KR20070031452A - Digital transmitter system employing self-generating predistortion parameter lists and adaptive controller - Google Patents

Abstract

A method and apparatus for digital predistortion of power amplifiers is disclosed. Successful digital predistortion parameter settings correlate with operating conditions that affect the distortion of amplifier 111. These operating conditions may include input power level, carrier frequency, temperature, DC supply voltage, and the like. Successful predistortion parameter settings along with corresponding operating conditions are stored in a list 224 indexed using multidimensional attribute vectors. The elements of the list are created automatically.

Predistortion, digital transmitter, adaptability, amplifier

Description

DIGITAL TRANSMITTER SYSTEM EMPLOYING SELF-GENERATING PREDISTORTION PARAMETER LISTS AND ADAPTIVE CONTROLLER}

The present invention relates to digital transmitters, RF power amplifiers and amplification methods. In particular, the present invention relates to digital transmitters, amplifiers, and related methods that use adaptive techniques to linearize gain transfer characteristics or reduce distortion emissions such as digital predistortion.

Wireless transmitters used in RF communication systems use RF power amplifiers as a key component and these amplifiers are a major factor of nonlinearity in the overall system. RF power amplifiers are devices for replicating an RF signal provided at an input and produce an output signal with a higher power level. The increase in power from the input to the output is called the 'gain' of the amplifier. When the gain is constant over the dynamic range of the input signal, the amplifier is called 'linear'. Amplifiers are limited in capacity by the power delivered due to gain and phase changes, especially saturation at high power, which makes all practical amplifiers non-linear when the input power level varies. The ratio of distortion power produced relative to the delivered signal power is a nonlinear measure of the amplifier.

In RF communication systems, the maximum allowable nonlinearity of the amplifier is specified by legislation such as FCC or ITU. Since amplifiers are inherently nonlinear when operating near saturation, linearity requirements often limit the rated power delivery capacity. In general, when operating near saturation, the linearity of the amplifier deteriorates quickly because the increased signal power delivered by the amplifier is proportionally less than the generated increased distortion power.

Various compensation approaches are conventionally provided to reduce distortion at the output of the system, which in turn increases the rated power delivery capacity. One approach is digital predistortion. In digital predistortion, an RF power amplifier is part of an RF transmitter in which the digital input signal is converted into an analog signal (DAC), the frequency is upconverted to produce an RF signal, and then amplified by the RF power amplifier. Predistortion is applied to the signal while in digital format to compensate for later nonlinearities in the transmission path.

The highest nonlinearity tends to be the gain of the RF power amplifier. The gain of the power amplifier is related to the amplitude and phase of the RF output signal relative to the RF input signal. Nonlinear gain means that the amplitude and / or phase components of the gain vary as a function of the input signal amplitude. In order to linearize the gain, the amplitude and phase components must be constant over the range of input signal amplitudes. Digital predistortion predicts the RF gain variation from the digital input signal and generates a digital gain that forms the overall gain of the transmitter, and compensates for these gain variations by a linear combination of digital and RF gains rather than RF gain alone. In order to accurately predict RF gain changes from the digital input signal, a digital model of the inverse of the RF gain is required. In general, based on a look-up table, polynomial extension, or both, the digital model has a parametric form where the inverse gain is the sum of the fundamental waveforms derived from the input signal weighted by the vector of complex predistortion parameters. Predistortion parameters are adjusted using an adaptive controller for optimal distortion correction and gain linearity.

Most end users of power amplifiers have specifications that limit the time that the adaptive portion can be used to achieve sufficient distortion correction. As a result, it is important to have good initial predistortion parameter settings when the adaptive controller begins to examine the best (or sufficient) overall distortion parameters. Some such specifications have times as low as 10 seconds.

There are a number of conventional approaches to predistortion. In previous approaches, the compensation gain was implemented using a static analog circuit that was manually adjusted to be optimal for nominal operating conditions. Later digital approaches are introduced to allow greater flexibility in the inverse gain model. Reference tables and polynomial extensions that depend only on instantaneous input size or power are used only to model the memory distortion. When the bandwidths of the input signal are increased, the models are generally expanded to be functions of both instantaneous and past input sizes, as a result of multicarrier signal formatting, and capture a type of distortion called "memory based". Amplifiers that exhibit memory-based distortion have a "memory effect."

The nonlinear gain of an RF amplifier is not only a function of instantaneous and near past input magnitudes (called the "envelope" of the input signal), but also other input and average input power, the number and carrier frequencies in the multi-carrier format, temperature, and It is influenced by environmental quantities such as the DC supply voltage (collectively referred to herein as "property"). These properties are largely independent of the power amplifier and are characterized as measurable amounts that change slowly or rarely with respect to changes in the instantaneous magnitude (envelope) of the input signal. In prior art approaches, adaptive methods are used to readjust the coefficients of the inverse gain model in response to changes in RF gain due to changes in input or environmental quantities. Means are needed for measuring the output signal and converting the output signal to a digital format. The digital input and digitized output signals are compared to estimate the residual distortion in the output signal where the adjustment of the predistortion parameters is calculated to reduce the output distortion. In such systems, the convergence time is dependent on the change in the amount of attributes and the magnitude and disruption of the sensitivity of the power amplifier gain for the attributes. During transients (converging predistortion parameters), the distortion at the output may exceed undesirable spectral mask specifications.

To avoid transient distortions, the prior art relies on reference tables (LUTs) indexed using one or more attribute quantities. When the inverse gain model is expressed using polynomial expansion, the entries of the LUT are the expansion coefficients that are optimally considered for the corresponding attribute (index) amount. If the inverse gain model is also represented by an LUT structure indexed by the instantaneous magnitude (envelope) of the input signal, a multidimensional table is formed. However, in the present invention, the dimensions indexed by the property and not by the envelope of the input signal are of interest.

As mentioned above, the conventional reference tables used a fixed structure. Inputs such as temperature are indexes to the array. The indices are equally spaced over the range in ascending order, and the corresponding predistortion parameters are stored in the array. The array of vectors forms a two dimensional LUT. This structure is suitable for memory chips because the index is the same as the address and the predistortion parameters are the same as the data. However, reference tables are generally based on exponential data (calibration) which requires considerable time to write the elements of the table. In addition, drift from component aging renders any reference table obsolete, requiring recalibration.

Another difficulty with reference tables is that it is extremely difficult to manage multidimensional arrays that require that many operating conditions be provided to affect optimal predistortion parameters. One can imagine the number of elements presented in equally spaced four dimensional arrays. For example, ten samples per dimension form 10000 elements.

