Method performed by electronic device, electronic device, storage medium, and program product

Abstract

Embodiments of the present disclosure provide a method performed by an electronic device, an electronic device, a storage medium and a program product, relating to the field of artificial intelligence. The method comprises: obtaining current transmission environment parameter information; obtaining current transmission state information based on the current transmission environment parameter information, historical transmission environment parameter information and historical transmission state information; wherein the transmission environment parameter information comprises at least one of the following: cell configuration information, radio measurement information and service type information. The accuracy of predicting transmission state information can be improved. Optionally, the above-mentioned method performed by the electronic device can be performed using an artificial intelligence model.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and more specifically, to a method executed by an electronic device, an electronic device, a storage medium, and a program product. Background Technology

[0002] Accurate prediction of transmission status can improve the user experience quality of network-aware services, but the prediction of transmission status in existing technical solutions is not accurate. Summary of the Invention

[0003] This disclosure provides a method executed by an electronic device, an electronic device, a storage medium, and a program product, aimed at improving the accuracy of predicting transmission status. The technical solution is as follows: According to one aspect of the present disclosure, a method performed by an electronic device is provided, the method comprising: Obtain the current transmission environment parameter information; Based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, the current transmission status information is obtained. The transmission environment parameter information includes at least one of the following: cell configuration information, wireless measurement information, and service type information.

[0004] In one feasible embodiment, obtaining the current transmission status information based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information may include: Based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, the first transmission status information is obtained. The adjustment amount is obtained based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information; Based on the first transmission status information and the adjustment amount, the current transmission status information is obtained.

[0005] In a feasible embodiment, the first transmission status information obtained based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information includes: Use artificial intelligence (AI) to select the first AI network that matches the current transmission environment parameter information; Using the first AI network, the first transmission status information is obtained based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information.

[0006] In one feasible embodiment, the above-mentioned use of artificial intelligence (AI) to select networks and obtain a first AI network that matches the current transmission environment parameter information includes: using AI to select networks, determining at least two first AI networks and at least two weights based on the current transmission environment parameter information, wherein the at least two first AI networks correspond one-to-one with the at least two weights.

[0007] The above-mentioned use of the first AI network to obtain the first transmission status information based on the current transmission environment parameter information, historical transmission environment parameter information and historical transmission status information includes: using at least two first AI networks to obtain at least two intermediate transmission status information based on the current transmission environment parameter information, historical transmission environment parameter information and historical transmission status information; and obtaining the first transmission status information based on the at least two intermediate transmission status information and at least two weights.

[0008] In one feasible embodiment, the adjustment amount obtained based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information may include: Based on the current transmission environment parameter information and parameter interval set, the first parameter interval is determined; the first parameter interval is the interval in the parameter interval set to which the transmission environment parameter in the current transmission environment parameter information belongs. Based on the correspondence between the parameter intervals in the parameter interval set and the transmission status information set, the transmission status information set corresponding to the first parameter interval is determined. The adjustment amount is obtained based on the transmission status information set corresponding to the first parameter interval and the transmission status information set corresponding to the second parameter interval; wherein, the second parameter interval is the interval in the parameter interval set to which the transmission environment parameter corresponding to the first time before the current time belongs.

[0009] In a feasible embodiment, determining the first parameter range based on the current transmission environment parameter information and parameter range set may include: Map the current transmission environment parameter information into a high-dimensional vector; The first parameter interval is obtained based on the high-dimensional vector and the set of parameter intervals.

[0010] In a feasible embodiment, the method provided by this disclosure may further include: obtaining a parameter interval set based on the transmission status information sets corresponding to adjacent preset intervals.

[0011] In a feasible embodiment, the parameter interval set obtained based on the transmission status information sets corresponding to adjacent preset intervals may include: The first residual is obtained based on the historical transmission status information set of the first preset interval and the predicted transmission status information set of the first preset interval. The second residual is obtained based on the historical transmission status information set of the second preset interval adjacent to the first preset interval and the predicted transmission status information set of the second preset interval. Whether to merge the first preset interval and the second preset interval is determined based on the first residual and the second residual.

[0012] According to another aspect of the present disclosure, an electronic device is provided, the electronic device including a memory, a processor and a computer program stored in the memory, the processor executing the computer program to implement the steps of the method performed by the electronic device described above.

[0013] According to another aspect of the present disclosure, a computer-readable storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the steps of the method performed by the electronic device described above.

[0014] According to another aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method performed by the electronic device described above.

[0015] The beneficial effects of the technical solutions provided in this disclosure are: This disclosure provides a method, electronic device, storage medium, and program product executed by an electronic device, which can improve the accuracy of prediction by combining transmission environment parameter information to predict transmission status information. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments of this disclosure will be briefly introduced below.

[0017] Figure 1 A flowchart illustrating a method performed by an electronic device according to an embodiment of this disclosure; Figures 1a-1e A schematic diagram of the architecture of some AI networks provided for embodiments of this disclosure; Figure 2 A schematic diagram of an architecture provided for an embodiment of this disclosure; Figure 3 Another architectural schematic diagram provided for an embodiment of this disclosure; Figure 4 A schematic diagram of the architecture of an Ant-Net network provided in an embodiment of this disclosure; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0018] The following description, with reference to the accompanying drawings, is provided to aid in a thorough understanding of the various embodiments of this disclosure as defined by the claims and their equivalents. This description includes various specific details to aid understanding but should be considered exemplary only. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the various embodiments described herein without departing from the scope and spirit of this disclosure. Furthermore, for clarity and brevity, descriptions of well-known functions and structures may be omitted.

