Household FTTR network self-optimization method and device, medium and equipment

Abstract

The invention discloses a home FTTR network self-optimization method, a home FTTR network self-optimization device, a medium and equipment. The method comprises the following steps: collecting home FTTR network data in real time; constructing a user movement track prediction model, outputting movement probability distribution of the user in a certain time in the future based on data, pre-judging a home area into which the user is about to enter, and dynamically selecting an optimal wireless access point in the area; constructing an optical fiber and wireless collaborative optimization model, performing dynamic adjustment on optical fiber path resources by taking network data and the selected optimal wireless access point as model input, and performing adaptive optimization on wireless parameters of the selected optimal wireless access point according to an adjustment result of the optical fiber path resources; and feeding back the adjustment result of the wireless parameters and the optical fiber path resources of the optimal wireless access point subjected to adaptive optimization to a data acquisition stage, so as to dynamically adjust the acquisition of state data, user behavior data and environment data, thereby realizing the self-optimization of the FTTR network.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of optical fiber communication, and particularly relates to a home FTTR network self-optimization method, device, medium and equipment. BACKGROUND

[0002] With the wide deployment of fiber-to-the-room (FTTR) technology in home networks, users' demand for high-bandwidth, low-latency and high-stability network services is increasing. However, the existing FTTR network system generally lacks dynamic perception ability for user behavior and changes in the home environment, resulting in network topology rigidity, rough wireless access point configuration, and difficulty in accurately responding to user movement and interference changes in resource allocation, making it difficult to achieve true adaptive optimization. Especially in complex home scenarios with multiple users and multiple terminals simultaneously accessing, traditional methods cannot perform coordinated scheduling of wireless and optical fiber resources based on real-time state, limiting the further improvement of FTTR network performance.

[0003] Therefore, there is an urgent need for an FTTR network self-optimization method that integrates user behavior prediction, environment perception and intelligent decision-making to achieve dynamic reconstruction of the network and adaptive configuration of resources. SUMMARY

[0004] In view of the deficiencies in the prior art, the purpose of the present application is to provide a home FTTR network self-optimization method, device, medium and equipment, which aims to realize real-time reconstruction of the home network topology, intelligent allocation of wavelength resources and interference coordination suppression.

[0005] To achieve the above-mentioned purpose, the application provides the following technical solutions: A home FTTR network self-optimization method, the method comprising: collecting state data, user behavior data and environment data of the home FTTR network in real time; constructing a user movement trajectory prediction model, inputting the user behavior data and environment data, outputting the movement probability distribution of the user within a certain time in the future, and according to the movement probability distribution, predicting the home area that the user will enter soon and dynamically selecting the optimal wireless access point in the area; constructing an optical fiber and wireless collaborative optimization model, taking the state data and the selected optimal wireless access point as the model input, dynamically adjusting the optical path resources, and adaptively optimizing the wireless parameters of the selected optimal wireless access point according to the adjustment result of the optical path resources; The wireless parameters of the adaptively optimized optimal wireless access point and the adjustment result of the optical path resources are fed back to the data collection stage to dynamically adjust the collection of state data, user behavior data and environment data, so as to realize the self-optimization of the FTTR network.

[0006] Optionally, the user movement trajectory prediction model comprises: a multimodal feature encoder, an environment constraint encoding module and a trajectory prediction module connected in sequence, wherein the multimodal feature encoder is configured to fuse user behavior data and environment perception information to extract a joint representation with spatial structure and temporal dynamic characteristics; the environment constraint encoding module is configured to model the physical limitations of room layout and environment state on user behavior to form a trainable global constraint vector; and the trajectory prediction module is configured to predict the future movement trajectory of the user based on the environment and behavior joint features.

[0007] Optionally, the multimodal feature encoder comprises: an input layer, a spatio-temporal feature extraction layer and an association layer connected in sequence, wherein the input layer is configured to receive and process the user behavior data and the environment data; the spatio-temporal feature extraction layer is configured to extract spatial features and temporal features from the user behavior data and the environment data, respectively; and the association layer is configured to dynamically associate the spatial features and the temporal features to generate trajectory features with spatial perception capability.

