Generalization enhancement of artificial intelligence / machine learning models with assistance information signals for new radio air interface

Abstract

Systems and methods for enhancing the generalization of artificial intelligence (AI) / machine learning (ML) models by introducing an AI / ML super model and inputting the AI / ML super model with assistance information are discussed herein. For example, a user equipment (UE) operating an AI / ML super model for network signal processing may receive network-side assistance information and process UE-side assistance information, where the network-side assistance information and the UE-side assistance information may be additional physical communication conditions or configurations. Then, the UE may measure reference signals to obtain reference signal received power (RSRP) values. Accordingly, the RSRP values, the network-side assistance information, and the UE-side assistance information may be inputted into the AI / ML super model to be used for the generation of an inference. In some cases, the RSRP values, the network-side assistance information, and the UE-side assistance information may be used to train the AI / ML super model.

Description

GENERALIZATION ENHANCEMENT OF ARTIFICIAL INTELLIGENCE / MACHINELEARNING MODELS WITH ASSISTANCE INFORMATION SIGNALS FOR NEWRADIO AIR INTERFACETECHNICAL FIELD

[0001] This application relates generally to wireless communication systems, including wireless communication systems implementing artificial intelligence (AI) / machine learning (ML) models.BACKGROUND

[0002] Wireless mobile communication technology uses various standards and protocols to transmit data between a base station and a wireless communication device. Wireless communication system standards and protocols can include, for example, 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) (e.g., 4G), 3GPP New Radio (NR) (e.g., 5G), and Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard for Wireless Local Area Networks (WLAN) (commonly known to industry groups as Wi-Fi®).

[0003] As contemplated by the 3GPP, different wireless communication systems' standards and protocols can use various radio access networks (RANs) for communicating between a base station of the RAN (which may also sometimes be referred to generally as a RAN node, a network node, or simply a node) and a wireless communication device known as a user equipment (UE). 3GPP RANs can include, for example. Global System for Mobile communications (GSM), Enhanced Data Rates for GSM Evolution (EDGE) RAN (GERAN), Universal Terrestrial Radio Access Network (UTRAN), Evolved Universal Terrestrial Radio Access Network (E-UTRAN), and / or Next-Generation Radio Access Network (NG-RAN).

[0004] Each RAN may use one or more radio access technologies (RATs) to perform communication between the base station and the UE. For example, the GERAN implements GSM and / or EDGE RAT, the UTRAN implements Universal Mobile Telecommunication System (UMTS) RAT or other 3GPP RAT, the E-UTRAN implements LTE RAT (sometimes simply referred to as LTE), and NG-RAN implements NR RAT (sometimes referred to herein as 5G RAT, 5G NR RAT, or simply NR). Incertain deployments, the E-UTRAN may also implement NR RAT. In certain deployments, NG-RAN may also implement LTE RAT.

[0005] A base station used by a RAN may correspond to that RAN. One example of an E-UTRAN base station is an Evolved Universal Terrestrial Radio Access Network (E- UTRAN) Node B (also commonly denoted as evolved Node B, enhanced Node B, eNodeB, or eNB). One example of an NG-RAN base station is a next generation Node B (also sometimes referred to as a g Node B or gNB).

[0006] A RAN provides its communication services with external entities through its connection to a core network (CN). For example, E-UTRAN may utilize an Evolved Packet Core (EPC) while NG-RAN may utilize a 5G Core Network (5GC).BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0007] To easily identify the discussion of any particular element or act, the most significant digit or digits in a reference number refer to the figure number in which that element is first introduced.

[0008] FIG. 1 illustrates an example of grouping AI / ML models into various supporting conditions with an overarching AI / ML model functionality.

[0009] FIG. 2 illustrates an example of a set of AI / ML models identified for a UE capability.

[0010] FIG. 3 illustrates an example of AI / ML model selection based on assistance information.

[0011] FIG. 4 illustrates an example of inputting an AI / ML super model with assistance information, according to embodiments herein.

[0012] FIG. 5 illustrates an example of inputting the training of the AI / ML super model with assistance information, according to embodiments herein.

[0013] FIG. 6 illustrates a method of a UE operating an AI / ML super model for network signal processing, according to embodiments herein.

[0014] FIG. 7 illustrates a method of a UE operating an AI / ML super model for network signal processing, according to embodiments herein.

[0015] FIG. 8 illustrates a method of a base station in communication with a UE operating an AI / ML super model for network signal processing, according to embodiments herein.

[0016] FIG. 9 illustrates an example architecture of a wireless communication system, according to embodiments disclosed herein.

[0017] FIG. 10 illustrates a system for performing signaling between a wireless device and a network device, according to embodiments disclosed herein.DETAILED DESCRIPTION

[0018] Various embodiments are described with regard to a UE. However, reference to a UE is merely provided for illustrative purposes. The example embodiments may be utilized with any electronic component that may establish a connection to a network and is configured with the hardware, software, and / or firmware to exchange information and data with the network. Therefore, the UE as described herein is used to represent any appropriate electronic component.

[0019] In some wireless communication systems, the use of artificial intelligence (Al)Zmachine learning (ML) models may achieve performance improvements over non AI / ML model operation in various circumstances. However, achieving performance improvements (e.g., achieving a reduction of measurement overhead, achieving reduced overhead through compression of channel state information (CSI), etc.) across all configurations and / or conditions may be challenging. Accordingly, it has been determined that adaptation to match the current radio propagation conditions may provide performance improvements. For example, for a beam measurement (BM) case and for a UE-sided AI / ML model, a downlink (DL) transmit (Tx) beam prediction is made for a first set (Set A) of one or more predicted DL Tx beams by an AI / ML model using actual measurement results of a (e.g., different and / or smaller) second set (Set B) of one or more measured DL Tx beams. Note that there may be some size flexibility on the set B beams in such cases. The smaller the size of the set B beams, the smaller the overhead is for measurements and, correspondingly, the larger the size of the set B of beams, the larger the overhead is for measurements.

[0020] In some wireless communication mechanisms, the UE may be pre-configured to measure a set B of beams irrespective of the dynamics of radio conditions.Corresponding to such cases, during deployment there are two possible outcomes that correspond to a mismatch between the current radio conditions and the current usage of the AI / ML model. In a first such outcome, the current radio conditions are favorablesuch that the achievement of the same throughput was otherwise possible using a smaller set of set B than what was actually used; thus, there is a waste of resources.

[0021] In a second such outcome, the radio conditions may be unfavorable, and the set B of beams is too small relative to the adverse radio conditions such that the spatial mapping / inference from set B to set A fails to preserve the system performance thus triggering interruptions (life cycle management (LCM) procedures of AI / ML model deactivation and / or fallback causing interruptions and / or unnecessary model transfer updating) that would not have occurred had a larger set B of beams been used.

[0022] Additionally, in some wireless communication systems, an AI / ML model for a CSI feedback use case may be considered. For example, in some cases, a two-sided AI / ML model (where each of a UE and a base station implement at least a portion of the AI / ML model) is used for CSI feedback compression (at a CSI feedback encoder) and CSI feedback decompression (at a CSI feedback decoder). For example, an AI / ML model may be made up of both an encoder (e.g., at the UE side) for encoding (e.g., compressing) one or more elements of CSI feedback into encoded CSI bits for transmission and a decoder (e.g., at the base station side) for decoding (decompressing) the encoded CSI bits feedback as received into decoded CSI feedback. Through the use of the AI / ML model (e.g., the use of the encoder at the UE and the decoder at the base station), CSI feedback transmission resource overhead may be reduced as compared to other cases. For example, in cases where the encoded CSI bits are smaller in size than a native representation of the CSI feedback, the transmission of the encoded CSI bits (e.g., instead of the transmission of the native representation of the CSI feedback) represents a more efficient use of transmission resources.

