Method and apparatus for determining subset of parameters from total parameters of digital predistortion in wireless communication system

Abstract

The disclosure relates to a 5G or 6G communication system for supporting a higher data transmission rate. Disclosed herein are method for a subset of parameters (Ps) from total parameters (Pt) for a Pt based digital pre-distorter (DPD). Initially, corresponding values of the Pt parameters are estimated, at a first time slot of first TIP slots. Further, the Pt parameters are sortec based on contribution of each of the Pt parameter to a first DPD output for the first time slot. Furthermore, the Ps parameters are determined by eliminating parameters with contribution lower than a predetermined threshold from the sorted Pt parameters in order that corresponding DPD outputs for the remaining slots of the first TIP slots are determined using corresponding values of the Ps parameters.

Description

METHOD AND APPARATUS FOR DETERMINING SUBSET OF PARAMETERS FROM TOTAL PARAMETERS OF DIGITAL PREDISTORTION IN WIRELESS COMMUNICATION SYSTEM

[0001] The present disclosure relates to the field of digital pre-distortion optimization, and more particularly relates to a method and apparatus for determining a subset of parameters (Ps) from total parameters (Pt) for a Ptbased digital pre-distorter (DPD).

[0002] 5G mobile communication technologies define broad frequency bands such that high transmission rates and new services are possible, and can be implemented not only in "Sub 6GHz" bands such as 3.5GHz, but also in "Above 6GHz" bands referred to as mmWave including 28GHz and 39GHz. In addition, it has been considered to implement 6G mobile communication technologies (referred to as Beyond 5G systems) in terahertz bands (for example, 95GHz to 3THz bands) in order to accomplish transmission rates fifty times faster than 5G mobile communication technologies and ultra-low latencies one-tenth of 5G mobile communication technologies.

[0003] At the beginning of the development of 5G mobile communication technologies, in order to support services and to satisfy performance requirements in connection with enhanced Mobile BroadBand (eMBB), Ultra Reliable Low Latency Communications (URLLC), and massive Machine-Type Communications (mMTC), there has been ongoing standardization regarding beamforming and massive MIMO for mitigating radio-wave path loss and increasing radio-wave transmission distances in mmWave, supporting numerologies (for example, operating multiple subcarrier spacings) for efficiently utilizing mmWave resources and dynamic operation of slot formats, initial access technologies for supporting multi-beam transmission and broadbands, definition and operation of BWP (BandWidth Part), new channel coding methods such as a LDPC (Low Density Parity Check) code for large amount of data transmission and a polar code for highly reliable transmission of control information, L2 pre-processing, and network slicing for providing a dedicated network specialized to a specific service.

[0004] Currently, there are ongoing discussions regarding improvement and performance enhancement of initial 5G mobile communication technologies in view of services to be supported by 5G mobile communication technologies, and there has been physical layer standardization regarding technologies such as V2X (Vehicle-to-everything) for aiding driving determination by autonomous vehicles based on information regarding positions and states of vehicles transmitted by the vehicles and for enhancing user convenience, NR-U (New Radio Unlicensed) aimed at system operations conforming to various regulation-related requirements in unlicensed bands, NR UE Power Saving, Non-Terrestrial Network (NTN) which is UE-satellite direct communication for providing coverage in an area in which communication with terrestrial networks is unavailable, and positioning.

[0005] Moreover, there has been ongoing standardization in air interface architecture / protocol regarding technologies such as Industrial Internet of Things (IIoT) for supporting new services through interworking and convergence with other industries, IAB (Integrated Access and Backhaul) for providing a node for network service area expansion by supporting a wireless backhaul link and an access link in an integrated manner, mobility enhancement including conditional handover and DAPS (Dual Active Protocol Stack) handover, and two-step random access for simplifying random access procedures (2-step RACH for NR). There also has been ongoing standardization in system architecture / service regarding a 5G baseline architecture (for example, service based architecture or service based interface) for combining Network Functions Virtualization (NFV) and Software-Defined Networking (SDN) technologies, and Mobile Edge Computing (MEC) for receiving services based on UE positions.

[0006] As 5G mobile communication systems are commercialized, connected devices that have been exponentially increasing will be connected to communication networks, and it is accordingly expected that enhanced functions and performances of 5G mobile communication systems and integrated operations of connected devices will be necessary. To this end, new research is scheduled in connection with eXtended Reality (XR) for efficiently supporting AR (Augmented Reality), VR (Virtual Reality), MR (Mixed Reality) and the like, 5G performance improvement and complexity reduction by utilizing Artificial Intelligence (AI) and Machine Learning (ML), AI service support, metaverse service support, and drone communication.

[0007] Furthermore, such development of 5G mobile communication systems will serve as a basis for developing not only new waveforms for providing coverage in terahertz bands of 6G mobile communication technologies, multi-antenna transmission technologies such as Full Dimensional MIMO (FD-MIMO), array antennas and large-scale antennas, metamaterial-based lenses and antennas for improving coverage of terahertz band signals, high-dimensional space multiplexing technology using OAM (Orbital Angular Momentum), and RIS (Reconfigurable Intelligent Surface), but also full-duplex technology for increasing frequency efficiency of 6G mobile communication technologies and improving system networks, AI-based communication technology for implementing system optimization by utilizing satellites and AI (Artificial Intelligence) from the design stage and internalizing end-to-end AI support functions, and next-generation distributed computing technology for implementing services at levels of complexity exceeding the limit of UE operation capability by utilizing ultra-high-performance communication and computing resources.

[0008] This summary is provided to introduce a selection of concepts, in a simplified format, that are further described in the detailed description of the invention. This summary is neither intended to identify key or essential inventive concepts of the invention nor is it intended for determining the scope of the invention.

[0009] According to one embodiment of the present disclosure, a method for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD is disclosed. The method includes estimating corresponding values of the Ptparameters, at a first time slot of first TIPslots, using a predefined technique. The method further includes sorting the Ptparameters based on contribution of each of the Ptparameter to a first DPD output for the first time slot. Furthermore, the method includes determining the Psparameters by eliminating parameters with contribution lower than a predetermined threshold from the sorted Ptparameterssuch that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the P parameters. In an embodiment, the corresponding values of the Psparameters is determined using the predefined technique.

[0010] According to one embodiment of the present disclosure, a method for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD is disclosed. The method includes estimating corresponding values of the Ptparameters, at a first time slot of first TIPslots, using a predefined technique. The method further includes determining the first DPD output for the first time slot using the corresponding values of the Ptparameters. Furthermore, the method includes determining the Psparameters from the Ptparameters by selecting the P parameters using the trained AI model, such that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the P parameters.

[0011] According to one embodiment of the present disclosure, a system for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD is disclosed. The system comprises a memory and a procerssor coupled with the memory. The processor is configured to estimate corresponding values of the Ptparameters, at a first time slot of first TIPslots, using a predefined technique. The processor is further configured to sort the Ptparameters based on contribution of each of the Ptparameter to a first DPD output for the first time slot. Furthermore, the processor is configured to determine the Psparameters by eliminating parameters with contributions lower than a predetermined threshold from the sorted Ptparameters such that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the P parameters. In an embodiment, the corresponding values of the Psparameters is determined using the predefined technique.