[0019] The terms and wording used in the following description and claims are not limited to their dictionary meanings, but are merely used by the inventors to enable a clear and consistent understanding of this disclosure. Therefore, it will be apparent to those skilled in the art that the following description of various embodiments of this disclosure is for illustrative purposes only and not for limiting the purpose of this disclosure as defined in the appended claims and their equivalents.

[0020] It should be understood that the singular forms of “a,” “an,” and “the” can also include plural references unless the context clearly indicates otherwise. Thus, for example, the reference to “component surface” includes referring to one or more such surfaces. When we say that an element is “connected” or “coupled” to another element, the element can be directly connected or coupled to the other element, or it can mean that the element and the other element are connected through an intermediate element. Furthermore, the use of “connected” or “coupled” herein can include wireless connections or wireless couplings.

[0021] The terms “comprising” or “may include” refer to the presence of a corresponding disclosed function, operation, or component that may be used in the various embodiments of this disclosure, rather than limiting the presence of one or more additional functions, operations, or features. Furthermore, the terms “comprising” or “having” may be interpreted as indicating certain characteristics, numbers, steps, operations, constituent elements, components, or combinations thereof, but should not be construed as excluding the possibility of the presence of one or more other characteristics, numbers, steps, operations, constituent elements, components, or combinations thereof.

[0022] The term "or" as used in the various embodiments of this disclosure includes any of the listed terms and all combinations thereof. For example, "A or B" may include A, may include B, or may include both A and B. When describing multiple (two or more) items, if the relationship between the multiple items is not explicitly defined, the multiple items may refer to one, more, or all of the multiple items. For example, the description "parameter A includes A1, A2, A3" can be implemented as parameter A includes A1 or A2 or A3, or it can be implemented as parameter A includes at least two of the three items A1, A2, and A3.

[0023] Unless otherwise defined, all terms used in this disclosure (including technical or scientific terms) have the same meaning as understood by one of those skilled in the art as described herein. Common terms as defined in dictionaries are to be interpreted as having a meaning consistent with the context in the relevant technical field and should not be interpreted ideally or overly formally unless expressly defined in this disclosure.

[0024] At least some of the functions of the device or electronic device provided in this disclosure embodiment can be implemented by an AI model (also referred to as an AI network), such as by implementing at least one module of a plurality of modules of the device or electronic device by an AI model. AI-related functions can be executed by non-volatile memory, volatile memory, and a processor.

[0025] The processor may include one or more processors. In this case, the one or more processors may be general-purpose processors, such as central processing unit (CPU), application processor (AP), etc., or pure graphics processing unit, such as graphics processing unit (GPU), visual processing unit (VPU), and / or AI-specific processors, such as neural processing unit (NPU).

[0026] The one or more processors control the processing of input data based on predefined operating rules or artificial intelligence (AI) models stored in non-volatile and volatile memory. These predefined operating rules or AI models are provided through training or learning.

[0027] Here, "providing through learning" refers to obtaining predefined operating rules or an AI model with desired characteristics by applying a learning algorithm to multiple learning datasets. This learning can be performed within the device or electronic device itself, in which the AI ​​is executed according to the embodiment, and / or can be implemented via a separate server / system.

[0028] AI models can contain multiple neural network layers. Each layer has multiple weight values, and each layer performs neural network computations by calculating the input data of that layer (such as the computation results of the previous layer and / or the input data of the AI ​​model) and the multiple weight values ​​of the current layer. Examples of neural networks include, but are not limited to, convolutional neural networks (CNNs), deep neural networks (DNNs), recurrent neural networks (RNNs), restricted Boltzmann machines (RBMs), deep belief networks (DBNs), bidirectional recurrent deep neural networks (BRDNNs), generative adversarial networks (GANs), and deep Q-networks.

[0029] A learning algorithm is a method of training a predetermined target device (e.g., a robot) using multiple learning data sets to enable, allow, or control the target device to make determinations or predictions. Examples of such learning algorithms include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0030] The methods provided in this disclosure may relate to one or more fields in the technical fields of speech, language, image, video, or data intelligence.

[0031] Optionally, in the context of data intelligence processing, according to this disclosure, in methods executed in electronic devices, the method for reasoning or predicting transmission state information can use an artificial intelligence model (also referred to as an artificial intelligence network) to obtain transmission state information by using transmission environment parameter information as input data to the artificial intelligence model. The processor of the electronic device can perform preprocessing operations on the data to transform it into a form suitable for use as input to the artificial intelligence model. The artificial intelligence model can be obtained through training. Here, "obtained through training" means obtaining a predefined operating rule or artificial intelligence model configured to perform desired features (or purposes) by training a basic artificial intelligence model with multiple training data using a training algorithm. Reasoning and prediction are techniques for logical reasoning and prediction based on determined information, including, for example, knowledge-based reasoning, optimization prediction, preference-based planning, or recommendation.

[0032] The following description of several optional embodiments illustrates the technical solutions of this disclosure and the technical effects produced by these solutions. It should be noted that the following embodiments can be referenced, learned from, or combined with each other. Identical terms, similar features, and similar implementation steps in different embodiments will not be repeated.

[0033] Figure 1 This illustration shows a flowchart of a method executed by an electronic device according to an embodiment of this disclosure. The method can be executed by a terminal or a server. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, in-vehicle device, etc.; the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms, but is not limited to these. Specific examples will be used in subsequent embodiments to illustrate the specific content executed by each device.