[0023] In some wireless communication systems, network-side additional conditions may refer to training assumptions associated with network implementation and / or network aspects (e.g., a network-side beam pattern), and UE-side additional conditions may refer to training assumptions associated with UE implementation and / or UE aspects (e.g., UE speed, UE-side beam pattern), as detailed in for example, 3GPP Technical Report (TR) 38.843 (Version 18.0.0 Published January 16. 2024). Additionally, additional conditions may refer to conditions that may be present in training data but may be unknown to either the UE or network, depending on where the AI / ML model is trained. For example, during training the beam width of the network set of beams may be a certain value and may be used to train an AI / ML model. However, the AI / ML model atthe UE, during inference, may not have this information, (e.g., for proprietary' reasons). Likewise, a UE-side additional condition (e.g., UE speed) may not be known to the network but may be assumed (by the network) when collecting data to be used for network-sided AI / ML model training.

[0024] Further, note that, for cases of AI / ML models for BM, beam prediction under different generalization performance assessment assumptions may be linked to additional conditions.

[0025] The assessment of generalization indicates the need for mitigation of performance issues that arise when assumptions made during training, (such as using an AI / ML model trained with a specific set of network-side or UE-side additional conditions), do not align with the assumptions made during inference. Further, this misalignment may lead to performance degradation. It should be understood that generalization performance assessments (or more generally generalization testing) is performed by artificially varying an additional condition of a certain AI / ML model to see how the AI / ML model performs under certain conditions and if the AI / ML model generalizes well to various possible operating conditions.

[0026] Additionally, in various wireless communication mechanisms, ensuring consistency between training and inference regarding identified additional conditions on the network-side for inference the UE may be important (especially for beam prediction use cases). As highlighted in, for example, 3GPP TR 38.843 (e.g.. in clause 4.2.3, Version 18.0.0 Published January 16, 2024), additional conditions may refer to aspects that are assumed for the training of an AI / ML model but are not a part of the UE capability7for an AI / ML model enabled feature and / or feature group (FG). Note that additional conditions may not be standardized and may differ from UE to UE or network to network depending on UE and / or network implementation (e.g., due to proprietary information). Moreover, additional conditions of an AI / ML model may be organized into two categories including network-side additional conditions and UE-side additional conditions. It should be understood that additional conditions may also be referred to herein as assistance information or auxiliary’ information.

[0027] Various approaches have been proposed to improve AI / ML model generalization (e.g., in clause 4.2.3 of 3GPP TR 38.843 Version 18.0.0 Published January 16, 2024). For example, in some approaches, model identification may be introduced to achieve alignment for the network-side additional condition between the network-side and theUE-side. In some other approaches, the AI / ML model training may be performed at the network and then, the AI / ML model may be transferred to the UE, where the AI / ML model may be trained based on the additional conditions that the network may be aware of. In yet some other approaches, information and / or an indication of the network-sided additional conditions may be provided to the UE. In yet some other approaches, consistency may be assisted by monitoring (as performed by the UE and / or the network) the performance of the UE-side AI / ML models and / or functionalities to select an AI / ML model and / or AI / ML model functionality.

[0028] It should be understood that, in some wireless communication systems, additional conditions were introduced as various AI / ML training variables may lack standardization. For example, in AI / ML-based beam measurement use cases, elements of network topology such as cell layout or antenna configuration, or internal variables of UEs (e g., UE speed), may influence the selection of an AI / ML model that may be utilized during AI / ML operations. Nonetheless, while these variables may be essential for the training of AI / ML models and thus possibly needed as an input for AI / ML model operation, they may not be standardized due to various concerns including proprietary information concerns. Therefore, categorizing these variables as additional conditions (or assistance information) may facilitate optimal AI / ML operation based on AI / ML model identification.

[0029] Among the various approaches proposed for ensuring consistency between the network-side additional conditions and the UE-side additional conditions, consistency obtained by monitoring may not be optimal. For example, consistency obtained through the assistance of monitoring may cause significant delays in selecting the optimal AI / ML model for the UE. Consider an example where multiple sets of beam measurement AI / ML models are developed for different UE speeds, and the network intends to utilize the monitoring procedure to select the optimal model. In such examples, the network may activate multiple AI / ML models, at the UE, one by one, each associated with different UE speeds, until the network identifies an AI / ML model that yields the best performance. During the period when the network is activating non-optimal model(s) at the UE, the UE performance may degrade and suffer, thus resulting in a delay in selecting the optimal AI / ML model.

[0030] FIG. 1 illustrates an example of grouping AI / ML models into various supporting conditions with an overarching AI / ML model functionality.

[0031] In some wireless communication systems, AI / ML models may be grouped together into non-overlapping sets of supporting conditions as part of a UE capability report signaling, where each set of supporting conditions may further include additional conditions that may be shared and other additional conditions that may be shared via virtual identifier (ID).

[0032] For example, in UE capability report signaling indicating support for one or more AI / ML model functionalities 102 (e.g., a beam measurement functionality, a CSI compression / decompression functionality, and / or a positioning functionality7, etc.), first supporting condition(s) 104 for a first category of AL / ML models (under “Model A” supporting conditions with UE capability for a beam measurement case for spatial prediction (referred to as "BM Case 1")) may include AI / ML model input characteristics where Set B beam may be a subset of Set A beams or a different set of beams, a model complexity7, performance metrics, monitoring metrics, a Doppler range, a link level cluster delay line (CDL) channel delay spread, an signal-to-noise ratio (SNR) range (e.g., a range of [0:5] decibel (dB)), Doppler (e.g., high or low Doppler, CDL, and / or a classifier beam ID). Note that various AI / ML models with different AI / ML model identifiers may reside under the same supporting conditions. For example, AI / ML model Ai and Ak reside under the Model A category of supporting conditions with Model Ai have a model identifier of 1 and Model Ak having a model identifier of k.

[0033] Then, as illustrated under the Model A supporting conditions, the use of Model Ai corresponds to the case of additional conditions 106 that may be shared including Set A beam and Set B beam configuration and / or mapping, a pattern of Set B beams, a number of Set A or Set B beams, various scenarios (e.g., urban macro (Uma), urban micro (Umi), indoor hotspot), inter-site distance (ISD), speed, and UE distribution. Note that the use of Model Ai is also illustrated to correspond to the case of additional conditions 108 that may be shared with a virtual ID including, for example, a 3 dB bandwidth, a beam shape, an antenna layout configuration, and / or antenna spacings. Note that the first supporting condition(s) 104, the additional conditions 106 that may be shared and the additional conditions 108 that may be shared via virtual ID, may be used in combination to achieve AI / ML model identification.

[0034] Additionally or alternatively, UE capability7report signaling may indicate support for second supporting condition(s) 110 for a second category of AI / ML models (under “Model B” supporting a beam measurement case for temporal prediction (referredto as "BM Case 2")) may include model input characteristics, Set B beams as a subset of Set A beams or different set of beams, model output characteristics (e.g., output types, prediction pattern for beam measurement case 2, number of top-K beams), model complexity, performance metrics, monitoring metrics, SNR range (e.g., [5: 10] dB), Doppler (e.g., high Doppler or low Doppler), additive white gaussian noise (AWGN), and / or an Layer 1 reference signal received power (Ll-RSRP) prediction. Again, note that various AI / ML models with different AI / ML model identifiers may reside under the same supporting conditions. For example, AI / ML model Bi and Bk reside under the Model B category7of supporting conditions with Model Bi having a model identifier of 1 and Model Bk having a model identifier of k.

