Receiver and method for life cycle management

Abstract

A method for life cycle management, LCM, of a first algorithm for channel estimation, wherein the method is performed by a receiver (100). The method comprises obtaining (210), as an output from the first algorithm based on a first set of input data, at least one first channel estimate, obtaining (220), as an output of a second algorithm, based on a second set of input data, at least one second channel estimate, determining (230), based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm, and in response to determining that the performance of the first algorithm meets at least one condition associated with an expected performance, performing (240) an LCM action associated to the first algorithm. Further there are a related receiver (100), a computer program (501) and a tangible, non-volatile computer readable medium (502).

Description

[0001] RECEIVER AND METHOD FOR LIFE CYCLE MANAGEMENT

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to a receiver and a method performed by the receiver. A related computer programs and computer readable storage medium are also disclosed.

[0004] BACKGROUND

[0005] Artificial Intelligence (Al) and Machine Learning (ML) have been investigated as promising tools to optimize a design of air-interface in wireless communication networks in both academia and industry.

[0006] A 3rd Generation Partnership Project (3GPP) Technical Report (TR) 38.843. “Study on Al / ML for NR air interface (Release 18)”, v2.0.0, Dec. 2023. refers to augmenting the air-interface with features enabling improved support of AI / ML based algorithms for enhanced performance and / or reduced complexity / overhead. The 3GPP TR 38.843 v2.0.0. refers to use cases such as Channel State Information (CSI) feedback enhancement, beam management and positioning accuracy, and life cycle management (LCM) of an AI / ML model.

[0007] An algorithm being e.g. trained and / or data-driven comprising an AI / ML model may provide performance benefits over classical model-based counterparts e.g. in case of channel estimation for CSI acquisition. In case of using the algorithm for the channel estimation for CSI acquisition, an improved channel estimate resulting in an improved beamforming performance, and / or improved link adaptation and scheduling may be achieved. The improved beamforming performance and link adaptation and scheduling may further result in an improved signal quality and throughput and latency. Further, an improved spectral efficiency and an improved energy efficiency may be achieved by producing a channel estimate of equal quality but using less radio resources.

[0008] Since the algorithm may adapt to a distribution of a training data set without being based on an explicit underlying data model, it is challenging to understand under which conditions the AI / ML model comprised in the algorithm performs well. Due to this property of the algorithm, the AI / ML model may be difficult to analyze and thus may require model monitoring techniques such as LCM. In the scenario of a serving or intended channel and / or interference realization deviating significantly from the training dataset, the property of the algorithm may result in performance degradation of the algorithm. The scenario may arise e.g., if the underlying data-model is trained on at least one first site but is deployed in at least one second site different to the first site, if the underlying data-model is trained on synthetic data, or if the characteristic of a site changes. Changes to the site may e.g. include changes in a mechanical tilt of the antennas, in new hardware installations, in an environment of a cell that the site serves, in a user distribution e.g. if a drone appears at high elevation in the site where users are typically distributed below a base station or if people gather on a square, and if the trained algorithm is deployed for the first time on a site.

[0009] SUMMARY

[0010] An object of the invention is to enable improved determining of channel estimate by determining the performance of a first algorithm for channel estimation.

[0011] A first aspect of the invention relates to a method for life cycle management, LCM, of a first algorithm for channel estimation, wherein the method is performed by a receiver, the method comprising obtaining, as an output from the first algorithm based on a first set of input data, at least one first channel estimate, obtaining, as an output of a second algorithm, based on a second set of input data, at least one second channel estimate, determining, based on the at least one first channel estimate and the at least one second channel estimate (benchmark channel estimate), a performance of the first algorithm, and in response to determining that the performance of the first algorithm meets at least one condition associated with an expected performance, performing an LCM action associated to the first algorithm. Thereby, the receiver is enabled to obtain enable an improved obtaining of channel estimates.

[0012] A second aspect of the invention relates to a receiver, whereby the receiver is configured to obtain, as an output from the first algorithm based on a first set of input data, at least one first channel estimate, obtain, as an output of a second algorithm, based on a second set of input data, at least one second channel estimate, determine, based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm, and in response to a determining that the performance of the first algorithm meets at least one condition associated with an expected performance, perform an LCM action associated to the first algorithm.

[0013] A third aspect of the invention relates to the receiver comprising processing circuitry and a computer readable storage medium, the computer readable storage medium containing instructions executable by the processing circuitry, whereby the receiver is configured to obtain, as an output from the first algorithm based on a first set of input data, at least one first channel estimate, obtain, as an output of a second algorithm, based on a second set of input data, at least one second channel estimate, determine, based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm, and in response to a determining that the performance of the first algorithm meets at least one condition associated with an expected performance, perform an LCM action associated to the first algorithm.

[0014] A fourth aspect of the invention relates to a computer program comprising instructions which, when executed on processing circuitry of a receiver, cause the processing circuitry of the receiver to carry out the method according to the first aspect or any embodiment therein.