[0012] According to one embodiment of the present disclosure, a system for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD is disclosed. The system comprises a memory and a processor coupled with the memory. The processor is configured to estimate corresponding values of the Ptparameters, at a first time slot of first TIPslots, using a predefined technique. The processor is further configured to determine the first DPD output for the first time slot using the corresponding values of the Ptparameters. Furthermore, The processor is configured to determine the Psparameters from the Ptparameters by selecting the P parameters using the trained AI model, such that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the P parameters.

[0013] To further clarify the advantages and features of the present invention, a more particular description of the invention will be rendered by reference to specific embodiments thereof, which is illustrated in the appended drawing. It is appreciated that these drawings depict only typical embodiments of the invention and are therefore not to be considered limiting its scope. The invention will be described and explained with additional specificity and detail with the accompanying drawings.

[0014] The foregoing and other features of embodiments will become more apparent from the following detailed description of embodiments when read in conjunction with the accompanying drawings. In the drawings, like reference numerals refer to like elements.

[0015] FIG. 1A depicts PA used in a legacy system, according to a related art;

[0016] FIG. 1B is a flow diagram depicting exemplary rectification of non-linearity of PA using a DPD, according to a related art;

[0017] FIG. 2 is a block diagram depicting a transmitting device implementing a system for for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD, according to embodiments of the present disclosure;

[0018] FIG. 3 is a schematic diagram depicting TIPslots of a predefined number of TIPslots, according to an existing art;

[0019] FIG. 4 is a flow diagram depicting evaluation of EVM during implementation of an exemplary DPD, according to embodiments of the present disclosure;

[0020] FIG. 5 is a flow diagram depicting the non-AI based approach for determining the Psparameters from the Ptparameters for the Ptbased DPD, according to embodiments of the present disclosure;

[0021] FIG. 6 is a schematic diagram depicting an approach for training the AI model 211 for determining determining the Psparameters from the Ptparameters for the Ptbased DPD, according to embodiments of the present disclosure;

[0022] FIG. 7 is a schematic diagram depicting another approach for training the AI model for determining determining the Psparameters from the Ptparameters for the Ptbased DPD, according to embodiments of the present disclosure;

[0023] FIG. 8 is a flow diagram depicting a method for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD, according to an embodiment of the present disclosure; and

[0024] FIG. 9 is a flow diagram depicting a method determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD, according to an embodiment of the present disclosure.

[0025] Further, skilled artisans will appreciate that elements in the drawings are illustrated for simplicity and may not have necessarily been drawn to scale. For example, the flow charts illustrate the method in terms of the most prominent steps involved to help improve understanding of aspects of the present invention. Furthermore, in terms of the construction of the device, one or more components of the device may have been represented in the drawings by conventional symbols, and the drawings may show only those specific details that are pertinent to understanding the embodiments of the present invention so as not to obscure the drawings with details that will be readily apparent to those of ordinary skill in the art having the benefit of the description herein.

[0026] For the purpose of promoting an understanding of the principles of the present disclosure, reference will now be made to the various embodiments and specific language will be used to describe the same. It will nevertheless be understood that no limitation of the scope of the present disclosure is thereby intended, such alterations and further modifications in the illustrated system, and such further applications of the principles of the present disclosure as illustrated therein being contemplated as would normally occur to one skilled in the art to which the present disclosure relates.

[0027] It will be understood by those skilled in the art that the foregoing general description and the following detailed description are explanatory of the present disclosure and are not intended to be restrictive thereof.

[0028] Whether or not a certain feature or element was limited to being used only once, it may still be referred to as "one or more features" or "one or more elements" or "at least one feature" or "at least one element." Furthermore, the use of the terms "one or more" or "at least one" feature or element do not preclude there being none of that feature or element, unless otherwise specified by limiting language including, but not limited to, "there needs to be one or more ..." or "one or more elements is required."

[0029] Reference is made herein to some "embodiments." It should be understood that an embodiment is an example of a possible implementation of any features and / or elements of the present disclosure. Some embodiments have been described for the purpose of explaining one or more of the potential ways in which the specific features and / or elements of the proposed disclosure fulfill the requirements of uniqueness, utility, and non-obviousness.

[0030] Use of the phrases and / or terms including, but not limited to, "a first embodiment," "a further embodiment," "an alternate embodiment," "one embodiment," "an embodiment," "multiple embodiments," "some embodiments," "other embodiments," "further embodiment", "furthermore embodiment", "additional embodiment" or other variants thereof do not necessarily refer to the same embodiments. Unless otherwise specified, one or more particular features and / or elements described in connection with one or more embodiments may be found in one embodiment or may be found in more than one embodiment, or may be found in all embodiments, or may be found in no embodiments. Although one or more features and / or elements may be described herein in the context of only a single embodiment, or in the context of more than one embodiment, or in the context of all embodiments, the features and / or elements may instead be provided separately or in any appropriate combination or not at all. Conversely, any features and / or elements described in the context of separate embodiments may alternatively be realized as existing together in the context of a single embodiment.

[0031] Any particular and all details set forth herein are used in the context of some embodiments and therefore should not necessarily be taken as limiting factors to the proposed disclosure.

[0032] The terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of steps does not include only those steps but may include other steps not expressly listed or inherent to such process or method. Similarly, one or more devices or sub-systems or elements or structures or components proceeded by "comprises... a" does not, without more constraints, preclude the existence of other devices or other sub-systems or other elements or other structures or other components or additional devices or additional sub-systems or additional elements or additional structures or additional components.

[0033] A detailed methodology is explained in the following paragraphs of the disclosure.

[0034] An objective of the present disclosure is to provide a low-complexity digital pre-distorter (DPD) to achieve improved performance and reduce latency in a system implementing such DPDs.

[0035] The above-mentioned objective is achieved by providing a methodology for providing the low-complexity DPD by determining a subset of top parameters out of a total number of parameters of GMP or Volterra series based DPD, which solves the non-linear complex problem in less amount of time without any loss in performance.

[0036] Embodiments of the present disclosure will be described below in detail with reference to the accompanying drawings.

[0037] For the sake of clarity, the first digit of a reference numeral of each component of the present disclosure is indicative of the Figure number, in which the corresponding component is shown. For example, reference numerals starting with digit "1" are shown at least in Fig. 1. Similarly, reference numerals starting with digit "2" are shown at least in Fig. 2.

[0038] In wireless communication systems, Power Amplifiers (PA) are used to boost an envelope baseband signal before transmission. This helps the signal reach a receiver in spite of attenuation experienced over the transmission channel. However, the PA used in legacy system generally exhibit non-linearity at a higher input voltage, as depicted in FIG. 1A, where an output signal (y(n)) from the PA 8 is shown as a non-linear function of an input signal (u(n)) to the PA 8.

[0039] In the legacy system, baseband quadrature amplitude modulated (QAM) signals are converted to a time domain signal using an orthogonal frequency domain multiplexing (OFDM) before transmission. Further, the amplitude of the OFDM, i.e., envelope signal is non-constant or variable and has significant peak-to-average-power-ratio (PAPR). The non-linearity introduced in OFDM based systems causes severe distortions for both in-band signal and out-band signal because of high PAPR. This not only affects the signal quality but also causes spectrum regrowth.