[0035] Then, as illustrated under the Model B supporting conditions, the use of Model Bi corresponds to the case of additional conditions 112 that may be shared including Set A beams and Set B beams configuration and / or mapping, a pattern of set B beams, a number of Set A beams and / or Set B beams, a time window configuration (e.g., in beam measurement case 2), a number of predicted future time instances, various scenario (e.g., Uma, Umi, indoor hotspot), an ISD, a UE speed, and / or a UE distribution. Note that the use of Model Bi is also illustrated to correspond to the case of additional conditions 114 that may be shared with a virtual ID including, for example, a 3 dB bandwidth, a beam shape, an antenna layout configuration, and / or antenna spacings. Note that the second supporting condition(s) 110, the additional conditions 112 that may be shared and the additional conditions 114 that may be shared via virtual ID, may be used in combination to achieve AI / ML model identification.

[0036] Additionally or alternatively, UE capability report signaling may indicate support for third supporting condition(s) 116 for a third category of AI / ML models (under “Model C,"’ used for CSI prediction) may include model input characteristics, model output characteristics, model complexity, performance metrics, monitoring metrics, an SNR range (e.g., [5: 10] dB), a Doppler indication, and an AWGN. Again, note that various AI / ML models with different AI / ML model identifiers may reside under the same supporting conditions. For example, AI / ML model Ci resides under the Model C category of supporting conditions with Model Ci having a model identifier of 1.

[0037] Then, as illustrated under the Model C supporting conditions, the use of Model Ci corresponds to the case of additional conditions 118 that may be shared including the site type, a down-title angle, transceiver unit (TXRU) mapping, and / or network planning.In some cases, the third supporting condition(s) 116 and the additional conditions 118 may be used in combination to achieve AI / ML model identification.

[0038] Note that the AI / ML model categories ("Model A", “Model B”, and “Model C”) are different AI / ML categories having different AI / ML model functionalities 102 (e.g., “Model A” having an AI / ML model functionality 102 of BM Case 1, “Model B” having an AI / ML AI / ML model functionality 102 of BM Case 2, and “Model C” having an AI / ML AI / ML model functionality 102 of CSI prediction).

[0039] It should be understood that for generalization testing 120, the network-side additional conditions and / or the UE-side additional conditions may be linked 122 to the generalizability of an AI / ML model. Therefore, meaning that the use of certain additional conditions (whether UE-side or network side) may increase the generalizability of an AI / ML model.

[0040] In some current wireless communication systems, the UE may use multiple AI / ML models where the AI / ML models may be stored at the UE or at an over the air (OTA) server. Note that the AI / ML models may be transferred between the UE and the OTA server, for training or even for AI / ML model use. Additionally, the functionality of an AI / ML model may refer to an AI / ML enabled feature and / or FG enabled by configurations, where the configurations may be supported based on conditions indicated by the UE capability'. In some cases, an AI / ML model may be identified by a configuration and / or condition associated with the UE capability’ and the additional conditions (whether network-sided or UE-sided).

[0041] FIG. 2 illustrates an example of a set of AI / ML models identified for a UE capability.

[0042] For example, a set of AI / ML models may be identified to support a certain UE capability 202 including, for example a first AI / ML model 210, a second AI / ML model 212, up to the Nth AI / ML model 214 (where the value “N’?may be large as to adequately support the UE capability7202). Additionally, each AI / ML model may correspond to various aspects associated or not associated with the UE capability or aspects of the UE capability7not specified. For example, the first AI / ML model 210 may include aspects associated with the UE capabilities 204, aspects not associated with UE capability 206, and aspects that are not specified 208. Additional conditions as discussed herein may fall into one of the categories of aspects not associated with UE capabilities 206 and aspects that are not specified 208. Note that a large number of AI / ML models ("N" number ofAI / ML models) may be needed to support the UE capability 302, all of which may need separate generalization testing performed to achieve adequate consistency, thus increasing system complexity. Further, note that, generally, each AI / ML model (e.g., AI / ML model k 222 from the set of AI / ML models under the same UE capability' 202) may be made up of pre-processing blocks 216, an AI / ML core 218 and post-processing blocks 220, where each AI / ML model may use physical signals (e.g., RSRP signals) as an input to the AI / ML model.

[0043] Further, in some wireless communication systems, assistance information may only be used for AI / ML model selection. For example, FIG. 3 illustrates an example of AI / ML model selection based on assistance information. An AI / ML model may be selected from a set of AI / ML models used to support a certain UE capability 302 (as discussed herein) and the additional conditions / assistance information may be used to identity' 304 an AI / ML model to be selected or the additional conditions / assistance information may be indicated 306 to the UE to help facilitate the selection of an AI / ML model. In either case, the selection of an AI / ML model 308 is based on the additional conditions / assistance information. Still note that a large number of AI / ML models (up to "N" AI / ML models 310) may be needed to support a single UE capability 302 so that an optimal AI / ML model may be chosen for the UE capability’ 302 regardless of the current additional conditions, all of which may need separate generalization testing performed on them to ensure adequate consistency (thus, problematically, increasing system complexity).

[0044] As a result, existing mechanisms to increase the consistency of additional conditions may be understood to arrive with various disadvantages. For example, if multiple models (e.g., per UE) with varying generalization capabilities and parameters for network-side additional conditions are trained by different UE vendors, it may be that substantial standardization efforts are needed. Each AI / ML model (each AI / ML model trained for a certain UE / network scenario / condition) may need to be generalized through standardized generalization testing methods, thus encountering issues when try ing to standardize proprietary information (possibly residing within certain generalization testing methods or certain AI / ML models). Note that a single base station may serve multiple UEs simultaneously, thus exponentially increasing the number of AI / ML models and thus exacerbating said disadvantage. Additionally, it may be that the UE complexity' increases (e.g., there may be multiple AI / ML models per UE per UEcapability to train and select from). In some cases, the complexity of LCM model management may also increase as the number of candidate AI / ML models may be large. Note that, problematically, deactivation, latency and delays may also occur when AI / ML model monitoring takes place or when an appropriate / optimal AI / ML model is not found (where in some such cases the UE may fall back to a previous AI / ML model or previous mechanisms).

[0045] Additionally, wireless communication systems may encounter the problem of achieving consistency between AI / ML models that are generating an inference and AI / ML models that are being trained. Note that achieving consistency for the networkside assistance information / additional conditions or for the UE-side assistance information / additional conditions may be related to AI / ML model generalizability. For example, current mechanisms may invoke either model monitoring procedures or model identification procedures for selecting the most optimal AI / ML model. However, both approaches may have disadvantages in terms of UE processing and memory complexity as well as LCM management functionality complexity with increased testing of the AI / ML model, which may need extensive standardization efforts.

[0046] Embodiments herein discuss the use of an "AI / ML super model’7that uses assistance information about applicable additional conditions as input to its inference engine. For example, the use of assistance information may optimize the tradeoff between disclosing proprietary information and enhancing the generalizability of the AI / ML super model. Additionally, embodiments herein discuss the training of an AI / ML super model with a mixed dataset from various base station additional conditions, UE additional conditions and / or base station / UE implementations. This wide applicability across varying additional conditions reduces the overall number of models needed to obtain adequate generalization, resulting in optimized system performance. As a result of the increased generalizability of the introduced AI / ML super model, beneficially, the number of models to be stored, managed, and / or tested may drastically be reduced, thus simplifying the LCM management complexity and system complexity as a whole.

[0047] FIG. 4 illustrates an example of inputting an AI / ML super model with assistance information, according to embodiments herein.

[0048] In some embodiments, an AI / ML super model 410 (that may be either a UE- sided AI / ML model or a network-sided AI / ML model) may be introduced and may have assistance information / additional conditions as an input. In some cases, an AI / ML supermodel 410 may be more generalized as compared to a (non-super) AI / ML model, thus fewer AI / ML super models 410 may be needed supporting the same UE capability 402. For example, a first AI / ML super model 404, a second Al / ML super model 406, up to a Lth AI / ML super model 408 may be identified to support a certain UE capability 402 where the “L” number of AI / ML super models identified to support a UE capability 402 is much smaller compared to the “N” number of (non-super) AI / ML models identified to support a UE capability (as discussed in FIG. 2 and FIG. 3).