[0015] A fifth aspect of the invention relates to a tangible, non-volatile computer readable medium comprising instructions that, when executed on processing circuitry of a receiver, cause the processing circuitry of the receiver to carry out the method according to the first embodiment or any embodiment therein.

[0016] BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic diagram illustrating an example of an environment.

[0018] Figure 2 is a flowchart illustrating a method 200 performed by a receiver. Figure 3A and 3B are flowcharts illustrating a method 300 performed by the receiver.

[0019] Figure 4 illustrates a schematic diagram illustrating first embodiment.

[0020] Figure 5 illustrates a block diagram illustrating embodiments of the receiver.

[0021] DETAILED DESCRIPTION

[0022] Figure 1 shows a schematic diagram illustrating an example of an environment in which embodiments presented herein can be applied. Figure 1 illustrates a receiver 100, a first entity 102, a second entity 104, a third entity 106 and at least one transmitter 110. The receiver 100 and the transmitter 110 is optionally communicatively connected via. external and / or internal entities using wired communications, e.g., based on Ethernet, and / or wireless communications, e.g., wireless fidelity (Wi-Fi), and / or a cellular network corresponding to one or a combination of 5thGeneration (5G) cellular networks, long-term evolution (LTE), LTE-advanced, Universal Mobile Telecommunications System (UMTS), or any other current or future wireless network, such as a future 3rd Generation Partnership Project (3GPP) 6thGeneration (6G) network.

[0023] The receiver 100 may be comprised in a network node such an access point (AP) (e.g., radio access points), base stations (BSs) (e.g., radio base stations, Node Bs, evolved Node Bs (eNBs) and NR NodeBs (gNBs)), O-RAN nodes or components of an O-RAN node (e.g., O-RU, O-DU, O-CU). Alternatively, or additionally the receiver 100 may be comprised in a communication device being any of a wide variety of a User Equipment (UE) including wireless devices arranged, configured, and / or operable to communicate wirelessly with network nodes and other communication devices.

[0024] The at least one transmitter 110 may be comprised in at least one network node or in at least one communication device different to the receiver 100.

[0025] The receiver 100 and the at least one transmitter 110 may be appropriate to provide network communications (e.g., optical, electrical, frequency allocations, and so forth). Moreover, the at least one transmitter 110 and the receiver 100 may be coupled to one or more antennas.

[0026] The first entity 102, the second entity 104, and the third entity 106 may correspond to at least one processing unit capable of executing an algorithm such as a cloud. In an example, the receiver 100 comprises the first entity 102, the second entity 104 and the third entity 106. Optionally the first entity 102 corresponds to the second entity 104 and / or the third entity 106. Optionally, the receiver 100 comprises the first entity 102, the second entity 104 and the third entity 106.

[0027] Figure 2 is a flowchart illustrating a method 200 performed by the receiver 100, as shown in Figure 1 and described in the text thereto.

[0028] The method 200 comprises a first step 210 of obtaining, as an output from the first algorithm based on a first set of input data, at least one first channel estimate. In an obtaining step 220, the receiver 100 obtains, as an output of a second algorithm, based on a second set of input data, at least one second channel estimate. In a determining step 230, the receiver 100 determines, based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm. The method 200 further comprises a step 240 of performing 240 an LCM action associated to the first algorithm in response to determining that the performance of the first algorithm meets at least one condition associated with an expected performance.

[0029] The solution presented herein addresses the above-mentioned challenges of assuring the performance of an algorithm despite the property to adapt to a distribution of a training dataset. The solution presented herein may enable a monitoring of the expected performance of the first algorithm. In an example, the first algorithm is trained based on training datasets associated to communication devices corresponding to e.g. terminals. In the example, the first algorithm is deployed on a side wherein an input data set is associated communication devices corresponding to e.g. drones. This may cause a performance degradation of the first algorithm and thus to more imprecise channel estimation. The solution presented herein may improve the channel estimation despite deployment on a different site. By obtaining 210 the at least one first channel estimate and obtaining 220 the at least one second channel estimate, the receiver 100 is enabled to determine in step 230 the performance of the first algorithm. Performing 240 the LCM action in response to determining that the performance of the first algorithm meets the at least one condition may enable an improved signal-to-noise ratio (SINR) by decreasing a risk of unexpected behavior of the first algorithm due to its underlying model. Alternatively, or additionally, performing 240 in response to determining may enable a reduced risk of performance degradation due to degradation of a model comprised in the first algorithm. Hence, in case the first algorithm is deployed in a communication network performing 240 may enable a better service availability. Alternatively, or additionally, performing 240 in response to determining may allow for simpler site-specific model training due to a decreased risk of a degraded performance of the first algorithm.