One technique for managing multiple indexing dimensions is to assume that the results can be separated. The separable conditions allow the use of separate arrays for each operating condition and the composite result will be the sum of the individual adjustments. (Unlike Taylor serial expansion with partial derivatives specified). Unfortunately, this approach is only valid for small (differential) alignment adjustments because any cross correlation between dimensions is ignored. The biggest error occurs at the edges of the multidimensional array. For example, the unwieldy edge in terms of temperature, the DC supply index spacing would be high and low voltage. These corner positions are typically checked by smart customers to determine if the amplifier is compliant with the specifications.

A related problem with array-based reference tables is the selection of sampling intervals (intervals between adjacent indexes) in the index space. In general, the sensitivity of the predistortion parameter settings varies over the index interval. The sampling density should be chosen based on the most sensitive area of the index space. The remaining areas will be oversampled. This oversampling problem becomes more important for multidimensional arrays.

Attempts have been made to make the reference tables self-correcting or self-generated. However, fixed array structures are difficult to manage. The key problem facing is 'update fragmentation'. Consider the four-dimensional array case mentioned previously. When the reference table is updated, only one element of 10000 is changed. If the cause of quality enhancement is overall (eg due to component drift), a total of 10000 factors are affected. However, changes must propagate as each index is visited. Simply because one of the indexes has aged, there is potential for neighboring indexes to have large differences.

Accordingly, there is a need for a system and method for fast preamplification parameter control in a digital predistortion amplifier system that avoids the above limitations of the prior art.

In a first aspect the present invention provides a digital RF transmitter system comprising an input, a transmission path and an observation path for receiving a digital input signal. The transmission path comprises a digital predistorter using adjustable predistortion parameters to predistort the digital input signal, a digital to analog converter, an RF upconverter, an RF power amplifier, and an output sampling combiner that receives the amplified RF output signal. Include. The observation path for measuring the sampled output of the combiner includes an RF down converter and an analog to digital converter that provides a sampled RF output signal in digital format. The digital RF transmitter system also includes an adaptive controller coupled to receive the digital input signal and the sampled digital output signal to determine residual distortion in the output signal and coupled to receive one or more parameters that characterize the operating conditions of the amplifier. do. The adaptive controller includes a predistortion parameter list with a plurality of list elements and provides predistortion parameters coupled to the digital predistorter and further adjusted to the digital predistorter to reduce distortion, each element operating the amplifier system. One or more predistortion parameters and one or more parameters that characterize the condition.

In a preferred embodiment of the digital RF transmitter system, the adjustable predistortion parameters are part of an inverse gain model that changes the gain of the transmitter path as a function of nonlinear modes of input magnitude. The adjustable predistortion parameters may include weights for unreserved waveforms and memory based waveforms. Parameters characterizing the operating conditions of the amplifier may include one or more temperatures, DC power supply, input signal power, and input signal carrier frequency. Preferably, the parameters characterizing the operating conditions of the amplifier include an attribute vector and a distance is defined between any two attribute vectors. The adaptive controller derives the current attribute vector from the input parameters, calculates the distance to the attribute vectors of the list elements, and selects the list element with the minimum distance for use as the predistortion parameters of the digital predistorter. The adaptive controller can measure distortion continuously using digital outputs and digital inputs and a set of predistortion parameters are retrieved from the predistortion parameter when the measured distortion exceeds a predetermined value. The adaptive controller also continuously measures the attribute vector and a set of predistortion parameters are retrieved from the predistortion parameter list when the change in the measured attribute vector exceeds a predetermined value. The adaptive controller uses the selected element as the initial set of predistortion parameters and calculates new predistortion parameters from the set of initial parameters using an iterative control algorithm. The adaptive controller then updates the predistortion parameter list with the new predistortion parameters after completing the iteration calculation. The distance between the nearest attribute vectors can be varied through the predistortion parameter list.

According to another aspect, the present invention provides an adaptive controller for controlling distortion compensation of an amplifier system. The adaptive controller includes one or more inputs to receive one or more attribute parameters corresponding to current operating conditions of the amplifier system. The adaptive controller further includes one or more processors coupled to the one or more inputs, the one or more processors having an associated list of predistortion parameter settings and adjusting the predistortion parameter settings to control the distortion correction of the amplifier system. It is programmed with a predistortion parameter list algorithm and a controller algorithm to provide. The predistortion parameter list algorithm generates a predistortion parameter list with the predistortion parameter settings calculated by the controller algorithm and associates each predistortion parameter setting with one or more attribute parameters.

In a preferred embodiment of the adaptive controller, the predistortion parameter list algorithm selects a predistortion parameter setting from the list of predistortion parameters for use at startup or when the output distortion is sufficiently large. The predistortion parameter list algorithm preferably calculates a distance between one or more attribute parameters corresponding to the current operating conditions and the attribute parameters associated with each of the predistortion parameter settings of the list and transposes the preposition corresponding to the attribute parameter with the minimum distance. Selecting a distortion parameter setting selects a stored predistortion parameter setting for use by the controller algorithm. The distance calculation may be weighted with different weights for different attribute parameters. Attribute parameters may include one or more of temperature, DC supply voltage, input signal power, and input signal carrier frequency. The distance d _attr between two sets of attribute parameters “n” and “0” can be defined by a weighted (L _inf ) norm distance measurement or a weighted (L ₂ ) norm distance measurement. have. The adaptive controller preferably further includes one or more inputs for receiving output distortion data. The output distortion data includes a baseband digital representation of the output signal provided from the observation path. The predistortion parameter settings may be polynomial expansion coefficients that include band limited nonlinear modes modeling memory effects in the amplifier system.

According to another aspect, the present invention provides an amplifier system having an adaptive control loop comprising a control loop input, a first signal path comprising a digital predistorter and an RF amplifier, and a second signal path feeding back an output of the RF amplifier. Provide a method for controlling. The method provides a list of predistortion parameter settings, each predistortion parameter setting having an associated operating condition. The method further comprises detecting current operating conditions of the amplifier system, comparing the current operating conditions against the conditions in the list of predistortion parameter settings, and selecting a predistortion parameter setting associated with the most similar operating condition in the list. Include.