[0040] Further, the non-constant or variable amplitude signal prompts the PA to operate at a lower input voltage than its saturation point. The difference between an operating point and the saturation point of the PA is called a PA-backoff. Thus, for a high PAPR signal, corresponding PA-backoff is kept higher, which affects the PA efficiency. Moreover, as the system bandwidth increases, PAs often exhibit memory effect. This means that output of a PA not only depends on the current sample but also on past few samples, i.e., the PA keeps the memory of past input samples and provide the output based on current and past few samples.

[0041] The above-described limitations associated with non-linearity and memory effect in PAs are addressed by utilizing a digital pre-distorter (DPD). A DPD modifies the baseband signal such that the modified signal when passed through the PA results in linear amplification of the baseband signal as depicted in FIG. 1B.

[0042] FIG. 1B is a flow diagram depicting exemplary rectification of the non-linearity of PA using a DPD 20, according to a related art. As shown in the figure, the DPD 20 modifies the input signal (u(n)) by determining an inverse of the distortion caused by the PA 8 and applying the inverse to (u(n)) resulting in a pre-distorted signal (x(n)). Consequently, when the pre-distorted signal (x(n)) passes through the PA 10, the distortion introduced by the PA is effectively canceled, producing a linear output signal (y(n)).

[0043] Conventionally, for PAs exhibiting memory effect, a DPD may be modeled using memory polynomial functions such as Volterra series which is used to model non-linear system with memory. The Volterra series is a combination of various coefficients with every type of cross-memory product terms for all the combinations of exponents. Further, in practice, various derivatives of the Volterra series are extensively used to model the DPD such as memory polynomial (MP), generalized memory polynomial (GMP), Wiener, Hammerstein, Wiener-Hammerstein, and parallel Wiener model. Among the various derivatives of the Volterra series, the GMP is most widely used to model a DPD. The DPD modeled using GMP is described using the equation 1.

[0044] [Equation 1]

[0045]

[0046] Where refers to the features of the GMP-DPD and refers to the parameters (ai) of the GMP-DPD. Thus, the equation (1) may also be represented as equation 2 below.

[0047] [Equation 2]

[0048]

[0049] Where N refers to the total number of parameters, i.e., all combinations ofk,m,l.

[0050] However, implementing the Volterra series or derivatives (such as GMP) thereof is highly complex as the number of parameters are in the hundreds and thousands. Such high complexity deeply affects the efficiency of the system implementing a memory polynomial based DPD due to significantly high complexity involved in finding the coefficients of said derivatives.

[0051] As mentioned above, the DPD may be the memory polynomial based DPD, such as GMP-DPD, where the GMP is characterized as product of a complex term with an exponent of an absolute term of the input signal. Further, referring to the equation (1), the parameters of the GMP-DPD are obtained using a least square (LS) method or a recursive least square (RLS) method with indirect learning approach (ILA). According to ILA, PA input x(n) and PA output y(n) are used to estimate the parameters of the GMP-DPD. Further, consider G as the average gain of the PA. Since the objective of the DPD is to obtain linearity between y(n) and u(n), y(n) / G is considered as u(n) to the DPD and is used for determining parameters of the GMP-DPD.

[0052] Referring to GMP-DPD equation 1, where are the parameters of the DPD and are the features of the DPD, assume a vector of all 'N' parameters as and all features as . Therefore, the DPD equation 1 can be written for a block of length L as equation 3 below.

[0053] [Equation 3]

[0054]

[0055] Further, assuming can be estimated as LS solution as: . Thus, for estimation of N parameters using the data of block size L, the number of complex multiplications are approximately . As shown in equation 1, the GMP-DPD can be written as product of two terms where the first term is a complex value and the second term is an absolute value. Further, the memory polynomials, such as GMP and Volterra involve a large number, i.e., in hundreds and thousands, of parameters, especially when dealing with complex nonlinear systems like power amplifiers. Such high dimensionality significantly impacts the complexity of the system and computational requirements for the DPD implementation.

[0056] Further, the memory polynomials have certain memory (M), and associated order (K) which contribute to the DPD output. Thus, the total number of parameters that can be estimated for GMP corresponds to . Moreover, when the Volterra series as depicted in equation 4below is used, the total number of parameters that can be estimated corresponds to KM.

[0057] [Equation 4]

[0058]

[0059] Furthermore, as system bandwidth further increases in 6thgeneration (6G), the nonlinear behavior and memory effects of PAs are expected to become more pronounced due to the higher frequencies and faster modulation schemes used in 6G. Thus, to compensate for these increased nonlinearities and memory effects, the DPDs used in 6G will require a larger number of parameters to accurately model the PA's behavior. Consequently, implementing such DPD with high complexity would result in slower system performance.

[0060] Therefore, in view of the above-mentioned problems, it is advantageous to provide an improved system and method that can overcome the above-mentioned problems and limitations associated with the existing implementation of DPDs.

[0061] FIG. 2 is a block diagram depicting a transmitting device 200 implementing a system 201 for for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD, according to embodiments of the present disclosure. Examples of the transmitting device 200 may include, but are not limited to, a base station, a mobile station, and any other wireless network device employing a non-linear amplifier and a digital pre-distorter (DPD) to compensate for distortions introduced by the non-linear amplifier.

[0062] In one or more embodiments, the transmitting device 200 comprises the system 201. The system 201 may include a power amplifier (PA) 203, a DPD 205, a memory 207, a processor 209, an artificial intelligence (AI) model 211, and a plurality of modules 213, and transceiver 230.

[0063] In one or more embodiments, the DPD 205 may be a total parameters(Pt) based DPD. In one or more embodiments, the DPD 205 and methods according to the principles of present disclosure may be implemented in the PA 203. According to embodiments of the present disclosure, the PA 203 may be a non-linear amplifier. In one or more embodiments, the DPD 205 and methods according to the principles of present disclosure may be implemented in any type of non-linear amplifier in the transmission device 200.

[0064] In one or more embodiments, the PA 203 may be modeled as a Gallium Arsenide (GaAs) or Gallium Nitride (GaN) PA which are widely used and advised by the 3GPP for research and development.

[0065] In one or more embodiments, the memory 207 stores instructions to be executed by the processor 209 for determining a subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD 205, as discussed throughout the disclosure. The memory 207 may include non-volatile storage elements. Examples of such non-volatile storage elements may include magnetic hard discs, optical discs, floppy discs, flash memories, or forms of electrically programmable memories (EPROM) or electrically erasable and programmable (EEPROM) memories. In addition, the memory 207 may, in some examples, be considered a non-transitory storage medium. The term "non-transitory" may indicate that the storage medium is not embodied in a carrier wave or a propagated signal. However, the term "non-transitory" should not be interpreted that the memory 207 is non-movable. In some examples, the memory 207 can be configured to store larger amounts of information than the memory. In certain examples, a non-transitory storage medium may store data that can, over time, change (e.g., in Random Access Memory (RAM) or cache). The memory 207 can be an internal storage unit, or it can be an external storage unit of the HMD 200, a cloud storage, or any other type of external storage.