[0008] Optionally, the environment constraint encoding module comprises: a constraint condition input layer, a static constraint mapping branch, a dynamic constraint mapping branch, a constraint fusion mechanism and an output layer, wherein the constraint condition input layer is configured to input the environment data and perform format and feature pre-processing on the environment data; the static constraint mapping branch is configured to perform spatial modeling on static data in the format and feature pre-processed environment data to obtain a static vector; the dynamic constraint mapping branch is configured to encode dynamic data in the format and feature pre-processed environment data to obtain a dynamic high-dimensional feature vector; the constraint fusion mechanism is configured to weight and fuse the static vector and the dynamic high-dimensional feature vector to obtain a fused feature; and the output layer is configured to output the fused feature.

[0009] Optionally, the trajectory prediction module comprises: a feature alignment and fusion layer, a spatio-temporal encoder and a prediction head connected in sequence, wherein the feature alignment and fusion layer is configured to perform dimension alignment and adaptive fusion on the spatially perceived trajectory features and the fused features to obtain fused features; the spatio-temporal encoder is configured to perform global dependency modeling and local feature extraction on the fused features to output an encoded feature sequence; and the prediction head is configured to predict the spatial position distribution of the user at a future time based on the encoded feature sequence.

[0010] Optionally, the user movement trajectory prediction model is trained by the following steps: obtaining historical user behavior data and environment data, constructing multi-modal training samples, and dividing into a training set and a validation set in proportion; setting training parameters, training the model through the training set until the cross-entropy loss function converges; verifying the trained model through the validation set, and in the verification process, when the model test index meets the threshold, the model verification passes; otherwise, adjust the training parameters to retrain the model until the model verification passes.

[0011] Optionally, the fiber and wireless coordination optimization model comprises a fiber resource dynamic allocator and a wireless parameter adaptive optimizer, wherein the fiber resource dynamic allocator is used to analyze the fiber path state, user demand and topology information in real time to dynamically adjust the fiber resource allocation; the wireless parameter adaptive optimizer is used to adjust the transmission power, channel allocation and beamforming parameters of the wireless access point in real time according to the fiber resource allocation result.

[0012] The application also provides a home FTTR network self-optimization device, which comprises: a collection module for collecting state data, user behavior data and environment data of the home FTTR network in real time; a dynamic selection module for constructing a user movement trajectory prediction model, inputting user behavior data and environment data, outputting the movement probability distribution of the user in a certain future time, and predicting the home area that the user will enter and dynamically selecting the optimal wireless access point in the area according to the movement probability distribution; an adjustment and optimization module for constructing a fiber and wireless coordination optimization model, taking the state data and the selected optimal wireless access point as the model input, dynamically adjusting the fiber path resource, and adaptively optimizing the wireless parameters of the selected optimal wireless access point according to the adjustment result of the fiber path resource; and a dynamic adjustment module for feeding back the wireless parameters of the adaptively optimized optimal wireless access point and the adjustment result of the fiber path resource to the data collection stage to dynamically adjust the collection of the state data, user behavior data and environment data, so as to realize the self-optimization of the FTTR network.

[0013] The application also provides a storage medium comprising instructions which, when executed on a computer, cause the computer to perform the method of any one of the preceding claims.

[0014] The application also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of the preceding claims when executing the program.

[0015] Compared with the prior art, the application has the following beneficial effects: The user mobile trajectory prediction model constructed by fusing multi-modal feature coding, environment constraint modeling and trajectory prediction mechanism can realize high-precision modeling of user behavior in the FTTR network and dynamic prediction of future trajectory, effectively improving the intelligent level of network resource allocation and access point configuration. Compared with the traditional static configuration or rule-driven method, the present application can perceive the environmental changes and user positions in real time, accurately predict the area that the user will enter, and dynamically adjust the wireless parameters and fiber resource allocation strategy, thereby significantly enhancing the adaptive optimization capability of the network and improving the overall communication quality and user experience. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a flowchart of a FTTR network self-optimization method provided by an embodiment of the present application; Figure 2 is a structural diagram of a user mobile trajectory prediction model provided by an embodiment of the present application; Figure 3 is a structural diagram of a FTTR network self-optimization device provided by another embodiment of the present application; Figure 4 is a structural diagram of a storage medium provided by another embodiment of the present application; Figure 5 is a structural diagram of an electronic device provided by another embodiment of the present application. DETAILED DESCRIPTION

[0017] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Although specific embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0018] It should be noted that certain terms are used in the specification and claims to refer to certain components. Those skilled in the art will understand that the same component can be referred to by different names. The specification and claims of this specification do not distinguish components based on the difference in name, but rather on the difference in function. As mentioned throughout the specification and claims, "comprising" or "including" is an open term, which should be interpreted as "including but not limited to". The subsequent description in the specification is a preferred embodiment of implementing the present application, which is for the purpose of illustrating the general principles of the specification, and is not intended to limit the scope of the present application. The scope of protection of the present application is defined by the appended claims.