[0030] The first algorithm for channel estimation may be a data-driven algorithm such as a trained algorithm. The first algorithm may use an architecture such as a deep learning architecture such as a Convolution Neural Network (CNN) or a Transformer. The first algorithm may be trained based on a dataset composed of at least one input set and a ground truth. During training, the first algorithm may be operative to modify parameters comprised in the first algorithm by minimizing a difference between an output of the first algorithm and a label indicative of a ground truth. The difference between the output and the ground truth may be computed, by the first algorithm, using at least one metric like a Mean Squared Error (MSE). If the metric is used, the parameters may be modified using a gradient descent. In an example, the input set corresponds to at least one Reference Signal, the ground truth corresponds to an expected true CSI and / or the parameters correspond to weights. The first algorithm may be constructed for prediction and / or denoising.

[0031] The at least one first channel estimate may correspond to estimated properties of the communication channel that a signal travels through between at least one transmitter 110 and the receiver 100. Accurate obtaining of the at least one first channel estimate enables the receiver 100 to compensate for variations in channel conditions due to factors like fading, interference, noise and multipath effect. The first channel estimate may be used for determining transmit precoder weights. In an example this includes solving a Minimum mean square error (MMSE) problem, computing a Singular Value Decomposition (SVD), and / or selecting entries from a codebook. Alternatively, or additionally, the first channel estimates may be used for determining a transmission format such as a rank and / or a Modulation and Coding Schemer (MCS), and / or computing a scheduling decision such as which User Element (UE) is to be scheduled. Thus, accurate obtaining of the at least one first channel estimate may enable the receiver 100 to improve decoding of a transmitted signal. In case the first algorithm is constructed for prediction, the at least one first channel estimate may be obtained for prediction. In this case, the at least one first cannel estimate is to be as similar as possible to the channel at a later point in time compared to the point in time the first set of input data is obtained. In the alternative or additional case, the first algorithm being constructed for denoising, the at least one first channel estimate may be obtained for denoising. In this case, the at least one first channel estimate may resemble as closely as possible an actual radio channel at a point in time where the first set of input data was obtained.

[0032] The second algorithm and / or a third algorithm may be a model-based algorithm, a MMSE channel estimator, or a data-driven algorithm such as a trained algorithm different to the first algorithm. The second algorithm and / or the third algorithm may differ from the first algorithm in terms of training data and / or a different model structure. In an example the second algorithm and / or third algorithm is implemented using a linear transform such as a discrete cosine transform, and a tap selection criterion such as an Akaike information criterion, followed by an inverse transform. The second algorithm and / or third algorithm may be for channel estimation. In an example the second algorithm and / or third algorithm is known to be robust. The third algorithm optionally corresponds to a conventional algorithm providing a good channel estimate without suffering from a model drift. Alternatively, the second algorithm optionally corresponds to the third algorithm and / or the first algorithm. The third algorithm optionally corresponds to the second algorithm and / or the first algorithm.

[0033] The at least one second channel estimate may correspond to an estimate approaching a ground truth, ideal and / or expected channel estimate using the second algorithm. The at least one second channel estimate may be estimates properties of the communication channel that a signal travels through between at least one transmitter 110 and the receiver 100. The at least one second channel estimate may pertain to a different domain compared to the at least one first channel estimate. In an example the domain of the at least one second channel estimate may cover a different set of subcarriers, time instances, and / or antennas such as on tx or rx side compared to the at least one first channel estimate. The domain is based on the at least one second set of input data and / or at least one second received signal compared to the at least one first channel estimate.

[0034] At least one third channel estimate may correspond to a benchmark channel estimate based on the at least one first set of input data. The benchmark channel estimate may correspond to an estimate approaching a ground truth, ideal and / or expected channel estimate using the third algorithm. The at least one third channel estimate may be estimated properties of the communication channel that a signal travels through between a transmitter 110 and the receiver 100. Optionally, at least one third channel estimate correspond to a predefined channel threshold. In an example the predefined channel threshold may be preconfigured in the receiver 100.

[0035] The first set of input data and / or the second set of input data may be associated to at least one first received signal and / or at least one second received signal. In an example, the first set of input data and / or the second set of input data corresponds at least one first processed reference signal and / or at least one second processed reference signal. The at least one first and / or second processed reference signal may be based on the at least one first received signal of a first signal transmission and / or based on the at least one second received signal of a second signal transmission such as at least one time domain signal, obtained by the receiver 100. The first set of input data may be used for a purpose of operating a system for channel estimation. The at least one first received signal and / or the at least one second received signal may be at least one reference signal such as a Sounding Reference Signal (SRS), New Radio (NR)-SRS, Physical Uplink Shared Channel (PUSCH) Demodulation Reference Signal (DMRS), Physical Uplink Control Channel (PUCCH), a Random Access Channel (RACH) preamble, channel state information (CSI), NR-CSI, Physical Downlink Shared Channel (PDSCH) DMRS, a reference signal used in a 6G network and / or a reference signal used in Long-Term Evolution (LTE) network.

