Measurement testing filtering, and prediction
A parallel processing framework for wireless telecommunication systems optimizes testing by evaluating multiple machine learning configurations, addressing inefficiencies in current frameworks and improving accuracy and efficiency in measurement reduction.
Patent Information
- Application Number
- PCT/EP2026/061821
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-09
- Filing Date
- 2026-04-24
- Publication Date
- 2026-11-12
AI Technical Summary
Current testing frameworks for measurement reduction in wireless telecommunication systems, particularly in 5G and 6G networks, lack efficiency and flexibility in evaluating different machine learning configurations for measurement reduction, leading to increased testing time and resource utilization.
A parallel processing framework is introduced to optimize testing by simultaneously evaluating multiple machine learning configurations for measurement reduction, utilizing different skipping patterns and measurement reduction ratios, enabling faster and more accurate performance evaluations.
The parallel processing framework significantly reduces testing time and resource consumption while identifying the most promising configurations for measurement reduction, enhancing the accuracy and efficiency of machine learning models in wireless telecommunication systems.
Smart Images

Figure EP2026061821_12112026_PF_FP_ABST
Abstract
Description
TITLE:MEASUREMENT TESTING FILTERING, AND PREDICTIONFIELD:
[0001] Some example embodiments may generally relate to mobile or wireless telecommunication systems. For example, certain example embodiments may relate to measurement testing, filtering, and prediction.BACKGROUND:
[0002] Examples of mobile or wireless telecommunication systems may include the Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (UTRAN), Long Term Evolution (LTE) Evolved UTRAN (E-UTRAN), LTE-Advanced (LTE-A), Mu IteFire, LTE-A Pro, fifth generation (5G) radio access technology or new radio (NR) access technology and / or sixth generation (6G) radio access technology. Fifth generation (5G) and sixth generation (6G) wireless systems refer to the next generation (NG) of radio systems and network architecture. 5G and 6G network technology is mostly based on new radio (NR) technology, but the 5G / 6G (or NG) network can also build on E-UTRAN radio. It is estimated that NR may provide bitrates on the order of 10-20 Gbit / s or higher, and may support at least enhanced mobile broadband (eMBB) and ultra-reliable low-latency communication (URLLC) as well as massive machine-type communication (mMTC). NR is expected to deliver extreme broadband and ultra-robust, low-latency connectivity and massive networking to support the Internet of Things (loT).SUMMARY:
[0003] According to some aspects, there is provided the subject matter of the independent claims. Some further aspects are defined in the dependent claims.BRIEF DESCRIPTION OF THE DRAWINGS:
[0004] For proper understanding of example embodiments, reference should be made to the accompanying drawings, wherein:
[0005] FIG. 1 illustrates an example intra-frequency measurement reduction ratio in time (MRRT) with two different skipping patterns.
[0006] FIG. 2 illustrates an example parallel testing procedure, according to certain example embodiments.
[0007] FIG. 3 illustrates an example MRRT with 50% and different skipping patterns, according to certain example embodiments.
[0008] FIG. 4 illustrates an example of parallel processing, according to certain example embodiments.
[0009] FIG. 5 illustrates an example flowchart, according to certain example embodiments.
[0010] FIG. 6 illustrates an example signal flow diagram, according to certain example embodiments.
[0011] FIG. 7 illustrates an example flow diagram of a method, according to certain example embodiments.
[0012] FIG. 8 illustrates an example flow diagram of another method, according to certain example embodiments.
[0013] FIG. 9 illustrates an example flow diagram of a another method, according to certain example embodiments.
[0014] FIG. 10 illustrates an example flow diagram of a another method, according to certain example embodiments.
[0015] FIG. 11 illustrates an example flow diagram of a another method, according to certain example embodiments.
[0016] FIG. 12 illustrates an example flow diagram of a another method, according to certain example embodiments.
[0017] FIG. 13 illustrates a set of apparatuses according to certain example embodiments.
[0018] FIG. 14 illustrates an example of a 5G / 6G network and system architecture, according to certain example embodiments.
[0019] FIG. 15 illustrates an example 6G architecture, according to certain example embodiments.DETAILED DESCRIPTION:
[0020] It will be readily understood that the components of certain example embodiments, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. The following is a detailed description of someexample embodiments of systems, methods, apparatuses, and computer program products for to testing framework optimization through parallel processing. For example, certain example embodiments may be directed to testing framework optimization for artificial intelligence machine learning (AIML) through parallel processing.
[0021] The features, structures, or characteristics of example embodiments described throughout this specification may be combined in any suitable manner in one or more example embodiments. For example, the usage of the phrases “certain embodiments,” “an example embodiment,” “some embodiments,” or other similar language, throughout this specification refers to the fact that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment. Thus, appearances of the phrases “in certain embodiments,” “an example embodiment,” “in some embodiments,” “in other embodiments,” or other similar language, throughout this specification do not necessarily refer to the same group of embodiments, and the described features, structures, or characteristics may be combined in any suitable manner in one or more example embodiments. Further, the terms “base station”, “cell”, “node”, “gNB”, “network” or other similar language throughout this specification may be used interchangeably.
[0022] As used herein, “at least one of the following: ” and “at least one of ” and similar wording, where the list of two or more elements are joined by “and” or “or,” mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements.
[0023] In the specifications of the 3rdGeneration Partnership Project (3GPP), measurement reduction may be realized in a temporal domain prediction with measurement reduction ratio in time (MRRT) parameter(s) in the temporal domain. However, for a given selected MRTT value (e.g., 50%), it may be possible to apply different skipping patterns. For instance, FIG.1 illustrates an example intra-frequency with an MRRT with two different skipping patterns.