[0019] For the convenience of understanding the embodiments of the present application, the following will be further explained and described with specific embodiments as examples in conjunction with the accompanying drawings, and each drawing does not constitute a limitation on the embodiments of the present application.

[0020] Figure 1 is a flowchart of a self-optimization method of a home FTTR network provided by an exemplary embodiment of the present application, as shown in Figure 1 The method comprises the following steps: S100: collecting state data, user behavior data and environment data of the home FTTR network in real time, wherein the state data includes fiber path state (including bandwidth occupancy, optical signal strength, splitting ratio, fault alarm) and wireless access point state (including signal strength, channel interference, number of connected devices, QoS indicators), the user behavior data includes user moving track and current position, and the environment data includes room layout (grid map or 3D point cloud), obstacle distribution and real-time interference signal.

[0021] S200: constructing a user moving track prediction model, inputting user behavior data and environment data, outputting moving probability distribution of the user within a certain time in the future (for example, within 1 to 5 minutes), and predicting the home area that the user will enter according to the moving probability distribution and dynamically selecting the optimal wireless access point in the area; S300: constructing a fiber and wireless cooperative optimization model, taking the state data and the selected optimal wireless access point as model input, dynamically adjusting the fiber path resources, and adaptively optimizing the wireless parameters (such as transmission power, channel allocation, beamforming) of the selected optimal wireless access point according to the adjustment result of the fiber path resources; S400: feeding back the wireless parameters of the adaptively optimized optimal wireless access point and the adjustment result of the fiber path resources to step S100 to dynamically adjust the collection of state data, user behavior data and environment data, so as to realize the self-optimization of the FTTR network.

[0022] In another exemplary embodiment, as shown in Figure 2 The user moving track prediction model comprises a multimodal feature encoder, an environment constraint encoding module and a track prediction module connected in sequence.

[0023] In this embodiment, the multi-modal feature encoder includes an input layer, a space-time feature extraction layer, and a correlation layer connected in sequence, wherein the input layer is used to input user behavior data and environment data; the space-time feature extraction layer includes a space encoding module and a time sequence encoding module, the space encoding module is used to encode the environment data into a continuous differentiable implicit field, and supports arbitrary coordinate query (the implicit field indirectly represents the shape of an object by defining a scalar value at each point in space, which usually represents the distance from the point to the nearest surface, or indicates whether the point is inside the object (as in the case of occupancy networks), therefore, by querying the scalar value at any coordinate, the positional relationship of the point relative to the object can be determined). Specifically, the space encoding module includes a space basis function layer, a dynamic field update network, and a physical constraint injection layer, wherein the space basis function layer takes the three-dimensional coordinates of the room and learnable spatial anchors (N anchors, each anchor is a trainable parameter) as input, and parameterizes the space through a learnable basis function as shown below, and outputs a low-dimensional space feature vector :

[0024] wherein, is the low-dimensional space feature vector, represents the original space encoding (such as the probability of passing, signal intensity distribution), represents the i th learnable spatial anchor; represents the weight coefficient; represents the smoothing factor; represents the three-dimensional coordinates of the room; N represents the number of spatial anchors.

[0025] Specifically, the space basis function layer parameterizes the spatial layout of the home environment through a learnable basis function (anchor), converts the complex room layout (such as 3D point cloud or grid map) into a low-dimensional feature vector, realizes the continuous differentiable implicit field representation, and directly queries the attributes (such as the probability of passing, signal intensity distribution) of any coordinate point in the space based on the characteristics of the implicit field, without relying on discrete grid or fixed resolution data structure, thereby providing high-precision, dynamically adjustable spatial basis for subsequent user movement prediction and fiber-wireless collaborative optimization.