[0034] The following describes a method performed by an electronic device according to an embodiment of this disclosure.

[0035] Specifically, such as Figure 1 As shown, the method performed by an electronic device according to the embodiments of this disclosure includes the following steps: S101 and S103.

[0036] Step S101: Obtain the current transmission environment parameter information.

[0037] For example, the transmission environment parameter information may include, but is not limited to, at least one of the following: cell configuration information, radio measurement information, and service type information. In this disclosure, the content of current and historical transmission environment parameter information can be referred to the description of this transmission environment parameter information, and will not be repeated here.

[0038] For example, cell configuration information may include, but is not limited to, at least one of the following: cell bandwidth, number of carrier units, and center frequency information.

[0039] For example, wireless measurement information may include, but is not limited to, at least one of the following: Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Received Signal Strength Indicator (RSSI), Signal to Interference plus Noise Ratio (SINR), and Signal-to-Noise Ratio (SNR).

[0040] For example, service types may include, but are not limited to, at least one of the following: web browsing, video playback, file download, and online gaming. Service types can be obtained by the access point (AP), for example, through the applications and services currently running on the UE. Service types can also be numerical vectors obtained through encoding, such as when the AP encodes the obtained service types to obtain a numerical vector.

[0041] Optionally, the transmitted environmental parameter information may also include, but is not limited to, at least one of the following: wireless status information and wireless connection events.

[0042] For example, radio condition information may include, but is not limited to, at least one of the following: scheduling delay, number of resource blocks (RBs), block error rate (BLER), modulation and coding scheme (MCS) level, and number of bits transmitted on each RB.

[0043] For example, throughput is determined by the resources allocated in the time and frequency domains, as well as spectral efficiency. Time-domain resources (such as scheduling delay) and frequency-domain resources (such as the number of resource blocks) are scheduling-related variables. Spectral efficiency refers to the rate at which information can be transmitted over a given bandwidth, which is closely related to channel conditions. Scheduling uncertainty, the diversity of wireless channels, and their combinations are among the main factors affecting throughput variation and exhibit dynamic changes.

[0044] For example, wireless connectivity events may include, but are not limited to, at least one of the following: handover (HO) and carrier aggregation (CA) configuration changes. Handover and carrier aggregation configuration changes can significantly affect throughput fluctuations. Wireless connectivity events are discrete and sparse in time, and cannot be directly used as input for throughput prediction.

[0045] Step S102: Obtain the current transmission status information based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information.

[0046] For example, transmission status information may include, but is not limited to, at least one of the following: bandwidth, latency, packet loss rate, and throughput. In this disclosure, the content of current transmission status information and historical transmission status information can be referred to the description of transmission status information, and will not be repeated here.

[0047] For example, historical transmission environment parameter information refers to the transmission environment parameter information over a period of time prior to the current moment. Historical transmission status information is the transmission status information corresponding to the historical transmission environment parameter information; it is the transmission status information obtained under the historical transmission environment parameter information.

[0048] In the method provided in this disclosure, combining transmission environment parameter information to predict transmission status information can improve the accuracy of prediction.

[0049] In some embodiments, historical transmission environment parameter information that is similar to or consistent with the current transmission environment parameter information can be searched, and the transmission status information under the similar (or consistent) historical transmission environment parameter information can be directly used as the current transmission status information.

[0050] Alternatively, in some embodiments, step S102 above, which obtains the current transmission status information based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, may include steps A1-A3.

[0051] Step A1: Based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, obtain the first transmission status information.

[0052] For example, find historical transmission environment parameter information that is similar to or consistent with the current transmission environment parameter information, and use the transmission status information under the similar (or consistent) historical transmission environment parameter information as the first transmission status information.

[0053] Step A2: Based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, obtain the adjustment amount.

[0054] For example, the adjustment amount can be obtained in a variety of ways. For instance, historical transmission environment parameter information that is similar to or consistent with the current transmission environment parameter information can be found, and the adjustment amount can be predicted based on the distribution difference between the transmission state information under the similar (or consistent) historical transmission environment parameter information and the transmission state information of the previous moment (or a period of time before the current moment, or a recent period of time). This adjustment amount reflects the change in transmission state caused by the change in transmission environment.

[0055] Step A2 can be an optional step, for example, if the adjustment amount corresponding to the current transmission environment parameter information has been calculated beforehand.

[0056] Step A3: Based on the first transmission status information and the adjustment amount, obtain the current transmission status information.

[0057] For example, by adjusting the first transmission status information by adjusting the amount, the current transmission status information can be obtained, which can further improve the prediction accuracy.

[0058] When significant changes occur in the transmission environment, more accurate transmission status information predictions can be obtained by adjusting the adjustment amount. This method involves identifying transmission statuses under historical transmission environments similar to the current one, and then predicting future transmission status information adjustments based on these similar historical transmission statuses, fully considering possible changes in transmission status under similar environments. Combining the initial transmission status information (first transmission status information) with the adjustment amount further improves prediction accuracy.

[0059] In some embodiments, step A1 above, based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, obtains the first transmission status information, which may include steps A11-A12.

[0060] Step A11: Use AI to select a network and obtain the first AI network that matches the current transmission environment parameter information.

[0061] For example, the AI ​​selection network can also be called a model selector or a predictive model selector, etc., used to select a model for predicting transmission state information based on the current transmission environment parameter information (which can be called the first AI network or the predictive AI network).