[0036] The second set of input data may relate at least in part to the first set of input data. The second set of input data may be associated to a part of the at least one first received signal. In an example, the second set of input data is pertaining to a part of a bandwidth of the at least one first received signal and / or of at least one second received signal. The second received signal may be received after the first received signal, e.g. at least one Orthogonal Frequency Division Multiplexing (OFDM) symbol after the first received signal. In an example, in case the first set of input data is pertaining to a full band based on a first full band Sounding Reference Signal (SRS) transmission, the second set of input data is pertaining to a part of the band based on a second SRS transmission occupying a small bandwidth immediately after the first full SRS transmission e.g., after one OFDM symbol. In that example, the first received signal corresponds to the first SRS transmission and the second received signal corresponds to the second SRS transmission. The second received signal may be received at a different point in time as the first received signal. Alternatively, or additionally the second received signal may be obtained based on at least one subcarrier. Optionally the second received signal may be obtained with a higher frequency resolution than the first set of input data (i.e. a reduced comb). Alternatively, or additionally, the second received signal may be obtained with a higher Signal to Noise Ratio (SNR) e.g. due to a reduced interference or a higher transmission power than the first received signal.

[0037] The performance of the first algorithm may be a metric for a similarity between the at least one first channel estimate and the at least one second channel estimate. The metric is for example a cosine similarity, a mean square error, a mean of a norm of a per subcarrier difference between the at least one first channel estimate or the at least one second channel estimate.

[0038] The expected performance may be a metric for a similarity between the at least one second channel estimate and the at least one third channel estimate. The metric is for example a cosine similarity, a mean square error, a mean of a norm of a per subcarrier difference between the at least one third channel estimate or the at least one second channel estimate. Alternatively, the expected performance may correspond to a predefined performance threshold. The predefined performance threshold may be relative to the at least one second channel estimate and may be obtained on a given metric such as Normalized Mean Square Error (NMSE). Optionally, the predefined performance threshold may be preconfigured in the receiver 100.

[0039] Figure 3A and 3B are flowcharts illustrating exemplary embodiments 300 of the method 200 performed by the receiver 100, as shown in Figure 2. Figure 3A and 3B refer to the receiver 100 as shown in Figure 1 and described in the text relating thereto. Figure 3A and 3B are illustrated on two pages for convenience, but they are understood as one single flowchart. The steps of Figure 3A continue where the steps of figure 3A end.

[0040] Referring to Figure 3A, an optional step 301 comprises obtaining 301 , by the receiver 100, the first algorithm. Obtaining 301 , by the receiver 100, may comprise retrieving the first algorithm or receiving the first algorithm.

[0041] An optional step 302 comprises obtaining 302, by the receiver 100, the second algorithm. Obtaining 302, by the receiver 100, may comprise retrieving the second algorithm or receiving the second algorithm.

[0042] An optional step 303 comprises obtaining 303, by the receiver 100, the third algorithm. Obtaining 302, by the receiver 100, may comprise retrieving the third algorithm or receiving the third algorithm.

[0043] The steps of obtaining 301-302 may optionally comprise loading the first, second, and / or third algorithm from a memory.

[0044] An optional step 304 comprises obtaining 304, by the receiver 100, the first set of input data. Obtaining 304, may comprise receiving and / or measuring at least one first received signal by the receiver 100 optionally from the at least one transmitter 110. Obtaining 304 may comprise, performing, by the receiver 100, an OFDM demodulation based on the first received signal and performing, by the receiver 100, matched filtering with a reference signal sequence for obtaining the at least one processed signal based on the first received signal. An optional step 305 comprises obtaining 305, by the receiver 100, the second set of input data. Obtaining may comprise receiving and / or measuring, by the receiver 100 optionally from the at least one transmitter 110, the second received signal. Receiving the second received signal may be at a later point in time than receiving the first received signal, over at least one time instance, over at least one subcarrier, with a higher transmission power, with a reduced interference and / or by receiving the first received signal. Receiving the second received signal at a later point in time than the second received signal may allow for an improved channel estimate. Receiving the second received signal power by receiving the first received signal and e.g. using a part of the first received signal, such as a part of the bandwidth may enable a higher spectral density and thus a higher Signal-to-lnterference-plus-Noise Ratio (SI NR). Receiving and / or collecting the second received signal over multiple time instance by e.g. using multiple adjacent OFDM symbols, may enable an improved processing gain. Receiving the second received signal over more subcarriers by using e.g. comb 2 instead of comb 8, may enable an improved processing gain. Receiving the second received signal with a higher transmission power may comprise mandating, by the receiver 100, the transmitter to increase its transmission power. Receiving the second received signal with a reduced interference may comprise coordinating with neighboring nodes of the receiver 100 for reducing interference of relevant resources. Obtaining 304, may comprise performing, by the receiver 100, an OFDM demodulation based on the second received signal and performing, by the receiver 100, matched filtering with a reference signal sequence for obtaining the at least one processed signal based on the second received signal.

[0045] Alternatively to sequence of steps listed above, the step 301 is performed after the step 303, step 304 and / or step 305. Alternatively, or additionally, the step 302 is performed after the step 304 and / or step 305. Alternatively, or additionally the step 303 is performed after the step 305.