[0024] For testing, a device under test (DUT) may match a skipping pattern for a given configured MRRT value. However, there is currently no agreement related to skipping pattern indication as to whether the skipping pattern is part of a machine learning (ML) model configuration. Under testing, at each iteration, the DUT may perform measurements only at the identified transmission slots following a skipping pattern within a current observation window (OW). The measurement samples collected within the OW may be input to an artificialintelligence machine learning (AIML) model to output predicted samples within a prediction window (PW), and the predicted samples may then be reported to test equipment (TE). The DUT may also be configured to measure and report ground truth measurements within the PW.
[0025] In addition to determining the predicted samples and ground truth measurements, the TE may determine an error (E) based on a selected evaluation metric (EM). In certain example embodiments, E may indicate to which extent the prediction is different from the ground truth measurement. In some example embodiments, the EM may be, for example, a mean square error value to enable the verification of whether the prediction is close to the ground truth. In other example embodiments, E may be based on the evaluation metric E<T, where T is a threshold error value that may be set to 5%, for example. In certain example embodiments, if E< T is satisfied, it is assumed that the test has passed. However, there can be a requirement that the E for the first evaluation metric (EM), E1 (for absolute reference signal received power (RSRP accuracy), and E for a second EM, E2 (for relative RSRP accuracy), both must satisfy the condition, T, or at least one of them. Further, separate threshold values may be applied such as, for example, T1, T2 for E1 and E2.
[0026] The test procedure by the TE may be looped over multiple iterations following the selected skipping pattern, where a “Max” value can be predefined at the TE corresponding to the maximum number of test iterations. When both E1 and E2 are configured by the TE, the TE may determine E, and then conditions E1 and E2 may be evaluated independently and indicate if the test passes or fails the considered iteration. The test results with respect to E1 and E2 may be stored in the TE, and the same test may be repeated until the iteration reaches the “Max” value. In addition to the error testing procedure, if the ratio of tests passed to tests failed reaches a certain predefined threshold, the testing may be deemed as ended. For example, if the ratio is 0.9 or 90% of the tests passed out of the total number of tests, the testing may end.
[0027] Various options may be available in identifying the skipping pattern information by the UE. For example, in one option, the skipping pattern may be part of the ML inference configuration such that the skipping pattern information is available along with OW and PW parameters. In another option, the skipping pattern may not be part of the ML inference configuration. In this example, the UE may autonomously select the skipping pattern for agiven configured MRRT value. However, in this example, an agreement may be needed on the specific test configuration to explicitly indicate to the DUT on the skipping pattern to use and the identification of the measurement slots.
[0028] To test the measurement reduction, the TE may configure the DUT with MRRT and corresponding skipping pattern(s). The trained model to test may expect a fixed number of input samples (within the OW) to realize an inference and provide a prediction.
[0029] A test occurrence may allow the evaluation of the performance of a given ML model which relates to a preselected measurement configuration (with OW length, MRRT, and skipping pattern). Thus, testing different ML configurations (with different parameter values and different ML trained models) may require a new testing setup each time. Accordingly, to address these drawbacks, certain example embodiments may provide a testing framework / procedure to test multiple configurations while ensuring faster and improved performance evaluations. According to certain example embodiments, it may also be possible to select the most representative configurations to test.
[0030] Certain example embodiments may provide a procedure and related signaling enhancement that targets testing multiple RAN4 mobility reduction measurements with minimum DUT measurements. In doing so, it may be possible to reuse, as much as possible, the DUT measurements to perform parallel tests of multiple configurations.
[0031] FIG. 2 illustrates an example parallel testing procedure, according to certain example embodiments. The testing procedure may be applicable for temporal domain prediction case B, which targets measurement reduction within AIML mobility use cases. As illustrated in FIG.2, at 200, the DUT may perform a complete set of measurements within a time window indicated by the TE (e.g., the OW for the considered AIML mobility functionality). At 205, the DUT may receive configurations for different ML models to test. Additionally, depending on the set of ML models to test and related configurations (e.g., particular skipping patterns for a given MRRT value), the DUT may realize different filtering on the complete set of measurements during the considered time window. At 210, 215, and 220, as a result of the different filtering steps (e.g., MRRT at 20%, 30%, 40%, 50%, and other percentages, etc.), it may be possible to obtain different sets of filtered data related to different ML model (e.g., neural network model) configurations (which may include the number of samples). According to certain example embodiments, the value of the MRRT parameter may be determined basedon capabilities of the DUT, and is not limited to any of the example percentages described herein. As such, at 225, 230, and 235, it may be possible to trigger different model inferences in parallel to account for the different filtered datasets as inputs.
[0032] At 240, 245, and 250, performance of the different models may be evaluated individually considering the same ground truth samples within the PW. The different performance metrics that are evaluated for the different models at a given test iteration may be gathered at the DUT or TE.
[0033] In certain example embodiments, at 255, the TE may perform a closed loop step to tune the configuration of the next testing framework. For example, at 260, the TE may select the most promising configurations based on the performances of the previously tested configurations. For instance, in some example embodiments, the most promising configurations may correspond to configurations that provide the highest accuracy for the model prediction (e.g., the lowest error values computed between the predictions and the ground truth). According to certain example embodiments, this closed loop step may serve as an additional requirement of the testing framework by not specifying in the beginning the exact configuration, but rather exploring and identifying the most promising configuration setups (e.g., skipping patterns for a given MRRT).
[0034] According to certain example embodiments, a sequence-to-sequence prediction model may be used to leverage past measurement samples and predict future measurement samples across multiple time steps. By doing so, it may be possible for certain example embodiments to reduce UE measurements by relying on the predicted measurements. The level of measurement reduction for these examples may be setup through MRRT parameter(s). However, in other example embodiments, a fixed MRRT value may relate to different skipping patterns that specify precisely in a sequence of time iterations what to measure and what to omit. Additionally, different skipping patterns may product different prediction performance results for the same MRRT value. The impact of the different skipping patterns may be more visible for high MRRT values, which may eventually lead to a higher number of possible skipping patterns.