[0066] In one or more embodiments, the processor 209 communicates with the the PA 203, the DPD 205, memory 207, the AI model 211, and transceiver 230. The processor 209 is configured to execute instructions stored in the memory 210 and to perform various processes for determining the subset of parameters (Ps) from the total parameters (Pt) for the Ptbased DPD 205 in the transmitting device 200, as discussed throughout the disclosure. The processor 209 may include one or a plurality of processors, maybe a general-purpose processor, such as a Central Processing Unit (CPU), an Application Processor (AP), or the like, a graphics-only processing unit such as a Graphics Processing Unit (GPU), a Visual Processing Unit (VPU), and / or an Artificial Intelligence (AI) dedicated processor such as a Neural Processing Unit (NPU).

[0067] In one or more embodiments, the processor 209 is implemented by processing circuitry such as logic gates, integrated circuits, microprocessors, microcontrollers, memory circuits, passive electronic components, active electronic components, optical components, hardwired circuits, or the like, and may optionally be driven by firmware. The circuits may, for example, be embodied in one or more semiconductor chips, or on substrate supports such as printed circuit boards and the like.

[0068] In one or more embodiments, the AI model 211 refers to a predetermined machine learning-based model for enabling determination of the subset of parameters (Ps) from the total parameters (Pt) for the Ptbased DPD 205. The AI model 211 may be trained according to embodiments of the present disclosure. In an embodiment, the AI model 211 may be trained on the transmitting device 200. In an embodiment, the AI model 211 may be pre-trained externally and thereafter stored in the transmitting device 200.

[0069] The AI model 211 may consist of a plurality of neural network layers. Each layer has a plurality of weight values and performs a layer operation through the calculation of a previous layer and an operation of a plurality of weights. Examples of neural networks include but are not limited to, convolutional neural network (CNN), deep neural network (DNN), recurrent neural network (RNN), restricted Boltzmann Machine (RBM), deep belief network (DBN), bidirectional recurrent deep neural network (BRDNN), generative adversarial networks (GAN), and deep Q-networks.

[0070] The learning technique is a method for training a predetermined target device (for example, a robot) using a plurality of learning data to cause, allow, or control the target device to decide or predict. Examples of learning techniques include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0071] As mentioned above, the AI model 211 may be obtained by training. Here, "obtained by training" means that a predefined operation rule or artificial intelligence model configured to perform a desired feature (or purpose) is obtained by training a basic AI model with multiple pieces of training dataset by a training technique. The AI model may include a plurality of neural network layers. Each of the plurality of neural network layers includes a plurality of weight values and performs neural network computation by computation between a result of computation by a previous layer and the plurality of weight values.

[0072] In one or more embodiments, the plurality of modules 213 may include a program, a subroutine, a portion of a program, a software component, or a hardware component capable of performing a stated task or function. As used herein, the plurality of modules 213 may be implemented on a hardware component such as a server independently of other modules 213, or a module can exist with other modules 213 on the same server, or within the same program. The plurality of modules 213 may be implemented on a hardware component, such as processor one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, and / or any devices that manipulate signals based on operational instructions. The plurality of modules 213 when executed by the processor(s) 205 may be configured to perform any of the functionalities discussed herein. In an embodiment, a subset of the plurality of modules 213 may be implemented within the transmitting device 200, while another subset of the plurality of modules 213 may be implemented remotely for training the AI model 211. In another embodiment, the plurality of modules 213 may be implemented on the transmitting device 200.

[0073] In an embodiment, the plurality of modules 213 may be implemented using the AI model 211 which may include a plurality of neural network layers. Examples of neural networks include but are not limited to, Convolutional Neural Network (CNN), Deep Neural Network (DNN), Recurrent Neural Network (RNN), and Restricted Boltzmann Machine (RBM). Further, 'learning' may be referred to in the disclosure as a method for training a predetermined target device (for example, a robot) using a plurality of learning data to cause, allow, or control the target device to make a determination or prediction. Examples of learning techniques include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning. At least one of a plurality of CNN, DNN, RNN, RBM models and the like may be implemented to thereby achieve execution of the present subject matter's mechanism through an AI model.

[0074] A function associated with an AI module may be performed through the non-volatile memory, the volatile memory, and the processor. The processor may include one or a plurality of processors. At this time, one or a plurality of processors may be a general-purpose processor, such as a central processing unit (CPU), an application processor (AP), or the like, a graphics-only processing unit such as a graphics processing unit (GPU), a visual processing unit (VPU), and / or an AI-dedicated processor, such as a neural processing unit (NPU). One or a plurality of processors control the processing of the input data in accordance with a predefined operating rule or artificial intelligence (AI) model stored in the non-volatile memory and the volatile memory. The predefined operating rule or artificial intelligence model is provided through training or learning.

[0075] The plurality of modules 213 may include a set of instructions that may be executed according to the embodiments of the present disclosure to detect and eliminate grid pattern noise from images captured using the transmitting device 200. The plurality of modules 213 is described in detail in the forthcoming paragraphs.

[0076] According to embodiments of the present disclosure, the plurality of modules 213 include an estimation module 215, a determination module 217, a sorting module 219, a validity assessment module 221, and a training module 222.

[0077] In one or more embodiments, the estimation module 215 estimate, for each TIPslots of a predefined number of TIPslots as shown in FIG. 3, corresponding values of the Ptparameters, at a first time slot of first TIPslots, using a predefined technique. In an embodiment, the predefined technique may correspond to a technique implementing a least square (LS) method or a recursive least square (RLS) method using an indirect learning approach (ILA). The predefined number of TIPslots may be specific to behavior and hardware type of the PA, heating of the device etc. Thereafter, the estimation module 215 determines a first DPD output for the first time slot using the estimated corresponding values of the Ptparameters.

[0078] In an embodiment, the sorting module 219 sorts the Ptparameters based on contribution of each of the Ptparameter to the first DPD output for the first time slot.

[0079] In an embodiment, the determination module 217 determines the Psparameters, using a non-AI based approach, by eliminating parameters with contribution lower than a predetermined threshold from the sorted Ptparameters such that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the Psparameters. According to embodiments of the present disclosure, the corresponding values of the Psparameters are determined using the predefined technique.

[0080] In an embodiment, in an AI-based approach, the determination module 217 determines the Psparameters using the trained AI model 211. In an example, the AI model 211 may be pre-trained using the training module 223 according to embodiments of the present disclosure.

[0081] In an embodiment, when the Psparameters are determined using the non-AI based approach, the validity assessment module 221 assessing validity of the determination of the Psparameters based on one or more predefined conditions associated with performance of the DPD. According to embodiments of the present disclosure, the one or more predefined conditions include, for a second DPD output, an error vector magnitude (EVM) lower than an EVM threshold (ThEVM) and an adjacent channel power leakage ratio (ACLR) lower than an ACLR threshold (ThACLR). In an embodiment, the second DPD output is determined using the corresponding values of the Psparameters.

[0082] The EVM refers to an error between an ideal transmitted signal before passing through the DPD 205 and PA 203, and received signal after passing through the DPD 205 and PA 203 at transmitting device 200 side and equalization at a receiving device side. An exemplary evaluation of EVM is described in conjunction with FIG. 4.