[0062] The first AI network matched with the current transmission environment parameters can refer to an AI network suitable for predicting the transmission status based on the current transmission environment parameters. For example, different predictive AI networks may have different advantages. Taking the service type in the transmission environment parameters as an example, some predictive AI networks are more suitable for predicting the transmission status of video streaming services, while others are more suitable for predicting the transmission status of file download services. Selecting a matching predictive AI network based on the service type of the current transmission environment can improve the accuracy of the prediction. Thus, selecting a more suitable AI network for predicting the transmission status under the actual conditions of the transmission environment can improve the accuracy of the prediction.

[0063] Based on the characteristics of the wireless transmission environment, input data can be processed in a targeted manner to improve the prediction accuracy of AI models.

[0064] The features obtained by the user equipment may include, but are not limited to, one or more of the following: throughput, percentage of time-domain resource allocation, number of redundancy blocks (RBs), number of MIMO layers, MCS, RSRP, SINR, and BLER. For example, the time series of feature i can be represented as... t represents the current time index. This refers to throughput characteristics.

[0065] Feature selection: Based on the local Pearson correlation coefficient, the correlation between features is calculated. Factors affecting transmission status are time-varying and dynamic. By using local segmented calculations to address the time-varying dynamic characteristics of these factors, key relevant features can be avoided. and For example, and Local similarity can be represented as: Where M represents the total number of local subsequences, m represents the index of the m-th subsequence, and cov() represents the covariance function. express The standard deviation of the m-th subsequence express The standard deviation of the m-th subsequence. The length of the local subsequence is related to the observation window length of the AI ​​model.

[0066] Local normalization: Applying global normalization to input data can lead to performance degradation due to the non-stationarity of the throughput time series. Applying local normalization to input data can solve the problem of throughput distribution drift. Let... For the i-th feature within the observation window T, its locally normalized throughput can be expressed as: .in, and The parameters are learnable, and the normalized input vector is denoted as . , express The local mean within the observation window T express The local standard deviation within the observation window T, where This represents the number of input features. AI networks can be designed specifically for different transmission environments. Take the Transformer as an example.

[0067] Combination Figure 1a We designed a weighted embedding network. The normalized input data was mapped into a high-dimensional vector through the AI ​​network. or This depends on the design of the expert model. D is the embedding dimension, N is the total number of input features, and T is the observation window length. A learnable parameter vector P is used. To record location information. The embedded vectors are clustered according to the correlation between features, and the same class shares the same attention module value and key.

[0068] Combination Figure 1b A wireless resource awareness model is designed. An attention dimension is applied to the feature dimension, and a feedforward layer is used to extract temporally varying features. The model input data can be represented as follows: , where m represents the D-dimensional embedding vector contained within the observation window T, and m represents the m-th encoder layer, with D being the embedding dimension. The model input data can be represented as: Where m represents the m-th layer encoder. The throughput prediction process can be expressed as: , where τ represents the prediction window length.

[0069] Combination Figure 1c A handover awareness model is designed. Since handover events cause significant changes in the transmission state, they introduce some non-stationarity. However, these non-stationarities exhibit similar temporal patterns. Therefore, preserving these non-stationarities of handover events can improve the accuracy of transmission state prediction during handover events. Historical handover patterns are analyzed. and their statistics and Two factors can be obtained through an AI model. and . , The corresponding multi-head attention was changed to .in, This is the query matrix for the attention module. Let V be the transpose of the key matrix, V be the value matrix, and D be the embedding dimension. It represents a vector consisting entirely of 1s.

[0070] Combination Figure 1d A carrier aggregation sensing model was designed. The input data within the observation window T was categorized into several types: Type 1, where only the primary cell (Pcell) was activated; Type 2, where some secondary cells (Scells) existed within the observation window; and Type 3, where both primary and secondary cells existed within the observation window. For Type 1 data, an attention module suitable for its bandwidth was used. For Type 2 data, a default attention structure was used. For Type 3 data, different attention modules were applied to different component carriers (CCs), and their data were aggregated.

[0071] For example, suppose E This is a collection of wireless connection events. Considering... E Due to the sparsity of the event, positional encoding can be used to mark the location of the event, and this encoding can be represented as a binary mask matrix, denoted as . Where N is the total number of AI networks, Let T represent the set of real numbers, and T be the length of the observation window. This mask matrix is ​​used to activate a suitable AI network from the AI ​​network library.

[0072] For example, suppose the vector E of connection events is used to select the appropriate sub-model. Figure 1e As shown, the event vector E is first reduced in dimensionality using average pooling. This latent vector is then transformed by the AI ​​encoding module into... A vector of dimension. The weight coefficients correspond to the weights of each sub-model. The method for selecting the expert model can be expressed by the formula: .

[0073]

[0074] in, Represents the mask vector. This represents the total number of expert models.

[0075] Step A12: Using the first AI network, based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, obtain the first transmission status information.

[0076] For example, using a first AI network adapted to the current transmission environment parameter information, the first transmission state information can be predicted based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission state information, which can improve the accuracy of the prediction.

[0077] In some embodiments, step A11 above uses AI to select a network and obtain a first AI network that matches the current transmission environment parameter information, which may include step A111.

[0078] Step A111: Use AI to select networks and determine at least two first AI networks and at least two weights based on the current transmission environment parameter information.

[0079] For example, at least two first AI networks correspond one-to-one with at least two weights.