[0035] FIG. 3 illustrates an example MRRT with 50% and different skipping patterns, according to certain example embodiments. In the testing of certain example embodiments, the MRRT value may be set, for example, to [50, 60, 66, 75, 80]%, and the UE speed may beset to [30, 60, 90, 120] km / h, resulting in a total number of 20 combinations. Additionally, as illustrated in FIG. 3, for each MRRT value, different skipping patterns may be considered, which may enable the ability to test the impact of reduction slot placement within the OW on the performance accuracy. Thus, the number of tests may increase considerably to cover different assumptions on the conditions (e.g., including UE speed, line of sight (LOS), and non-line of sight (NLOS)), and specific ML model parameters (e.g., PW, OW, MRRT, and related skipping pattern). The example illustrated in FIG. 2 highlights the different skipping patterns for MRRT of 50%. According to certain example embodiments, the number of possible combinations to skip 5 slots among 10 total slots within the OW is =5!(1I0Ol15), = 252 total number of possible skipping patterns.
[0036] Due to the high number of possible skipping patterns, certain example embodiments may provide parallel processing for testing the framework which specifically targets multiple skipping patterns (with the same or different MRRT values). The parallelization provided by certain example embodiments may enable testing of different skipping patterns which can eventually relate to different MRRT values, but with the constraint of covering the same total time window (7), which may serve as the OW. With parallel processing, it may also be possible to enable the DUT to perform all the measurements within T and realize multiple filtering operations to extract precisely the slots marked as measured in the skipping pattern to test each time.
[0037] FIG. 4 illustrates an example of parallel processing, according to certain example embodiments. In particular, FIG. 4 illustrates an example of parallel processing of the testing with two MRRT values at 50% and 60%. As also illustrated in FIG. 4, the parallel processing of the testing framework may use the same set of measured samples (e.g., total of 10) to extract four different skipping patterns. In the example of FIG. 4, patterns #1 and #2 correspond to MRRT = 50%, whereas patterns #3 and #4 correspond to MRRT = 60%. As illustrated in FIG. 4, the outcomes of the filtering operation may result in 5 samples for the first two patterns (#1 and #2), and 6 samples for the last two patterns (#3 and #4). Thus, different ML models may be tested in parallel (e.g., at the same time), and each ML model may result in a different prediction output. As illustrated in FIG. 4, to optimize further testing, the ML models in certain example embodiments may be tested within a same PW to enable usingthe same ground truth PW samples to evaluate all the models at the same time (as shown in FIG. 3, PW is 2). According to certain example embodiments, parallel testing through appropriate data filtering may allow access to the performance of the ML model in terms of accuracy (e.g., absolute and / or relative RSRP accuracy) and not of delay, for which the precise configuration with measured and skipped measurements may be done in practice.
[0038] FIG. 5 illustrates an example flowchart, according to certain example embodiments. In certain example embodiments, the procedures illustrated in FIG. 5 may be performed by either the DUT or TE. Accounting for DUT capabilities, the TE may determine to proceed with a parallel processing that includes determining the configurations that may share the same time window for the initial measurements (T (e.g., OW)), as well as the PW (to use single set of ground truth dataset for the performance evaluation). According to some example embodiments, the DUT (e.g., UE) may inform the TE (e.g., network) about its capability of running up to #N ML models, and may provide configurations and details supported by the DUT. At 500, based on the TE indication and provided configurations, the DUT may perform all the required measurements within the time window of size T over which the different skipping patterns may be applied. At 505, the DUT may realize (e.g., implement or execute filtering operation) the total #N filtering operations to extract the different #N of datasets that correspond to the different skipping patterns, and related MRRT to be tested in parallel. At 510, the #N ML inferences may be realized (e.g., implemented, run, etc.) in parallel to account for the different input datasets. That is, all the different ML models may be run in parallel (e.g., at the same time) to each other. At 515, after running each ML model, a prediction result may be output from each ML model, and reported to the TE.
[0039] At 520, based on the TE configuration, the DUT may measure ground truth data (within PW) and report it to the TE. In certain example embodiments, depending on the MRRT and skipping patterns, the ground truth data collection relates to the PW which is common for the ML models tested in parallel. At 525, the TE may identify a particular configuration for parallel processing, including ML models and related parameters, and transmit the configuration to the DUT.
[0040] At 530, the TE may determine #N performance results (e.g., error evaluated such as, for example mean square error between predicted measurements and ground truth measurements) for each parallel test based on the received model outputs and relatedcommon ground truth samples. At 535, the TE may determine whether the #N parallel tests pass or fail based on the performance results of each of the parallel tests. At 540, the TE may determine which additional configuration are more interesting / promising to evaluate in future rounds of testing based on the obtained performance results over the different testing configurations.
[0041] FIG. 6 illustrates an example signal flow diagram, according to certain example embodiments. At 610, the TE 605 (e.g., network) may enable a first AIML model for temporal domain prediction case B for testing that involves the TE 605 and DUT 600 (e.g., UE). At 615, the DUT 600 may share its capabilities with the TE 605 with an indication about the DUT’s 600 capability / ability to support multiple parallel ML model inferences. According to certain example embodiments, if the DUT 600 is capable of supporting multiple parallel ML model inferences, the DUT 600 may also indicate to the TE 605, and if so, a number of ML model inferences that can be supported along with other related parameters.
[0042] At 620, the TE 605 may configure the testing procedure that should be run by the DUT 600 based on the shared DUT 600 capabilities information. Along with the configured testing procedure, the TE 605 may also configure different ML models to implement at the DUT 600 and related PW, OW, MRRT, and skipping patterns. In certain example embodiments, because the DUT 600 may realize the first complete set of measurements, the TE 605 may indicate to the DUT 600 precisely the time window Tto completely measure the complete set of samples, and for the DUT 600 to perform filtering of the set of measurements within the time window T.