In a preferred embodiment of the method for controlling the amplifier system, the relevant operating conditions are configured as multidimensional attribute vectors. Comparing the conditions of the list of predistortion parameter settings with the current operating conditions preferably comprises measuring the distance between the current attribute vector and the respective attribute vectors of the list. Selecting a predistortion parameter setting associated with the most similar operating conditions in the list includes determining an attribute vector with a minimum distance from the current operating condition attribute vector. The method may further comprise calculating a new predistortion parameter setting using an iterative adaptive controller algorithm, wherein the predistortion parameter setting associated with the most similar operating condition is an initial predistortion parameter setting for the adaptive controller algorithm. Used as The method further includes updating the predistortion parameter list with the new predistortion parameter setting calculated by the adaptive controller algorithm. The size of the predistortion parameter list may be dynamic. In addition, the spacing of the stored predistortion parameter settings, as defined by the attribute vector distance, may vary throughout the list. For example, a higher density of predistortion parameter settings is provided in regions of the list where distortion correction is most sensitive to one or more operating conditions that include an attribute vector.

According to another aspect, the present invention provides a method for maintaining a list of predistortion parameter settings of an adaptive digital predistortion amplifier system, the list comprising a plurality of elements, each element of the predistortion parameter setting and the amplifier system. Has a set of operating condition parameters corresponding to the operating conditions. The method includes selecting an element of the predistortion parameter and determining an element of the predistortion parameter list with corresponding operating conditions very similar to the selected element. The method includes determining if two elements are similar enough to be considered redundant and deleting the oldest of the two elements of the predistortion parameters if the elements are redundant.

In a preferred embodiment of the method of maintaining a list of predistortion parameter settings of an adaptive digital predistortion amplifier system, selecting an element of the predistortion parameter list selects the oldest element of the list that was not previously affected by the list maintenance process. It includes a step. Determining an element of the predistortion parameter list with operating conditions most similarly to the selected element preferably comprises determining a distance measurement for operating condition parameter values of each of the remaining elements of the predistortion parameter list; Selecting the element with the minimum distance. The distance measurement may comprise a weighted difference between operating condition parameter values. The operating condition parameters of the amplifier system may include one or more temperatures, DC power supply, input signal power and input signal carrier frequency. Determining whether the elements are similar enough to be considered redundant includes determining a distance measure between the predistortion parameter settings and comparing the predistortion parameter distance to the overlap distance threshold. Alternatively, determining whether the elements are similar enough to be considered redundant includes comparing the distance between the operating condition parameters of the two elements to an old distance threshold. The method includes repeating the list maintenance process for each element of the predistortion parameter list.

According to another aspect, the present invention provides a method for generating a hierarchical list of predistortion parameter settings of an adaptive digital predistortion amplifier system. The list includes a number of elements, each element having a predistortion parameter setting and a corresponding set of parameters corresponding to operating conditions of the amplifier system, and having a hierarchical structure comprising at least two levels. The method includes selecting an element at a first level of the predistortion parameter list. The method determines the element of the first level of the predistortion parameter list with the corresponding operating conditions most similar to the selected element and demotes the oldest of the two elements to the lower level of the hierarchical predistortion parameter list. It comprises the step of.

In a preferred embodiment of the method for generating a hierarchical list of predistortion parameter settings of an adaptive digital predistortion amplifier system, the step of determining an element of the predistortion parameter list with the corresponding operating conditions most similar to the selected element comprises: Determining a distance measurement for operating conditions of each of the remaining elements of the first level of the distortion parameter list and selecting the element with the minimum distance. The method may further comprise determining whether two elements are redundant, wherein the older element is demoted only if the local elements are redundant. The method includes repeating the list process for each level of the hierarchical list. The method may also include deleting older entries if the list maintenance process is at the lowest level of the hierarchy. A demoted element is preferably associated when a subset list entry of duplicate elements is not demoted. Elements that are demoted and have a subset list are merged with a subset list of non-demotiond duplicate elements.

According to another aspect, the present invention provides a method for controlling an amplifier system having an adaptive control loop comprising a control loop input, a first signal path, a second signal path, and a control loop output, wherein the first and second At least one of the signal paths includes an amplifier and a predistorter. The method provides a hierarchical list of predistortion parameter settings having at least two levels, each predistortion parameter setting having an associated operating condition and some or all of the highest level of the predistortion parameter settings being of a lower level. Has a subset predistortion parameter settings. The method further includes detecting current operating conditions of the amplifier system and comparing the current operating conditions to the highest level conditions of the hierarchical list of predistortion parameter settings. The method further includes selecting a predistortion parameter setting associated with the most similar operating condition at the highest level of the list. The method further includes comparing the current operating conditions to a subset condition of the selected highest level of predistortion parameter setting. The method further includes selecting a predistortion parameter setting of the subset having the most similar operating condition. The method further includes selecting a higher or lower level predistortion parameter setting having an operating condition most similar to the current operating condition.

In a preferred embodiment, the method for controlling the amplifier system includes repeating the processing for each level of the hierarchical list until the next lower subset is empty. The highest level may have settings of the predistortion parameter of coarser spacing than the lower level. For example, any two predistortion parameter settings may have a predistortion parameter distance and the highest level has a predistortion parameter distance between settings that are greater than the lower level. The predistortion parameter distance may comprise a weighted difference between the predistortion parameter settings. For example, the predistortion parameter settings may include inverse gain model settings and the weight may include the predistortion parameter sensitivity.

Other features and advantages of the invention are described in the following detailed description.

1 is a schematic block diagram of a digital transmitter system including an RF power amplifier and a digital predistortion module according to the present invention.

2 is a schematic block diagram of a control system of a digital transmitter system according to the present invention.

3 is a flow diagram of a process control algorithm illustrating removal of a predistortion parameter list to remove duplicate elements in accordance with the present invention.

4 is a flow diagram of a process control algorithm illustrating second removal of a predistortion parameter list to remove outdated elements, in accordance with the present invention.

5 is a flow diagram of a process control algorithm illustrating predistortion parameter estimation and control processing including interactions between predistortion parameter list processing and adaptive controller processing in accordance with the present invention.

6 is a flow diagram of a process control algorithm illustrating the generation of a hierarchical predistortion parameter list structure in accordance with the present invention.

7 is a schematic diagram of the hierarchical predistortion parameter list structure before demoting an entry.

8 is a schematic diagram of a hierarchical predistortion parameter list structure after demoting an entry.

A block diagram of an RF transmitter system using a power amplifier (PA) that is linearized using digital predistortion in accordance with a preferred embodiment of the present invention is shown in FIGS. 1 and 2. 1 shows a basic digital predistortion transmitter and FIG. 2 shows a control system.