[0080] For example, when using AI to select networks, the number of prediction networks selected based on the current transmission environment parameter information can be one or more. For example, N first AI networks and N weights can be determined, where N is an integer greater than or equal to 1. The value of N can be preset or selected by the AI ​​network selection based on the actual transmission environment parameters.

[0081] For example, there can be one or more first AI networks that are applicable to the current transmission environment parameter information. The AI ​​selects N first AI networks and the weights corresponding to each first AI network based on the current transmission environment parameter information.

[0082] Optionally, the number of AI selection networks can be one or more, and one or more AI selection networks can be used to select the first applicable AI network.

[0083] In some embodiments, step A12 above uses a first AI network to obtain first transmission status information based on current transmission environment parameter information and historical transmission status information, and may include steps A121-A122.

[0084] Step A121: Using at least two first AI networks, based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, obtain at least two intermediate transmission status information.

[0085] For example, at least two selected first AI networks are used to predict the transmission status information corresponding to the current transmission environment parameter information, respectively, to obtain at least two intermediate transmission status information. The at least two intermediate transmission status information can correspond one-to-one with at least two first AI networks.

[0086] For example, the transmission environment parameter information may include one or more parameters. For each transmission environment parameter in the current transmission environment parameter information, at least two first AI networks are used to predict the transmission state and obtain at least two intermediate transmission state information.

[0087] Step A122: Obtain the first transmission state information based on at least two intermediate transmission state information and at least two weights.

[0088] For example, by weighted fusion of at least two intermediate transmission state information and corresponding at least two weights to obtain first transmission state information, and then using multiple predictive AI networks for prediction, the prediction accuracy can be further improved.

[0089] Combination Figure 2 For example, consider the number of AI selection networks and the number of prediction AI networks.

[0090] like Figure 2 As shown, the AI ​​selection network includes AI selection network 1, AI selection network 2, and AI selection network 3. Their main function is to select the prediction AI network based on the current transmission environment parameters. Different AI selection networks can correspond to different transmission environment parameters. The prediction AI network includes prediction AI network 1 to prediction AI network N.

[0091] The current transmission environment parameters are input into the corresponding AI selection network to obtain the weight vector of the prediction AI network. Combined with... Figure 2 Cell configuration information is input into AI selection network 1 to obtain weight vector 1 for predicted AI networks 1 to N. Wireless measurement information is input into AI selection network 2 to obtain weight vector 2 for predicted AI networks 1 to N. Service type information is input into AI selection network 3 to obtain weight vector 3 for predicted AI networks 1 to N. By superimposing weight vector 1, weight vector 2, and weight vector 3, a set of weight vectors is obtained. This weight vector is a one-dimensional vector with length N, and each index... The corresponding values ​​are the weight coefficients of the prediction AI network for the subsequent corresponding numbers. .

[0092] Historical transmission environment parameters, historical transmission status information (such as historical throughput sequences), and current transmission environment parameters are input into the prediction AI network. For each historical transmission environment parameter, its correlation with historical transmission state information can be calculated. This yields a correlation matrix between all historical transmission environment parameters and historical transmission state information, denoted as... And also input it into the predictive AI network. Each predictive AI network can obtain the predicted throughput, denoted as . Where t represents the current time, Indicates the prediction window length.

[0093] For each predictive AI network, multiplying the prediction result (such as intermediate transmission state information) by the corresponding weight coefficient and then performing weighted fusion yields the predicted first transmission state information, such as the initial throughput sequence, denoted as . ,in, Indicates the current moment. Indicates the prediction window length. Let t represent the weight corresponding to AI network i in the prediction, and t represent the historical time.

[0094] Therefore, selecting a more suitable predictive AI network based on the actual transmission environment can improve the accuracy of predictions. At the same time, considering the combined results of multiple predictive AI networks can enhance the robustness of predictions.

[0095] In some embodiments, step A2 above, based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, obtains the adjustment amount, and may include steps A21-A23.

[0096] Step A21: Determine the first parameter range based on the current transmission environment parameter information and parameter range set.

[0097] For example, the set of parameter ranges includes one or more parameter ranges.

[0098] For example, the first parameter range can be the range in the parameter range set to which the transmission environment parameters in the current transmission environment parameter information belong.

[0099] For example, each parameter in the transmission environment parameter information can have a corresponding set of parameter intervals. For each parameter in the current transmission environment parameter information, the parameter interval to which the parameter belongs (called the first parameter interval) can be determined from the set of parameter intervals.

[0100] Optionally, the parameter range set can be preset or generated based on the transmission status information corresponding to the transmission environment parameters.

[0101] Step A22: Based on the correspondence between the parameter intervals in the parameter interval set and the transmission status information set, determine the transmission status information set corresponding to the first parameter interval.

[0102] For example, the transmission status information set corresponding to the first parameter interval is found from the correspondence between one or more parameter intervals and one or more transmission status information sets.

[0103] Step A23: Based on the transmission status information set corresponding to the first parameter interval and the transmission status information set corresponding to the second parameter interval, obtain the adjustment amount.

[0104] For example, the set of transmission status information corresponding to the first parameter interval can refer to the set of transmission status information corresponding to the first parameter interval.

[0105] For example, the second parameter interval is the interval within the parameter interval set to which the input environment parameters corresponding to the first time point before the current time point belong. The first time point can be the time point before the current time point or any recent time point.