[0043] At 625, the DUT 600 may perform RSRP measurements during the complete OW (e.g., time window 7). At 630, the DUT 600 may perform multiple filtering operations in parallel (e.g., at the same time) to extract the datasets (e.g., subsets) following the different MRRT / skipping patterns, and the extracted datasets may subsequently serve as input for each corresponding ML model for testing. At 635, the DUT 600 may run multiple ML inferences corresponding to each of the multiple ML models with the input data extracted at 630. According to certain example embodiments, after each ML model has been implemented, the output obtained (e.g., prediction results) from implementing each ML model may be obtained during the PW. At 640, the different obtained prediction results (e.g., RSRP) for #N ML models may be transmitted to the TE 605. At 645, the TE 605 may transmit a radioresource control (RRC) configuration to the DUT 600 forthe ground truth measurements (e.g., RSRP measurements) to be performed by the DUT 600 within the PW. According to some example embodiments, transmission of the RRC configuration may be performed together with operation 620.
[0044] At 650, the DUT 600 may perform ground truth measurements (e.g., RSRP) in the PW. According to certain example embodiments, the DUT 600 may measure the RSRP within the PW once for all the different configurations. At 655, the DUT 600 may share the ground truth measurements with the TE 605. At 660, the TE 605 may determine a performance metric (e.g., absolute or relative RSRP accuracy, which corresponds to the error (E)) for each concurrent ML model using the different received prediction output for each ML model (operation 640), and the single ground truth samples (operation 655). At 665, the TE 605 may evaluate whether the test passes or fails for each ML model based on, for example, the error (E).
[0045] FIG. 7 illustrates an example flow diagram of a method, according to certain example embodiments. In an example embodiment, the method of FIG. 7 may be performed by a network entity, or a group of multiple network elements in a 3GPP system, such as LTE or 5G-NR. For instance, in an example embodiment, the method of FIG. 7 may be performed by a DUT or UE, similar to apparatus 10 illustrated in FIG. 13.
[0046] As illustrated in FIG. 7, the method may include, at 700, receiving, from a network element, a configuration for testing measurement prediction functionality for at least two measurement reduction ratios. The method may also include, at 705, performing, based on the configuration, a set of measurements a predefined time window. The method may further include, at 710, measurements to extract at least two subsets of measurements corresponding to the at least two measurement reduction ratios. Further, the method may include, at 715, generating a plurality of predicted measurements by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0047] According to certain example embodiments, the predefined time window may be an observation time window of the at least one learning model. According to some example embodiments, the method may also include reporting the plurality of predicted measurementsto the network element. According to other example embodiments, the set of measurements may correspond to reference signal received power measurements.
[0048] In certain example embodiments, the filtering operations may be performed based on different measurement reduction ratio in time parameters or different skipping patterns of slots. In some example embodiments, the predefined time window may be the same for the set of measurements as for another set of measurements. In other example embodiments, the at least one learning model may be executed in parallel with another learning model, and each of the plurality of predicted measurements may be generated within a same prediction time window. In further example embodiments, the method may also include receiving, from the network element, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
[0049] FIG. 8 illustrates an example flow diagram of a method, according to certain example embodiments. In an example embodiment, the method of FIG. 8 may be performed by a network entity, or a group of multiple network elements in a 3GPP system, such as LTE or 5G-NR. For instance, in an example embodiment, the method of FIG. 8 may be performed by a DUT or UE, similar to apparatus 10 illustrated in FIG. 13.
[0050] As illustrated in FIG. 8, the method may include, at 800, transmitting, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement reduction ratios. The method may also include, at 805, receiving, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window. The method may further include, at 810, receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, the method may include, at 815, determining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
[0051] According to certain example embodiments, the method may also include determining whether the different learning models pass or fail based on the performance metric. According to some example embodiments, the method may further include transmitting, to the user equipment, another set of measurement reduction ratios at leastpartly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
[0052] FIG. 9 illustrates an example flow diagram of a method, according to certain example embodiments. In an example embodiment, the method of FIG. 9 may be performed by a network entity, or a group of multiple network elements in a 3GPP system, such as LTE or 5G-NR. For instance, in an example embodiment, the method of FIG. 9 may be performed by a DUT or UE, similar to apparatus 10 illustrated in FIG. 13.
[0053] As illustrated in FIG. 9, the method may include, at 900, receiving, from a network element, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns. The method may also include, at 905, performing, based on the configuration, a set of measurements within a predefined time window. The method may further include, at 910, performing, based on the configuration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement skipping patterns. In addition, the method may include, at 915, generating a plurality of predicted measurements by executing by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0054] According to certain example embodiments, the predefined time window may be an observation time window of the at least one learning model. According to some example embodiments, the method may also include reporting the plurality of predicted measurements to the network element. According to other example embodiments, the set of measurements may correspond to reference signal received power measurements.
[0055] In certain example embodiments, the filtering operations may be performed based on different measurement reduction ratio in time parameters or different skipping patterns of slots. In some example embodiments, the predefined time window may be the same for the set of measurements as for another set of measurements. In other example embodiments, the at least one learning model may be executed in parallel with another learning model, and each of the plurality of predicted measurements may be generated within a same prediction time window.
[0056] FIG. 10 illustrates an example flow diagram of another method, according to certain example embodiments. In an example embodiment, the method of FIG. 10 may be performedby a network entity, or a group of multiple network elements in a 3GPP system, such as LTE or 5G-NR. For instance, in an example embodiment, the method of FIG. 10 may be performed by a TE, network, orgNB, similar to apparatus 20 illustrated in FIG. 13.
[0057] As illustrated in FIG. 10, the method may include, at 1000, transmitting, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns. The method may also include, at 1005, receiving, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration. According to certain example embodiments, the plurality of predicted measurements may be obtained during a same prediction time window. The method may further include, at 1010, receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, the method may include, at 1015, determining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
[0058] According to certain example embodiments, the method may also include determining whether the different learning models pass or fail based on the performance metric. According to other example embodiments, the method may further include transmitting, to the user equipment, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
[0059] FIG. 11 illustrates an example flow diagram of another method, according to certain example embodiments. In an example embodiment, the method of FIG. 11 may be performed by a network entity, or a group of multiple network elements in a 3GPP system, such as LTE or 5G-NR. For instance, in an example embodiment, the method of FIG. 11 may be performed by a TE, network, orgNB, similar to apparatus 20 illustrated in FIG. 13.
