Interference prediction method, device, equipment, medium and product of drive test neighbor cell signal

Abstract

The invention discloses an interference prediction method, device, equipment, medium and product for a drive test neighbor cell signal, and relates to the technical field of communication. The method comprises the following steps: acquiring a to-be-processed drive test data set; the to-be-processed drive test data set comprises signal measurement data of the current service base station and the adjacent base station; performing dimension reduction processing and feature extraction on the to-be-processed drive test data set to obtain an interference feature data set; reconstructing the interference characteristic data set to obtain a reconstructed drive test data set corresponding to the drive test data set to be processed; and determining an adjacent region signal interference intensity graph of the current service base station according to the reconstructed drive test data set. Interference prediction on the to-be-processed drive test data set is realized through interference feature extraction and data reconstruction, complex multi-level signal interference features can be effectively captured, and prediction accuracy of adjacent region signal interference intensity is improved.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method, apparatus, device, medium, and product for predicting interference in drive-tested neighboring cell signals. Background Technology

[0002] The collection and analysis of drive test data is an important part of mobile communication network optimization, and the prediction of neighboring cell signal interference plays a crucial role in improving the overall performance of the communication network.

[0003] Most current signal interference prediction methods are based on single-layer neural networks or statistical models based on simple regression. These methods often fail to capture multi-layered interference features and cannot effectively handle complex interference patterns in mobile networks. When the data contains a lot of noise or uncertainty, the prediction accuracy of statistical models based on simple regression drops significantly. When faced with large-scale data and complex dynamic environments, it is difficult to achieve accurate interference prediction of drive-test neighboring cell signals. Summary of the Invention

[0004] This invention provides a method, apparatus, device, medium, and product for predicting interference in road test neighboring cell signals, in order to solve the problem that traditional signal interference prediction methods are difficult to achieve accurate interference prediction of road test neighboring cell signals when faced with large-scale data and complex dynamic environments.

[0005] In a first aspect, embodiments of the present invention provide an interference prediction method for drive-tested neighboring cell signals, comprising:

[0006] Obtain the drive test dataset to be processed; the drive test dataset to be processed includes signal measurement data of the current serving base station and neighboring base stations;

[0007] The drive test dataset to be processed is subjected to dimensionality reduction and feature extraction to obtain an interference feature dataset; the interference feature dataset is reconstructed to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; and the neighboring cell signal interference intensity map of the current serving base station is determined based on the reconstructed drive test dataset.

[0008] Secondly, embodiments of the present invention provide an interference prediction device for drive-test neighbor cell signals, comprising:

[0009] The drive test data acquisition module is used to acquire the drive test dataset to be processed; the drive test dataset to be processed includes signal measurement data of the current serving base station and neighboring base stations;

[0010] The interference intensity prediction module is used to perform dimensionality reduction and feature extraction on the drive test dataset to be processed to obtain an interference feature dataset; to reconstruct the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; and to determine the neighboring cell signal interference intensity map of the current service base based on the reconstructed drive test dataset.

[0011] Thirdly, embodiments of the present invention provide an electronic device, the electronic device comprising:

[0012] At least one processor;

[0013] and a memory communicatively connected to the at least one processor;

[0014] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the interference prediction method for drive-test neighbor cell signals according to any embodiment of the present invention.

[0015] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer instructions, which are used to cause a processor to execute and implement the interference prediction method for drive-test neighbor cell signals according to any embodiment of the present invention.

[0016] Fifthly, embodiments of the present invention provide a computer program product including a computer program, which, when executed by a processor, implements the interference prediction method for drive-test neighboring cell signals according to any embodiment of the present invention.

[0017] The technical solution of this invention involves acquiring a drive test dataset to be processed; this dataset includes signal measurement data of the current serving base station and neighboring base stations; performing dimensionality reduction and feature extraction on the drive test dataset to obtain an interference feature dataset; reconstructing the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; and determining the neighboring signal interference intensity map of the current serving base station based on the reconstructed drive test dataset. By extracting interference features and reconstructing data, interference prediction of the drive test dataset to be processed is achieved, effectively capturing complex, multi-level signal interference features. This solves the problem that traditional signal interference prediction methods struggle to achieve accurate interference prediction of neighboring signals in the face of large-scale data and complex dynamic environments, thus improving the accuracy of neighboring signal interference intensity prediction.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of an interference prediction method for drive-tested neighboring cell signals provided in Embodiment 1 of the present invention;

[0021] Figure 2 This is a flowchart of an interference prediction method for drive-test neighbor cell signals provided in Embodiment 2 of the present invention;

[0022] Figure 3 This is a schematic diagram of the structure of an interference prediction device for drive-test neighbor cell signals provided in Embodiment 3 of the present invention;

[0023] Figure 4 A schematic diagram of the structure of an electronic device for implementing the interference prediction method for road test neighboring cell signals according to embodiments of the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] Example 1

[0027] Figure 1This is a flowchart of a method for predicting interference in drive-by neighboring cell signals according to Embodiment 1 of the present invention. This embodiment is applicable to predicting the interference intensity of drive-by neighboring cell signals. This method can be executed by a drive-by neighboring cell signal interference prediction device, which can be implemented in hardware and / or software and can be configured in an electronic device. For example... Figure 1 As shown, the method includes:

[0028] S110. Obtain the drive test dataset to be processed; the drive test dataset to be processed contains signal measurement data of the current serving base station and neighboring base stations.