[0046] A step 310 comprises obtaining 310, by the receiver 100, as an output from the first algorithm, based on a first set of input data, at least one first channel estimate. The step 310 may correspond to step 210 of method 200. Obtaining 310 may comprise applying, by the receiver 100, the first algorithm, based on the first set of input data. Applying, by the receiver 100, may comprise executing the first algorithm with the first set of input data as input and obtaining the at least one first channel estimate as an output. In an alternative example, the first algorithm is not executed by the receiver 100, but by the first entity 102, wherein obtaining 310 may comprise receiving the at least one first channel estimate, from the first entity 102. In the alternative example, obtaining 310 may further comprise sending, to the first entity 102, the first set of input data and / or the first algorithm.

[0047] A step 320 comprises, obtaining 320, by the receiver 100, as an output of the second algorithm, based on a second set of input data, at least one second channel estimate. The step 320 may correspond to step 220 of method 200. Obtaining 320 may comprise applying, by the receiver 100, the second algorithm, based on the second set of input data. Applying, by the receiver 100, may comprise executing the second algorithm with the second set of input data as input and obtaining the at least one second channel estimate as the output. In an alternative example, the first algorithm is not executed by the receiver 100, but by the second entity 104, wherein obtaining 310 may comprise receiving the at least one channel estimate, from the second entity 104. In the alternative example, obtaining 310 may further comprise sending, to the second entity 104, the second set of input data and / or the second algorithm.

[0048] A step 330 comprises, determining 330, based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm. The step 330 may correspond to step 230 of method 200. Determining 330, may comprise comparing the at least one first channel estimate to the at least one second channel estimate. Determining 330 may comprise computing the performance on the domain of the at least one second channel estimate. For example, computing may comprise computing a difference between a subset of subcarriers comprised in the at least one second channel estimate, a subset of at least one antennas comprised in the at least one second channel estimate and / or at least one beam comprised in the at least one second channel estimate, and / or a subset of time instances comprised in the at least one second channel estimate to the at least one first channel estimate. Referring to Figure 3B, an optional step 331 comprises, obtaining 331 , based on the first set of input data, the at least one third channel estimate.

[0049] Obtaining 331 the at least one third channel estimate, may comprise a step 331a of obtaining 331a as an output of a third algorithm based on the first set of input, the at least one third channel estimate. Obtaining 331a may comprise applying, by the receiver 100, the third algorithm, based on the first set of input data. Applying, by the receiver 100, may comprise executing the third algorithm with the first set of input data as input and obtaining the at least one third channel estimate as the output. In an alternative example, the third algorithm is not executed by the receiver 100, but by the third entity 106, wherein obtaining 331 a may comprise receiving the at least one channel estimate, from the third entity 106. In the alternative example, obtaining 310 may further comprise sending, to the third entity 106, the first set of input data and / or the third algorithm.

[0050] Alternatively, or additionally, obtaining 331 the at least one third channel estimate, may comprise the step 331 b of obtaining 331 b at least one predefined channel threshold, wherein the at least one third channel estimate correspond to the at least one predefined channel threshold. The at least one predefined channel threshold may be preconfigured in the receiver 100.

[0051] An optional step 332 comprises determining 332, based on the at least one second channel estimate and the at least one third channel estimate, the expected performance. Step 332 may be performed in response to step 331. Determining 332 may comprise computing the performance on the domain of the at least one second channel estimate. For example, computing may comprise computing a difference of the subset of subcarriers comprised in the at least one third channel estimate, the subset of at least one antennas comprised in the at least one third channel estimate and / or at least one beam comprised in the at least one third channel estimate, and / or a subset of time instances comprised in the at least one third channel estimate to the second channel estimate.

[0052] Determining 332 may optionally comprise a step 332a of comparing 332a the at least one third channel estimate to the at least one second channel estimate. Step 332a may be performed in response to step 331a and / or step 331 b. In case, step 332a is performed in response to step 331 b, comparing may comprise that a difference between the at least one third channel estimate to the at least one second channel estimate exceeds, stays below, or meets the at least one predefined channel threshold.

[0053] Alternatively, or additionally determining 332 may optionally comprise a step 332b of obtaining 332b the predefined performance threshold, wherein the expected performance corresponds to the predefined performance threshold.

[0054] An optional step 333 comprises determining 333, based on the performance of the first algorithm and the expected performance, a performance metric, wherein the performance metric is associated with the performance of the first algorithm and the expected performance. Determining may comprise comparing a first difference between the performance of the first algorithm and the expected performance based on the at least one third channel estimate and the at least one second channel estimate with a threshold, wherein the first difference is the performance metric. In the example of the expected performance being the predefined performance threshold, determining may alternatively or additionally comprise comparing a second difference between the performance of the first algorithm and the predefined performance threshold with the threshold, wherein the second difference and optionally the first difference corresponds to the performance metric. The threshold may be a predefined threshold based on e.g. a simulation.