[0106] For example, based on the transmission state information set corresponding to the first parameter interval and the transmission state information set corresponding to the second parameter interval, feature information about the relationship between these two distributions is extracted, including the mean, variance, quantiles, and deviation. An adjustment amount is then obtained based on the extracted feature information. This adjustment amount can reflect the drift in transmission state information caused by changes in the transmission environment. This drift is difficult to obtain through time-series analysis and prediction, and can be used to correct for changes in the current transmission state information, thereby improving prediction accuracy.

[0107] Optionally, the change between the parameter range to which the current transmission environment parameter information belongs and the parameter range to which the historical transmission environment parameter information belongs can be detected, and based on the change, it can be determined whether to perform the step of determining the adjustment amount corresponding to the current transmission environment parameter information.

[0108] For example, if the parameter range of the historical transmission environment parameter information includes the parameter range of the current transmission environment parameter information, it means that the corresponding adjustment amount has been calculated, and the step of calculating the adjustment amount does not need to be performed. If the parameter range of the historical transmission environment parameter information does not include the parameter range of the current transmission environment parameter information, it means that the corresponding adjustment amount has not been calculated, and the step of calculating the adjustment amount needs to be performed.

[0109] In some embodiments, step A21 above, which determines the first parameter range based on the current transmission environment parameter information and parameter range set, may include steps A211-A212.

[0110] Step A211: Use the current transmission environment parameter information to map it into a high-dimensional vector.

[0111] For example, a second AI network can be used to map the current transmission environment parameter information into a high-dimensional vector.

[0112] Step A212: Obtain the first parameter interval using a set of high-dimensional vectors and parameter intervals.

[0113] For example, the first parameter interval can be obtained by using a third AI network based on high-dimensional vectors and a set of parameter intervals.

[0114] After mapping transmission environment parameters to a high-dimensional space, if transmission environment parameters that have not appeared in the past, especially qualitative vectors such as cell bandwidth, appear now, the parameter range that is close to the current transmission environment parameters in the past can be found by comparing the similarity in the high-dimensional space. For example, if cell bandwidths of 20M and 40M have only appeared in the past, and the current cell bandwidth is 50M, the 40M cell bandwidth can be used as the first parameter range, and the historical transmission state information of the 40M cell bandwidth can be used as a reference to predict the current transmission state information.

[0115] The following is combined with Figure 3 An example is provided to illustrate how the adjustment amount is obtained.

[0116] For each parameter in the current transmission environment parameter information Using AI network a, it is mapped to a high-dimensional vector. Each parameter has a corresponding parameter range. Each interval corresponds to a set of transmission status information. The mapped high-dimensional vector... And a set of parameter intervals, input into AI network b, to obtain the first parameter interval corresponding to the parameter, such as the optimal parameter interval, and obtain the set of historical transmission status information corresponding to the first parameter interval.

[0117] The transmission state information set obtained at the previous moment (which could be the transmission state information set corresponding to the second parameter interval mentioned above) and the transmission state information set corresponding to the first parameter interval are input into the distribution information extraction module to extract feature information about the relationship between these two distributions, including mean, variance, quantiles, deviation, etc. The obtained feature information is then input into the AI ​​network c to obtain the adjustment amount. .

[0118] Using a similar method, for all parameter intervals in the parameter interval set, the adjustment set can be obtained. .

[0119] Finally, the first transmission status information will be obtained. plus adjustment set To obtain the current transmission status information .

[0120] In some embodiments, the method provided in this disclosure may further include step B1.

[0121] Step B1: Obtain the parameter interval set based on the transmission status information sets corresponding to adjacent preset intervals. This can be performed before step A2.

[0122] For example, for each transmission environment parameter, the transmission environment parameter is divided into multiple preset intervals, and the transmission status information is segmented according to the preset intervals to obtain the transmission status information set corresponding to different preset intervals. The parameter interval set is obtained based on the transmission status information set corresponding to adjacent preset intervals.

[0123] In some embodiments, step B1 above, based on the set of transmission status information of adjacent preset intervals, obtains a set of parameter intervals, which may include steps B11-B13.

[0124] Step B11: Obtain the first residual based on the historical transmission status information set of the first preset interval and the predicted transmission status information set of the first preset interval.

[0125] For example, there can be multiple preset intervals, such as a first preset interval and a second preset interval.

[0126] For example, the predicted transmission status information set for the first preset interval can be obtained by using a fourth AI network to predict the transmission status information set for the first preset interval.

[0127] For example, the historical transmission status information set of the first preset interval may refer to the actual transmission status information set of the first preset interval.

[0128] Step B12: Obtain the second residual based on the historical transmission status information set of the second preset interval adjacent to the first preset interval and the predicted transmission status information set of the second preset interval.

[0129] For example, the predicted transmission status information set for the second preset interval can be obtained by using a fourth AI network to predict the transmission status information set for the second preset interval.

[0130] For example, the historical transmission status information set of the second preset interval can refer to the actual transmission status information set of the second preset interval.

[0131] Step B13: Determine whether to merge the first preset interval and the second preset interval based on the first residual and the second residual.

[0132] For example, if the difference between the first residual and the second residual is less than a preset threshold, the first preset interval and the second preset interval are merged into one interval, and the historical transmission status information sets corresponding to the first preset interval and the second preset interval are merged into one set. If the difference is greater than or equal to the preset threshold, they can be left unmerged. Based on the obtained intervals and transmission status information sets, and other intervals and other transmission status information sets, intervals and transmission status information sets are obtained again. By traversing all sets, the parameter interval set of the transmission environment parameter is obtained, and then the parameter interval set corresponding to each transmission environment parameter and the transmission status information set corresponding to each parameter interval are obtained.