[0029] The unprocessed road test dataset can be understood as road test data awaiting processing. Road test data can be considered as data collected during driving in a real road environment by a test vehicle equipped with sensors and computing units. In this embodiment, interference prediction of neighboring cell signals is mainly performed on the unprocessed road test dataset of the base station. The unprocessed road test dataset includes signal measurement data of the current serving base station and at least one neighboring base station obtained over a period of time.

[0030] In this embodiment, the signal measurement data includes the base station location, spectrum usage, timestamp information, and signal strength and signal-to-noise ratio between the current serving base station and neighboring base stations, i.e., D={P ij (t), SNR ij (t), L i (x i ,y i ), L j (x j , y j ), U i (t), T}。 P ij (t) represents the signal strength between the current serving base station i and the j-th neighboring base station at time t, and the SNR. ij (t)=P ij (t) / N ij (t), N ij (t) represents the noise intensity received by the current serving base station i from the j-th neighboring base station at time t, and the SNR. ij (t) represents the signal-to-noise ratio between the current serving base station i and the j-th neighboring base station at time t. i (x i ,y i ) represents the location of the current serving base station i, L j (x j , y j ) represents the location of the j-th neighboring base station; U i(t) represents the proportion of frequency resources currently used by serving base station i at time t; timestamp T = {t1, t2, …, t} n ,} is used to ensure the integrity of timing information in signal measurement data.

[0031] In this embodiment, for noise intensity N ij The source of (t) can be determined based on availability priority: if noise measurement is available at the road test site, the measured noise intensity should be used; if only statistics such as total received power and bandwidth are available, the noise intensity N should be estimated. ij (t). Then, using the obtained noise intensity N ij (t) and signal strength P ij (t) Calculate the signal-to-noise ratio (SNR) ij (t). If N cannot be directly obtained in some measurement scenarios... ij (t), or the estimated SNR can be directly used in the preprocessing and decoding output. ij (t) is used as an equivalent input feature in subsequent determinations.

[0032] S120. Perform dimensionality reduction and feature extraction on the drive test dataset to be processed to obtain an interference feature dataset; reconstruct the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; determine the neighboring cell signal interference intensity map of the current serving base station based on the reconstructed drive test dataset.

[0033] The interference feature dataset can be considered as a dataset representing the characteristics of interference signals. The reconstructed drive test dataset can be considered as a drive test dataset reconstructed from the interference feature dataset of the drive test dataset to be processed. The neighboring cell signal interference intensity map can be considered as a map drawn based on the interference intensity of neighboring cell base station signals received by the current serving base station.

[0034] In this embodiment, the dimensionality of the road test dataset to be processed is reduced and the interference features are extracted to obtain an interference feature set; then the interference feature set is reconstructed to obtain a reconstructed road test dataset; finally, the interference intensity is determined based on the reconstructed road test dataset, and an interference intensity map is formed.

[0035] One implementation approach is to use a fully trained hierarchical variational encoder model to perform feature extraction, data reconstruction, and interference intensity determination. Specifically, the hierarchical variational encoder model first performs dimensionality reduction and feature extraction on the drive test data to be processed to obtain low-order feature data. Interference features are then extracted from the low-order feature data to obtain an interference feature dataset. Next, the interference feature dataset is reconstructed to obtain a reconstructed drive test dataset. Finally, the interference intensity of the neighboring cell signals of the current serving base station is determined based on the reconstructed drive test dataset.

[0036] The technical solution of this embodiment involves acquiring a drive test dataset to be processed; this dataset includes signal measurement data of the current serving base station and neighboring base stations; performing dimensionality reduction and feature extraction on the drive test dataset to obtain an interference feature dataset; reconstructing the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; and determining the neighboring cell signal interference intensity map of the current serving base station based on the reconstructed drive test dataset. By extracting interference features and reconstructing data, interference prediction of the drive test dataset to be processed can be achieved, effectively capturing complex, multi-level signal interference features and improving the accuracy of neighboring cell signal interference intensity prediction.

[0037] In an optional embodiment of this example, after obtaining the signal interference prediction results of neighboring base stations, the method further includes:

[0038] A1. Obtain the neighbor cell signal strength map from the drive test dataset to be processed.

[0039] Among them, the neighboring cell signal strength map can be considered as a map drawn based on the signal strength of the neighboring cell base stations of the current serving base station.

[0040] Specifically, the signal strength of each neighboring base station of the current serving base station is obtained from the drive test dataset to be processed and plotted as a neighboring cell signal strength map of the current serving base station.

[0041] A2. Based on the neighboring cell signal interference intensity map and the neighboring cell signal intensity map, determine the candidate serving base stations that meet the handover conditions, and determine the handover priority queue for each candidate serving base station.

[0042] In this embodiment, the switching condition may include the neighboring cell signal interference strength being higher than a set signal strength threshold S. th Furthermore, the interference intensity of neighboring cell base stations is lower than the interference intensity threshold I. th The neighboring cell signal interference strength of a neighboring base station can be considered as the signal strength of interference experienced by the neighboring base station when the neighboring base station acts as a serving base station. A candidate serving base station can be considered as a neighboring base station that meets the handover conditions.

[0043] In this embodiment, there may be one or more candidate serving base stations. Therefore, it is also necessary to determine the handover order of the target serving base station based on the handover priority of the candidate serving base stations, and then determine the target serving base station based on the handover order. When determining the handover priority of the candidate serving base stations, it can be determined based on the signal strength and interference strength of neighboring base stations, and the handover cost of the base station can also be comprehensively considered, such as the cost of computing resources occupied by the base station.