[0060] As illustrated in FIG. 11, the method may include, at 1100, receiving, from a network element, a configuration for testing measurement prediction functionality for at least two prediction configurations, wherein the at least two prediction configurations indicate a same prediction time window. The method may also include, at 1105, reporting, to the network element, information on predicted measurements based on the at least two prediction configurations. The method may further include, at 1110, performing ground truthmeasurements at the same prediction window of the predicted measurements. In addition, the method may include, at 1115, reporting the ground truth measurements to the network element.
[0061] According to certain example embodiments, two key performance indicators ,may be evaluated for the at least two prediction configurations with the same ground truth measurements. According to some example embodiments, the method may also include performing, based on the configuration, filtering operations on a set of measurements to extract at least two subsets of measurements. According to other example embodiments, the set of measurements may correspond to reference signal received power measurements.
[0062] In certain example embodiments, the filtering operations may be performed based on the least two prediction configurations. In some example embodiments, the method may also include generating the predicted measurements by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0063] FIG. 12 illustrates an example flow diagram of another method, according to certain example embodiments. In an example embodiment, the method of FIG. 12 may be performed by a network entity, or a group of multiple network elements in a 3GPP system, such as LTE or 5G-NR. For instance, in an example embodiment, the method of FIG. 12 may be performed by a TE, network, orgNB, similar to apparatus 20 illustrated in FIG. 13.
[0064] As illustrated in FIG. 12, the method may include, at 1200, transmit, to a user equipment, a configuration for testing measurement prediction functionality for at least two prediction configurations. The at least two prediction configurations indicate a same prediction window. The method may also include, at 1205, receive, from the user equipment, information on predicted measurements based on the at least two prediction configurations. The method may further include, at 1210, the method may include receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. The method may further include, at 1215 determine, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with at least two measurement reduction ratios.
[0065] According to certain example embodiments, the method may also include determining whether the different learning models pass or fail based on the performance metric. According to other example embodiments, the method may further includetransmiting, to the user equipment, a set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
[0066] FIG. 13 illustrates a set of apparatuses 10 and 20 according to certain example embodiments. In certain example embodiments, apparatuses 10 and 20 may be elements in a communications network or associated with such a network. For example, apparatus 10 may be a UE, or other similar radio communication computer device, and apparatus 20 may be a BS, gNB, network, or other similar computing device.
[0067] In some example embodiments, apparatuses 10 and 20 may include one or more processors, one or more computer-readable storage medium (for example, memory, storage, or the like), one or more radio access components (for example, a modem, a transceiver, or the like), and / or a user interface. In some example embodiments, apparatuses 10 and 20 may be configured to operate using one or more radio access technologies, such as GSM, LTE, LTE-A, NR, 5G, WLAN, WiFi, NB-loT, Bluetooth, NFC, MulteFire, and / or any other radio access technologies. It should be noted that one of ordinary skill in the art would understand that apparatuses 10 and 20 may include components or features not shown in FIG. 13.
[0068] As illustrated in the example of FIG. 13, apparatuses 10 and 20 may include or be coupled to a processor 12 and 22 for processing information and executing instructions or operations. Processors 12 and 22 may be any type of general or specific purpose processor. In fact, processors 12 and 22 may include one or more of general-purpose computers, special purpose computers, microprocessors, DSPs, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), and processors based on a multi-core processor architecture, as examples. While a single processor 12 and 22 is shown in FIG.13, multiple processors may be utilized according to other example embodiments. For example, it should be understood that, in certain example embodiments, apparatuses 10 and 20 may include two or more processors that may form a multiprocessor system (e.g., in this case processors 12 and 22 may represent a multiprocessor) that may support multiprocessing. According to certain example embodiments, the multiprocessor system may be tightly coupled or loosely coupled (e.g., to form a computer cluster).
[0069] Processors 12 and 22 may perform functions associated with the operation of apparatuses 10 and 20 including, as some examples, precoding of antenna gain / phaseparameters, encoding and decoding of individual bits forming a communication message, formatting of information, and overall control of the apparatuses 10 and 20, including processes and examples illustrated in FIGs. 1-12.
[0070] Apparatuses 10 and 20 may further include or be coupled to a memories 14 and 24 (internal or external), which may be respectively coupled to processors 12 and 22 for storing information and instructions that may be executed by processors 12 and 22. Memories 14 and 24 may be one or more memories and of any type suitable to the local application environment, and may be implemented using any suitable volatile or nonvolatile data storage technology such as a semiconductor-based memory device, a magnetic memory device and system, an optical memory device and system, fixed memory, and / or removable memory. For example, memories 14 and 24 can be comprised of any combination of random access memory (RAM), read only memory (ROM), static storage such as a magnetic or optical disk, hard disk drive (HDD), or any other type of non-transitory machine or computer readable media. The instructions stored in memories 14 and 24 may include program instructions or computer program code that, when executed by processors 12 and 22, enable the apparatuses 10 and 20 to perform tasks as described herein.
[0071] In certain example embodiments, apparatuses 10 and 20 may further include or be coupled to (internal or external) a drive or port that is configured to accept and read an external computer readable storage medium, such as an optical disc, USB drive, flash drive, or any other storage medium. For example, the external computer readable storage medium may store a computer program or software for execution by processors 12 and 22 and / or apparatuses 10 and 20 to perform any of the methods and examples illustrated in FIGs. 1-12.