[0026] The dynamic field update network takes the low-dimensional space feature vector output by the space basis function layer and real-time sensor data The dynamic field updating network takes, as input, dynamic information (e.g., new obstacle coordinates, temperature and humidity, etc.), and specifically, the dynamic field updating network comprises, in sequence, a concatenation layer, a hidden layer 1 (fully connected layer + activation function ReLU), a hidden layer 2 (fully connected layer + activation function ReLU), and an output layer (fully connected layer + nonlinear transformation), wherein the concatenation layer is configured to concatenate the low-dimensional spatial feature vector and real-time sensor data to obtain a spatial feature vector , i.e.,

[0027] wherein denotes a vector concatenation operation, and the output dimension is d+k .

[0028] The hidden layer 1 is responsible for multi-modal data fusion and primary nonlinear transformation, and solves "what has changed". The hidden layer 1 takes the spatial feature data as input, and introduces nonlinearity through the activation function ReLU to capture complex patterns of environmental changes (such as "signal attenuation due to rising temperature and humidity") from the spatial feature data . The hidden layer 1 is specifically represented as follows:

[0029] wherein denotes the preliminary fused features output by the hidden layer 1, the dimension is , and is used to retain more information; denotes the weight matrix of the hidden layer 1, the dimension is , and the input is mapped to a hidden space.

[0030] The hidden layer 2 is responsible for deepening feature abstraction and dynamic adjustment, and solves "how does the change affect the whole". The hidden layer 2 takes the output of the hidden layer 1 as input, and further extracts high-order features therefrom, for example: abstracting "regional passability" from "local obstacle coordinates", or inferring "future interference trend" from "signal strength".

[0031] The hidden layer 2 is specifically represented as follows:

[0032] wherein denotes the output of the hidden layer 2, the dimension is , and ; denotes the weight matrix of the hidden layer 2, the dimension is ; denotes the bias vector, the dimension is .

[0033] It should be noted that the high-order features output by the hidden layer 2 have a dimension lower than , so that the output layer focuses on key information.

[0034] In summary, the hidden layer 1 is used for coarse-grained fusion, and the hidden layer 2 is used for fine-grained adjustment. For example, for the user movement prediction scenario, the hidden layer 1 can find that “an obstacle is added to the right side of the dining table”, and the hidden layer 2 can determine that “the weight of the right side of the dining table needs to be reduced, but the left side path is kept” by combining the user historical trajectory. Through hierarchical nonlinear mapping, both can convert the original environment data into high-value features that can drive the field update, guarantee the real-time performance of network self-optimization, for example, when the user approaches the newly added obstacle, the dynamic field update network can immediately adjust the implicit field to reduce the signal strength weight in this area, and ensure that the model can respond to environmental changes in real time.

[0035] The output layer depends on the high-order features output by the hidden layer 2 , completes the final adjustment, and outputs the updated field parameters , for example: if the hidden layer 2 outputs “obstacle influence is significant”, the output layer will greatly correct the original space field parameters; if the hidden layer 2 outputs “change is negligible”, the output layer will fine-tune or keep the original parameters, and the output layer is represented as follows:

[0036] wherein represents the updated field parameters, which have the same dimension as the low-dimensional space feature vector ; represents the weight matrix of the output layer, which has a dimension of , and maps the features back to the original space dimension; represents the bias vector, which has a dimension of , and is used to fine-tune the output. The physical constraint injection layer takes the updated field parameters and the binary mask M as input, and forces the obstacle coordinates (forces the obstacle area to have a 0 passage probability) through the function as shown below:

[0037] wherein is the adjusted field parameters, representing the passage probability or signal intensity distribution of each coordinate point in the space, and the obstacle area is forced to be 0; represents the activation function, which is used to map the updated field parameters ​Compress to [0,1]; ⊙ represents element-wise multiplication; M represents a binary mask (0 represents the obstacle area, 1 represents the passable area), used to force the passability probability of the obstacle position to be 0.

[0038] By introducing a physical constraint injection layer, it is possible to ensure that the model prediction and optimization results conform to the actual physical constraints, thereby improving the system reliability. For example, when predicting the user's movement trajectory, the physical constraint injection layer directly masks the coordinates of obstacles, avoiding the generation of paths through walls.