[0044] For example, the handover priority score for the current serving base station to hand over to each candidate serving base station is determined, and then a handover priority queue is formed by the handover priorities of each candidate serving base station in descending order of handover priority scores. The handover priority score O for the current serving base station i to hand over to the j-th candidate serving base station is... ij It can be:

[0045] ;

[0046] in, Let the signal strength be the j-th candidate serving base station of the current serving base station i. Let the signal interference strength be the j-th candidate serving base station of the current serving base station i. The handover cost for switching from the current serving base station i to the j-th candidate serving base station; To represent the weight of signal strength in relation to the switching order, The weight of signal interference intensity in relation to the handover order. The weight of the switching cost relative to the switching order.

[0047] A3. Determine the target serving base station according to the handover priority queue, and switch the serving base station to the target serving base station.

[0048] Specifically, the highest priority candidate serving base station is determined as the target serving base station according to the handover priority queue, and the serving base station is switched to the target serving base station; if the highest priority candidate serving base station cannot be used, the next priority candidate serving base station is determined as the target serving base station according to the handover priority queue.

[0049] As an optional embodiment of this example, the method further includes: collecting a drive test dataset to be processed within a preset time period along a preset planned path; performing dimensionality reduction and feature extraction on the drive test dataset to be processed at each measurement time within the preset time period to obtain an interference feature dataset; reconstructing the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; determining the neighboring cell signal interference intensity map of the current serving base station at each measurement time based on the reconstructed drive test dataset; for each measurement time, obtaining a neighboring cell signal intensity map from the drive test dataset to be processed; determining candidate serving base stations that meet the handover conditions based on the neighboring cell signal interference intensity map and the neighboring cell signal intensity map; determining the handover priority of the candidate serving base stations; determining the candidate serving base station with the highest handover priority as the target serving base station at the measurement time; determining a serving base station handover sequence along the preset planned path based on the target serving base station at each measurement time; and generating a handover path map based on the serving base station handover sequence and the preset planned path.

[0050] Understandably, this handover path map can be dynamically adjusted in real time based on the collected drive test dataset, and the weight of the handover priority score can also be adjusted based on the network performance indicators and user experience after the base station handover.

[0051] In this embodiment, the handover path diagram can be represented as a dynamic connection between base stations, with nodes representing base stations and edges representing handover paths. Handover events can be marked on the handover path diagram to display the changes in interference intensity before and after the handover. The changes in interference between base stations after the handover are displayed in a visual manner.

[0052] Example 2

[0053] Figure 2 This is a flowchart of an interference prediction method for drive-test neighboring cell signals provided in Embodiment 1 of the present invention. This embodiment specifies the processing and interference prediction steps of the drive-test dataset to be processed in the above embodiments. Specifically, the process involves: performing dimensionality reduction and feature extraction on the drive test dataset to be processed to obtain an interference feature dataset; reconstructing the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; and determining the neighboring cell signal interference intensity map of the current serving base station based on the reconstructed drive test dataset. This includes: inputting the drive test dataset to be processed into the encoder sub-model of the hierarchical variational encoder model through a fully trained hierarchical variational encoder model to perform dimensionality reduction and feature extraction on the drive test dataset to obtain an interference feature dataset; the encoder sub-model includes multiple cascaded encoding layers; inputting the interference feature dataset into the decoder sub-model of the hierarchical variational encoder model to reconstruct the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; the decoder sub-model includes multiple cascaded decoding layers; and inputting the reconstructed drive test dataset into the interference intensity decision sub-model of the hierarchical variational encoder model to perform interference intensity decision on the reconstructed drive test data to obtain the neighboring cell signal interference intensity map of the current serving base station.

[0054] like Figure 2 As shown, the method includes:

[0055] S210. Obtain the drive test dataset to be processed; the drive test dataset to be processed contains signal measurement data of the serving base station and neighboring base stations.

[0056] S220. By training a complete hierarchical variational encoder model, the road test dataset to be processed is input into the encoder sub-model in the hierarchical variational encoder model to perform dimensionality reduction and feature extraction on the road test dataset to be processed, and obtain the interference feature dataset; the encoder sub-model includes multiple cascaded coding layers.

[0057] In this embodiment, the hierarchical encoder model (H-VAE) includes an encoder sub-model, a decoder sub-model, and an interference intensity decision sub-model. The encoder sub-model includes K cascaded coding layers, where K ≥ 2.

[0058] First, the road test dataset to be processed is input into the encoder sub-model of the hierarchical variational encoder model. The encoder sub-model sequentially inputs the road test dataset into the bottom encoding layer. The bottom encoding layer performs dimensionality reduction and feature extraction on the road test dataset, capturing the low-level feature set of the dataset. The low-level feature set is a collection of low-level features represented by low-order latent variables, such as... Where W1 is the weight matrix of the bottom coding layer, and b1 is the bias vector of the bottom coding layer. is the activation function of the encoder sub-model. ~N(0,1) represents the noise term of the standard normal distribution in the bottom coding layer, V1 is the variance estimate of the bottom coding layer, used to capture the uncertainty in low-order features, and Z1 is the bottom feature set. for.

[0059] Then, the bottom feature set Z1 is passed layer by layer to the next coding layer, and the feature set Z1 of the previous layer is extracted layer by layer through K coding layers. K-1 The features in the data are used to obtain the interference feature set, which is a collection of high-level features represented by higher-order latent variables, such as... Among them, W K b is the weight matrix of the higher-level coding layer. K V is the bias vector of the higher-level coding layer. K For variance estimation of higher-level coding layers, Z K Z is the interference feature set, used to represent complex interference features in the data. K-1 These are the hidden variables of the K-1 layer coding layer.