[0049] Consider an example of an AI / ML model using beam widths as an input. In current mechanisms, multiple AI / ML models may be configured and then chosen from according to the certain beam width use case (one AI / ML model with a beam width of 10, a second AI / ML model with a beam width of 30, a third AI / ML model with a beam width of 60, a fourth AI / ML model with a beam width of 90, and so on). However, according to embodiments herein, only one or a small few AI / ML super models may be needed to cover the full range of beam widths as the AI / ML super model may be inputted / training with the intended beam width use case and may vary its output according to the input information to be applicable to the certain beam width use case. Note that a selection between or identification of an AI / ML super model out of a small set of AI / ML super models may still be performed (if more than one AI / ML super model is configured), however the set of AI / ML super models supporting a UE capability may be smaller than a set of AI / ML models supporting a UE capability7. Note that beam width is one example of an assistance information / additional condition input. Other types of additional conditions / assistance information may be provided as inputs to the AI / ML super model to support other UE capabilities.

[0050] As discussed herein, due to the increase in generalizability of the AI / ML super model 410, there may be a need for fewer models (i.e., AI / ML super models 410) to achieve an adequate amount of consistency and / or generalization while still maintaining or even optimizing performance (e.g., consistency in assistance information / additional conditions). As a result, there may be a reduction in deactivations, latency, and delays incurred during AI / ML super model 410 monitoring, a reduction in UE complexity, and a reduction in LCM model management complexity. For example, currently, assistance information may only be used for AI / ML model selection. However, embodiments herein input the AI / ML super model 410 with assistance information 418. Note that the assistance information 418 may be used in the AI / ML super model's inference engine.Additionally note that the AI / ML super model 410 may contain pre-processing blocks 412, an AI / ML core 414, and post-processing blocks 416.

[0051] In some cases, the assistance information 418 used as inputs to the AI / ML super model 410 may be split up into two categories: network-side assistance information (that the network transmits to the UE) and UE-side assistance information (that is stored and processed at the UE). The network-side assistance information / additional conditions that may be applicable across one or more various UE capabilities may include but is not limited to, for example, a Set A beam pattern, a Set B beam pattern, scenario information (e.g., Uma, Umi, typical values of delay spread (DS), Doppler, and / or always on service (AOS)), 3dB beam bandwidth. Tx beam shape, Tx beam angle, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, beam timing window information, antenna site information, antenna tilt information, network planning information, TXRU information, SNR information, and Tx beam ID information. UE-side assistance information / additional conditions that may be applicable across one or more various UE capabilities may include, but is not limited to, for example, UE speed information, UE receive (Rx) beam information, UE implementation information, UE distribution information, UE scenario information, and Doppler information. Note that, more generally, the network-side assistance information and the UE-side assistance information include additional physical communication conditions as described in the aforementioned examples.

[0052] In some cases, to achieve assistance information 418 that may be inputted into the AI / ML super model 410 (without overloading the AI / ML super model 410 with copious amounts of data points of the assistance information 418 inputted into the AI / ML super model 410), the assistance information 418 may be first processed (ran through processing blocks 420) so that it is quantized and encoded for inputting into the AI / ML super model 410. Note that the quantized and encoded data is illustrated as and / or may be referred to as ‘"Z 422’’ that may be represented in the latent space. It should be understood that, in some cases, Z 422 (the quantized and / or encoded assistance information) may be seen as a tuning parameter used for tuning the AI / ML super model 410. Additionally. Z 422 may be a quantized and / or encoded combination of values of assistance information and / or additional conditions or may be a combination of vectors of the assistance information and / or additional conditions.

[0053] Once processed, the assistance information 418 (including both network-side assistance information and UE-side assistance information or alternatively one of network-side assistance information or UE-side assistance information, “Z 422”) may be inputted into the AI / ML super model 410 in addition to various physical signals 424 (such as RSRP values obtained by measuring reference signals from the network). Accordingly, the AI / ML super model 410 may generate an inference. Note that the AI / ML super model 410 may be used for various use cases according to the overarching UE capability 402 that the AI / ML super model 410 is identified for and / or supports (e g., beam prediction). Additionally note that the AI / ML super model 410 may still adhere to basic AI / ML model operations such as AI / ML model monitoring and undergoing AI / ML model generalization testing. It should be understood that the AI / ML super model 410 may have an increased generalizability as compared to a (non-super) AI / ML model, thus the AI / ML super model 410 may be considered an optimal AI / ML model for various wireless communication use cases and may be optimally applied in view of various additional conditions.

[0054] FIG. 5 illustrates an example of inputting the training of the AI / ML super model with assistance information, according to embodiments herein.

[0055] Similarly, to how assistance information may be an input to the AI / ML super model to generate an inference (discussed in FIG. 4), in some embodiments, assistance information may be used in training of the AI / ML super model 518. As previously- discussed, in some cases, the AI / ML super model 518 may be more generalized as compared to a (non-super) AI / ML model, thus fewer AI / ML super models 518 may be needed to support or be identified for the same UE capability 502. For example, a first AI / ML super model 504, a second AI / ML super model 506, up to an Lth AI / ML super model 508 may be identified to support a certain UE capability 502 where the “L” number of AI / ML super models identified to support a UE capability 502 is much smaller compared to the “N” number of (non-super) AI / ML models identified to support a UE capability (where “N” is previously discussed in FIG. 2 and FIG. 3).

[0056] Again, similarly to AI / ML super model inference embodiment discussed herein in relation to FIG. 4, in some cases, for the training of the AI / ML super model 518, the assistance information 510 may be split up into two categories: network-side assistance information (that the network transmits to the UE) and UE-side assistance information (that is stored and processed at the UE), both including similar additional physicalcommunication conditions as previously discussed for AI / ML super model inference (for example in FIG. 4).

[0057] Additionally, to achieve assistance information 510 that may be used to train the AI / ML super model 518 (without overloading the AI / ML super model 518 with copious amounts of data points of the information being used to train the AI / ML super model 518), the assistance information 510 may be first processed (ran through processing blocks 512) so that it is quantized and encoded for use in training the AI / ML super model 518 (referred to as “Z 514”). Once processed, the assistance information 510 (including both network-side assistance information and UE-side assistance information or alternatively one of network-side assistance information or UE-side assistance information, “Z 514”) may be used to train the AI / ML super model 518.

[0058] In some cases, a mixed dataset 516 may be used in the training of the AI / ML super model 518 where the mixed dataset 516 includes various different quantized and / or encoded assistance information (multiple different Z 514 values resulting in ‘"Zi, Z2,..., ZL” being used to train the AI / ML super model 518). For example, the mixed dataset 516 may include various combinations of assistance information / additional conditions, implementation information and / or a combination of UE-side and / or network-side assistance information (e.g., beam width, beam pattern, UE speed and Doppler may be used in the mixed dataset 516 together to train the AI / ML super model 518). In some examples, the mixed dataset 516 may include a high variation of one or more additional condition and / or assistance information, as to train the AI / ML super model 518 to be applicable to the whole range of the additional condition / assistance information (e.g., high UE speed to low UE speed or high Doppler to low Doppler). It should be understood that the mixed dataset 516 may be used to capture multiple applicable additional conditions and / or assistance information that may be important in the training of the AI / ML super model 518, thus increasing the AI / ML super model 518 generalizability as compared to a (non-super) AI / ML model.

[0059] Then, once trained, the AI / ML super model 518 may be used for various use cases according to the UE capability 502 that it may support or be identified for (e.g., beam prediction), with physical signals 520 (e.g., RSRP measurement values) used as inputs. Note that the trained AI / ML super model 518 may still adhere to basic AI / ML model operations such as AI / ML model monitoring and undergoing AI / ML model generalization testing.