[0055] An optional step 335 comprises determining 335 that the performance of the first algorithm meets at least one condition. Determining 335 may comprise determining a status of a model of the first algorithm.

[0056] Determining 335 may comprise an optional step 335a of determining that the performance of the first algorithm is exceeding, below, or meeting the expected performance, wherein the at least one condition comprises the performance of the first algorithm is exceeding, staying below, or meeting the expected performance.

[0057] Additionally, or alternatively to step 335a, determining 335 may comprise an optional step 335b of determining that the performance metric is exceeding, staying below, or meeting the threshold, wherein the at least one condition comprises the performance metric is exceeding, staying below, or meeting a threshold. The step 335b may be in response to step 333. In an example determining 335b comprises determining that the first difference is exceeding the threshold and / or the second difference is exceeding the threshold.

[0058] Additionally, or alternatively, determining 335 may comprise sending a lookup request to a database, and receiving, from the database, a response to the lookup request. The response may comprise at least one previous performance metric, or information on at least one determining that the condition associated to the first algorithm has been previously met.

[0059] Additionally, or alternatively, determining 335 may comprise storing in the database the performance metric, and / or information that the performance of the first algorithm meets the at least one condition.

[0060] In an example, the step 320 is performed before the step 310 or after the step 310. In an example the optional step 332 is performed before step 330 and / or step 310.

[0061] Optionally, a next step 340 is performed in response to determining that the first algorithm meeting the at least one condition two or more times.

[0062] The step 340 comprises performing the LCM action associated to the first algorithm in response to determining that the performance of the first algorithm meets at least one condition associated with an expected performance. The step 340 may correspond to step 240 of method 200.

[0063] Performing 340 the LCM action comprises a step 340a of disabling 340a a performing of the first algorithm. Disabling 340a may comprise switching from the performing of the first algorithm to a performing of the second algorithm. In an example, the LCM action is adopted per link.

[0064] A performing of the first algorithm and a performing of the second algorithm corresponds to an action of performing the first algorithm such as an execution of the first algorithm and an execution, or a trigger of execution of the first algorithm, and an action of performing the second algorithm such as an execution of the second algorithm and an execution, or a trigger of execution of the second algorithm. Alternatively, or additionally, performing 340 the LCM action comprises a step 340b of enabling 340b the performing of the first algorithm. Enabling 340b may comprise switching from the performing of e.g., the second algorithm to the performing of the first algorithm.

[0065] Alternatively, or additionally, performing 340 the LCM action comprises a step 340c of updating 340c the first algorithm. Updating 340c may comprise training and / or fine tuning the first algorithm based on training data. For example, updating may be performed using transfer learning techniques.

[0066] Alternatively, or additionally, performing 340 the LCM action comprises a step 340d of issuing 340d an indication that the performance of the first algorithm meets at least one condition. Issuing 340d may comprise sending the indication e.g. to a management function in a node, the receiver 100 is e.g. comprised in.

[0067] The step 305, step 320, step 330, optionally step 331 , optionally step 332, optionally step 333, optionally step 335 and the step 340 may be performed in a periodic and / or predetermined manner. In this example, the step 305, step 320, step 330, optionally step 331 , optionally step 332, optionally step 333, optionally step 335 and the step 340 are not performed as often as the steps 304 and 310, which may enable decreased overhead. Alternatively, and / or additionally, the step 305, step 320, step 330, optionally step 331 , optionally step 332, optionally step 333, optionally step 335 and the step 340 are performed in response to determining a trigger such as power angular characteristics differ over a threshold in compared to historical data and / or link performance differs over a threshold in compared to historical data for similar channel situations.

[0068] A Figure 4 illustrates a schematic diagram illustrating a first embodiment of the steps described with reference to Figure 3A and 3B and the method 200, performed by the receiver 100 as shown in Figure 1 and described in the text relating thereto.

[0069] The first embodiment comprises performing the step 304, wherein the first set of input data corresponds to at least one processed reference signal based on the at least one first received signal. The first embodiment further comprises performing the step 301 , wherein the first algorithm is retrieved.

[0070] The first embodiment further comprises performing the step 310, wherein the first algorithm is executed by the receiver 100, using the first set of input data as an input to the first algorithm and wherein the at least one first channel estimate is obtained as an output to executing the first algorithm.

[0071] The first embodiment further comprises performing the step 305, wherein the second set of input data is associated to a part of at least one first received signal, and wherein the first set of input data is associated to the first received signal and wherein the part of the at least one processed reference signal is based on a part of a bandwidth of the at least one first received signal.

[0072] The first embodiment further comprises performing the step 302, wherein the second algorithm is retrieved.

[0073] The first embodiment further comprises performing the step 320, wherein the second algorithm is executed by the receiver 100, using the second set of input data as an input to the second algorithm, and wherein at least one second channel estimate is obtained as an output to executing the second algorithm.

[0074] The first embodiment further comprises, optionally, performing the step 303, wherein the third algorithm may correspond to the second algorithm. Optionally, step 302 is performed.