[0060] S230. Input the interference feature dataset into the decoder sub-model of the hierarchical variational encoder model to reconstruct the interference feature dataset and obtain the reconstructed road test dataset corresponding to the road test dataset to be processed.

[0061] In this embodiment, the interference feature dataset is input into the decoder sub-model, and data reconstruction is performed on the interference feature dataset to obtain the reconstructed road test dataset, such as... ;in, This is the weight matrix of the decoder sub-model. This is the bias vector of the decoder sub-model. For the activation function of the decoder sub-model, To reconstruct road test data.

[0062] S240. Input the reconstructed drive test dataset into the interference intensity decision sub-model in the hierarchical variational encoder model, perform interference intensity decision on the reconstructed drive test data, and obtain the neighboring cell signal interference intensity map of the current serving base station.

[0063] In this embodiment, a fully connected layer is added after the decoder sub-model as an interference strength decision sub-model. This sub-model is used to determine the interference strength of the reconstructed drive test data, obtain the interference strength of the neighboring cell signals of the current serving base station, and determine the neighboring cell interference strength map based on the interference strength of the neighboring cell signals, such as... ;in, To predict the interference intensity of the signal from neighboring base station j received by the current serving base station i, To reconstruct the drive test data for the signal of neighboring base station j of the currently serving base station i in the drive test dataset, The weight matrix of the interference intensity decision sub-model is... This is the bias vector for the interference intensity decision sub-model.

[0064] (1) The technical solution of this invention involves obtaining a drive test dataset to be processed; the drive test dataset to be processed includes signal measurement data of the serving base station and neighboring base stations; by training a complete hierarchical variational encoder model, the drive test dataset to be processed is input into the encoder sub-model in the hierarchical variational encoder model, and the drive test dataset to be processed is subjected to dimensionality reduction and feature extraction to obtain an interference feature dataset; the encoder sub-model includes multiple cascaded coding layers; the interference feature dataset is input into the decoder sub-model of the hierarchical variational encoder model, and the interference feature dataset is reconstructed to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; the reconstructed drive test dataset is input into the interference intensity decision sub-model in the hierarchical variational encoder model, and the interference intensity decision is performed on the reconstructed drive test data to obtain the interference intensity map of the neighboring signal of the current serving base station. By constructing a multi-layered hierarchical variational autoencoder model, which employs a multi-layer encoder and decoder structure, the lower layer captures simple signal features (such as signal strength and signal-to-noise ratio), while the higher layer extracts complex interference patterns. Unlike traditional single-layer models, the hierarchical structure can more accurately capture complex features in neighboring signal interference at different levels. Especially when signal interference involves multi-dimensional data, the hierarchical encoder in this proposal can dynamically extract multiple interference dimensions, making the model more flexible and significantly improving prediction accuracy.

[0065] Traditional statistical models based on single-layer neural networks or simple regression for interference prediction have another drawback: they lack the ability to effectively handle uncertainties in the data. Road test data often contains a lot of noise and outliers. Traditional prediction models are prone to overfitting or underfitting when faced with these uncertainties, resulting in poor robustness and generalization ability of the prediction results. Existing technologies have low adaptive ability when dealing with dynamic changes in mobile communication networks and cannot adjust the model structure and parameters according to real-time updated data, resulting in insufficient accuracy in signal interference prediction.

[0066] To address the aforementioned problems, based on any of the above embodiments, this embodiment further provides a training step for a hierarchical variational encoder model, including:

[0067] B1. Obtain the hierarchical variational encoder model to be trained and the drive test data sample set; the drive test data sample set includes a training sample set and a validation sample set; the training sample set and the validation sample set include at least one sample tuple, and the sample tuple includes real signal measurement data and real interference intensity map of neighboring base stations.

[0068] B2. Input the training samples from the training sample set into the encoder sub-model of the hierarchical variational encoder model to be trained, and obtain the training sample interference feature dataset output by the encoder sub-model of the hierarchical variational encoder model; input the training sample interference feature dataset into the decoding sub-model to obtain the training sample reconstruction data; input the training sample reconstruction data into the interference intensity decision sub-model to obtain the predicted interference intensity map.

[0069] In this embodiment, training samples from the training sample set are input into the encoder sub-model of the hierarchical variational encoder model to be trained. The encoder sub-model performs dimensionality reduction, low-order feature extraction, and high-order feature extraction layer by layer through K encoding layers to obtain a training sample interference feature dataset. Then, the training sample interference feature dataset is input into the decoding sub-model to reconstruct the training sample interference feature dataset, obtaining reconstructed training sample data. Finally, the reconstructed training sample data is input into the interference intensity decision sub-model to determine the interference intensity, obtaining a predicted interference intensity map.

[0070] B3. Based on the training sample set, the training sample reconstruction data, and the predicted interference intensity map, optimize the model parameters of the hierarchical variational encoder model using the maximum evidence lower bound optimization method.

[0071] In this embodiment, the first total loss function value is calculated based on the training sample set, the training sample reconstruction data, and the predicted interference intensity map, and the model parameters of the hierarchical variational encoder model are optimized based on the first total loss function value and the evidence lower bound optimization method.