[0060] Embodiments discussed herein ensure consistency between training and inference regarding network-side additional conditions, thus AI / ML model monitoring may have a lower priority as compared to other AI / ML model operations and / or aspects. Additionally, in order to ease the burden for testing various AI / ML models with different network additional conditions, it may be beneficial to train the UE-side AI / ML model with a mixed dataset from various base stations (various base station settings), thus reducing the number of AI / ML models (selected by network-sided additional conditions) needed to guarantee generalization while maintaining system performance. Further, embodiments herein discuss the use of network-side assistance information and UE-side assistance information as an core AI / ML model input signal, thus optimizing the tradeoff between disclosing proprietary information and enhancing the generalizability of the AI / ML model (an AI / ML super model).

[0061] FIG. 6 illustrates a method 600 of a UE operating an AI / ML super model for network signal processing, according to embodiments herein. The illustrated method 600 includes receiving 602, from a base station, first network-side assistance information. The method 600 further includes processing 604, at the UE, first UE-side assistance information. The method 600 further includes measuring 606 reference signals from a wireless network to obtain first RSRP values. The method 600 further includes inputting 608 the first RSRP values, the first network-side assistance information, and the first UE-side assistance information into the AI / ML super model, wherein the first networkside assistance information and the first UE-side assistance information comprise additional physical communication conditions or configurations. The method 600 further includes generating 610 a first inference using the AI / ML super model.

[0062] In some embodiments of the method 600, the first network-side assistance information comprises one or more of Tx beam information, Tx beam pattern information. Tx beam bandwidth information, Tx beam shape information, Tx beam angle information, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, Tx beam timing window information, antenna site information, antenna tilt information, network planning information. TXRU information. SNR information. Tx beam ID information, and Doppler information.

[0063] In some embodiments of the method 600, the first network-side assistance information is quantized and encoded for input to the AI / ML super model.

[0064] In some embodiments, the method 600 further comprises quantizing and encoding the first UE-side assistance information prior to inputting the first UE-side assistance information into the AI / ML super model.

[0065] In some embodiments of the method 600, the first UE-side assistance information comprises one or more of UE speed information, UE Rx beam information, UE implementation information, UE distribution information, UE scenario information, and Doppler information.

[0066] In some embodiments, the method 600 further comprises performing monitoring of the AI / ML super model.

[0067] In some embodiments, the method 600 further comprises determining that the UE cannot adequately generate a previous inference by using the AI / ML super model with prior RSRP values, prior network-side assistance information and prior UE-side assistance information as inputs, and wherein the first RSRP values, the first networkside assistance information, and the first UE-side assistance information are inputted into the AI / ML super model in response to the determining that the UE cannot adequately generate the prior inference by using the AI / ML super model. Some such embodiments further comprise sending, to the base station, in response to determining that the UE cannot adequately generate the prior inference by using the AI / ML super model, a request for the first network-side assistance information.

[0068] FIG. 7 illustrates a method 700 of a UE operating an AI / ML super model for network signal processing, according to embodiments herein. The illustrated method 700 includes receiving 702, from a base station, first network-side assistance information. The method 700 further includes processing 704, at the UE, first UE-side assistance information. The method 700 further includes measuring 706 reference signals from a wireless network to obtain first RSRP values. The method 700 further includes performing 708 training of the AI / ML super model using the first RSRP values, the first network-side assistance information, and the first UE-side assistance information, wherein the first netw'ork-side assistance information and the first UE-side assistance information comprise additional physical communication conditions or configurations.

[0069] In some embodiments of the method 700, the first UE-side assistance information and the first network-side assistance information comprises a mixed dataset. In some such embodiments, the mixed dataset comprises one or more of conditional information and implementation information.

[0070] In some embodiments of the method 700, the first network-side assistance information comprises one or more of Tx beam information. Tx beam pattern information, Tx beam bandwidth information, Tx beam shape information, Tx beam angle information, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, Tx beam timing window information, antenna site information, antenna tilt information, network planning information. TXRU information. SNR information. Tx beam ID information, and Doppler information.

[0071] In some embodiments of the method 700, the first network-side assistance information is quantized and encoded for input to the AI / ML super model.

[0072] In some embodiments, the method 700 further comprises quantizing and encoding the first UE-side assistance information prior to inputting the first UE-side assistance information into the AI / ML super model.

[0073] In some embodiments of the method 700, the first UE-side assistance information comprises one or more of UE speed information, UE Rx beam information, UE implementation information. UE distribution information, UE scenario information, and Doppler information.

[0074] In some embodiments, the method 700 further comprises performing monitoring of the AI / ML super model.

[0075] In some embodiments, the method 700 further comprises determining that the AI / ML super model is not adequately trained using prior RSRP values, prior networkside assistance information, and prior UE-side assistance information, and wherein the AI / ML super model is trained using the first RSRP values, the first network-side assistance information, and the first UE-side assistance information in response to the determining that the AI / ML super is not adequately trained using the prior RSRP values, the prior network-side assistance information, and the prior UE-side assistance information. Some such embodiments further comprise sending, to the base station, in response to determining that the AI / ML super is not adequately trained using the prior RSRP values, the prior network-side assistance information, and the prior UE-side assistance information, a request for the first network-side assistance information.

[0076] FIG. 8 illustrates a method 800 of a base station in communication with a UE operating an AI / ML super model for network signal processing, according to embodiments herein. The illustrated method 800 includes processing 802 first network-side assistance information, for use in AI / ML super model inference and AI / ML super model training. The method 800 further includes transmitting 804, to the UE. the first network-side assistance information.

[0077] In some embodiments of the method 800, the first network-side assistance information comprises one or more of Tx beam information, Tx beam pattern information, Tx beam bandwidth information, Tx beam shape information, Tx beam angle information, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, Tx beam timing window information, antenna site information, antenna tilt information, network planning information. TXRU information. SNR information. Tx beam ID information, and Doppler information.

[0078] In some embodiments, the method 800 further comprises quantizing and encoding the first network-side assistance information.

[0079] In some embodiments, the method 800 further comprises receiving, from the UE, a request for second network-side assistance information, processing the second network-side assistance information, for use in the AI / ML super model inference and the AI / ML super model training, and transmitting, to the UE, the second network-side assistance information.

[0080] FIG. 9 illustrates an example architecture of a wireless communication system 900, according to embodiments disclosed herein. The following description is provided for an example wireless communication system 900 that operates in conjunction with the LTE system standards and / or 5G or NR system standards as provided by 3GPP technical specifications.

[0081] As shown by FIG. 9, the wireless communication system 900 includes UE 902 and UE 904 (although any number of UEs may be used). In this example, the UE 902 and the UE 904 are illustrated as smartphones (e.g., handheld touchscreen mobile computing devices connectable to one or more cellular networks), but may also comprise any mobile or non-mobile computing device configured for wireless communication.

[0082] The UE 902 and UE 904 may be configured to communicatively couple with a RAN 906. In embodiments, the RAN 906 may be NG-RAN, E-UTRAN, etc. The UE 902 and UE 904 utilize connections (or channels) (shown as connection 908 and connection 910, respectively) with the RAN 906, each of which comprises a physical communications interface. The RAN 906 can include one or more base stations (such asbase station 912 and base station 914) that enable the connection 908 and connection 910.

[0083] In this example, the connection 908 and connection 910 are air interfaces to enable such communicative coupling, and may be consistent with RAT(s) used by the RAN 906, such as, for example, an LTE and / or NR.