As shown in Fig. 1, the digital predistortion linearization transmitter has a conventional structure using a transmission path and an observation (or feedback) path. Transmission paths include digital signal input 103, digital predistortion (DPD) module 105, digital-to-analog conversion (DAC) device 107, RF upconverter 109, RF power amplifier 111, output combiner 113 and the RF output signal 125. The observation path includes an output combiner 113, an RF downconversion device 115, an analog to digital conversion (ADC) device 117, and a digital version 119 of the sampled output signal. The digital input signal 103 and the digital output signal 119 are inputs to the predistortion parameter estimator / adaptive controller module 121 (simply referred to as an "adaptive controller"). The output of adaptive controller 121 is fed to digital predistortion module 105 to linearize transmitter 127 including digital to analog converter 107, RF upconverter 109, and RF power amplifier 111. Is a set of predistortion parameters 123 used.

Adaptive controller 121 is shown in FIG. As shown in FIG. 2, the adaptive controller may include a processor 202 that executes both the adaptive controller and the predistortion parameter list algorithms described in detail below. The predistortion parameter lists are stored in a suitable memory 224 and configured and accessed in the manner described in more detail below. Alternatively, separate processors may be provided for adaptive controller and predistortion parameter list functions. Predistortion parameters 123 are used by digital predistortion module 105 to perform digital predistortion. More specifically, the adjustable predistortion parameters are preferably part of an inverse gain model that changes the transmitter path gain as a function of the input magnitude of the nonlinear modes. The adjustable predistortion parameters also include weights for both unreserved waveforms and memory based waveforms. For example, the predistortion parameters may be polynomial expansion coefficients that include band limited nonlinear modes that model memory effects of the amplifier system. The implementation may use the techniques of the provisional patent application US 60 / 485,246 filed July 3, 2003 and the utility patent application US 10 / 881,476 filed June 30, 2004, the disclosure of which is hereby incorporated by reference in its entirety. Are integrated. Alternatively, various different digital predistortion methods may be used and associated with the distortion parameters provided using techniques known to those skilled in the art.

The processor receives the input data 103 through the delay unit 215 and also receives the output data 119. The delay unit 215 is selected such that the input data 103 and the output data 119 are time aligned in the processor. The time alignment delay unit 215 may be implemented as part of the processor. The processor also receives inputs corresponding to the current operating conditions of the amplifier system. For example, inputs to temperature, DC power supply, and input RF signal carrier frequencies 226, 228, 230 may be provided to be converted into digital form by analog to digital converters 236, 238, 240. Other operating condition inputs 234 are provided and converted into digital form by the A / D converters 242.

First, the general operating principles of the digital predistortion transmitter system will be described. The adaptive digital predistortion control system provides fast convergence characteristics by storing and reusing successful predistortion parameter settings. The system has the ability to acquire the effect that operating conditions such as temperature, DC supply, input power level, and carrier frequency have optimal (steady-state) predistortion parameters. As a result, the predistortion control system can respond to changes in these operating conditions faster than an adaptive controller operating alone.

More specifically, the adaptive controller function of the processor is to measure residual distortion in the output signal and adjust the predistortion parameters of the predistortion module. The controller adjusts the predistortion parameters in an iterative manner and examines the minimum residual distortion. During the transient period when the check is not complete, the predistortion parameter error (the difference between the current and steady state predistortion parameters) degrades the performance of the predistortion power amplifier and degrades the linearity of the system. In order to minimize transient quality drop, good initial predistortion parameters are required. In addition, the potential for instability due to the diverging adaptive controller is reduced by good initial predistortion parameters.

To achieve a set of good initial predistortion parameters, the predistortion power amplifier control system processor 202 maintains a list 224 of previous successful predistortion parameter settings. Before requiring the use of the adaptive controller, the processor 202 examines the predistortion parameter list for conventional predistortion parameter settings used under similar operating conditions. Operating conditions are represented as multidimensional attribute vectors. The attribute 'distance' defined below is calculated by comparing the stored attributes with the current attributes. In some cases, the initial predistortion parameters from the list provide sufficient distortion correction to prevent the use of the adaptive controller.

Attribute distance is also used to predict the state change of the predistortion power amplifier system. When sudden changes in the current attribute vector are detected, the predistortion power amplifier system adjusts task scheduling. (The term 'current' is used herein to denote 'current time' and should not be interpreted as 'electron flow'). The maintenance tasks are finished for predistortion parameter estimation.

Distortion correction includes controlling the predistortion parameter settings of the digital predistortion module to minimize the distortion power detected in the output signal in FIG. 1. The adaptive control function may use conventional digital predistortion techniques found in the prior art.

There are many operating conditions that affect the optimal predistortion parameter settings. These include environmental conditions, application specific conditions and system specific conditions. For example, temperature, input power level x (t), carrier frequency, and DC supply voltage are relevant operating conditions for most applications and they are provided as inputs 226, 234, 230 and 228 as shown in FIG. . Time can be considered as a parameter affecting predistortion parameters due to component aging. All of these parameters are measurable within the digital predistortion transmitter system and can be monitored by the processor 202. In the control system of the present invention, the relevant measurable parameters are used to form the attribute vector. When the controller converges to a steady state, the attribute vectors and associated predistortion parameter settings are stored in the predistortion parameter list of memory 224. The correlation of attribute vector to predistortion parameter setting is achieved by a predistortion parameter list. The disclosed system combines adaptive controller processing and predistortion parameter list processing to allow fast convergence to steady state predistortion parameters.

Subsequently, with reference to Figs. 1 to 8, detailed embodiments of the present invention will be described.

First, a preferred embodiment of the manner in which attribute vector and predistortion parameter settings are represented and stored in a predistortion parameter list is described. A distance metric for measuring the similarity of attribute vectors is described. Processing to remove the predistortion parameter list is also discussed to maintain a manageable number of list entries while maintaining coverage of the attribute interval. Next, the use of an adaptive controller and a predistortion parameter list for adapting the predistortion parameters is described. Also described is the self-generation of the elements in the predistortion parameter list. An alternative predistortion parameter list structure based on a hierarchical list structure is then described with reference to FIGS. 6 to 8.

As noted above, sets of attribute parameters or attribute vectors are used to allow the digital predistortion transmitter system to learn from past operations. By correlating the attribute vectors with past predistortion parameter settings, the convergence of the predistortion parameter estimate is faster and more robust. Attribute parameters that affect the digital predistortion transmitter system for a given application are determined when defining the attribute vector. Temperature, average input power, and center frequency are all time-varying, which are important parameters and gains for cellular applications. Other parameters, such as modulation format and number of carriers, also achieve gain, but if these parameters are constant over time, they provide a small value as part of the attribute vector and can be excluded from the attribute vector.