[0072] In one implementation of this embodiment, the model parameters of the hierarchical variational encoder model are updated based on the maximum evidence lower bound optimization method, according to the training sample set, the training sample reconstructed data, and the predicted interference intensity map, including:

[0073] The reconstruction loss function value is determined based on the reconstructed data from the training samples and the measured data from the real signal; the prediction loss function value is determined based on the real interference intensity map and the predicted interference intensity map; the regularization loss function value is determined based on the posterior distribution of the encoder sub-model and the preset prior distribution; the first total loss function value is determined based on the reconstruction loss function value, the prediction loss function value, and the regularization loss function value; and the model parameters of the hierarchical variational encoder model are updated based on the first total loss function value.

[0074] The reconstruction loss function represents the difference between the reconstructed data and the actual signal measurement data. The prediction loss function represents the difference between the predicted interference intensity map obtained by the hierarchical variational encoder model and the labeled actual interference intensity map in the training samples. The regularization loss function is used to normalize the latent space, prevent overfitting, and ensure the continuity of model training.

[0075] In this embodiment, the calculation method of the reconstruction loss function value can be expressed as follows:

[0076] ;in, To reconstruct the signal measurement data of neighboring base station j of the currently serving base station i in the dataset, To reconstruct the loss function value.

[0077] The calculation method for the prediction loss function value can be expressed as follows:

[0078] ;in, Let be the actual interference intensity received by the current serving base station i from the neighboring base station j. To predict the value of the loss function.

[0079] The regularization loss function value can be calculated as follows:

[0080] ;in, For interference feature set The posterior distribution, For interference feature set The prior distribution, It is usually a standard normal distribution.

[0081] The first total loss function value can be calculated as follows: ;in, The weighting coefficients used to predict the value of the loss function. These are the weighting coefficients for the regularization loss function, used to balance the importance of various losses.

[0082] Using an optimizer (such as Adam), the gradient of the first total loss function value with respect to all model parameters (such as the weight matrix W, bias vector b, and variance estimate V of the encoder sub-model, decoder sub-model, and interference intensity decision sub-model) is calculated via backpropagation. Then, all model parameters are updated based on the calculated gradients. After multiple iterations, the model parameters are continuously optimized and eventually converge to a set of model parameters that can accurately predict the interference intensity.

[0083] B4. Freeze the model parameters of the hierarchical variational encoder model trained with training samples, and input the validation sample set into the hierarchical variational encoder model with frozen model parameters to obtain a validation interference intensity map; calculate the second total loss function value based on the validation interference intensity map and the validation sample set, and feed the second total loss function value back to the Bayesian optimizer.

[0084] In this embodiment, the second total loss function value serves as an evaluation value of the model's generalization ability and is used for subsequent hyperparameter optimization. The calculation method for the second total loss function value is the same as that for the first total loss function value, and will not be repeated here. The difference lies in that the first total loss function value is calculated based on the prediction results of the training sample set, while the second total loss function value is calculated based on the prediction results of the validation sample set.

[0085] B5. Based on the Bayesian optimizer, determine the current optimal hyperparameters according to the second total loss function value.

[0086] In this embodiment, the hyperparameters of the Bayesian optimizer are automatically tuned, and the optimal combination of hyperparameters for the hierarchical variational encoder model is found through an external optimization loop.

[0087] In one implementation of this embodiment, determining the current optimal hyperparameters based on the Bayesian optimizer and the second total loss function value includes:

[0088] B51. Obtain a set of hyperparameters for the hierarchical variational encoder model to be trained, and determine the prior distribution of each hyperparameter.

[0089] In this embodiment, a set of hyperparameters for a hierarchical variational autoencoder model is defined. Where α represents the learning rate, and d Z represents the dimension of the interference feature set, and K represents the number of network layers in the hierarchical variational encoder model. represents the set of hyperparameters. Define the prior distribution : , , ;in, This is the upper bound of the learning rate α. This is the lower bound of the learning rate α; The mean of the dimension of the interference feature set. The variance of the dimension of the interference feature set. This is the upper bound of the number of network layers. This is the lower bound of the number of network layers.

[0090] B52. Initialize the Bayesian optimizer; the Bayesian optimizer includes a surrogate model and a collection function, the surrogate model includes a target loss function between the current hyperparameters and the value of the second total loss function; the collection function is the expectation of the improvement between the target loss function of the current hyperparameters and the minimum observation loss function of the current optimal hyperparameters.

[0091] In this embodiment, a Bayesian optimizer is initialized, which includes a surrogate model (such as a Gaussian process) to model the target loss function relative to the current hyperparameters Θ and the second total loss function value of the model on the validation set. and a collection function This is used to guide users in their searches.

[0092] The core objective of the Bayesian optimizer is to minimize the second total loss function value of the model on the validation set. Assume the currently evaluated optimal hyperparameters are... The corresponding minimum observation loss function is:

[0093] ;

[0094] Based on this, the expected improvement (EI) is adopted as the acquisition function. To evaluate and select hyperparameters. Acquisition function. Calculate the new parameter Θ relative to the current optimal parameter. The expected performance improvement. The core objective of the Bayesian optimizer is to minimize the value of the second total loss function. Performance improvement means a reduction in loss, therefore the improvement amount... Defined as:

[0095] ;

[0096] Acquisition function That is, the amount of improvement. The expectation, which is expressed through a surrogate model (such as a Gaussian process) on the loss function. The posterior distribution is obtained as follows:

[0097] ;

[0098] in, It is the probability distribution of the second loss function value of the surrogate model at the current hyperparameter Θ; It contains all historical observation data up to the nth iteration; This represents the improvement of the prediction loss of the current hyperparameter Θ compared to the minimum observation loss function of the current optimal hyperparameter; max(0, ...) is used to focus only on the positive improvement; To determine the expected value of the positive improvement, we select the hyperparameter Θ that maximizes this expected value as the current optimal hyperparameter (i.e., the next current hyperparameter).