[0084] In some embodiments, the UE 902 and UE 904 may also directly exchange communication data via a sidelink interface 916. The UE 904 is shown to be configured to access an access point (shown as AP 918) via connection 920. By way of example, the connection 920 can comprise a local wireless connection, such as a connection consistent with any IEEE 802. 11 protocol, wherein the AP 918 may comprise a Wi-Fi® router. In this example, the AP 918 may be connected to another network (for example, the Internet) without going through a CN 924.

[0085] In embodiments, the UE 902 and UE 904 can be configured to communicate using orthogonal frequency division multiplexing (OFDM) communication signals with each other or with the base station 912 and / or the base station 914 over a multicarrier communication channel in accordance with various communication techniques, such as. but not limited to, an orthogonal frequency division multiple access (OFDMA) communication technique (e.g., for downlink communications) or a single carrier frequency division multiple access (SC-FDMA) communication technique (e g., for uplink and ProSe or sidelink communications), although the scope of the embodiments is not limited in this respect. The OFDM signals can comprise a plurality of orthogonal subcarriers.

[0086] In some embodiments, all or parts of the base station 912 or base station 914 may be implemented as one or more software entities running on server computers as part of a virtual network. In addition, or in other embodiments, the base station 912 or base station 914 may be configured to communicate with one another via interface 922. In embodiments where the wireless communication system 900 is an LTE system (e.g., when the CN 924 is an EPC), the interface 922 may be an X2 interface. The X2 interface may be defined between two or more base stations (e.g., two or more eNBs and the like) that connect to an EPC, and / or between two eNBs connecting to the EPC. In embodiments where the wireless communication system 900 is an NR system (e.g., when CN 924 is a 5GC), the interface 922 may be an Xn interface. The Xn interface is defined between two or more base stations (e g., two or more gNBs and the like) that connect to5GC, between a base station 912 (e.g., a gNB) connecting to 5GC and an eNB, and / or between two eNBs connecting to 5GC (e.g.. CN 924).

[0087] The RAN 906 is shown to be communicatively coupled to the CN 924. The CN 924 may comprise one or more network elements 926, which are configured to offer various data and telecommunications services to customers / subscribers (e.g., users of UE 902 and UE 904) who are connected to the CN 924 via the RAN 906. The components of the CN 924 may be implemented in one physical device or separate physical devices including components to read and execute instructions from a machine-readable or computer-readable medium (e.g., a non-transitory machine-readable storage medium).

[0088] In embodiments, the CN 924 may be an EPC, and the RAN 906 may be connected with the CN 924 via an SI interface 928. In embodiments, the SI interface 928 may be split into two parts, an SI user plane (Sl-U) interface, which carries traffic data between the base station 912 or base station 914 and a serving gateway (S-GW), and the SI -MME interface, which is a signaling interface between the base station 912 or base station 914 and mobility management entities (MMEs).

[0089] In embodiments, the CN 924 may be a 5GC. and the RAN 906 may be connected with the CN 924 via an NG interface 928. In embodiments, the NG interface 928 may be split into two parts, an NG user plane (NG-U) interface, which carries traffic data between the base station 912 or base station 914 and a user plane function (UPF), and the SI control plane (NG-C) interface, which is a signaling interface between the base station 912 or base station 914 and access and mobility management functions (AMFs).

[0090] Generally, an application server 930 may be an element offering applications that use internet protocol (IP) bearer resources with the CN 924 (e.g., packet switched data services). The application server 930 can also be configured to support one or more communication services (e.g., VoIP sessions, group communication sessions, etc.) for the UE 902 and UE 904 via the CN 924. The application server 930 may communicate with the CN 924 through an IP communications interface 932.

[0091] FIG. 10 illustrates a sy stem 1000 for performing signaling 1034 between a wireless device 1002 and a network device 1018, according to embodiments disclosed herein. The system 1000 may be a portion of a wireless communications system as herein described. The wireless device 1002 may be, for example, a UE of a wirelesscommunication system. The network device 1018 may be, for example, a base station (e.g.. an eNB or a gNB) of a wireless communication system.

[0092] The wireless device 1002 may include one or more processor(s) 1004. The processor(s) 1004 may execute instructions such that various operations of the wireless device 1002 are performed, as described herein. The processor(s) 1004 may include one or more baseband processors implemented using, for example, a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a controller, a field programmable gate array (FPGA) device, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein.

[0093] The wireless device 1002 may include a memory' 1006. The memory 1006 may be a non-transitory computer-readable storage medium that stores instructions 1008 (which may include, for example, the instructions being executed by the processor(s) 1004). The instructions 1008 may also be referred to as program code or a computer program. The memory 1006 may also store data used by, and results computed by, the processor(s) 1004.

[0094] The wireless device 1002 may include one or more transceiver(s) 1010 that may include radio frequency (RF) transmitter circuitry and / or receiver circuitry that use the antenna(s) 1012 of the wireless device 1002 to facilitate signaling (e.g.. the signaling 1034) to and / or from the wireless device 1002 with other devices (e.g., the network device 1018) according to corresponding RATs.

[0095] The wireless device 1002 may include one or more antenna(s) 1012 (e.g., one, two, four, or more). For embodiments with multiple antenna(s) 1012, the wireless device 1002 may leverage the spatial diversity of such multiple antenna(s) 1012 to send and / or receive multiple different data streams on the same time and frequency resources. This behavior may be referred to as, for example, multiple input multiple output (MIMO) behavior (referring to the multiple antennas used at each of a transmitting device and a receiving device that enable this aspect). MIMO transmissions by the wireless device 1002 may be accomplished according to precoding (or digital beamforming) that is applied at the wireless device 1002 that multiplexes the data streams across the antenna(s) 1012 according to known or assumed channel characteristics such that each data stream is received with an appropriate signal strength relative to other streams and at a desired location in the spatial domain (e.g., the location of a receiver associated withthat data stream). Certain embodiments may use single user MIMO (SU-MIMO) methods (where the data streams are all directed to a single receiver) and / or multi user MIMO (MU-MIMO) methods (where individual data streams may be directed to individual (different) receivers in different locations in the spatial domain).

[0096] In certain embodiments having multiple antennas, the wireless device 1002 may implement analog beamforming techniques, whereby phases of the signals sent by the antenna(s) 1012 are relatively adjusted such that the (joint) transmission of the antenna(s) 1012 can be directed (this is sometimes referred to as beam steering).

[0097] The wireless device 1002 may include one or more interface(s) 1014. The interface(s) 1014 may be used to provide input to or output from the wireless device 1002. For example, a wireless device 1002 that is a UE may include interface(s) 1014 such as microphones, speakers, a touchscreen, buttons, and the like in order to allow for input and / or output to the UE by a user of the UE. Other interfaces of such a UE may be made up of transmitters, receivers, and other circuitry (e.g., other than the transceiver(s) 1010 / antenna(s) 1012 already described) that allow for communication between the UE and other devices and may operate according to known protocols (e g., Wi-Fi®, Bluetooth®, and the like).

[0098] The wireless device 1002 may include an AI / ML super model module 1016. The AI / ML super model module 1016 may be implemented via hardware, software, or combinations thereof. For example, the AI / ML super model module 1016 may be implemented as a processor, circuit, and / or instructions 1008 stored in the memory 1006 and executed by the processor(s) 1004. In some examples, the AI / ML super model module 1016 may be integrated within the processor(s) 1004 and / or the transceiver(s) 1010. For example, the AI / ML super model module 1016 may be implemented by a combination of software components (e.g., executed by a DSP or a general processor) and hardware components (e.g., logic gates and circuitry) within the processor(s) 1004 or the transceiver(s) 1010.