To determine the similarity of two attribute vectors, a distance measure is used. The difference (square or absolute difference) between each parameter is weighted based on the sensitivity associated with the digital predistortion transmitter system gain. That is, parameters with greater effect on digital predistortion transmitter system gain are weighted more heavily. These sensitivities can be evaluated using sub-specs or experiments for specific execution and application requirements.

In particular, assume that the attribute vector for the predistortion parameter list element 'n' is defined as follows.

(Equation 1)

Where p _k is the value of the attribute parameter 'k' (such as temperature). To facilitate the execution of distance measurements on processor 202, weighted (L _inf ) norms can be used; That is, the distance between the elements 'n' and '0' indicated by d _attr (n, 0) is defined as follows.

(Equation 2)

Where w _k is a weight for the parameter 'k'. Alternative distance measurements, such as the L ₂ criterion, may also be used. The weighted L ₂ norm measurement between elements 'n' and '0', denoted by d _attr (n, 0), is defined as follows.

(Equation 3)

This is also a value for measuring the similarity of the predistortion parameter settings by defining the distance d _DPD between the two predistortion parameter settings. Once again, L _inf gambling can be used.

(Equation 4)

Where Δ _k (n, 0) is the difference between the 'k' distortion position parameters between the 'n' list element and the '0' list element, and s _k is the respective sensitivity. Once again, alternative distance measurements such as L ₂ norm can be used. The L ₂ norm measurements of d _DPD are as follows.

(Equation 5)

The predistortion parameter list structure may be dynamic. Both the list entries and the number of entries can change dynamically. More specifically, along with the attribute vector, processor 202 keeps track of past successful predistortion parameter settings. Before executing the adaptive controller function, processor 202 checks for residual distortion associated with the current predistortion parameter setting. If adequate in terms of distortion correction quality, no action is required. If not appropriate, the predistortion parameter setting of the listed element with the attribute vector closest to the current operating condition is retrieved. The residual distortion correction for the new setting is then checked. If distortion correction is still not suitable, the adaptive controller creates new predistortion parameter settings. After the system converges with the aid of the adaptive controller, a new predistortion parameter setting is added to the predistortion parameter list.

It is important to limit the number of elements in the predistortion parameter list to limit processor computation time. The simplest way is to set an upper limit on the number of elements and overwrite the oldest element if a new predistortion parameter setting is received. An alternative method is to use removal. If processor 202 is not used to process priority instructions, deletion may be performed. Deletion occurs when the predistortion parameter setting is approximately equal to the neighbor (i.e., the neighbor of element 'k' has the lowest d _attr (n, k) and is considered redundant if d _DPD (n, k) is small). Eliminate duplicates by deleting older elements. As a result, the number of elements required to represent the attribute interval area is proportional to the change in the predistortion parameter setting. The list-based approach produces the most compact representation of attribute predistortion parameter mapping.

One implementation of the predistortion parameter list deletion process flow is shown in FIG. 3 by way of example only. As shown, the processing flow begins at 302 when the processor 202 is not occupied with higher priority tasks. The process flow proceeds to 304 to select the oldest element from the predistortion parameter list. The process then uses the attribute vector distance measure d _attr as described above to first calculate the distance for the remaining elements of the list and select the element with the minimum distance 308 at 306 to obtain the most from the rest of the list. Determine the nearest factor. Next, the processing flow at 310 calculates the predistortion parameter distance d _DPD as described above for the selected nearest element. If the predistortion parameter settings are the same or close enough, the older element is deleted at 312. Close enough depends on the distortion sensitivity and distortion tolerance for each predistortion parameter settings. For example, a minimum distance (d _redundant ) can be used and if d _DPD is less than or equal to d _redundant , older elements are deleted and both elements are retained if d _DPD is greater than d _redundant . The processing flow at 314 checks if the entire list has been examined and if the remaining elements are not repeated until they are sufficiently spaced in relation to the predistortion parameter tolerance (or until the lower limit of the list size is reached).

The second deletion process is shown in FIG. This can be used to remove old elements from the predistortion parameter list. For example, component aging can change the relationship between the best predistortion parameter setting and a given attribute vector. Attribute distances between small elements identify potentially old elements. For example, if an element has an incorrect predistortion parameter setting, the residual distortion becomes too large, requiring the use of an adaptive controller function. The adaptive controller will find a new predistortion parameter setting. As a result, two different predistortion parameter settings will be listed for a given attribute vector (or two very similar vectors). Deleting older elements resolves any conflicts and also keeps the current list.

As shown in FIG. 4, with reference to the specific processing flow for this second deletion, the second deletion processing begins at 402 when the processor does not execute higher priority tasks (including the first deletion processing). At 404 the processing flow selects the oldest element of the predistortion parameter list that has not been preprocessed for the second deletion. At 406 the distance d _attr is calculated for this oldest element for each of the remaining elements of the list. Thus, at 408 the process determines the element with the minimum distance d _attr for the oldest element. If this minimum distance is less than or a predetermined distance (d _outdated) equal, at 410 the processing deletes the older of the two elements than from the predistortion parameter list. If the minimum distance is greater than d _outdated , then these two elements are considered sufficiently different and both remain in the predistortion parameter list. At 412 the processing flow checks if more elements remain to be checked, and if so, the processing flow returns to 404 to check the next oldest element in the list. The second deletion process ends at 414 when all elements are examined.

Default predistortion parameter settings may be maintained for when the predistortion parameter list is empty. In addition, the factory default predistortion parameter settings can be maintained individually so that they are not deleted.

Next, with reference to FIG. 5, the predistortion parameter estimation and control process is described. The control process is shown in FIG.

As shown generally in FIG. 5, the predistortion parameter evaluation algorithm uses two parallel processes 500 and 501 as well as a process flow that controls the interaction between these two processes. The first process, indicated as 501, creates a new predistortion parameter setting for initial use in controller processor 500. This process 501 is used at the beginning and is run continuously so that it is used for control processing when the output distortion is too large. More specifically, first processing flow 501 continuously monitors current operating condition parameters to obtain a current attribute vector as indicated by 504. For example, the process can determine the current temperature, DC power, carrier frequency and input power to determine a current attribute vector for current operating conditions. The processing flow then examines the predistortion parameter list for the list element with the smallest attribute distance from the current vector at 506. This list element predistortion parameter setting is retrieved at 510. Whether the retrieved predistortion parameter setting is used to update the predistortion parameters for adaptive controller process 500 may be controlled by the distortion measurement process 514 and changes in the retrieved element. At startup and whenever the measured distortion is too large, a predistortion parameter list setting is retrieved and used at 512 to update the predistortion parameters. Adaptive controller processing 500 then begins and the adaptive controller then calculates the predistortion parameter modification at 516 using an iterative controller algorithm as previously described.

