[0217] In Example 54, the first sampling rate in the device described in Example 53 is equal to or higher than the sampling rate of the baseband signal.

[0218] In Embodiment 55, the predistortion circuit in the apparatus described in Embodiment 53 or Embodiment 54 is further configured to: receive information about the spectral allocation of data of the baseband signal in the spectrum; if the spectral allocation meets a first decision criterion, calculate samples of the predistorted baseband signal at a second sampling rate; and if the spectral allocation meets a second decision criterion, calculate samples of the predistorted baseband signal at a first sampling rate.

[0219] In Example 56, if the spectrum allocation meets the second decision criterion, the upsampling filter in the device described in Example 55 is deactivated.

[0220] Example 57 is a wireless transmitter that includes a means for predistorting a baseband signal according to any one of Examples 53 to 56.

[0221] In embodiment 58, the wireless transmitter described in embodiment 57 further includes a power amplifier coupled to the output of a means for predistorting the baseband signal.

[0222] Example 59 is a mobile device that includes a wireless transmitter according to Example 57 or Example 58.

[0223] In Embodiment 60, the mobile device described in Embodiment 59 further includes at least one antenna coupled to the wireless transmitter.

[0224] Example 61 is a method for predistorting a baseband signal, comprising: determining samples of the baseband signal at a first sampling rate; calculating samples of the predistorted baseband signal at a second sampling rate based on predistortion parameters and the samples of the baseband signal, wherein the second sampling rate is lower than the first sampling rate; and calculating samples of the predistorted baseband signal at the first sampling rate based on the samples of the predistorted baseband signal at the second sampling rate.

[0225] In Example 62, the first sampling rate in the method described in Example 61 is equal to or higher than the sampling rate of the baseband signal.

[0226] In embodiment 63, the method described in embodiment 61 or embodiment 62 further includes: receiving information about the spectrum allocation of data of the baseband signal in the spectrum, wherein if the spectrum allocation meets a first decision criterion, samples of the predistorted baseband signal are calculated at a second sampling rate; and if the spectrum allocation meets a second decision criterion, samples of the predistorted baseband signal are calculated at a first sampling rate.

[0227] In Example 64, the method described in Example 63 uses an upsampling filter to calculate samples of the predistorted baseband signal at the first sampling rate based on samples of the predistorted baseband signal at the second sampling rate. The method further includes deactivating the upsampling filter if the spectrum allocation meets the second decision criterion.

[0228] Example 65 is a non-transitory computer-readable medium having a program stored thereon, the program having program code for performing the method described in any one of Examples 61 to 64 when the program is executed on a computer or processor.

[0229] Example 66 is a computer program having program code configured to perform the method described in any one of Examples 61 to 64 when the program is executed on a computer or processor.

[0230] The aspects and features mentioned and described in one or more of the previously detailed examples and figures may also be combined with one or more other examples in order to replace similar features in other examples or to introduce the feature into other examples.

[0231] When a computer program is executed on a computer or processor, examples may also be or relate to a computer program having program code for performing one or more of the methods described above. The steps, operations, or processes of various methods described above can be performed by a programmed computer or processor. These examples may also cover program storage devices such as digital data storage media that are machine-readable, processor-readable, or computer-readable, and that encode machine-executable, processor-executable, or computer-executable instructions. These instructions perform or cause some or all of the actions in the methods described above to be performed. Program storage devices may include or be, for example, digital memories, magnetic storage media such as disks and tapes, hard disk drives, or optically readable digital data storage media. Further examples may also cover computers, processors, or control units programmed to perform the methods described above or the actions of a (Field-Programmable Array of Logic) ((F)PLA) or (Field-Programmable Gate Array of Logic) ((F)PGA), said computers, processors, or control units being programmed to perform the actions of the methods described above.

[0232] The specification and accompanying drawings illustrate only the principles of this disclosure. Furthermore, all examples set forth herein are intended primarily for illustrative purposes only, to aid the reader in understanding the principles of this disclosure and the inventors' concepts for advancing the art. All statements listing the principles, aspects, and examples of this disclosure, and specific examples thereof, are intended to include their equivalents.

[0233] A function block labeled "means for..." can refer to a circuit configured to perform a specific function. Therefore, "means for..." can be implemented as "means configured or adapted for...", such as a device or circuit configured or adapted for a given task.