[0099] The AI / ML super model module 1016 may be used for various aspects of the present disclosure, for example, aspects of any of FIG. 1, FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 6, FIG. 7, and / or FIG. 9. The AI / ML super model module 1016 is configured to cause the wireless device 1002 to receive, from a network device 1018, first networkside assistance information. The AI / ML super model module 1016 is further configured to cause the wireless device 1002 to process, at the wireless device 1002, first UE-sideassistance information and measure reference signals from a wireless network to obtain first RSRP values. The AI / ML super model module 1016 is further configured to cause the wireless device 1002 to input the first RSRP values, the first network-side assistance information, and the first UE-side assistance information into the AI / ML super model, wherein the first network-side assistance information and the first UE-side assistance information comprise additional physical communication conditions or configurations. Subsequently, the AI / ML super model module 1016 is further configured to cause the wireless device 1002 to generate a first inference using the AI / ML super model. In some cases, the AI / ML super model module 1016 is configured to cause the wireless device 1002 to train the AI / ML super model using the first RSRP values, the first network-side assistance information, and the first UE-side assistance information into the AI / ML super model, wherein the first network-side assistance information and the first UE-side assistance information comprise additional physical communication conditions or configurations. Additionally, in some cases, the AI / ML super model module 1016 is configured to cause the wireless device 1002 to determine that the wireless device 1002 cannot adequately generate a previous inference or that the AI / ML super model is not adequately previously trained and input into the AI / ML super model and / or train the AI / ML super model on the first RSRP values, first network-side assistance information and first UE-side assistance information.

[0100] The network device 1018 may include one or more processor(s) 1020. The processor(s) 1020 may execute instructions such that various operations of the network device 1018 are performed, as described herein. The processor(s) 1020 may include one or more baseband processors implemented using, for example, a CPU, a DSP, an ASIC, a controller, an FPGA device, another hardware device, a firmware device, or any combination thereof configured to perform the operations described herein.

[0101] The network device 1018 may include a memory 1022. The memory 1022 may be a non-transitory computer-readable storage medium that stores instructions 1024 (which may include, for example, the instructions being executed by the processor(s) 1020). The instructions 1024 may also be referred to as program code or a computer program. The memory’ 1022 may also store data used by. and results computed by. the processor(s) 1020.

[0102] The network device 1018 may include one or more transceiver(s) 1026 that may include RF transmitter circuitry and / or receiver circuitry that use the antenna(s) 1028 ofthe network device 1018 to facilitate signaling (e.g., the signaling 1034) to and / or from the network device 1018 with other devices (e.g.. the wireless device 1002) according to corresponding RATs.

[0103] The network device 1018 may include one or more antenna(s) 1028 (e.g., one, two, four, or more). In embodiments having multiple antenna(s) 1028, the network device 1018 may perform MIMO, digital beamforming, analog beamforming, beam steering, etc., as has been described.

[0104] The network device 1018 may include one or more interface(s) 1030. The interface(s) 1030 may be used to provide input to or output from the network device 1018. For example, a network device 1018 that is a base station may include interface(s) 1030 made up of transmitters, receivers, and other circuitry' (e.g., other than the transceiver(s) 1026 / antenna(s) 1028 already described) that enables the base station to communicate with other equipment in a core network, and / or that enables the base station to communicate with external networks, computers, databases, and the like for purposes of operations, administration, and maintenance of the base station or other equipment operably connected thereto.

[0105] The network device 1018 may include an AI / ML super model module 1032. The AI / ML super model module 1032 may be implemented via hardw are, software, or combinations thereof. For example, the AI / ML super model module 1032 may be implemented as a processor, circuit, and / or instructions 1024 stored in the memory 1022 and executed by the processor(s) 1020. In some examples, the AI / ML super model module 1032 may be integrated within the processor(s) 1020 and / or the transceiver(s) 1026. For example, the AI / ML super model module 1032 may be implemented by a combination of software components (e.g., executed by a DSP or a general processor) and hardware components (e.g., logic gates and circuitry) within the processor(s) 1020 or the transceiver(s) 1026.

[0106] The AI / ML super model module 1032 may be used for various aspects of the present disclosure, for example, aspects of any of FIG. 1, FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 8, and / or FIG. 9. The AI / ML super model module 1032 is configured to cause the network device 1018 to process first network-side assistance information, for use in AI / ML super model inference and AI / ML super model training and transmit, to the wireless device 1002. the first network-side assistance information. In some cases, the AI / ML super model module 1032 is configured to cause the network device 1018 toreceive, from a wireless device 1002, a request for the first network-side assistance information, process the first network-side assistance information and transmit, to the wireless device 1002, the first network-side assistance information.

[0107] Embodiments contemplated herein include an apparatus comprising means to perform one or more elements of the method 600 and method 700. This apparatus may be, for example, an apparatus of a UE (such as a wireless device 1002 that is a UE, as described herein).

[0108] Embodiments contemplated herein include one or more non -transitory computer-readable media comprising instructions to cause an electronic device, upon execution of the instructions by one or more processors of the electronic device, to perform one or more elements of the method 600 and method 700. This non-transitory computer-readable media may be, for example, a memory of a UE (such as a memory 1006 of a wireless device 1002 that is a UE. as described herein).

[0109] Embodiments contemplated herein include an apparatus comprising logic, modules, or circuitry to perform one or more elements of the method 600 and method 700. This apparatus may be, for example, an apparatus of a UE (such as a wireless device 1002 that is a UE, as described herein).

[0110] Embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of the method 600 and method 700. This apparatus may be, for example, an apparatus of a UE (such as a wireless device 1002 that is a UE, as described herein).[OHl] Embodiments contemplated herein include a signal, as described in or related to one or more elements of the method 600 and method 700.

[0112] Embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution of the program by a processor is to cause the processor to carry out one or more elements of the method 600 and method 700. The processor may be a processor of a UE (such as a processor(s) 1004 of a wireless device 1002 that is a UE, as described herein). These instructions may be, for example, located in the processor and / or on a memory of the UE (such as a memory 1006 of a wireless device 1002 that is a UE, as described herein).

[0113] Embodiments contemplated herein include an apparatus comprising means to perform one or more elements of the method 800. This apparatus may be. for example, an apparatus of a base station (such as a network device 1018 that is a base station, as described herein).

[0114] Embodiments contemplated herein include one or more non -transitory computer-readable media comprising instructions to cause an electronic device, upon execution of the instructions by one or more processors of the electronic device, to perform one or more elements of the method 800. This non-transitory computer-readable media may be. for example, a memory of a base station (such as a memory 1022 of a network device 1018 that is a base station, as described herein).

[0115] Embodiments contemplated herein include an apparatus comprising logic, modules, or circuitry to perform one or more elements of the method 800. This apparatus may be, for example, an apparatus of a base station (such as a network device 1018 that is a base station, as described herein).

[0116] Embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of the method 800. This apparatus may be, for example, an apparatus of a base station (such as a network device 1018 that is a base station, as described herein).

[0117] Embodiments contemplated herein include a signal as described in or related to one or more elements of the method 800.

[0118] Embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution of the program by a processing element is to cause the processing element to carry out one or more elements of the method 800. The processor may be a processor of a base station (such as a processor(s) 1020 of a network device 1018 that is a base station, as described herein). These instructions may be, for example, located in the processor and / or on a memory of the base station (such as a memory 1022 of a network device 1018 that is a base station, as described herein).

[0119] For one or more embodiments, at least one of the components set forth in one or more of the preceding figures may be configured to perform one or more operations, techniques, processes, and / or methods as set forth herein. For example, a basebandprocessor as described herein in connection with one or more of the preceding figures may be configured to operate in accordance with one or more of the examples set forth herein. For another example, circuitry associated with a UE, base station, network element, etc. as described above in connection with one or more of the preceding figures may be configured to operate in accordance with one or more of the examples set forth herein.