For example, this process flow can be controlled by status flags. If the element retrieved at 510 changes, it is evident that the status flag indicates that the system does not repeat. The process flow also continuously measures the output distortion level at 514. If the distortion is too large, the status flag is checked. If the flag is clear, a significant change in the attribute vector is indicated and the predistortion parameters are updated using the new settings retrieved from the predistortion parameter list. After retrieving the new setting, the status flag is set to begin repetitive adaptive controller processing. If the flag is already set, adaptive controller processing 500 is required to calculate the predistortion parameter corrections at 516 by differential adjustment. The predistortion parameters are updated at 518 and the iterative process is repeated until the distortion is sufficiently reduced. If the distortion measured after completing the adaptive controller processing is small, the predistortion parameter setting and the current attribute vector are stored in the predistortion parameter list at 520. The predistortion parameter estimation routine is then completed. At this point, the predistortion parameter list may be deleted as previously described. At system shutdown or after timeout, the best predistortion parameter setting of the session may be selected at 522 and stored and used to initiate a quick start.

6-8, an additional optional feature of predistortion parameter list processing is shown using a hierarchy of elements. There is a trade off in selecting the number of elements in the list in the above approach. The advantage of allowing multiple elements is the dense coverage of the attribute spacing. However, the advantage of small elements is that less time is required to determine the element with the minimum attribute distance from the current vector. Using a hierarchical predistortion parameter list structure allows for both dense coverage and fast inspection.

The use of deletion can be used to generate hierarchical predistortion parameter lists as shown in FIGS. During the previously described duplicate deletion, when two elements are determined to be 'similar', older elements are deleted as duplicates and only the other elements survive. In a hierarchical approach to list management, duplicate elements are not deleted; Instead, it 'degrades' to a lower level subset below the surviving element. The generation of lower subsets is repetitive and allows to be defined as needed levels (in most cases, a subset level of zero or one is appropriate).

The basic deletion process flow used to generate hierarchical predistortion parameter lists is shown in FIG. 6. Deletion at 602 begins when the processor is not occupied by higher priority tasks. At 604, duplicate pairs of entries are identified. This process 604 is in accordance with the same distance calculation described in connection with FIG. 3 (in 304, 306, 308 and 310). At 606, older entries are marked as duplicate entries and the other is maintained at the current level of the hierarchical list structure. At 608 the delete processing flow checks if the list is at the lowest level. If so, at 610, duplicate entries are deleted. If the list is not at the lowest level, then duplicate entries are demoted to the next lower level at 612. The lower level position is marked as a sublist entry of the higher level survival entry. The next delete process flow proceeds to the next lower level at 614 and processing 602 begins at the level for the current entry sublist.

Duplicate elements may have subsets. This type of hierarchy is illustrated in FIGS. 7 and 8. Before the duplicate element 710 is demoted, the lower level subset list 714 is merged (at the same level) with the subset list 712 of surviving elements. The duplicate element 710 is then demoted to a lower level 704. Merging subsets at the same levels makes the subsets too large in a short period of time; However, when the deletion process begins, the subsets size will return to the desired value.

In the hierarchical list, the highest level 702 preferably has the coarsest sampling (largest value for threshold d _DPD ). Lower levels have very fine resolutions (smaller d _DPD ). By adjusting the threshold d _DPD for each level, it is possible to adjust the number of entries of a given predistortion parameter list at a given level. By increasing d _DPD , the number of list entries decreases. It is desirable to have all lists at various levels of the hierarchical structure with an approximately equal number of entries.

The processing flow for the hierarchical lists may follow the processing flow 501 of FIG. 5 as described above. However, the search for the element with the minimum attribute distance at 506 is limited to one set; This set is called 'active'. When search 506 begins, the top level set is active. The search for the predistortion parameter setting closest to the current attribute vector finds the closest entry in the highest level list, and then searches the subset list of entries. Once the element with the minimum attribute distance in the top level is identified, the predistortion parameters are retrieved (as described above with respect to FIG. 5). However, instead of requiring an adaptive controller as described above, the next lower level subset of elements becomes active. The next lower level subset is then retrieved with the element with the minimum attribute distance, and the new predistortion parameter setting is retrieved. The process is repeated cyclically until the next lower subset is empty. At this point an adaptive controller is required and the adaptive controller uses the retrieved predistortion parameter settings. This process is repeated cyclically until the bottom level is reached. If any intermediate settings provide sufficient distortion compensation, the process stops before requiring an adaptive controller. The subset search must include the parent entry (or parent entries if two or more levels are below the top) because it may be the best match.

The complexity of the search time is proportional to the product of the number of entries per list N and the number of levels in hierarchy L. In contrast, the exhaustive search is proportional to the power of N versus L, which will generally be higher enough. (This assumes that each list has N entries. It is noted that level L has N times as many subset lists as L-1, which means that level L has N for the power of L entries as a whole. do).

It is noted that deleted elements as part of the second deletion process described in connection with FIG. 4 above are out of date and therefore should not be observed at a lower level set.

To summarize the foregoing, the disclosed system combines multidimensional predistortion parameter list processing and adaptive controller processing to adjust the transmission gain of a digital predistortion amplifier system. The two processes are combined in a new way to improve the dynamic response of the system. The multidimensional predistortion parameter list used in the disclosed approach has a different structure compared to the array based reference table, thus avoiding the problem of the reference tables described above. Instead of storing elements using an array structure, the elements are collected as a set. Each element has the following: (a) a set of parameters or attributes corresponding to operating conditions affecting the amplifier; And (b) the best set of predistortion parameters found under the operating conditions. A metric is defined that defines the 'distance' between two elements, which is based on the differences between the attributes of the elements. If the digital predistortion system detects excessive distortion, the properties associated with the current operating conditions are measured. Then, the element of the predistortion parameter list with the minimum distance from the current attributes is identified and its corresponding predistortion parameter setting is retrieved from the repository. If the new predistortion parameter setting is not appropriate, the adaptive controller is activated to further reduce the distortion. Once the predistortion parameter setting is close enough to the optimum value, the predistortion parameter setting together with the current attributes are combined to form a new element in the set. Thus, the predistortion parameter lists are self generated.