[0133] By analyzing the degree of difference between two adjacent intervals, a more accurate set of parameter intervals can be obtained, which can improve the accuracy of prediction.

[0134] Taking SNR in wireless measurement information as an example, SNR is related to modulation level and can affect throughput. However, the relationship between SNR and throughput is not a direct mapping.

[0135] We can first determine the preset range of SNR, and then divide the throughput at a certain prediction granularity, such as 5dB, to obtain the throughput set corresponding to each preset range: .

[0136] For a set of throughput corresponding to a preset interval , using a length of A sliding window is used to divide the throughput dataset into a set of throughput samples for prediction. Each throughput sample... The former sample points Input AI network (such as the fourth AI network), where Indicates the first For each sample, the throughput prediction result is obtained. Compare the predicted results of the sample with the actual throughput of that sample. By comparison, we obtain the residual set: ,in, Indicates the length of the observation window. This represents the prediction window length. Similarly, we can obtain the AI ​​network's predictions for adjacent intervals within a preset interval. The corresponding throughput set prediction results, and based on these results and intervals The actual throughput set, and the resulting residual set .

[0137] Calculate the residual set and residual set The degree of difference between .For example . Represents the throughput set The residual distribution Represents the throughput set The residual distribution. Degree of variability. The smaller the value, the smaller the difference between the distributions of the two residuals.

[0138] if If the throughput is less than a preset threshold, it can be considered that there is no significant difference between the distributions of these two throughput sets, and the AI ​​models obtained based on these two throughput sets have no performance difference. Therefore, these two intervals (e.g., preset intervals) can be considered as such. and They are merged into one interval.

[0139] By traversing all the preset intervals, the final set of parameter intervals can be obtained. The distribution of the transmission status information sets between different parameter intervals in the parameter interval set has obvious differences.

[0140] By performing similar operations on all transmission environment parameters, we can obtain a set of transmission status information corresponding to different intervals of different transmission environment parameters.

[0141] Optionally, the method provided in this disclosure may further include: preprocessing current transmission environment parameter information, historical transmission environment parameter information, and / or historical transmission status information. This facilitates subsequent use by the AI ​​network.

[0142] For example, based on the preprocessed current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, the current transmission status information can be obtained, which can further improve the accuracy of prediction.

[0143] For example, a set of parameter intervals can be obtained based on the set of transmission status information corresponding to adjacent preset intervals after preprocessing.

[0144] For example, different methods can be used to handle missing values ​​depending on the type of missing value, as shown in Examples 1-3.

[0145] Example 1: Data loss due to inconsistent sampling periods. If the missing data is a measurable quantitative variable, such as RSRP or RSRQ, linear interpolation can be used to fill in the missing values. If the missing data is a non-measurable qualitative variable, such as the number of carrier aggregations configured in the cell or the maximum bandwidth configured in the cell, the missing data can be filled in using the data from the most recent time point adjacent to the missing time point.

[0146] Example 2 illustrates data loss due to unavailability. For instance, the business type might be missing; in some cases, it may be impossible to obtain the specific business type used by the running application. Similarly, user GPS data might be unavailable. In such cases, missing data can be masked to exclude it from subsequent AI network computation processes.

[0147] Example 3: In a Communication Processor (CP), some variables, such as throughput, will be recorded as missing values ​​if their value is 0. For such variables, expert knowledge needs to be used to pre-screen them, and then the missing values ​​should be replaced with 0.

[0148] For example, the relevant variables can be preprocessed, as in Examples a-b.

[0149] Example a: For some throughput-related variables, both their numerical values ​​and corresponding upper limits need to be considered to more comprehensively reflect information such as cell scheduling. For instance, for a cell with a bandwidth of 100MB, when the UE is performing some low-priority tasks or when the cell is particularly congested, the UE may only be allocated 20MB of bandwidth resources. This is different from allocating 20MB of bandwidth resources to the UE in a cell with a bandwidth of 20MB; although the allocated PRB resources are the same, the underlying scheduling situation is different. For these scheduling-related variables, absolute values ​​can be retained, and additional proportional information can be calculated, that is, the actual value at the same time is divided by the upper limit obtained from the configuration information at that time.

[0150] Therefore, combining and processing these variables in advance can reduce the difficulty of feature mining by AI networks, thereby reducing model complexity while improving accuracy.

[0151] Example b: For some non-numerical variables, such as business type, an AI network can be used to encode and map them into numerical vectors, which facilitates subsequent processing by the AI ​​network.

[0152] Figure 4 This is a schematic diagram of the architecture of an Ant-Net network provided in an embodiment of this disclosure.

[0153] Taking environmental parameter information, including wireless connection events and wireless status information, as an example. Combined with... Figure 4The wireless connection event is processed through the event processing module. For example, location encoding is used to mark the location where the event occurs and encode it into a binary mask matrix. Then, AI models suitable for wireless connection events in the AI ​​model library are used to predict the transmission status information. For example, AI model 1 can be a recurrent neural network (RNN), AI model k can be a multi-layer perceptron (MLP), and AI model N can be a transformer-based model. These are just examples.

[0154] The event processing module processes the wireless status information, for example, mapping it to a high-dimensional vector, calculating its correlation with historical wireless status information to obtain a correlation matrix, and inputting the correlation matrix into an AI model (e.g., AI model N). This model combines historical transmission status information, historical wireless status information, and current wireless status information to predict the transmission status information. Finally, the prediction results from each AI model are weighted and fused to obtain the predicted transmission status information. Optionally, the predicted transmission status information can be corrected based on an adjustment amount to further improve the accuracy of the prediction.