[0075] The first embodiment further comprises performing the step 331a, wherein the second or third algorithm is executed by the receiver 100, using the first set if input data as an input to the second or third algorithm, and wherein at least one third channel estimate is obtained as an output to executing the second or third algorithm.

[0076] The first embodiment further comprises performing the step 330 of obtaining the performance of the first algorithm.

[0077] The first embodiment further comprises performing the step 332a of determining the expected performance based on the at least one third channel estimate and the at least one second channel estimate. The first embodiment further comprises performing the step 335a of determining that the performance of the first algorithm meets the at least one condition, wherein the performance of the first algorithm is below the expected performance. In an example, the step 335a comprises determining that the first algorithm has an NMSE with a delta Decibel (dB) smaller than the second algorithm. In the example, determining that the first algorithm has an NMSE with the delta dB smaller than the second algorithm comprises detecting that the first algorithm is performing better, wherein the delta is a positive real number.

[0078] The first embodiment further comprises, performing the step 340c of updating the first algorithm, wherein updating comprises using transfer learning techniques.

[0079] Referring to Figure 3A and 3B, a second embodiment comprises performing steps 304, and 305, wherein the second set of input data is associated to the at least one second received signal, wherein the first set of input data is associated to at least one first received signal. The second received signal is received with a higher frequency solution than the first received signal and / or with a higher SNR than the first received signal.

[0080] The second embodiment further comprises performing steps 310 and 320, wherein the receiver 100 sends the at least one first and at least one second input data to the first entity 102. The first entity 102 receives the at least one first and at least one second input data from the receiver 100. The first entity 102 executes the first algorithm based on the first set of input data and the second algorithm based on the second set of input data and obtains the at least one first channel estimate as an output of the executing of the first algorithm and the at least one second channel estimate as an output of the executing of the second algorithm. The first entity 102 sends to the receiver 100 the at least one first channel estimate and the at least one second channel estimate. The receiver 100 receives, from the first entity 102, the at least one first channel estimate and the at least one second channel estimate.

[0081] The second embodiment further comprises performing steps 330 and 332b, wherein step 332b comprises obtaining the predefined performance threshold, wherein obtaining comprises retrieving the predefined performance threshold being preconfigured in the receiver 100. The second embodiment further comprises performing step 335a of determining that the performance of the first algorithm meets the at least one condition, wherein the at least one condition comprises that the performance of the first algorithm is meeting or is above the predefined performance threshold. In an example the performance of the first algorithm corresponds to an NMSE higher or equal to the predefined performance threshold.

[0082] The second embodiment further comprises performing step 340b in response to step 335a.

[0083] Referring to Figure 3A and 3B, a third embodiment comprises performing steps 304, 305, and 301 to 303, wherein the first, second and third algorithm differ from each other and wherein the second set of input data is based on a second reference signal transmitted with a higher SNR compared to the first reference signal. In the third embodiment, step 305 is performed in response to determining the trigger.

[0084] Further, the third embodiment comprises performing step 310, 320, 330 and 331 b, wherein the predefined channel threshold is obtained in step 331 b. Obtaining 331 b comprises retrieving the predefined channel threshold which is preconfigured in the receiver 100.

[0085] Further, the third embodiment comprised performing step 332 and 332a of determining the expected performance by comparing the at least one predefined channel threshold with the at least one second channel estimate.

[0086] Further, the third embodiment comprises performing step 333 of determining the performance metric. Determining comprises comparing a first difference between the performance of the first algorithm and the expected performance with a threshold, wherein the first difference is the performance metric.

[0087] Further, the third embodiment comprises performing step 335 and 335b of determining that the performing of the first algorithm meets at least one condition, wherein the condition comprises the performance metric exceeding or meeting the threshold.

[0088] In the third embodiment step 340a 340d are performed in response to step 335b. Figure 5 illustrates a block diagram illustrating embodiments of the receiver 100 in further detail. In practice, the steps 210 to 240 of the method 200 performed by the receiver 100 are performed by processing circuitry 504, embodied in one or more processors and / or microprocessors arranged to execute a computer program 501 that is downloaded to a computer program product 505, here in the form of a suitable computer readable storage medium 502 associated with the microprocessor. The computer readable storage medium 502 may be a memory, such as a Random Access Memory (RAM) or a Read-Only Memory (ROM), or a tangible non-volatile computer readable storage medium, such as flash memory or a hard disk drive, or any combination thereof. The computer program 501 comprises computerexecutable instructions stored or downloaded to the computer readable storage medium 502 and are executable by the processing circuitry 504. Alternatively, the computer program 501 may be transferred to the computer readable storage medium 502 using a suitable computer program product, such as a memory stick or in a memory of a device. Thus, the computer program 501 may be stored in any suitable manner in the computer program product. The processing circuity 504 is arranged to cause the receiver 100 to carry out the steps 210 to 240 of method 200 in accordance with any of the of the described embodiments for steps 210 to 240. The processing circuitry 504 is in one embodiment one or more general-purpose processors wherein each one of the general purpose processors includes one or more cores, but may alternatively be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a Complex Programmable Logic Device (CPLD), etc. An I / O interface 503 is provided for communicating with external and / or internal entities using wired communications, e.g., based on Ethernet, and / or wireless communications, e.g., WiFi, and / or a cellular network corresponding to one or a combination of 5G cellular networks, LTE, LTE-advanced, UMTS, or any other current or future wireless network, such as a future 3GPP 6G network, as long as the principles described below are applicable.