[0083] FIG. 4 is a flow diagram depicting the evaluation of EVM during implementation of an exemplary DPD 205, according to embodiments of the present disclosure. For instance, consider a radio frequency (RF) transmitter (Tx), which transmits OFDM signal of bandwidth 10 MHz, subcarrier frequency (SCS) 15 kHz, and a fast Fourier transform (FFT) size 2048. As shown in the figure, a random bit stream corresponding to the OFDM signal is generated and passed through a modulation block, such as a 256-QAM, to generate QAM symbols (S(f)). Further, the generated QAM symbols S(f) are multiplexed through an inverse discrete Fourier transform (IDFT) block, and then cyclic prefix (CP) is added and time domain signal (s(n)) is obtained. Thereafter, the time domain signal (s(n)) is up-sampled before passing through the PA functional blocks.

[0084] Further, to keep the PAPR low or the PA-backoff low, a crest factor reduction (CFR) block is used. The CFR-block clips the input signal voltage corresponding to s(n) beyond a threshold and an input u(n) to the DPD 205 is obtained. The DPD 205 generates a polynomial function of u(n). For evaluating EVM, the PA 203 output y(n) is first down-sampled to obtain s~(n), and then the CP is removed to obtain S~(f). Further, is obtained by equalizing the S~(f). According to an embodiment, the EVM is evaluated using equation 5 below.

[0085] [Equation 5]

[0086]

[0087] In one or more embodiments, in order to validate the performance of the DPD 205, the EVM is required to be lower than ThEVM.

[0088] Referring to the validity assessment module 221, for an upper frequency end and a lower frequency end of a frequency channel associated with the signals transmitted from the transmitting device 200, the ACLR refers to ACLR is used as a measure of the amount of power leaking into adjacent channels and is defined as the ratio of the filtered mean power centered on the assigned channel frequency to the filtered mean power centered on an adjacent channel frequency. Furthermore, the performance of the DPD 205 is validated when the ACLR of the second DPD is lower than the ThACLR, in addition to being lower than the ThEVM. According to embodiments of the present disclosure, the ThACLRand ThEVMcorrespond to the minimum threshold values required to achieve an improved quality of signal generated using the transmitting device 200 implementing the DPD 205. Further, the non-AI based approach and AI-based approach for determining the Psparameters from the Ptparameters for the Ptbased DPD 205 using one or more modules of the plurality of modules 213 are described below in the forthcoming paragraphs.

[0089] FIG. 5 is a flow diagram depicting the non-AI based approach for determining the Psparameters from the Ptparameters for the Ptbased DPD 205, according to embodiments of the present disclosure. The non-AI based approach is an iterative approach for determining the Psparameters and includes a operations 501 through 509 performed for each TIPslots of a predefined number of TIPslots using one or more modules of the plurality of modules 213, according to embodiments of the present disclosure.

[0090] At operation 501, the estimation module 215 estimates corresponding values of the Ptparameters, at the first time slot of the first TIPslots, using the predefined technique. Examples of the predefined technique includes, but are not limited to, techniques using LS method and RLS method using ILA.

[0091] At operation 502, the estimation module 215 determines the first DPD output using the corresponding values of Ptparameters. In an embodiment, the first DPD output may be determined using the memory polynomial function such as GMP-DPD defined in equation 1. In an embodiment, the first DPD output may be determined using the memory polynomial function such as Volterra series based DPD as defined in equation 4.

[0092] At operation 503, the sorting module 219 sorts the Ptparameters based on contribution of each of the Ptparameter to the first DPD output for the first time slot.

[0093] At operation 504, the determination module 217 determines the Psparameters by eliminating parameters with contribution lower than a predetermined threshold from the sorted Ptparameters.

[0094] At operation 505, the determination module 217 determines corresponding values of the Psparameters using the predefined technique.

[0095] At operation 506, the determination module 217 determines second DPD output using the corresponding values of the Psparameters. In an embodiment, the second DPD output may be determined using the memory polynomial function such as GMP-DPD defined in equation 1. In an embodiment, the second DPD output may be determined using the memory polynomial function such as Volterra series based DPD as defined in equation 4.

[0096] At operation 507, the validity assessment module 221 validity of the determination of the Psparameters based on the one or more predefined conditions associated with performance of the DPD. According to embodiments of the present disclosure, the one or more predefined conditions include, for the second DPD output, the EVM lower than the ThEVMand the ACLR lower than the ThACLR. Thus, the validity assessment module 221 determines whether the EVM and ACLR of the second DPD output are lower than the ThEVMand the ThACLR, respectively.

[0097] According to embodiments of the present disclosure, the Psis adjusted by either increasing the number of Psor decreasing the number of Psbased on the assessed validity of the determination of the Psparameters.

[0098] At operation 508, when the one or more preconditions are met and the determination of parameters is determined to be valid, the number of Psis decreased such that the Psis greater than a first threshold associated with a minimum number of Psparameters below which the performance of the DPD is determined to be invalid.

[0099] At operation 509, when any one of the one or more preconditions are not met and the determination of parameters is determined to be invalid, the number of Psis increased such that the increased Psis less than the Ptparameters and lower than a second threshold associated with a maximum number of Psparameters at which the performance of the DPD is valid.

[0100] According to embodiments of the present disclosure, upon determination of the Psparameters at the first slot of the first TIPslots, corresponding DPD outputs for the remaining slots of the the first TIPslots may be determined using corresponding values of the Psparameters, wherein the corresponding values of the Ps parameters is determined using the predefined technique.

[0101] In an example, consider total parameters Ptof DPD 205 as 105. Further, consider ThEVMis 3% such that if EVM of a DPD output is less than 3%, signal quality corresponding to the said DPD output is considered good. In a first iteration, as shown below in table 1, of the non-AI based approach, the Ps= 83 parameters out of Pt= 105 parameters are selected.

[0102] ParametersEVM (RMS)ACPR105 (all parameters)1.25 %-51 dBc90 (1stIter)0.52 %-51 dBc83 (1stIter) (GOOD)0.71 %-51 dBc75 (1stIter) : (POOR)11.9 %-21 dBc50 (1stIter)13.8 %-20 dBc22 (1stIter)14.38 %-16 dBc

[0103] In a second iteration, as shown below in Table 2, of the non AI-based approach, the Ps= 50 parameters out of Pt= 83 parameters are selected.

[0104] ParametersEVM (RMS)ACPR83 (1stIter)0.71 %-51 dBc75 (2ndIter) (Good)0.50 %-52 dBc50 (2ndIter) (Good)0.59 %-51 dBc

[0105] According to an embodiment of the present disclosure, the ranking of Ps= 83 parameters selected in the first iteration from Pt= 105 parameters is not the same as the ranking of Ps= 50 parameters selected in the second iteration from Pt= 83 parameters. Furthermore, in a third iteration, as shown below in Table 3, of the non-AI based approach, the Ps= 15 parameters out of Pt= 50 parameters are selected.