[0072] In some example embodiments, apparatuses 10 and 20 may also include or be coupled to one or more antennas 15 and 25 for receiving a downlink signal and for transmitting via an UL from apparatuses 10 and 20. Apparatuses 10 and 20 may further include a transceivers 18 and 28 configured to transmit and receive information. The transceivers 18 and 28 may also include a radio interface (e.g., a modem) coupled to the antennas 15 and 25. The radio interface may correspond to a plurality of radio access technologies including one or more of GSM, LTE, LTE-A, 5G, NR, WLAN, NB-loT, Bluetooth, BT-LE, NFC, RFID, UWB, and the like. The radio interface may include other components, such as filters, converters (for example, digital-to-analog converters and the like), symbol demappers, signalshaping components, an IFFT module, and the like, to process symbols, such as OFDMA symbols, carried by a downlink or an UL.
[0073] For instance, transceivers 18 and 28 may be configured to modulate information on to a carrier waveform for transmission by the antennas 15 and 25 and demodulate information received via the antenna 15 and 25 for further processing by other elements of apparatuses 10 and 20. In other example embodiments, transceivers 18 and 28 may be capable of transmitting and receiving signals or data directly. Additionally or alternatively, in some example embodiments, apparatus 10 may include an input and / or output device (I / O device). In certain example embodiments, apparatuses 10 and 20 may further include a user interface, such as a graphical user interface or touchscreen.
[0074] In certain example embodiments, memories 14 and 24 store software modules that provide functionality when executed by processors 12 and 22. The modules may include, for example, an operating system that provides operating system functionality for apparatuses 10 and 20. The memory may also store one or more functional modules, such as an application or program, to provide additional functionality for apparatuses 10 and 20. The components of apparatuses 10 and 20 may be implemented in hardware, or as any suitable combination of hardware and software. According to certain example embodiments, apparatuses 10 and 20 may optionally be configured to communicate each other (in any combination) via a wireless or wired communication links 70 according to any radio access technology, such as NR.
[0075] According to certain example embodiments, processors 12 and 22 and memories 14 and 24 may be included in or may form a part of processing circuitry or control circuitry. In addition, in some example embodiments, transceivers 18 and 28 may be included in or may form a part of transceiving circuitry.
[0076] For instance, in certain example embodiments, apparatus 10 may be controlled by memory 14 and processor 12 to receive, from a network element, a configuration fortesting measurement prediction functionality for at least two measurement reduction ratios. Apparatus 10 may also be controlled by memory 14 and processor 12 to perform, based on the configuration, a set of measurements a predefined time window. Apparatus 10 may further be controlled by memory 14 and processor 12 to perform, based on the configuration, filtering operations on the set of measurements to extract at least two subsets ofmeasurements corresponding to the at least two measurement reduction ratios. In addition, apparatus 10 may be controlled by memory 14 and processor 12 to generate a plurality of predicted measurements by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0077] In other example embodiments, apparatus 20 may be controlled by memory 24 and processor 22 to transmit, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement reduction ratios. Apparatus 20 may also be controlled by memory 24 and processor 22 to receive, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window. Apparatus 20 may further be controlled by memory 24 and processor 22 to receive, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, apparatus 20 may be controlled by memory 24 and processor 22 to determine, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
[0078] In other example embodiments, apparatus 10 may be controlled by memory 14 and processor 12 to receive, from a network element, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns. Apparatus 10 may also be controlled by memory 14 and processor 12 to perform, based on the configuration, a set of measurements within a predefined time window. Apparatus 10 may further be controlled by memory 14 and processor 12 to perform, based on the configuration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement skipping patterns. In addition, apparatus 10 may be controlled by memory 14 and processor 12 to generate a plurality of predicted measurements by executing by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0079] In other example embodiments, apparatus 20 may be controlled by memory 24 and processor 22 to transmit, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns. Apparatus 20 may also be controlled by memory 24 and processor 22 to receive, from the user equipment, areport comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window. Apparatus 20 may further be controlled by memory 24 and processor 22 to receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, apparatus 20 may be controlled by memory 24 and processor 22 to determine, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
[0080] In other example embodiments, apparatus 10 may be controlled by memory 14 and processor 12 to receive, from a network element, a configuration for testing measurement prediction functionality for at least two prediction configurations. According to certain example embodiments, the at least two prediction configurations indicate a same prediction time window. Apparatus 10 may also be controlled by memory 14 and processor 12 to report, to the network element, information on predicted measurements based on the at least two prediction configurations. Apparatus 10 may further be controlled by memory 14 and processor 12 to perform ground truth measurements at the same prediction window of the predicted measurements. In addition, apparatus 10 may be controlled by memory 14 and processor 12 to report the ground truth measurements to the network element.
[0081] In other example embodiments, apparatus 20 may be controlled by memory 24 and processor 22 to transmit, to a user equipment, a configuration for testing measurement prediction functionality for at least two prediction configurations. According to certain example embodiments, the at least two prediction configurations indicate a same prediction window. Apparatus 20 may also be controlled by memory 24 and processor 22 to receive, from the user equipment, information on predicted measurements based on the at least two prediction configurations. Apparatus 20 may further be controlled by memory 24 and processor 22 to receive, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, apparatus 20 may be controlled by memory 24 and processor 22 to determine, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with at least two measurement reduction ratios.
[0082] In some example embodiments, an apparatus (e.g., apparatus 10 and / or apparatus 20) may include means for performing a method, a process, or any of the variants discussed herein. Examples of the means may include one or more processors, memory, controllers, transmitters, receivers, and / or computer program code for causing the performance of the operations.