[0234] The functions of the various elements shown in the figure include any functional blocks labeled "device," "device for providing a signal," "device for generating a signal," etc., which can be implemented in the form of dedicated hardware such as "signal provider," "signal processing unit," "processor," "controller," etc., as well as hardware capable of executing software in association with appropriate software. When provided by a processor, these functions can be provided by a single dedicated processor, a single shared processor, or multiple separate processors, some or all of which may be shared. However, the terms "processor" or "controller" are not limited to hardware solely capable of executing software, but can include digital signal processor (DSP) hardware, network processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), read-only memory (ROM) for storing software, random access memory (RAM), and non-volatile memory. Other conventional and / or custom hardware may also be included.

[0235] Block diagrams may, for example, illustrate high-level circuit diagrams that implement the principles of this disclosure. Similarly, flowcharts, process diagrams, state transition diagrams, pseudocode, etc., may represent various processes, operations, or steps that may, for example, be substantially represented in a computer-readable medium and thus executed by a computer or processor, whether or not such a computer or processor is explicitly shown. The methods disclosed in this specification or claims may be implemented by devices having means for performing each corresponding action of these methods.

[0236] It should be understood that the disclosure of various actions, processes, operations, steps, or functions in this specification or claims is not to be construed as being in a specific order, unless otherwise expressly or implicitly stated, for example, for technical reasons. Therefore, the disclosure of various actions or functions will not limit these actions or functions to a specific order, unless such actions or functions are not interchangeable for technical reasons. Furthermore, in some examples, a single action, function, process, operation, or step may include or be divided into multiple sub-actions, sub-functions, sub-processes, sub-operations, or sub-steps. Unless expressly excluded, such sub-actions may be included and are part of the disclosure of that single action.

[0237] Furthermore, the following claims are incorporated herein by reference, each of which may be considered an independent example. While each claim may be considered an independent example, it should be noted that although specific combinations of dependent claims with one or more other claims may be referred to in the claims, other examples may also include combinations of dependent claims with the subject matter of each other dependent or independent claim. Such combinations are expressly proposed herein unless it is stated that no particular combination is intended to be used. Furthermore, it is intended that features of the claims be included in any other independent claim, even if that claim does not directly depend on the independent claim.

Claims

1. A predistortion circuit, comprising: A first sub-circuit is configured to calculate a first portion of a predistorted baseband signal based on a received baseband signal and generate a corresponding first predistorted baseband signal having a first sampling rate, wherein the first sampling rate is based on a first bandwidth of the first predistorted baseband signal, wherein the first predistorted baseband signal includes a received baseband signal having a first nonlinearity inversely superimposed on the received baseband signal, and A second sub-circuit is configured to calculate a second portion of a predistorted baseband signal based on a received baseband signal and generate a corresponding second predistorted baseband signal having a second sampling rate different from the first sampling rate, wherein the second sampling rate is based on a second bandwidth of the second predistorted baseband signal, and wherein the second predistorted baseband signal includes a received baseband signal having a second nonlinearity that is inversely superimposed on the received baseband signal. The predistortion circuit is configured to generate the predistortion baseband signal based on one or more of the first predistortion baseband signal and the second predistortion baseband signal.

2. The predistortion circuit according to claim 1, wherein the first sub-circuit and the second sub-circuit are coupled in series, wherein the output of the first sub-circuit is coupled to the input of the second sub-circuit.

3. The predistortion circuit according to claim 2, wherein the second predistortion baseband signal is the predistortion baseband signal.

4. The predistortion circuit of claim 2, wherein the second sub-circuit includes a sampling rate converter having an input coupled to the output of the first sub-circuit, wherein the sampling rate converter is configured to convert the first sampling rate of the first predistortion baseband signal to the second sampling rate.

5. The predistortion circuit of claim 2 further includes one or more additional sub-circuits series-coupled between the output of the first sub-circuit and the input of the second sub-circuit, wherein each of the one or more additional sub-circuits is configured to calculate different corresponding portions of the predistortion baseband signal and generate corresponding predistortion baseband signals with different corresponding rates.

6. The predistortion circuit of claim 5, wherein each of the one or more additional sub-circuits includes a corresponding rate converter, the corresponding rate converter being configured to: Receives the corresponding signal output from the preceding sub-circuit of the corresponding sub-circuit coupled into series; and The corresponding rate of the corresponding signal is converted into the corresponding rate of the corresponding sub-circuit.

7. The predistortion circuit according to claim 1, wherein the first sampling rate and the second sampling rate correspond to different signal bandwidths.