In order to limit the computational complexity of examining the minimum distance element, it is desirable to limit the size of the element set. In order to identify duplicate elements, the similarity of the elements is measured by attribute distance and predistortion parameter separation (performed during rest time). If the set size exceeds a preset number, the oldest of a pair of similar elements is deleted. By limiting the list size, the time for determining the element with the minimum distance to the current attributes is controllable. If additional elements are required for greater coverage, it is possible to form a hierarchy of list levels. Instead of deleting the oldest similar element, it 'degrades' to a lower subset of levels below the surviving element. If the demoted element includes its lower level subsets, the element is merged with subsets of the surviving element. The generation of lower subsets is repetitive and allows many levels to be defined (in most cases, a subset level of zero or one is appropriate).

Since the hierarchical structure is limited in size for each active set, the check for the minimum distance element is computationally sufficient. Initially, the highest level set is activated. If the element with the smallest distance from the highest level set does not sufficiently reduce distortion, the subset (if present) is activated. The minimum distance match from the lower level subset is checked for distortion correction quality. The subsets are checked repeatedly until the residual distortion is low enough or the next lower subset is empty. In the latter case, the adaptive controller will be activated to improve distortion correction.

The combined operation of the adaptive predistortion controller and the disclosed predistortion parameter list is dynamic from learning and varying input power levels, changing (or hopping) carrier frequencies, changing temperature or DC supply, or component aging. Provide the disclosed transmitter system with the ability to improve performance in the presence of conditions. The system can accommodate any number of attributes (multidimensional index space) without a significant increase in complexity. Hierarchical set management allows any number of elements to be stored without significantly increasing the worst case delay in finding elements with minimum attribute distances.

An additional advantage of the disclosed predistortion parameter list is that the attribute space can be sampled unevenly. In general, the sensitivity of the predistortion parameter settings varies over the attribute space. The disclosed system will naturally form elements of higher density in regions with higher sensitivity, as desired.

The predistortion parameter lists provided by the present invention are useful for dynamic waveforms. For example, one application for a power amplifier system is for a hopping signal station. The predistortion parameter settings for each carrier in the application may be stored to allow for fast hopping (relative to conventional PA settling time).

In this respect, it will be appreciated that the present invention provides a number of desirable features. The combined use of predistortion parameter lists and adaptive controller processing provides fast convergence of a digital predistortion transmitter system using a power amplifier. The self-generating nature of the predistortion parameter list allows the system to learn from past experiences, reducing the inspection time required by the adaptive controller. Transient surges of distortion energy are reduced and the adaptive controller is more robust than when good initial estimates of predistortion parameter settings are provided. The hierarchical structure of the predistortion parameter list makes the search for the minimum distance element computationally efficient while providing wide convergence of the attribute space.

While the present invention has been described in connection with the presently preferred embodiments, those skilled in the art will recognize that many more variations can be made within the scope of the invention. Accordingly, it is to be understood that the above description is for the purpose of illustration only and not of limitation.

Claims (50)