[0083] Certain example embodiments may be directed to an apparatus that includes means for receiving, from a network element, a configuration for testing measurement prediction functionality for at least two measurement reduction ratios. The apparatus may also include means for performing, based on the configuration, a set of measurements a predefined time window. The apparatus may further include means for performing, based on the configuration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement reduction ratios. In addition, the apparatus may include means for generating a plurality of predicted measurements by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0084] Other example embodiments may be directed to an apparatus that includes means for transmitting, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement reduction ratios. The apparatus may also include means for receiving, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window. The apparatus may further include means for receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, the apparatus may include means for determining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
[0085] Other example embodiments may be directed to an apparatus that includes means for receiving, from a network element, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns. The apparatus may also include means for performing, based on the configuration, a set of measurements within a predefined time window. The apparatus may further include means for performing, based on theconfiguration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement skipping patterns. In addition, the apparatus may include means for generating a plurality of predicted measurements by executing by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
[0086] Other example embodiments may be directed to an apparatus that includes means for transmitting, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns. The apparatus may also include means for receiving, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration. According to certain example embodiments, the plurality of predicted measurements may be obtained during a same prediction time window. The apparatus may further include means for receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, the apparatus may include means for determining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
[0087] Other example embodiments may be directed to an apparatus that includes means for receiving, from a network element, a configuration for testing measurement prediction functionality for at least two prediction configurations. According to certain example embodiments, the at least two prediction configurations indicate a same prediction time window. The apparatus may also include means for reporting, to the network element, information on predicted measurements based on the at least two prediction configurations. The apparatus may further include means for performing ground truth measurements at the same prediction window of the predicted measurements. In addition, the apparatus may include means for reporting the ground truth measurements to the network element.
[0088] Other example embodiments may be directed to an apparatus that includes means for transmitting, to a user equipment, a configuration for testing measurement prediction functionality for at least two prediction configurations, wherein the at least two prediction configurations indicate a same prediction window. The apparatus may also include means for receiving, from the user equipment, information on predicted measurements based on the at least two prediction configurations. The apparatus may further include receive, from theuser equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window. In addition, the apparatus may include means for determining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with at least two measurement reduction ratios.
[0089] FIG. 14 illustrates an example of a 5G / 6G network and system architecture, according to certain example embodiments. Shown are multiple network functions that may be implemented as software operating as part of a network device or dedicated hardware, as a network device itself or dedicated hardware, or as a virtual function operating as a network device or dedicated hardware. The UE illustrated in FIG. 14 may be similar to UE 10. The user plane function (UPF) may provide services such as intra-RAT and inter-RAT mobility, routing and forwarding of data packets, inspection of packets, user plane quality of service (QoS) processing, buffering of downlink packets, and / or triggering of downlink data notifications. The application function (AF) may primarily interface with the core network to facilitate application usage of traffic routing and interact with the policy framework .
[0090] FIG. 15 illustrates an example 6G architecture, according to certain example embodiments. In particular, the 6G architecture in FIG. 9 may support LCM configured to natively support AI / ML, cloud-native functionalities. Additionally, 6G gNBs may be configured to support multi-RAT spectrum sharing (MRSS).
[0091] Certain example embodiments described herein provide several technical improvements, enhancements, and / or advantages. For instance, in some example embodiments, it may be possible to optimize a testing framework / procedure to test multiple configurations (e.g., MRRT and ML models), and ensure faster performance evaluations. In other example embodiments, it may be possible to select the most representative configuration to test based on the evaluations.
[0092] A computer program product may include one or more computer-executable components which, when the program is run, are configured to carry out some example embodiments. The one or more computer-executable components may be at least one software code or portions of it. Modifications and configurations required for implementing functionality of certain example embodiments may be performed as routine(s), which may beimplemented as added or updated software routine(s). Software routine(s) may be downloaded into the apparatus.
[0093] As an example, software or a computer program code or portions of it may be in a source code form, object code form, or in some intermediate form, and it may be stored in some sort of carrier, distribution medium, or computer readable medium, which may be any entity or device capable of carrying the program. Such carriers may include a record medium, computer memory, read-only memory, photoelectrical and / or electrical carrier signal, telecommunications signal, and software distribution package, for example. Depending on the processing power needed, the computer program may be executed in a single electronic digital computer or it may be distributed amongst a number of computers. The computer readable medium or computer readable storage medium may be a non-transitory medium.
[0094] In other example embodiments, the functionality may be performed by hardware or circuitry included in an apparatus (e.g., apparatus 10 or apparatus 20), for example through the use of an application specific integrated circuit (ASIC), a programmable gate array (PGA), a field programmable gate array (FPGA), or any other combination of hardware and software. In yet another example embodiment, the functionality may be implemented as a signal, a nontangible means that can be carried by an electromagnetic signal downloaded from the Internet or other network.
[0095] According to certain example embodiments, an apparatus, such as a node, device, or a corresponding component, may be configured as circuitry, a computer or a microprocessor, such as single-chip computer element, or as a chipset, including at least a memory for providing storage capacity used for arithmetic operation and an operation processor for executing the arithmetic operation.
[0096] One having ordinary skill in the art will readily understand that the disclosure as discussed above may be practiced with procedures in a different order, and / or with hardware elements in configurations which are different than those which are disclosed. Therefore, although the disclosure has been described based upon these example embodiments, it would be apparent to those of skill in the art that certain modifications, variations, and alternative constructions would be apparent, while remaining within the spirit and scope of example embodiments. Although the above embodiments refer to 5G NR and LTEtechnology, the above embodiments may also apply to any other present or future 3GPP technology, such as LTE-advanced, and / or fourth generation (4G) technology.