[0106] CoeffEVM (RMS)ACPR50 (2ndIter)0.59 %-51 dBc15 (3rd Iter) (Good)1.02 %-43 dBc10 (3rdIter) (Poor)12.01 %-22 dBc

[0107] Therefore, in the above example, using the non-AI based approach according to embodiments of the present disclosure, DPD 205 may be implemented using only 15 parameters out of a total 105 parameters without degrading the quality of the output signal.According to embodiments of the present disclosure, in an AI-based approach for determining determining the Psparameters from the Ptparameters for the Ptbased DPD 205. The AI-based approach may be implemented for determining the Psparameters from the Ptparameters for each TIPslots of the predefined number of TIPslots.

[0108] According to the AI-based approach, the estimation module 215 estimates corresponding values of the Ptparameters, at a first time slot of first TIPslots. Thereafter, a first DPD output is determined for the first time slot using the corresponding values of the Ptparameters. In an embodiment, the first DPD output may be determined using the memory polynomial function such as GMP-DPD defined in equation 1. In an embodiment, the first DPD output may be determined using the memory polynomial function such as Volterra series based DPD as defined in equation 4.

[0109] Further, the determination module 217 determines the Psparameters from the Ptparameters by selecting the P parameters using the AI model 211. According to embodiments of the present disclosure, the AI model 211 may be trained using the training module 223. Said training of the AI model 211 is described below in the forthcoming paragraphs in conjunction with FIGS. 6 and 7.

[0110] FIG. 6 is a schematic diagram depicting an approach for training the AI model 211 for determining determining the Psparameters from the Ptparameters for the Ptbased DPD 205, according to embodiments of the present disclosure. In an embodiment, the AI model 211 may be a concrete selector AI model (AI-CS).

[0111] As shown in the figure, the training module 223 provides corresponding features of the Ptparameters, i.e., a total N parameters, as input to an input layer of the AI model 211. Further, the training module 223 selects aset of parameters from the N parameters at a second layer 603 of the AI model 211 during the training process. In the first iteration of the training process, the second layer select the parameters randomly but as the training goes on, the parameter selection becomes the best possible set of parameters with minimum error between AI-model-output and the expected outputs / ground-truth. The second layer 603 may refer to a layer of concrete random variable (c.r.v), which is a continuous relaxation of the one hot vector and the relaxation is controlled by a variable which is called T (0,∞). Accordingly, the second layer 603 after the input layer 601 consists of c.r.v , where .

[0112] Where j=1..N, are drawn from Gumbel distribution and are learnable parameter. Further, as , , and becomes one hot vector resulting in selecting one of the feature out of N. Hence, the output of the second layer 603 is for i = {1...P} resulting in the P selected features.

[0113] Further, the training module 223, at a third layer 605 of the AI model 211, separates the corresponding real part and corresponding imaginary part of each of the initial set of parameters. Further, the training module 223, at a dense layer 607 of the AI model 211, simulates complex multiplication of the corresponding real part and the corresponding imaginary part by a first shared model (SM-1) and a second shared model (SM-2) to obtain a second DPD output as an AI output. According to embodiments of the present disclosure, the second DPD output is obtained by combining a real part of the second DPD output obtained from the SM-1 and an imaginary part of the second DPD output obtained from the SM-2.

[0114] Furthermore, the training module 223 trains the AI model 211 to determine the Psparameters by selecting the best combination of the parameters such that a mean square error between the first DPD output and the AI output is minimized.

[0115] In an example, consider a GMP-DPD with non-linearity order K = 4 and memory M = 14, which results in total parameters Pt= 602. The second layer 603 of the AI model 211 may select Ps= {10, 25, 50, 100}}, where the AI model 211 has an input complex vector of size 602, and the output size is 1. Further, 16 neurons for P = {50, 100} and 5 neurons for P = {10, 25} may be utilized in the SM-1 and SM-2 models. Further, linear activation is used and use of regularizer and biases is avoided in the layers so that only the features contribute to the DPD output prediction. Furthermore, an adaptive moment (ADAM) optimizer and mean square error as the loss function. Moreover, the AI model 211 is run the concrete random variables becomes one hot vectors with the mean probability of 0.99. Finally, the AI model 211 is able to select top 50 parameters of the total 602 parameters without any drop in performance.

[0116] FIG. 7 is a schematic diagram depicting another approach for training the AI model 211 for determining determining the Psparameters from the Ptparameters for the Ptbased DPD 205, according to embodiments of the present disclosure. In an alternative embodiment, the AI model 211 may be a perturbation AI model (AI-perturb).

[0117] As shown in the figure, the training module 223 provides the Ptparameters as input to an input layer 701 of the AI model 211. Further, the training module 223 separates corresponding real part and corresponding imaginary part of each of the Ptparameters. Furthermore, the training module 223 simulates, at a dense layer of the AI model 211, complex multiplication of the corresponding real part and the corresponding imaginary part by a third shared model (SM-3) and a fourth shared model (SM-4) to obtain a real part of the third DPD output from the SM-3 and an imaginary part of the third DPD output from the SM-4. Finally, the training module 223 trains the AI model 211 to determine the third DPD output by combining the real part from the SM-3 and the imaginary part from the SM-4.

[0118] Further, when the AI model 211 corresponds to the perturbation AI model and trained by the training module 223 in accordance with the embodiments of the present disclosure described in conjunction with FIG. 7, such trained AI model may be used to determine the Psparameters from the Ptparameters in Ptiterations corresponding to the Ptparameters.

[0119] For determining the Psparameters from the Ptparameters using the trained AI model 211, in each iteration of the Ptiterations, a predefined delta ( ) is added to corresponding feature of one of the Ptparameters. Thereafter, corresponding DPD outputs are determined by the trained AI model 211 for the corresponding iterations using the one of the Ptparameters with the predefined delta ( ), and corresponding perturbations in the corresponding DPD outputs are determined with respect to the first DPD output.

[0120] Further, the Ptparameters are sorted with respect to the corresponding perturbations in an order such that a parameter with a maximum corresponding perturbation is ranked highest and the Psparameters are determined from the sorted Ptparameters by selecting the Psparameters with least corresponding perturbations.

[0121] In the above example, where the GMP-DPD with non-linearity order K = 4 and memory M = 14 is used, which results in total parameters Pt= 602. The AI model 211 has an input complex vector of size 602, and the output size is 1. Further, 64 neurons and 16 neurons are used respectively in the two layers of the SM-3 and SM-4 models. Further, linear activation is used and the use of regularizers and biases is avoided in the layers so that only the features contribute to the DPD output prediction. Furthermore, an adaptive moment (ADAM) optimizer and mean square error as the loss function. Moreover, the AI model 211 is run for approximately for 500 epochs. Finally, the AI model 211 is able to select the top 50 parameters of the total 602 parameters without any drop in performance.

[0122] Further, the performance of GaN PAs and GaAs PA implementing the DPD 205 with any of the approaches according to embodiments of the present disclosure may be compared using Table 4 through Table 7. For example, Table 4 depicts a comparative analysis of EVM performance for GaN PA implementing DPD using the non-AI based approach, the AI based approach using AI-CS, and the AI based approach using AI-perturb.