8. The predistortion circuit according to claim 1, further comprising: A rate matching circuit configured to equalize the first sampling rate of the first predistorted baseband signal and the second sampling rate of the second predistorted baseband signal to the sampling rate of the predistorted baseband signal.

9. The predistortion circuit according to claim 8, wherein the first sub-circuit and the second sub-circuit are coupled in parallel, and wherein the predistortion circuit further comprises: A combination circuit configured to combine a rate-equalized first predistorted baseband signal and a rate-equalized second predistorted baseband signal into the predistorted baseband signal.

10. A method for generating a predistorted baseband signal, the method comprising: A first portion of the predistorted baseband signal is calculated based on the received baseband signal and a corresponding first predistorted baseband signal with a first sampling rate is generated, wherein the first sampling rate is based on a first bandwidth of the first predistorted baseband signal, and wherein the first predistorted baseband signal includes a received baseband signal having a first nonlinearity that is inversely superimposed on the received baseband signal. The second part of the predistorted baseband signal is calculated based on the received baseband signal and a corresponding second predistorted baseband signal with a second sampling rate different from the first sampling rate is generated, wherein the second sampling rate is based on the second bandwidth of the second predistorted baseband signal, wherein the second predistorted baseband signal includes a received baseband signal having a second nonlinearity that is inversely superimposed on the received baseband signal; and The predistorted baseband signal is generated based on one or more of the first predistorted baseband signal and the second predistorted baseband signal.

11. The method of claim 10, wherein generating the predistorted baseband signal comprises: The second predistorted baseband signal is generated based on the first predistorted baseband signal; and The second predistorted baseband signal is provided as the predistorted baseband signal.

12. The method of claim 10, further comprising: After calculating the first portion of the predistorted baseband signal and before calculating the second portion of the predistorted baseband signal, the first sampling rate of the first predistorted baseband signal is converted to the second sampling rate.

13. The method of claim 10, wherein generating the predistorted baseband signal comprises equalizing the first sampling rate of the first predistorted baseband signal and the second sampling rate of the second predistorted baseband signal to the sampling rate of the predistorted baseband signal.

14. The method of claim 13, wherein generating the predistorted baseband signal further comprises combining a rate-equalized first predistorted baseband signal and a rate-equalized second predistorted baseband signal.

15. A wireless device, comprising: At least one antenna; and Transmitter circuitry, coupled to the at least one antenna and configured to: A first portion of the predistorted baseband signal is calculated based on the received baseband signal, and a corresponding first predistorted baseband signal with a first sampling rate is generated, wherein the first sampling rate is based on a first bandwidth of the first predistorted baseband signal, and wherein the first predistorted baseband signal includes a received baseband signal having a first nonlinearity that is inversely superimposed on the received baseband signal. The second portion of the predistorted baseband signal is calculated based on the received baseband signal, and a corresponding second predistorted baseband signal with a second sampling rate different from the first sampling rate is generated. The second sampling rate is based on the second bandwidth of the second predistorted baseband signal. The second predistorted baseband signal includes a received baseband signal with a second nonlinearity that is inversely superimposed on the received baseband signal. The predistorted baseband signal is generated based on one or more of the first predistorted baseband signal and the second predistorted baseband signal.

16. The wireless device of claim 15, wherein the transmitter circuitry is further configured to: The second predistorted baseband signal is generated based on the first predistorted baseband signal; and The second predistorted baseband signal is provided as the predistorted baseband signal.

17. The wireless device of claim 15, wherein the transmitter circuitry is further configured to convert the first sampling rate of the first predistorted baseband signal to the second sampling rate after calculating the first portion of the predistorted baseband signal and before calculating the second portion of the predistorted baseband signal.

18. The wireless device of claim 15, wherein the transmitter circuitry is further configured to equalize the first sampling rate of the first predistorted baseband signal and the second sampling rate of the second predistorted baseband signal to the rate of the predistorted baseband signal before generating the predistorted baseband signal.

19. The wireless device of claim 18, wherein the transmitter circuitry is further configured to combine a rate-equalized first predistorted baseband signal and a rate-equalized second predistorted baseband signal to generate the predistorted baseband signal.

20. The wireless device of claim 15, further comprising: A power amplifier configured to receive a radio frequency version of the predistorted baseband signal and amplify the radio frequency version of the predistorted baseband signal for transmission via the at least one antenna, wherein the first nonlinearity and the second nonlinearity are associated with the power amplifier.

Images