[0097] Partial Glossary:
[0098] 3GPP 3rd Generation Partnership Project
[0099] 5G 5th Generation
[0100] 5GC 5G Core
[0101] 5GCN 5G Core Network
[0102] Al Artificial Intelligence
[0103] BS Base Station
[0104] BW Bandwidth
[0105] DL Downlink
[0106] DUT Device Under Test
[0107] eNB Enhanced Node B
[0108] E-UTRAN Evolved UTRAN
[0109] gNB 5G or Next Generation NodeB
[0110] ML Machine Learning
[0111] MRRT Measurement Reduction Ratio in Time
[0112] NR New Radio
[0113] OW Observation Window
[0114] PW Prediction Window
[0115] RF Radio Frequency
[0116] RRC Radio Resource Control
[0117] RSRP Reference Signal Received Power
[0118] TE Testing Equipment
[0119] UE User Equipment
[0120] UL Uplink
Claims
WE CLAIM:
1. An apparatus, comprising:at least one processor; andat least one memory storing instructions which, when executed by the at least one processor, cause the apparatus to at least:receive, from a network element, a configuration fortesting measurement prediction functionality for at least two measurement skipping patterns;perform, based on the configuration, a set of measurements within a predefined time window;perform, based on the configuration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement skipping patterns; andgenerate a plurality of predicted measurements by executing by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
2. The apparatus according to claim 1, wherein the predefined time window is an observation time window of the at least one learning model.
3. The apparatus according to claims 1 or 2, wherein the instructions, when executed by the at least one processor, further cause the apparatus to:report the plurality of predicted measurements to the network element.
4. The apparatus according to any one of claims 1-3, wherein the set of measurements correspond to reference signal received power measurements.
5. The apparatus according to any one of claims 1-4, wherein the filtering operations are performed based on different measurement reduction ratio in time parameters or different skipping patterns of slots.
6. The apparatus according to any one of claims 1-5, wherein the predefined time window is the same for the set of measurements as for another set of measurements.
7. The apparatus according to any one of claims 1-6,wherein the at least one learning model is executed in parallel with another learning model, andwherein each of the plurality of predicted measurements are generated within a same prediction time window.
8. The apparatus according to any one of claims 1-7, wherein the instructions, when executed by the at least one processor, further cause the apparatus to:receive, from the network element, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
9. An apparatus, comprising:at least one processor; andat least one memory storing instructions which, when executed by the at least one processor, cause the apparatus to at least:transmit, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns;receive, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window;receive, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window; anddetermine, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
10. The apparatus according to claim 9, wherein the instructions, when executed by theat least one processor, further cause the apparatus to:determine whether the different learning models pass or fail based on the performance metric.
11. The apparatus according to claims 9 or 10, wherein the instructions, when executed by the at least one processor, further cause the apparatus to:transmit, to the user equipment, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
12. A method, comprising:receiving, from a network element, a configuration fortesting measurement prediction functionality for at least two measurement skipping patterns;performing, based on the configuration, a set of measurements within a predefined time window;performing, based on the configuration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement skipping patterns; andgenerating a plurality of predicted measurements by executing by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
13. The method according to claim 12, wherein the predefined time window is an observation time window of the at least one learning model.
14. The method according to claims 12 or 13, further comprising:reporting the plurality of predicted measurements to the network element.
15. The method according to any one of claims 12-14, wherein the set of measurements correspond to reference signal received power measurements.
16. The method according to any one of claims 12-15, wherein the filtering operations are performed based on different measurement reduction ratio in time parameters or different skipping patterns of slots.
17. The method according to any one of claims 12-16, wherein the predefined time window is the same for the set of measurements as for another set of measurements.
18. The method according to any one of claims 12-17,wherein the at least one learning model is executed in parallel with another learning model, andwherein each of the plurality of predicted measurements are generated within a same prediction time window.
19. The method according to any one of claims 12-18, further comprising:receiving, from the network element, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
20. A method, comprising:transmitting, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement skipping patterns;receiving, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window;receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window; anddetermining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
21. The method according to claim 9, further comprising:determining whether the different learning models pass or fail based on the performance metric.
22. The method according to claims 9 or 10, further comprising:transmitting, to the user equipment, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
23. An apparatus, comprising:means for receiving, from a network element, a configuration fortesting measurement prediction functionality for at least two measurement skipping patterns;means for performing, based on the configuration, a set of measurements within a predefined time window;means for performing, based on the configuration, filtering operations on the set of measurements to extract at least two subsets of measurements corresponding to the at least two measurement skipping patterns; andmeans for generating a plurality of predicted measurements by executing by executing at least one learning model with the at least two subsets of measurements serving as input for the at least one learning model.
24. The apparatus according to claim 23, wherein the predefined time window is an observation time window of the at least one learning model.
25. The apparatus according to claims 23 or 24, further comprising:means for reporting the plurality of predicted measurements to the network element.
26. The apparatus according to any one of claims 23-25, wherein the set of measurements correspond to reference signal received power measurements.
27. The apparatus according to any one of claims 23-26, wherein the filtering operations are performed based on different measurement reduction ratio in time parameters or differentskipping paterns of slots.
28. The apparatus according to any one of claims 23-27, wherein the predefined time window is the same for the set of measurements as for another set of measurements.
29. The apparatus according to any one of claims 23-28,wherein the at least one learning model is executed in parallel with another learning model, andwherein each of the plurality of predicted measurements are generated within a same prediction time window.
30. The apparatus according to any one of claims 23-29, further comprising:means for receiving, from the network element, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
31. An apparatus, comprising:means for transmiting, to a user equipment, a configuration for testing measurement prediction functionality for at least two measurement skipping paterns;means for receiving, from the user equipment, a report comprising a plurality of predicted measurements based on the configuration, wherein the plurality of predicted measurements are obtained during a same prediction time window;means for receiving, from the user equipment, ground truth measurements of the plurality of predicted measurements during the same prediction time window; and means for determining, based on the plurality of predicted measurements and the ground truth measurements, a performance metric for different learning models associated with the at least two measurement reduction ratios.
32. The apparatus according to claim 31 , further comprising:means for determining whether the different learning models pass or fail based on the performance metric.
33. The apparatus according to claims 31 or 32, further comprising:means for transmitting, to the user equipment, another set of measurement reduction ratios at least partly based on a prediction accuracy corresponding to the at least two measurement reduction ratios.
34. A non-transitory computer readable medium comprising program instructions stored thereon for performing the method according to any of claims 12-22.
35. An apparatus comprising circuitry configured to cause the apparatus to perform the method according to any of claims 12-22.