Claims

CLAIMS:1 . A method (200, 300) for life cycle management, LCM, of a first algorithm for channel estimation, wherein the method is performed by a receiver (100), the method comprising: obtaining (210), as an output from the first algorithm based on a first set of input data, at least one first channel estimate; obtaining (220), as an output of a second algorithm, based on a second set of input data, at least one second channel estimate; determining (230), based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm; and in response to determining that the performance of the first algorithm meets at least one condition associated with an expected performance, performing (240) an LCM action associated to the first algorithm.

2. The method (200, 300) according to claim 1 , wherein the at least one condition comprises at least one of: the performance of the first algorithm is exceeding, staying below, or meeting the expected performance; and a performance metric is exceeding, staying below, or meeting a threshold, wherein the performance metric is associated with the performance of the first algorithm and the expected performance.

3. The method (200, 300) according to claim 2, wherein the method comprises: if the at least one criterion comprises at least the performance metric, determining (333), based on the performance of the first algorithm and the expected performance, the performance metric; andif the at least one criterion comprises at least the second metric, determining (334), based on the performance of the first algorithm, the excepted performance and the previous determined performance, the second metric.

4. The method (200, 300) according to any of the previous claims, wherein performing (240) the LCM action comprises at least one of: disabling (340a) a performing of the first algorithm; enabling (340b) a performing of the first algorithm; updating (340c) the first algorithm; and / or issuing (340d) an indication that the performance of the first algorithm meets at least one condition.

5. The method (200, 300) according to any of the previous claims, wherein the method comprises determining (332), based on the at least one second channel estimate and at least one third channel estimate, the expected performance.

6. The method (200, 300) according to claim 5, wherein the method comprises obtaining (331), based on the first set of input data, the at least one third channel estimate.

7. The method (200, 300) according to claim 6, obtaining (331), based on the first set of input data, the at least one third channel estimate: obtaining (331a), as an output of a third algorithm based on the first set of input data, the at least one third channel estimate; and / orobtaining (331 b) at least one predefined channel threshold, wherein the at least one third channel estimate correspond to the at least one predefined channel threshold.

8. The method (200, 300) according to claim 5, wherein determining (332) the expected performance comprises determining (332b) a predefined performance threshold, wherein the expected performance corresponds to a predefined performance threshold.

9. The method (200, 300) according to any of the previous claims, wherein the second set of input data is related at least in part to the first set of input data.

10. The method (200, 300) according to any of the previous claims, wherein the method comprises at least one of: obtaining (301) the first algorithm; obtaining (302) the second algorithm; obtaining (303) a third algorithm; obtaining (304) the first set of input data; and / or obtaining (305) the second set of input data.

11. The method (200, 300) according to claim 10, wherein the second algorithm corresponds to the third algorithm and / or the first algorithm.

12. A receiver (100), whereby the receiver (100) is configured to:obtain, as an output from the first algorithm based on a first set of input data, at least one first channel estimate; obtain, as an output of a second algorithm, based on a second set of input data, at least one second channel estimate; determine, based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm; and in response to a determining that the performance of the first algorithm meets at least one condition associated with an expected performance, perform an LCM action associated to the first algorithm.

13. The receiver (100), according to claim 12, configured to perform the method according to any of claims 2 to 11 .

14. A receiver (100), comprising processing circuitry (504) and a computer readable storage medium (502), the computer readable storage medium (502) containing instructions executable by the processing circuitry (504), whereby the receiver (100) is configured to: obtain, as an output from the first algorithm based on a first set of input data, at least one first channel estimate; obtain, as an output of a second algorithm, based on a second set of input data, at least one second channel estimate; determine, based on the at least one first channel estimate and the at least one second channel estimate, a performance of the first algorithm; and in response to a determining that the performance of the first algorithm meets at least one condition associated with an expected performance, perform an LCM action associated to the first algorithm.

15. The receiver (100), according to claim 14, configured to perform the method according to any of claims 2 to 11 .

16. A computer program (501) comprising instructions which, when executed on processing circuitry (504) of a receiver (100), cause the processing circuitry (504) of the receiver (100) to carry out the method according to any one of claims 1 to 11 .

17. A tangible, non-volatile computer readable medium (502) comprising instructions that, when executed on processing circuitry (504) of a receiver (100), cause the processing circuitry (504) of the receiver (100) to carry out the method according to any one of claims 1 to 11 .

Images