[0123] PA without DPDGMP-DPD (Pt=602)4.37%3.02%Selected Params (Ps)100502510Non-AI3.01%3.02%3.07%3.16%AI-CS3.01%3.02%3.05%3.11%AI-Perturb3.01%3.02%3.02%3.11%

[0124] Table 5 depicts a comparative analysis of ACLR performance in decibels (dB) for GaN PA implementing DPD using the non-AI based approach, the AI based approach using AI-CS, and the AI based approach using AI-perturb.

[0125] PA without DPDGMP-DPD (Pt=602)-42.36-69.02Selected Params (Ps)100502510Non-AI-72.92-63.78-54.26-52.77AI-CS-71.96-69.05-64.62-56.73AI-Perturb-70.71-69.32-67.11-52.43

[0126] Table 6 depicts a comparative analysis of EVM performance for GaA PA implementing DPD using the non-AI based approach, the AI based approach using AI-CS, and the AI based approach using AI-perturb.

[0127] PA without DPDGMP-DPD (Pt=602)5.31%3.11%Selected Params (Ps)100502510Non-AI4.61%5.87%8.68%51.21%AI-CS3.01%3.01%3.01%3.07%AI-Perturb3.00%3.02%3.01%3.21%

[0128] Table 7 depicts a comparative analysis of ACLR performance in decibels (dB) for GaA PA implementing DPD using the non-AI based approach, the AI based approach using AI-CS, and the AI based approach using AI-perturb.

[0129] PA without DPDGMP-DPD (Pt=602)-36.47-55.12Selected Params (Ps)100502510Non-AI-48.04-47.17-46.71-40.41AI-CS-54.13-54.46-54.44-52.23AI-Perturb-53.16-54.28-53.28-43.93

[0130] Further, using the systems and methods described herein for selecting the subset of parameters (Ps) from the total parameters (Pt) for the Ptbased digital pre-distorter (DPD), the complexity of the system may be reduced by approximately hundred times, as shown in Table 8 below. Table 8 depicts a complexity analysis of the using the total Ptparameters and Psparameters.

[0131] ParametersPt= 602Ps= 100Ps= 50Ps= 25Ps= 10LS estimate3630010102556511DPD calculation60.21052.51Total36360102026067.512

[0132] FIG. 8 is a flow diagram depicting a method 800 for determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD, according to an embodiment of the present disclosure. The method 800 includes a series of operations 801 through 805 executed by one or more components of the transmitting device 200, in particular the processor 209 for each TIPslots of a predefined number of TIPslots.At step 801, the processor 209 estimates corresponding values of the Ptparameters, at a first time slot of first TIPslots, using a predefined technique.

[0133] At step 802, the processor 209 sorts the Ptparameters based on contribution of each of the Ptparameter to a first DPD output for the first time slot. In an embodiment, the first DPD output is determined using the corresponding values of the Ptparameters.

[0134] At step 803, the processor 209 determines the Psparameters by eliminating parameters with contribution lower than a predetermined threshold from the sorted Ptparameters such that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the P parameters. In an embodiment, the corresponding values of the Psparameters is determined using the predefined technique.

[0135] At step 804, the processor 209 assesses validity of the determination of the Psparameters based on one or more predefined conditions associated with performance of the DPD.

[0136] At step 805, the processor 209 adjusts the Ps, based on the assessed validity, such that the performance of the DPD with the Psparameters is valid. In an embodiment, the Psis adjusted by either increasing the number of Psor decreasing the number of Ps.

[0137] FIG. 9 is a flow diagram depicting a method 900 determining the subset of parameters (Ps) from total parameters (Pt) for the Ptbased DPD, according to an embodiment of the present disclosure. The method 900 includes a series of operations 901 through 903 executed by one or more components of the transmitting device 200, in particular the processor 209 for each TIPslots of a predefined number of TIPslots.

[0138] At step 901, the processor 209 estimates corresponding values of the Ptparameters, at a first time slot of first TIPslots, using the predefined technique.

[0139] At step 902, the processor 209 determines the first DPD output for the first time slot using the corresponding values of the Ptparameters.

[0140] At step 903, the processor 209 determines the Psparameters from the Ptparameters by selecting the P parameters using the trained AI model, such that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the P parameters. In an embodiment, the corresponding values of the P parameters is determined using the predefined technique.

[0141] In an embodiment, the AI model may be trained by providing corresponding features of the Ptparameters as input to an input layer of the AI model, selecting the initial set of parameters at a second layer of the AI model, wherein the initial set of parameters is randomly selected from the Ptparameters, separating, at the third layer of the AI model, corresponding real part and corresponding imaginary part of each of the initial set of parameters, simulating, at the dense layer of the AI model, complex multiplication of the corresponding real part and the corresponding imaginary part by a first shared model (SM-1) and a second shared model (SM-2) to obtain a second DPD output as an AI output, and training the AI model to determine the Psparameters by adjusting the initial set of parameters such that a mean square error between the first DPD output and the AI-output is minimized.

[0142] In an embodiment, during the training of the AI model, the second DPD output is obtained by obtaining the real part of the second DPD output from the SM-1, the imaginary part of the second DPD output from the SM-2, and obtaining the second DPD output by combining the obtained real part from the SM-1 and the imaginary part from the SM-2.

[0143] In an embodiment, the AI model may be trained by providing the Ptparameters as input to an input layer of the AI model, separating corresponding real part and corresponding imaginary part of each of the Ptparameters, simulating, at a dense layer of the AI model, complex multiplication of the corresponding real part and the corresponding imaginary part by a third shared model (SM-3) and a fourth shared model (SM-4), obtaining the real part of the third DPD output from the SM-3, obtaining the imaginary part of the third DPD output from the SM-4, and training the AI model to determine the third DPD output by combining the real part from the SM-3 and the imaginary part from the SM-4.

[0144] In an embodiment, the Psparameters may be determined from the Ptparameters for Ptiterations corresponding to the Ptparameters, by adding, in each iteration, the predefined delta ( ) to the corresponding feature of one of the Ptparameters, determining, using the trained AI model, the corresponding DPD outputs for the corresponding iterations using the one of the Ptparameters with the predefined delta ( ), and determining the corresponding perturbations in the corresponding DPD outputs with respect to the first DPD output. Further, the Ptparameters are sorted with respect to the corresponding perturbations in an order such that a parameter with a maximum corresponding perturbation is ranked highest and the Psparameters are determined from the sorted Ptparameters by selecting the Psparameters with least corresponding perturbations.

[0145] At least by virtue of aforesaid, the present subject matter at least provides the following advantages:

[0146] The system and method described herein provide end of end solution for DPD and can easily be integrated with the legacy system.

[0147] The system and method described herein significantly reduce the complexity of the DPD, while keeping the performance intact and without any degradation

[0148] In this application, unless specifically stated otherwise, the use of the singular includes the plural, and the use of "or" means "and / or." Furthermore, the use of the terms "including" or "having" is not limiting. Any range described herein will be understood to include the endpoints and all values between the endpoints. Features of the disclosed embodiments may be combined, rearranged, omitted, etc., within the scope of the invention to produce additional embodiments. Furthermore, certain features may sometimes be used to advantage without a corresponding use of other features.

[0149] While at least one exemplary embodiment has been presented in the foregoing detailed description, it should be appreciated that a vast number of variations exist.