In a digital RF transmitter system, An input for receiving a digital input signal; A digital predistorter using adjustable predistortion parameters to predistort the digital input signal, a digital to analog converter, an RF upconverter, an RF power amplifier, and an output sampling combiner to receive the amplified RF output signal. Transmission path; An observation path comprising a sampled output of the combiner and an analog to digital converter for providing an RF downconverter and an RF output signal sampled in a digital format; And An adaptive controller coupled to receive the digital input signal and the sampled digital output signal to determine residual distortion in the output signal, the adaptive controller coupled to receive one or more parameters that characterize an operating condition of an amplifier; A predistortion parameter list having a plurality of list elements and providing predistortion parameters coupled to the digital predistorter and adjusted to the digital predistorter to further reduce the distortion, each element operating an amplifier system. A digital RF transmitter system having one or more parameters and one or more predistortion parameters that characterize a condition. The method of claim 1, The adjustable predistortion parameters are part of an inverse gain model that changes the gain of the transmitter path as a function of nonlinear modes of input magnitude. The method of claim 1, The adjustable predistortion parameters include weights for both memoryless waveforms and memory based waveforms. The method of claim 1, The parameters characterizing the operating conditions of the amplifier may include one or more of temperature, DC power supply, input signal power, and input signal carrier frequency. The method of claim 1, The parameters characterizing the operating condition of the amplifier include an attribute vector, and the distance is defined between any two attribute vectors. The method of claim 5, The adaptive controller derives a current attribute vector from input parameters, calculates the distance to the attribute vectors of the list elements, and selects a list element with the minimum distance for use as predistortion parameters of the digital predistorter. Digital RF transmitter system. The method of claim 1, The adaptive controller uses the digital output and the digital input to continuously measure distortion and a set of predistortion parameters are retrieved from the predistortion parameter list when the measured distortion exceeds a predetermined value. . The method of claim 5, The adaptive controller continuously measures the attribute vector and a set of predistortion parameters are retrieved from the predistortion parameter list when a change in the measured attribute vector exceeds a predetermined value. The method of claim 1, Wherein the adaptive controller uses the selected element as an initial set of predistortion parameters and calculates new predistortion parameters from the initial set of parameters using an iterative control algorithm. The method of claim 9, And wherein the adaptive controller updates the predistortion parameter list with new predistortion parameters after completing the iterative calculation. The method of claim 5, And the distance between the closest attribute vectors can vary over the predistortion parameter list. An adaptive controller for controlling distortion compensation of an amplifier system, One or more inputs for receiving one or more attribute parameters corresponding to current operating conditions of the amplifier system; And One or more processors coupled to the one or more inputs, the one or more processors having an associated list of predistortion parameter settings to provide adjustments to the predistortion parameter settings to control the distortion correction of the amplifier system. Programmed with a predistortion parameter list algorithm and a controller algorithm, the predistortion parameter list algorithm generates the predistortion parameter list with the predistortion parameter settings calculated by the controller algorithm and generates respective predistortion parameter settings and one or more attributes. An adaptive controller that associates the parameters. The method of claim 12, Wherein the predistortion parameter list algorithm selects a predistortion parameter setting from the predistortion list for use by the controller algorithm at startup or when the output distortion is sufficiently large. The method of claim 12, The predistortion parameter list algorithm calculates a distance between one or more attribute parameters corresponding to current operating conditions and attribute parameters associated with each of the predistortion parameter settings of the list and corresponds to the attribute parameter having a minimum distance. And selecting a stored predistortion parameter setting for use by the controller algorithm by selecting a predistortion parameter setting. The method of claim 14, Wherein the distance calculation is weighted with different weights for different attribute parameters. The method of claim 12, The attribute parameters may include one or more of temperature, DC supply voltage, input signal power, and input signal carrier frequency. The method of claim 14, The distance d _attr between two sets of attribute parameters "n" and "0" is _defined by a weighted L _inf norm distance measurement or a weighted L ₂ norm distance measurement, Adaptive Controller. The method of claim 12, The adaptive controller further comprises one or more inputs for receiving output distortion data. The method of claim 18, Wherein the output distortion data comprises a baseband digital representation of an output signal provided from an observation path. The method of claim 12, Wherein the predistortion parameter settings are coefficients in a polynomial extension including band-limited nonlinear modes that model memory effects in the amplifier system. A method for controlling an amplifier system having an adaptive control loop comprising a control loop input, a first signal path comprising a digital predistorter and an RF amplifier, and a second signal path feeding back an output of the RF amplifier. Providing a list of predistortion parameter settings, each predistortion parameter setting having an associated operating condition; Detecting current operating conditions of the amplifier system; Comparing current operating conditions to conditions in the list of predistortion parameter settings; And Selecting the predistortion parameter setting associated with the most similar operating condition in the list. The method of claim 21, And the associated operating conditions are configured as a multidimensional attribute vector. The method of claim 21, Comparing current operating conditions to conditions in the list of predistortion parameter settings comprises measuring a distance between the current attribute vector and respective attribute vectors of the list. The method of claim 21, Selecting the predistortion parameter setting associated with the most similar operating condition in the list comprises determining an attribute vector with a minimum distance from the current operating condition attribute vector. The method of claim 21, Calculating a new predistortion parameter setting using an iterative adaptive controller algorithm, wherein the predistortion parameter setting associated with the most similar operating condition is used as an initial predistortion parameter setting for an adaptive controller algorithm. Control method. The method of claim 25, Updating the predistortion parameter list with the new predistortion parameter setting calculated by the adaptive controller algorithm. The method of claim 21, And the size of the predistortion parameter list is dynamic. The method of claim 23, And the spacing of the stored predistortion parameter settings defined by the attribute vector distance is variable throughout the list. The method of claim 23, Higher density predistortion parameter settings are provided in list regions and the distortion correction is most sensitive to one or more operating conditions that include the attribute vector. A method of maintaining a list of predistortion parameter settings of an adaptive digital predistortion amplifier system, the list comprising a number of elements, each element having a predistortion parameter setting and a set of operating condition parameters operating in the amplifier system. In the list maintaining method corresponding to the conditions, Selecting an element of the predistortion parameter list; Determining an element of the predistortion parameter list having operating conditions most similarly to the selected element; Determining whether two elements are sufficiently similar to be considered redundant; And Deleting the oldest of the two elements of the predistortion parameter list if the elements are duplicates. The method of claim 30, Selecting an element of the predistortion parameter list comprises selecting the oldest element of the list that was previously unaffected by the list maintenance process. The method of claim 30, Determining the element of the predistortion parameter list with operating conditions most similarly to the selected element comprises determining a distance measurement for operating condition parameter values of each of the remaining elements of the predistortion parameter list. And selecting the element with the minimum distance. The method of claim 32, And the distance measure comprises a weighted difference between operating condition parameter values. The method of claim 30, Operating condition parameters of the amplifier system include one or more of temperature, DC power supply, input signal power, and input signal carrier frequency. The method of claim 30, Determining whether the elements are sufficiently similar to be considered redundant includes determining a distance measure between the predistortion parameter settings and comparing the predistortion parameter distance to a duplicate distance threshold. Way. The method of claim 30, Determining whether elements are similar enough to be considered redundant includes comparing the distance between the operating condition parameters of the two elements against an old distance threshold. The method of claim 30, Repeating the list maintenance process for each element of the predistortion parameter list. A method of generating a hierarchical list of predistortion parameter settings of an adaptive digital predistortion amplifier system, the list comprising a number of elements, each element corresponding to a parameter corresponding to the predistortion parameter setting and operating conditions of the amplifier system. 10. A method of generating a hierarchical list, having a corresponding set of elements and having a hierarchical structure comprising at least two levels, Selecting an element at a first level of the predistortion parameter list; Determining the element in the first level of predistortion parameter list having operating conditions most similarly to the selected element; And Demoting the oldest of the two elements to a lower level of the hierarchical predistortion parameter list. The method of claim 38, Determining the element of the predistortion parameter list with operating conditions most similarly to the selected element comprises: measuring distances for operating conditions of each of the remaining elements of the predistortion parameter list of the first level; And determining the element with the minimum distance. The method of claim 38, Determining whether the two elements are duplicates, wherein the older element is only demoted if the elements are duplicates. The method of claim 38, Repeating the list process for each level of the hierarchical list. The method of claim 38, Deleting older entries if the list maintenance process is at the lowest level hierarchy. The method of claim 38, And the demoted element is associated as a subset list entry of non-demotiond duplicate elements. The method of claim 43, An element that is demoted and has one subset list is merged with the subset list of non-demotiond duplicate elements. An adaptive control loop comprising a control loop input, a first signal path, a second signal path, and a control loop output, wherein at least one of the first and second signal paths controls an amplifier system including an amplifier and a predistorter. In the way, Providing a hierarchical list of predistortion parameter settings having at least two levels, wherein each predistortion parameter setting has an associated operating condition and at some level all or some of the predistortion parameter settings are subordinate at a lower level. Providing the hierarchical list having set predistortion parameter settings; Detecting current operating conditions of the amplifier system and comparing the current operating conditions to the hierarchical list conditions of the highest level of predistortion parameter settings; Selecting a predistortion parameter setting associated with the most similar operating condition at the highest level of the list; Comparing current operating conditions to subset conditions of the selected highest level predistortion parameter setting; Selecting a predistortion parameter setting of the subset having the most similar operating condition; And Selecting a higher or lower level predistortion parameter setting having an operating condition most similar to the current operating condition. The method of claim 45, Repeating the processing for each level of the hierarchical list until the next lower subset is empty. The method of claim 45, And wherein the highest level has predistortion parameter settings of spacing that is coarser than the lower level. The method of claim 45, Wherein any two predistortion parameter settings have a predistortion parameter distance and the highest level has a greater predistortion parameter distance between the settings than the lower level. 49. The method of claim 48 wherein And the predistortion parameter distance comprises a weighted difference between the predistortion parameter settings. The method of claim 45, The predistortion parameter settings include inverse gain model settings and the weighting comprises a predistortion parameter sensitivity.

Images