[0099] B53. Based on the Bayesian optimizer, determine the current optimal hyperparameters according to the acquisition function and the surrogate model.

[0100] In this embodiment, an iterative loop based on a Bayesian optimizer is performed until the maximum number of evaluations is reached or performance convergence is achieved. Candidate hyperparameters are intelligently generated using the posterior distribution, and real observation data H is accumulated. This is based on a surrogate model and a data acquisition function. We recommend the current optimal hyperparameter that best represents the expected performance as the next optimal hyperparameter. .

[0101] ;

[0102] B6. Pass the current optimal hyperparameters to the hierarchical variational encoder model, return to execute the relevant steps of inputting the training samples from the training sample set into the encoder sub-model of the hierarchical variational encoder model to be trained, obtaining the training sample interference feature dataset output by the encoder sub-model of the hierarchical variational encoder model, and updating the Bayesian optimizer until the iteration termination condition is reached, to obtain the target optimal hyperparameters and a fully trained hierarchical variational encoder model with the target optimal hyperparameters.

[0103] In this embodiment, the next current hyperparameter is... The input is passed to step B2, steps B2 through B5 are re-executed, and the corresponding minimum observation loss function is returned. , will new observation data { Add historical datasets and update the surrogate model of the Bayesian optimizer accordingly, then update the posterior distribution based on Bayes' theorem. :

[0104] ;

[0105] in, To the current hyperparameters The following road test dataset The likelihood function represents the ability of the hierarchical variational autoencoder model to adapt to interference from neighboring signal areas under given parameters. This represents the marginal distribution that is normalized over all possible combinations of hyperparameters.

[0106] After the Bayesian optimization loop completes, a historical dataset containing all hyperparameter evaluation records is obtained. :

[0107] ;

[0108] From historical datasets The hyperparameter that minimizes the observation loss function on the validation set is selected as the final target optimal hyperparameter. :

[0109] ;

[0110] Thus, the target optimal hyperparameters and those with the target optimal hyperparameters are obtained. A fully trained hierarchical variational encoder model.

[0111] This invention provides a training method for a hierarchical variational encoder model. By dynamically adjusting hyperparameters through Bayesian optimization, it avoids the inefficiencies of traditional grid search and random search. Bayesian optimization efficiently determines the optimal hyperparameter combination through prior and posterior distribution updates, significantly accelerating model training convergence. In scenarios handling large-scale road test data, Bayesian optimization effectively solves the underfitting or overfitting problems caused by improper hyperparameter selection, improving the overall robustness of the model and significantly enhancing signal interference prediction performance. The method utilizes the latent variable mechanism in the hierarchical variational autoencoder to model the uncertainty in road test data. By maximizing the lower bound of evidence to optimize the posterior distribution of latent variables, the model can effectively preserve important patterns in the signal when processing complex data containing noise or outliers, avoiding overfitting to noise information. Compared to traditional methods, latent variables can better express the potential structural and pattern changes in neighboring signal interference, improving the model's ability to handle uncertain data and making the prediction results more robust and generalizable.

[0112] Example 3

[0113] Figure 3 This is a schematic diagram of the structure of a drive-test neighbor cell signal interference prediction device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a road test data acquisition module 310 and an interference intensity prediction module 320; wherein:

[0114] The drive test data acquisition module 310 is used to acquire a drive test dataset to be processed; the drive test dataset to be processed includes signal measurement data of the current serving base station and neighboring base stations.

[0115] The interference intensity prediction module 320 is used to perform dimensionality reduction and feature extraction on the drive test dataset to be processed to obtain an interference feature dataset; to reconstruct the interference feature dataset to obtain a reconstructed drive test dataset corresponding to the drive test dataset to be processed; and to determine the neighboring cell signal interference intensity map of the current service base based on the reconstructed drive test dataset.

[0116] This invention provides an interference prediction device for drive-test neighboring cell signals. The device acquires a drive-test dataset to be processed, which includes signal measurement data of the current serving base station and neighboring base stations. It performs dimensionality reduction and feature extraction on the drive-test dataset to obtain an interference feature dataset. The interference feature dataset is then reconstructed to obtain a reconstructed drive-test dataset corresponding to the drive-test dataset to be processed. Finally, the device determines the interference intensity map of the neighboring cell signals of the current serving base station based on the reconstructed drive-test dataset. By extracting interference features and reconstructing data, interference prediction of the drive-test dataset to be processed is achieved, effectively capturing complex, multi-level signal interference features and improving the accuracy of neighboring cell signal interference intensity prediction.

[0117] Optionally, the interference intensity prediction module 320 is specifically used for:

[0118] By training a complete hierarchical variational encoder model, the road test dataset to be processed is input into the encoder sub-model in the hierarchical variational encoder model to perform dimensionality reduction and feature extraction on the road test dataset to be processed, thereby obtaining the interference feature dataset; the encoder sub-model includes multiple cascaded encoding layers.

[0119] The interference feature dataset is input into the decoder sub-model of the hierarchical variational encoder model to reconstruct the interference feature dataset and obtain the reconstructed drive test dataset corresponding to the drive test dataset to be processed; the decoder sub-model includes multiple cascaded decoding layers;

[0120] The reconstructed drive test dataset is input into the interference intensity decision sub-model in the hierarchical variational encoder model to perform interference intensity decision on the reconstructed drive test data, thereby obtaining the neighboring cell signal interference intensity map of the current serving base station.