[0039] In summary, the spatial coding module, through parametric modeling, dynamic updating, and the fusion of physical constraints, abstracts the complex home environment into a high-precision, low-dimensional implicit field representation, thereby laying a reliable spatial perception foundation for the real-time self-optimization (such as resource allocation and interference suppression) of the home FTTR network.

[0040] The timing encoding module uses the user's historical trajectory sequence. (time interval) The system uses short-term memory (STM) as input to model short-term dynamics (such as turning) and long-term habits (such as daily routes) of user movement. The temporal coding module includes a short-term memory layer, a long-term memory layer, a causal attention mechanism, and a fusion layer. The STM layer extracts recent trajectory features of the user through sliding window convolution (convolution kernel: 1D temporal convolution, kernel size = 5 (last 5 steps trajectory)). (Such as acceleration, changes in direction), the long-term memory layer uses a key-value memory matrix. (Key: Location area (e.g., "living room-dining room corridor"), Value: User's statistical behavior in that area (e.g., average speed, frequently used routes)) to obtain user habit characteristics. The user habit features Specifically, it is expressed as follows:

[0041] in, Indicates characteristics of long-term memory; Represents a memory matrix, storing... m 10 long-term behavioral patterns, each pattern using d 3D vector representation; Indicates the first Each key encodes location region features (e.g., "living room-dining room corridor"). Indicates the first Each value stores the user's statistical behavior in this area (such as average speed and frequently used routes). This represents the query vector, used to retrieve relevant memories. This represents the normalized weight, used to calculate the similarity between the query and each key; the weights sum to 1.

[0042] The causal attention mechanism is used for screening trajectory segments related to the current state in the history and generating context features , the context features are specifically represented as follows:

[0043]

[0044] wherein, represents the context features output by the causal attention mechanism; represents the hidden state of the current time step ; represents the hidden state of the historical time step , used for storing past trajectory information; represents a learnable weight matrix, which calculates the relevance of the current state and the historical state; represents the attention weight, which represents the importance of the historical state to the current prediction; represents the normalization along the historical time.

[0045] The fusion layer is used for fusing the outputs of the short-term memory layer, the long-term memory layer, and the causal attention mechanism through a perception MLP to obtain time sequence features , the time sequence features are specifically represented as follows:

[0046] In summary, the time sequence encoding module analyzes the user historical trajectory data, combines short-term dynamics (such as acceleration, direction change) and long-term habits (such as regional stay preference, commonly used path), uses a sliding window convolution to extract recent behavior features, stores long-term behavior patterns (such as average speed, position area statistics) through a key-value memory matrix, and introduces a causal attention mechanism to screen historical segments related to the current state. Finally, high-precision time sequence features are generated by fusion. The effect is to accurately model the user movement rule, enhance the confidence of future trajectory prediction (such as predicting the area the user will enter soon), thereby dynamically optimizing the selection of wireless access points and resource allocation, and significantly improving the real-time response ability and service quality of the FTTR network.

[0047] The association layer introduces a cross-attention mechanism to dynamically associate the outputs of the spatial encoding module and the time sequence encoding module. The cross-attention mechanism takes the time sequence features as the query vector, and the adjusted field parameters as the key or value vector, and outputs trajectory features with spatial perception ability , which is specifically represented as:​

[0048] wherein, represents a cross-attention mechanism.

[0049] The association layer generates user future moving path prediction by fusing space-time features through a cross-attention mechanism, contains a space-aware trajectory heat map (such as an area that the user is about to enter, a path that bypasses an obstacle), and can provide dynamic behavior guidance for subsequent collaborative optimization models, such as pre-allocating optimal wireless access points for high-probability areas or adjusting signal coverage direction.

[0050] The environment constraint encoding module includes a constraint condition input layer, a static constraint mapping branch, a dynamic constraint mapping branch, a constraint fusion mechanism, and an output layer. The constraint condition input layer is used to input environment data (including static data and dynamic data) and perform format and feature pre-processing on the environment data. Specifically, for the static data (such as room layout) in the environment data, a gridding processing method is used to convert the planar structure diagram of the room layout into a binary matrix G (0 = passable, 1 = obstacle); for the dynamic data (such as temperature and humidity, new obstacle coordinates) in the environment data, normalization processing is first performed, for example, the temperature is standardized to the interval [0, 1], and then all dynamic data is spliced into a dynamic data vector .