Claims

1.A method by a device for determining a subset of parameters (Ps) from total parameters (Pt) for a Ptbased digital pre-distorter (DPD), the method comprising:estimating corresponding values of the Ptparameters, at a first time slot of first TIPslots;sorting the Ptparameters based on contribution of each of the Ptparameter to a first DPD output for the first time slot, wherein the first DPD output is determined using the corresponding values of the Ptparameters; anddetermining the Psparameters by eliminating parameters with contribution lower than a predetermined threshold from the sorted Ptparameters in order that corresponding DPD outputs for the remaining slots of the first TIPslots are determined using corresponding values of the Ptparameters,2.The method of the claim 1, further comprising:assessing validity of the determination of the Ps parameters based on at least one predefined condition associated with performance of the DPD; andadjusting Ps, based on the assessed validity, in order that the performance of the DPD with the Ps parameters is valid, wherein the Ps is adjusted by either increasing the number of Ps or decreasing the number of Ps.3.The method of the claim 2, wherein the Psparameters are:greater than a first threshold associated with a minimum number of Ps parameters below which the performance of the DPD is determined to be invalid; andlower than a second threshold associated with a maximum number of Ps parameters at which the performance of the DPD is valid.4.The method of the claim 2, wherein the at least one predefined condition include, for a second DPD output, an error vector magnitude (EVM) lower than an EVM threshold (ThEVM) and an adjacent channel power leakage ratio (ACLR) lower than an ACLR threshold (ThACLR), wherein the second DPD output is determined using the corresponding values of the Psparameters.5.A method by a device for determining a subset of parameters (Ps) from a total parameters(Pt) for a Ptbased digital pre-distorter (DPD), the method comprising:estimating corresponding values of the Pt parameters, at a first time slot of first TIP slots;determining a first DPD output for the first time slot using the corresponding values of the Pt parameters; anddetermining the Ps parameters from the Pt parameters by selecting the P parameters using a trained artificial intelligence (AI) model, in order that corresponding DPD outputs for the remaining slots of the first TIP slots are determined using corresponding values of the P parameters6.The method of the claim 5, wherein the AI model is trained by:providing corresponding features of the Pt parameters as input to an input layer of the AI model;selecting an initial set of parameters at a second layer of the AI model, wherein the initial set of parameters is randomly selected from the Pt parameters;separating, at a third layer of the AI model, corresponding real part and corresponding imaginary part of each of the initial set of parameters;simulating, at a dense layer of the AI model, complex multiplication of the corresponding real part and the corresponding imaginary part by a first shared model (SM-1) and a second shared model (SM-2) to obtain a second DPD output as an AI output; andtraining the AI model to determine the Ps parameters by adjusting the initial set of parameters such that a mean square error between the first DPD output and the AI-output is minimized.7.The method of the claim 6, wherein during the training of the AI model (211), the second DPD output is obtained by:obtaining a real part of the second DPD output from the SM-1;obtaining an imaginary part of the second DPD output from the SM-2; andobtaining the second DPD output by combining the obtained real part from the SM-1 and the imaginary part from the SM-2.8.The method of the claim 5, wherein the AI model is trained by:providing the Pt parameters as input to an input layer of the AI model;separating corresponding real part and corresponding imaginary part of each of the Pt parameters;simulating, at a dense layer of the AI model, complex multiplication of the corresponding real part and the corresponding imaginary part by a third shared model (SM-3) and a fourth shared model (SM-4);obtaining a real part of the third DPD output from the SM-3;obtaining an imaginary part of the third DPD output from the SM-4; andtraining the AI model to determine the third DPD output by combining the real part from the SM-3 and the imaginary part from the SM-4.9.The method of the claim 8, wherein determining the Psparameters from the Ptparameters comprises:for Ptiterations corresponding to the Ptparameters:adding, in each iteration, a predefined delta () to corresponding feature of one of the Pt parameters;determining, using the trained AI model (211), corresponding DPD outputs for the corresponding iterations using the one of the Pt parameters with the predefined delta (); anddetermining corresponding perturbations in the corresponding DPD outputs with respect to the first DPD output;sorting the Pt parameters with respect to the corresponding perturbations in an order such that a parameter with a maximum corresponding perturbation is ranked highest; anddetermining the Ps parameters from the sorted Pt parameters by selecting the Ps parameters with least corresponding perturbations.10.A device for determining a subset of parameters (Ps) from total parameters (Pt) for a Ptbased digital pre-distorter (DPD), the device comprising:a transceiver; anda processor coupled with the transceiver and configured toestimate corresponding values of the Pt parameters, at a first time slot of first TIP slots;sort the Pt parameters based on contribution of each of the Pt parameter to a first DPD output for the first time slot, wherein the first DPD output is determined using the corresponding values of the Pt parameters; anddetermine the Ps parameters by eliminating parameters with contribution lower than a predetermined threshold from the sorted Pt parameters in order that corresponding DPD outputs for the remaining slots of the first TIP slots are determined using corresponding values of the P parameters.11.The device of the claim 10, wherein the processor further configured to:assess validity of the determination of the Ps parameters based on one or more predefined conditions associated with performance of the DPD; andadjust Ps, based on the assessed validity, such that the performance of the DPD with the Ps parameters is valid, wherein the Ps is adjusted by either increasing the number of Ps or decreasing the number of Ps.12.The device of the claim 11, wherein Psparameters are:greater than a first threshold associated with a minimum number of Ps parameters below which the performance of the DPD is determined to be invalid; andlower than a second threshold associated with a maximum number of Ps parameters at which the performance metrics of the DPD is valid.13.The device of the claim 11, wherein the one or more predefined conditions include, for a second DPD output, an error vector magnitude (EVM) lower than an EVM threshold (ThEVM) and an adjacent channel power leakage ratio (ACLR) lower than an ACLR threshold (ThACLR), wherein the second DPD output is determined using the corresponding values of the Psparameters.14.A device for determining a subset of parameters (Ps) from a total parameters (Pt) for a Ptbased digital pre-distorter (DPD) model, the device comprising:a transceiver; anda processor coupled with the transceiver and configured to:estimate corresponding values of the Pt parameters, at a first time slot of first TIP slots;determine a first DPD output for the first time slot using the corresponding values of the Pt parameters; anddetermine the Ps parameters from the Pt parameters by selecting the P parameters using a trained artificial intelligence (AI) model, such that corresponding DPD outputs for the remaining slots of the first TIP slots are determined using corresponding values of the P parameters, wherein the corresponding values of the P parameters is determined using the predefined technique.15.The device of the claim 14, wherein the AI model is trained by:providing corresponding features of the Pt parameters as input to an input layer of the AI model;selecting an initial set of parameters at a second layer of the AI model, wherein the initial set of parameters is randomly selected from the Pt parameters;separating, at a second layer of the AI model, corresponding real part and corresponding imaginary part of each of the initial set of parameters;simulating, at a dense layer of the AI model, complex multiplication of the corresponding real part and the corresponding imaginary part by a first shared model (SM-1) and a second shared model (SM-2) to obtain a second DPD output as an AI input; andtraining the AI model to determine the Ps parameters by adjusting the initial set of parameters such that a mean square error between the first DPD output and the AI-output is minimized.

Images