[0121] Optionally, the device further includes a model training module, the model training module comprising:

[0122] The sample set acquisition unit is used to acquire the hierarchical variational encoder model to be trained and the drive test data sample set; the drive test data sample set includes a training sample set and a validation sample set; the training sample set and the validation sample set include at least one sample tuple, and the sample tuple includes real signal measurement data and real interference intensity of neighboring base stations;

[0123] The interference intensity prediction module is used to input training samples from the training sample set into the encoder sub-model of the hierarchical variational encoder model to be trained, and obtain the training sample interference feature dataset output by the encoder sub-model of the hierarchical variational encoder model; input the training sample interference feature dataset into the decoding sub-model to obtain training sample reconstruction data; and input the training sample reconstruction data into the interference intensity decision sub-model to obtain the predicted interference intensity map.

[0124] The model parameter update unit is used to update the model parameters of the hierarchical variational encoder model based on the maximum evidence lower bound optimization method according to the training sample set, the training sample reconstructed data and the predicted interference intensity map.

[0125] The verification feedback unit is used to freeze the model parameters of the updated hierarchical variational encoder model, input the verification sample set into the hierarchical variational encoder model with frozen model parameters, and obtain a verification interference intensity map; calculate the second total loss function value based on the verification interference intensity map and the verification sample set, and feed the second total loss function value back to the Bayesian optimizer;

[0126] The current optimal hyperparameter determination unit is used to determine the current optimal hyperparameters based on the Bayesian optimizer and the value of the second total loss function.

[0127] The iterative update unit is used to pass the current optimal hyperparameters to the hierarchical variational encoder model, return to execute the relevant steps of inputting the training samples from the training sample set into the encoder sub-model of the hierarchical variational encoder model to be trained, obtaining the training sample interference feature dataset output by the encoder sub-model of the hierarchical variational encoder model, and updating the Bayesian optimizer until the iteration termination condition is reached, thereby obtaining the target optimal hyperparameters and a fully trained hierarchical variational encoder model with the target optimal hyperparameters.

[0128] Optionally, the model parameter update unit is specifically used for:

[0129] The reconstruction loss function value is determined based on the reconstructed data from the training samples and the actual signal measurement data.

[0130] The prediction loss function value is determined based on the actual interference intensity map and the predicted interference intensity map;

[0131] The regularization loss function value is determined based on the posterior distribution of the encoder sub-model and the preset prior distribution;

[0132] The first total loss function value is determined based on the reconstruction loss function value, the prediction loss function value, and the regularization loss function value.

[0133] The model parameters of the hierarchical variational encoder model are updated based on the first total loss function value.

[0134] Optionally, the current optimal hyperparameter determination unit is specifically used for:

[0135] Obtain a set of hyperparameters for the hierarchical variational encoder model to be trained, and determine the prior distribution of each hyperparameter;

[0136] Initialize the Bayesian optimizer; the Bayesian optimizer includes a surrogate model and a collection function, the surrogate model includes a target loss function between the current hyperparameters and the value of the second total loss function; the collection function is the expectation of the improvement between the target loss function of the current hyperparameters and the minimum observation loss function of the current optimal hyperparameters;

[0137] Based on the Bayesian optimizer, the current optimal hyperparameters are determined according to the acquisition function and the surrogate model.

[0138] Optionally, the device further includes:

[0139] The neighbor cell signal strength map acquisition module is used to acquire the neighbor cell signal strength map from the drive test dataset to be processed after obtaining the signal interference prediction results of the neighbor cell base station.

[0140] The handover priority queue determination module is used to determine candidate serving base stations that meet the handover conditions based on the neighboring cell signal interference intensity map and the neighboring cell signal intensity map, and to determine the handover priority queue of the candidate serving base stations.

[0141] The serving base station module is used to determine the target serving base station according to the handover priority queue and switch the serving base station to the target serving base station.

[0142] The interference prediction device for road test neighboring cell signals provided in this embodiment of the invention can execute the interference prediction method for road test neighboring cell signals provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0143] Example 4

[0144] Figure 4A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0145] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0146] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0147] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as interference prediction methods for drive-test neighboring cell signals.

[0148] In some embodiments, the interference prediction method for drive-side neighboring cell signals can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the interference prediction method for drive-side neighboring cell signals described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the interference prediction method for drive-side neighboring cell signals by any other suitable means (e.g., by means of firmware).

[0149] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0150] In some embodiments, the interference prediction method for drive-side neighboring cell signals can be implemented as a computer program, which is implicitly included in a computer program product. When executed by a processor, the computer program implements the interference prediction method for drive-side neighboring cell signals of the present invention. The computer program product can be understood as a software product that primarily implements its solution through a computer program. The computer program used to implement the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program can be executed entirely on a machine, partially on a machine, partially on a remote machine as a standalone software package, or entirely on a remote machine or server.

[0151] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0152] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0153] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0154] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0155] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0156] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for predicting interference in drive-tested neighboring cell signals, characterized in that, include: Obtain the drive test dataset to be processed; the drive test dataset to be processed includes signal measurement data of the current serving base station and neighboring base stations; The road test dataset to be processed is subjected to dimensionality reduction and feature extraction to obtain an interference feature dataset; The interference feature dataset is reconstructed to obtain a reconstructed road test dataset corresponding to the road test dataset to be processed; The neighboring cell signal interference intensity map of the current serving base station is determined based on the reconstructed drive test dataset.