[0051] The static constraint mapping branch is used to perform spatial modeling on the static data in the format and feature pre-processed environment data to obtain a static vector. Specifically, the static constraint mapping branch takes the gridded binary matrix G as input and processes the gridded binary matrix G through a lightweight CNN (2 layers of convolution (extracting local spatial features) + 1 layer of full connection (extracting global abstract features)) to obtain a spatial graph feature (e.g., "living room-kitchen path is clear"), and further compresses the spatial graph feature into a static vector through global averaging.

[0052] The dynamic constraint mapping branch takes the dynamic data vector as input and maps the dynamic data vector to a high-dimensional space through a full connection layer to generate a dynamic high-dimensional feature vector (e.g., "current location has temporary obstacles").

[0053] The constraint fusion mechanism is used to perform weighted fusion on the static vector and the dynamic high-dimensional feature vector to obtain a fusion feature and output through the output layer. The fusion feature is represented as follows:

[0054] wherein, denotes the weight, which is generated by learnable parameters or attention mechanism.

[0055] In summary, the environment constraint encoding module contains static environment constraints (such as room layout obstacles), dynamic environment changes (such as newly added obstacles, temperature and humidity), and user behavior time sequence features (such as moving habits), which can provide global physical restrictions and user behavior background for subsequent collaborative optimization models, for example, to ensure that resource allocation does not violate actual environmental restrictions (such as avoiding wall areas).

[0056] The trajectory prediction module includes a feature alignment and fusion layer, a space-time encoder, and a prediction head connected in turn, wherein the feature alignment and fusion layer takes the spatially perceived trajectory feature and the fused feature as input, first maps the two features to the same dimension through a fully connected layer, then adaptively fuses the two features mapped to the same dimension through a gating mechanism, and element-wise adds the space-time position encoding (including the combined encoding of the timestamp and the spatial coordinates (such as room three-dimensional coordinates)) to form the fused feature , which is a joint representation containing environmental constraints and user behavior. The space-time encoder includes a multi-head attention layer, a space-time convolution layer, and a layer normalization layer, which are connected in turn to form a complete path for space-time feature modeling. Among them, the multi-head attention layer takes the fused feature as input and models the global dependency relationship in the feature, thereby enhancing the information interaction capability between time steps based on the time sequence feature. Specifically, the multi-head attention layer processes the feature through multiple parallel attention heads (each head has independent Query, Key, and Value weights), and after concatenating the output of each attention head, a linear transformation is performed to obtain the final output of the multi-head attention :

[0057] wherein, denotes the context-enhanced feature representation, which integrates the long-term dependency relationship between different time steps in the time series and is the input basis of the space-time convolution.

[0058] The space-time convolution layer takes the context-enhanced feature representation as input, captures local time and spatial features through a three-dimensional convolution kernel Conv3D, and finally obtains local dynamic behavior features .

[0059] The normalization layer is used to capture the local dynamic behavior features captured by the spatio-temporal convolution layer Normalization is performed to improve model training stability and speed up convergence, and finally output an encoding feature sequence containing context semantics and spatio-temporal structure The feature sequence contains global attention context-dependent features and local spatio-temporal convolution features of the input sequence, which can provide high-quality input for the subsequent prediction head module.

[0060] The prediction head includes an input layer, a processing layer, and an output layer. The input layer is used to input the feature sequence output by the spatio-temporal encoder The processing layer includes two stacked LSTM networks, which receive the feature sequence Through the internal memory unit and forget gate, input gate, output gate, etc., the LSTM network can effectively capture the temporal dependency in the feature sequence In addition, a Dropout layer is connected after each LSTM network to prevent overfitting. The output layer uses a fully connected layer, which contains a number of neurons depending on the number of target locations or regions to be predicted. For example, if the user's movement between three rooms is to be predicted, 3 neurons are set, each corresponding to a possible location or region. The output of this layer is processed by the Softmax function and converted into a probability distribution, representing the likelihood of the user appearing in each location or region. For example, the prediction head predicts that the user has an 80% probability of entering room B, a 15% probability of entering room A, and a 5% probability of entering room C within the next minute. Such prediction results can provide strong support for the self-optimization of the home FTTR network, such as dynamically selecting the optimal wireless access point and adjusting the signal coverage direction.