[0155] This disclosure also provides an electronic device including a processor, and optionally, a transceiver and / or memory coupled to the processor, the processor being configured to perform the steps of the method provided in any optional embodiment of this disclosure.

[0156] Figure 5 The diagram shows a structural schematic of an electronic device to which an embodiment of the present invention applies, such as... Figure 5 As shown, Figure 5 The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of this disclosure. Optionally, the electronic device may be a network node.

[0157] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with this disclosure. Processor 4001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0158] Bus 4002 may include a pathway for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0159] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium capable of carrying or storing computer programs and capable of being read by a computer, without limitation herein.

[0160] The memory 4003 is used to store computer programs that execute embodiments of the present disclosure, and is controlled by the processor 4001 to execute them. The processor 4001 is used to execute the computer programs stored in the memory 4003 to implement the steps shown in the foregoing method embodiments.

[0161] This disclosure provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.

[0162] This disclosure also provides a computer program product, including a computer program that, when executed by a processor, can implement the steps and corresponding content of the aforementioned method embodiments.

[0163] The terms “first,” “second,” “third,” “fourth,” “1,” “2,” etc. (if present) in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in a sequence other than that shown in the figures or text.

[0164] It should be understood that although arrows indicate various operation steps in the flowcharts of the embodiments of this disclosure, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of the embodiments of this disclosure, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages based on the actual implementation scenario. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured as required, and the embodiments of this disclosure do not limit this.

[0165] The above text and accompanying drawings are provided as examples only to help the reader understand this disclosure. They are not intended and should not be construed as limiting the scope of this disclosure in any way. Although certain embodiments and examples have been provided, it will be apparent to those skilled in the art, based on the content disclosed herein, that changes can be made to the illustrated embodiments and examples, and other similar implementations based on the technical concept of this disclosure can be adopted without departing from the scope of this disclosure, and these modifications and modifications are also within the protection scope of the embodiments of this disclosure.

Claims

1. A method performed by an electronic device, characterized in that, include: Obtain the current transmission environment parameter information; Based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information, the current transmission status information is obtained. The transmission environment parameter information includes at least one of the following: cell configuration information, wireless measurement information, and service type information.

2. The method according to claim 1, wherein, The process of obtaining the current transmission status information based on the current transmission environment parameter information, historical transmission environment parameter information, and historical transmission status information includes: Based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission status information, the first transmission status information is obtained; Based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission status information, the adjustment amount is obtained; Based on the first transmission status information and the adjustment amount, the current transmission status information is obtained.

3. The method according to claim 2, wherein, The first transmission status information is obtained based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission status information, including: The system uses artificial intelligence (AI) to select a network and obtains the first AI network that matches the current transmission environment parameter information. Using the first AI network, the first transmission status information is obtained based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission status information.

4. The method according to claim 3, wherein, The step of using artificial intelligence (AI) to select a network and obtaining a first AI network that matches the current transmission environment parameter information includes: Using the AI ​​to select networks, based on the current transmission environment parameter information, at least two first AI networks and at least two weights are determined, with the at least two first AI networks corresponding one-to-one with the at least two weights; Wherein, the step of using the first AI network to obtain the first transmission status information based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission status information includes: Using the at least two first AI networks, at least two intermediate transmission state information are obtained based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission state information; The first transmission status information is obtained based on the at least two intermediate transmission status information and the at least two weights.

5. The method according to claim 2, wherein, The adjustment amount is obtained based on the current transmission environment parameter information, the historical transmission environment parameter information, and the historical transmission status information, including: Based on the current transmission environment parameter information and the parameter interval set, a first parameter interval is determined; the first parameter interval is the interval to which the transmission environment parameter in the current transmission environment parameter information belongs in the parameter interval set. Based on the correspondence between the parameter intervals in the parameter interval set and the transmission status information set, the transmission status information set corresponding to the first parameter interval is determined. The adjustment amount is obtained based on the transmission status information set corresponding to the first parameter interval and the transmission status information set corresponding to the second parameter interval; wherein, the second parameter interval is the interval in the parameter interval set to which the transmission environment parameter corresponding to the first time before the current time belongs.

6. The method according to claim 5, wherein, The determination of the first parameter range based on the current transmission environment parameter information and parameter range set includes: Map the current transmission environment parameter information into a high-dimensional vector; The first parameter interval is obtained based on the high-dimensional vector and the set of parameter intervals.

7. The method according to claim 5 or 6, wherein, The method further includes: The parameter interval set is obtained based on the transmission status information sets corresponding to adjacent preset intervals.

8. The method according to claim 7, wherein, The parameter interval set is obtained based on the transmission status information sets corresponding to adjacent preset intervals, including: The first residual is obtained based on the historical transmission status information set of the first preset interval and the predicted transmission status information set of the first preset interval. The second residual is obtained based on the historical transmission status information set of the second preset interval adjacent to the first preset interval and the predicted transmission status information set of the second preset interval. Whether to merge the first preset interval and the second preset interval is determined based on the first residual and the second residual.

9. An electronic device, wherein, It includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method of any one of claims 18.

10. A computer-readable storage medium, wherein, It stores a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 18.

11. A computer program product, wherein, Includes a computer program that, when executed by a processor, implements the steps of the method of any one of claims 18.

Images