2. The method according to claim 1, characterized in that, The road test dataset to be processed is subjected to dimensionality reduction and feature extraction to obtain an interference feature dataset; The interference feature dataset is reconstructed to obtain a reconstructed road test dataset corresponding to the road test dataset to be processed; The neighboring cell signal interference intensity map of the current serving base station is determined based on the reconstructed drive test dataset, including: By training a complete hierarchical variational encoder model, the road test dataset to be processed is input into the encoder sub-model in the hierarchical variational encoder model to perform dimensionality reduction and feature extraction on the road test dataset to be processed, thereby obtaining the interference feature dataset; the encoder sub-model includes multiple cascaded encoding layers. The interference feature dataset is input into the decoder sub-model of the hierarchical variational encoder model to reconstruct the interference feature dataset and obtain the reconstructed drive test dataset corresponding to the drive test dataset to be processed; the decoder sub-model includes multiple cascaded decoding layers; The reconstructed drive test dataset is input into the interference intensity decision sub-model in the hierarchical variational encoder model to perform interference intensity decision on the reconstructed drive test data, thereby obtaining the neighboring cell signal interference intensity map of the current serving base station.

3. The method according to claim 2, characterized in that, The training steps for the hierarchical variational encoder model include: Obtain the hierarchical variational encoder model to be trained and the drive test data sample set; the drive test data sample set includes a training sample set and a validation sample set; the training sample set and the validation sample set include at least one sample tuple, the sample tuple including real signal measurement data and real interference intensity of neighboring base stations; The training samples in the training sample set are input into the encoder sub-model of the hierarchical variational encoder model to be trained, and the training sample interference feature dataset output by the encoder sub-model of the hierarchical variational encoder model is obtained; the training sample interference feature dataset is input into the decoding sub-model to obtain the training sample reconstruction data; the training sample reconstruction data is input into the interference intensity decision sub-model to obtain the predicted interference intensity map. The model parameters of the hierarchical variational encoder model are updated based on the training sample set, the training sample reconstructed data, and the predicted interference intensity map, using the maximum evidence lower bound optimization method. The model parameters of the updated hierarchical variational encoder model are frozen, and the validation sample set is input into the hierarchical variational encoder model with frozen model parameters to obtain a validation interference intensity map; the second total loss function value is calculated based on the validation interference intensity map and the validation sample set, and the second total loss function value is fed back to the Bayesian optimizer; Based on the Bayesian optimizer, the current optimal hyperparameters are determined according to the value of the second total loss function; The current optimal hyperparameters are passed to the hierarchical variational encoder model. The process then returns to the step of inputting the training samples from the training sample set into the encoder sub-model of the hierarchical variational encoder model to be trained, obtaining the training sample interference feature dataset output by the encoder sub-model of the hierarchical variational encoder model, and updating the Bayesian optimizer until the iteration termination condition is met, thereby obtaining the target optimal hyperparameters and a fully trained hierarchical variational encoder model with the target optimal hyperparameters.

4. The method according to claim 3, characterized in that, Based on the training sample set, the reconstructed training sample data, and the predicted interference intensity map, the model parameters of the hierarchical variational encoder model are updated using the maximum evidence lower bound optimization method, including: The reconstruction loss function value is determined based on the reconstructed data from the training samples and the actual signal measurement data. The prediction loss function value is determined based on the actual interference intensity map and the predicted interference intensity map; The regularization loss function value is determined based on the posterior distribution of the encoder sub-model and the preset prior distribution; The first total loss function value is determined based on the reconstruction loss function value, the prediction loss function value, and the regularization loss function value. The model parameters of the hierarchical variational encoder model are updated based on the first total loss function value.

5. The method according to claim 3, characterized in that, Based on the Bayesian optimizer, the current optimal hyperparameters are determined according to the second total loss function value, including: Obtain a set of hyperparameters for the hierarchical variational encoder model to be trained, and determine the prior distribution of each hyperparameter; Initialize the Bayesian optimizer; the Bayesian optimizer includes a surrogate model and a collection function, the surrogate model includes a target loss function between the current hyperparameters and the value of the second total loss function; the collection function is the expectation of the improvement between the target loss function of the current hyperparameters and the minimum observation loss function of the current optimal hyperparameters; Based on the Bayesian optimizer, the current optimal hyperparameters are determined according to the acquisition function and the surrogate model.

6. The method according to claim 1, characterized in that, After obtaining the signal interference prediction results from neighboring base stations, the following is also included: Obtain the neighbor cell signal strength map from the drive test dataset to be processed; Based on the neighboring cell signal interference intensity map and the neighboring cell signal intensity map, candidate serving base stations that meet the handover conditions are determined, and the handover priority queue of the candidate serving base stations is determined. The target serving base station is determined according to the handover priority queue, and the serving base station is switched to the target serving base station.

7. An interference prediction device for drive-tested neighboring cell signals, characterized in that, include: The drive test data acquisition module is used to acquire the drive test dataset to be processed; the drive test dataset to be processed includes signal measurement data of the current serving base station and neighboring base stations; The interference intensity prediction module is used to perform dimensionality reduction and feature extraction on the road test dataset to be processed, and obtain the interference feature dataset. The interference feature dataset is reconstructed to obtain a reconstructed road test dataset corresponding to the road test dataset to be processed; The neighboring cell signal interference intensity map of the current service base is determined based on the reconstructed drive test dataset.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the interference prediction method for drive-test neighbor cell signals according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the interference prediction method for drive-test neighbor cell signals according to any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the interference prediction method for drive-test neighbor cell signals as described in any one of claims 1-6.

Images