[0061] The prediction head can efficiently extract information from spatio-temporal features and accurately predict user movement by combining the powerful sequence modeling capability of the LSTM network and the effective regularization method of the Dropout technique.

[0062] In another example embodiment, the fiber and wireless collaborative optimization model comprises a fiber resource dynamic allocator and a wireless parameter adaptive optimizer, wherein the fiber resource dynamic allocator is configured to analyze fiber path status (bandwidth occupancy, optical signal strength, splitting ratio), user demand (QoS indicator, device number) and topology information (fiber node location, link connection relationship) in real time to dynamically adjust fiber resource allocation. The fiber resource dynamic allocator comprises a state perception layer, a multi-objective optimization calculation layer and a dynamic adjustment module, wherein the state perception layer is configured to input fiber path status (bandwidth occupancy, optical signal strength, splitting ratio), user demand (QoS indicator, device number) and topology information (fiber node location, link connection relationship) and integrate them into a state vector S. The multi-objective optimization calculation layer calculates an optimal solution according to a preset objective function as shown below to enable users with high latency or insufficient bandwidth to be allocated bandwidth in priority:

[0063] wherein, , , respectively represent weight coefficients for balancing the priority of different optimization objectives; represents the proportion of users meeting the demand, such as 90% of users with latency ; represents the bandwidth of the link that is not effectively utilized, with the unit being Mbps; represents the splitting ratio adjustment range of the optical splitter .

[0064] The dynamic adjustment module adopts a hybrid DBA strategy as shown below to balance the bandwidth demand of different service types in the network, maximize the utilization of fiber resources, and ensure the quality of service (QoS) of critical services:

[0065] wherein, represents the bandwidth guarantee value of the user , such as 8 Mbps for a video conference; represents the proportion of remaining bandwidth allocation, such as represents the allocation of 80% of the remaining bandwidth; represents the actual latency of the user ; represents the maximum allowed latency of the user , such as the requirement of a video conference ; represents the minimum bandwidth guarantee of the user , such as 50 kbps for a voice call.

[0066] ​The hybrid DBA strategy combines both fixed bandwidth allocation and dynamic bandwidth allocation. In the fixed bandwidth allocation, a fixed time slot is reserved for high priority traffic, for example, 8 Mbps is allocated for each video conference user regardless of whether it is active or not, thus ensuring that its bandwidth requirement is always met. In the dynamic bandwidth allocation, after the fixed bandwidth allocation, the remaining bandwidth is allocated dynamically according to real-time traffic demand to optimize the experience of non-real-time traffic, for example, when a user initiates a file download request, the ONU reports the bandwidth demand to the OLT, and the OLT allocates a temporary time slot according to the priority and current load. The optical fiber resource dynamic allocator can finally generate an optical fiber resource allocation matrix and the remaining optical fiber bandwidth, and pass it to the wireless parameter adaptive optimizer.

[0067] The wireless parameter adaptive optimizer is used to adjust the transmission power, channel allocation and beamforming parameters of the wireless access point in real time according to the optical fiber resource allocation results. The wireless parameter adaptive optimizer receives the output of the optical fiber resource dynamic allocator and constructs a lotus optimization input vector X combining with the wireless environment parameters (channel quality, used for mobility). Secondly, the wireless parameter adaptive optimizer can dynamically adjust the transmission power of the AP according to the remaining optical fiber bandwidth :

[0068] wherein, , represents the minimum and maximum value of the transmission power of the AP, respectively, in dBm; represents the remaining optical fiber bandwidth, in Mbps; represents the total optical fiber bandwidth, such as 10 Gbps.

[0069] When the optical fiber resources are sufficient, the transmission power is increased to enhance the coverage; when the resources are tight, the power is reduced to save resources.

[0070] In addition, the wireless parameter adaptive optimizer jointly optimizes the channel allocation and the modulation order according to the following formula:

[0071] wherein, represents the signal-to-noise ratio of the AP; represents the modulation order; represents the interference power of the channel to ; represents the spectral efficiency corresponding to the modulation order.