[0120] Any of the above described embodiments may be combined with any other embodiment (or combination of embodiments), unless explicitly stated otherwise. The foregoing description of one or more implementations provides illustration and description, but is not intended to be exhaustive or to limit the scope of embodiments to the precise form disclosed. Modifications and variations are possible in light of the above teachings or may be acquired from practice of various embodiments.

[0121] Embodiments and implementations of the systems and methods described herein may include various operations, which may be embodied in machine-executable instructions to be executed by a computer system. A computer system may include one or more general-purpose or special-purpose computers (or other electronic devices). The computer system may include hardware components that include specific logic for performing the operations or may include a combination of hardware, software, and / or firmware.

[0122] It should be recognized that the systems described herein include descriptions of specific embodiments. These embodiments can be combined into single systems, partially combined into other systems, split into multiple systems or divided or combined in other ways. In addition, it is contemplated that parameters, attributes, aspects, etc. of one embodiment can be used in another embodiment. The parameters, attributes, aspects, etc. are merely described in one or more embodiments for clarity, and it is recognized that the parameters, attributes, aspects, etc. can be combined with or substituted for parameters, attributes, aspects, etc. of another embodiment unless specifically disclaimed herein.

[0123] It is well understood that the use of personally identifiable information should follow privacy policies and practices that are generally recognized as meeting or exceeding industry or governmental requirements for maintaining the privacy of users. In particular, personally identifiable information data should be managed and handled so asto minimize risks of unintentional or unauthorized access or use, and the nature of authorized use should be clearly indicated to users.

[0124] Although the foregoing has been described in some detail for purposes of clarity, it will be apparent that certain changes and modifications may be made without departing from the principles thereof. It should be noted that there are many alternative ways of implementing both the processes and apparatuses described herein. Accordingly, the present embodiments are to be considered illustrative and not restrictive, and the description is not to be limited to the details given herein, but may be modified within the scope and equivalents of the appended claims.

Claims

CLAIMS1. A method of a user equipment (UE) operating an artificial intelligence (AI) / machine learning (ML) super model for network signal processing, the method comprising: receiving, from a base station, first network-side assistance information; processing, at the UE. first UE-side assistance information; measuring reference signals from a wireless network to obtain first reference signal received power (RSRP) values; inputting the first RSRP values, the first network-side assistance information, and the first UE-side assistance information into the AI / ML super model, wherein the first network-side assistance information and the first UE-side assistance information comprise additional physical communication conditions or configurations; and generating a first inference using the AI / ML super model.

2. The method of claim 1, wherein the first network-side assistance information comprises one or more of transmit (Tx) beam information, Tx beam pattern information, Tx beam bandwidth information, Tx beam shape information, Tx beam angle information, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, Tx beam timing window information, antenna site information, antenna tilt information, network planning information, transceiver unit (TXRU) information, signal-to-noise (SNR) information, Tx beam identifier (ID) information, and Doppler information.

3. The method of claim 1, wherein the first network-side assistance information is quantized and encoded for input to the AI / ML super model.

4. The method of claim 1 further comprising quantizing and encoding the first UE-side assistance information prior to inputting the first UE-side assistance information into the AI / ML super model.

5. The method of claim 1, wherein the first UE-side assistance information comprises one or more of UE speed information. UE receive (Rx) beam information, UE implementation information, UE distribution information, UE scenario information, and Doppler information.

6. The method of claim 1 further comprising performing monitoring of the AI / ML super model.

7. The method of claim 1 further comprising: determining that the UE cannot adequately generate a previous inference by using the AI / ML super model with prior RSRP values, prior network-side assistance information and prior UE-side assistance information as inputs; and wherein the first RSRP values, the first network-side assistance information, and the first UE-side assistance information are inputted into the AI / ML super model in response to the determining that the UE cannot adequately generate a prior inference by using the AI / ML super model.

8. The method of claim 7 further comprising sending, to the base station, in response to determining that the UE cannot adequately generate the prior inference by using the AI / ML super model, a request for the first network-side assistance information.

9. A method of a user equipment (UE) operating an artificial intelligence (AI) / machine learning (ML) super model for network signal processing, the method comprising: receiving, from a base station, first network-side assistance information; processing, at the UE, first UE-side assistance information; measuring reference signals from a wireless network to obtain first reference signal received power (RSRP) values; performing training of the AI / ML super model using the first RSRP values, the first network-side assistance information, and the first UE-side assistance information, wherein the first network-side assistance information and the first UE-side assistance information comprise additional physical communication conditions or configurations.

10. The method of claim 9. wherein the first UE-side assistance information and the first network-side assistance information comprises a mixed dataset.

11. The method of claim 10, wherein the mixed dataset comprises one or more of conditional information and implementation information.

12. The method of claim 9, wherein the first network-side assistance information comprises one or more of transmit (Tx) beam information, Tx beam pattern information, Tx beam bandwidth information. Tx beam shape information, Tx beam angleinformation, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, Tx beam timing window information, antenna site information, antenna tilt information, network planning information, transceiver unit (TXRU) information, signal-to-noise (SNR) information, Tx beam identifier (ID) information, and Doppler information.

13. The method of claim 9, wherein the first network-side assistance information is quantized and encoded for input to the AI / ML super model.

14. The method of claim 9 further comprising quantizing and encoding the first UE-side assistance information prior to inputting the first UE-side assistance information into the AI / ML super model.

15. The method of claim 9, wherein the first UE-side assistance information comprises one or more of UE speed information. UE receive (Rx) beam information, UE implementation information, UE distribution information, UE scenario information, and Doppler information.

16. The method of claim 9 further comprising performing monitoring of the AI / ML super model.

17. The method of claim 9 further comprising: determining that the AI / ML super is not adequately trained using prior RSRP values, prior network-side assistance information, and prior UE-side assistance information; and wherein the AI / ML super model is trained using the first RSRP values, the first network-side assistance information, and the first UE-side assistance information in response to the determining that the AI / ML super is not adequately trained using the prior RSRP values, the prior network-side assistance information, and the prior UE-side assistance information.

18. The method of claim 17 further comprising sending, to the base station, in response to determining that the AI / ML super is not adequately trained using the prior RSRP values, the prior network-side assistance information, and the prior UE-side assistance information, a request for the first network-side assistance information.

19. A method of a base station in communication with a user equipment (UE) operating an artificial intelligence (AI) / machine learning (ML) super model for network signal processing, the method comprising: processing first network-side assistance information, for use in AI / ML super model inference and AI / ML super model training; and transmitting, to the UE, the first network-side assistance information.

20. The method of claim 19, wherein the first network-side assistance information comprises one or more of transmit (Tx) beam information, Tx beam pattern information, Tx beam bandwidth information. Tx beam shape information, Tx beam angle information, frequency information, band information, antenna layout information, antenna spacing information, antenna configuration information, Tx beam timing window information, antenna site information, antenna tilt information, network planning information, transceiver unit (TXRU) information, signal-to-noise (SNR) information, Tx beam identifier (ID) information, and Doppler information.

21. The method of claim 19 further comprising quantizing and encoding the first network-side assistance information.

22. The method of claim 19 further comprising: receiving, from the UE, a request for second network-side assistance information; processing the second network-side assistance information, for use in the AI / ML super model inference and the AI / ML super model training; and transmitting, to the UE, the second network-side assistance information.

23. An apparatus comprising means to perform the method of any of claim 1 to claim 22.

24. A computer-readable media comprising instructions to cause an electronic device, upon execution of the instructions by one or more processors of the electronic device, to perform the method of any of claim 1 to claim 22.

25. An apparatus comprising logic, modules, or circuitry to perform the method of any of claim 1 to claim 22.

26. A baseband processor for a user equipment (UE) that is configured to cause the UE to perform one or more elements of any one of claim 1 to claim 18.

27. A baseband processor for a base station that is configured to cause the base station to perform one or more elements of any one of claim 19 to claim 22.

Images