[0072] High SNR channel allocation high order modulation (such as 64QAM), low SNR channel uses QPSK to reduce the bit error rate.

[0073] The wireless parameter adaptive optimizer also dynamically adjusts the beam direction vector through the DRL algorithm :

[0074] wherein, represents the beam direction vector; represents a regularization coefficient, such as 0.1, for preventing excessive beam gain from causing interference; represents the signal-to-noise ratio of the user .

[0075] Finally, the optimized wireless parameters (transmission power , channel allocation table, beam direction vector ) are issued to each AP to adjust the wireless network configuration in real time.

[0076] In addition, the packet loss rate PLR, delay D wireless , and other indicators of the wireless link are monitored in real time. If it is found that the QoS is not up to standard (such as PLR> PLR threshold ), the ORDA is triggered to re-allocate the fiber resources. If the wireless link causes the QoS to drop due to interference or insufficient coverage, the ORDA can dynamically release part of the fiber bandwidth to the wireless network, or adjust the splitting ratio to enhance the light signal strength. According to the change in the priority of the service (such as the demand for burst video stream), the weight coefficients , , of the optimization objective function are dynamically adjusted to ensure the priority of the key service.

[0077] For example, when the fiber side detects that a user requests a high-definition video stream, the ORDA preferentially allocates the guaranteed bandwidth and releases the remaining bandwidth. The WPAO allocates the remaining fiber bandwidth to the wireless network, increases the AP transmission power , and enables 64QAM modulation. The DRL adjusts the beamforming direction to focus on the area where the user is located, reducing interference. If the wireless link causes the SNR to drop due to user movement, the WPAO triggers the ORDA to re-allocate the fiber resources, or reduces the modulation order to ensure reliability.

[0078] In another exemplary embodiment, in step S200, the user movement trajectory prediction model is trained through the following steps: S201: Obtain historical user behavior data and environmental data, construct multi-modal training samples, and divide them into a training set and a validation set in proportion, for example, 7:3; S202: Set the training parameters, for example, set the learning rate to 0.001, the batch size to 32, and the number of training rounds to 100, and iteratively train the model through the training set until the constructed multi-classification cross-entropy loss function converges; S203: Validate the trained model using a validation set. During the validation process, the model is validated when the prediction accuracy and average position error meet the thresholds (e.g., prediction accuracy greater than or equal to 85%, and error less than or equal to a specified number of pixels, such as 30 pixels). Otherwise, the model is returned to the training phase, and the training parameters are adjusted (e.g., the learning rate is adjusted to 0.005, or the training parameter is adjusted to 200) to retrain the model until the model is validated.

[0079] In another exemplary embodiment, the multi-class cross-entropy loss function is expressed as follows:

[0080]

[0081] in, This represents the multi-class cross-entropy loss value, used to measure the overall difference between the predicted trajectory and the actual trajectory; This indicates the time step of the predicted sequence, i.e., the number of steps to predict the future trajectory; Indicates the total number of categories (e.g., number of rooms or grids) of candidate regions or locations for the trajectory. Indicates at time step Next, the The weight factor of a class reflects its importance in time and space; Indicates the true label at time step Is it in the first The probability of each position; Indicates the model at time step Predicting the user belongs to the first The probability of each position; Represents a numerical stability constant to prevent occur; This represents a hyperparameter used to control the balance between spatial and temporal weights; Indicates the first The spatial priority of each location; Indicates the first The importance of each time step on the timeline.

[0082] The proposed multi-class cross-entropy loss function fully integrates the importance of spatial location with the dynamic weight information of prediction time points, exhibiting high adaptability and scene awareness. This is achieved by introducing a spatiotemporal weight factor. The function not only gives higher punishment to the user's resident area or key network nodes (such as main rooms, AP coverage blind area), but also dynamically adjusts the error sensitivity according to the proximity of the predicted time, so as to realize accurate modeling of key positions and key moments. Compared with the traditional cross-entropy loss, the function is more suitable for the user trajectory prediction requirements in the FTTR network of the family, can significantly improve the response ability and prediction stability of the model to complex behaviors in the real scene, and effectively enhance the forward-looking and intelligent degree of network resource allocation.

[0083] In another exemplary embodiment, the present application also provides a self-optimization device for a FTTR network of a family, as shown in Figure 3 As shown, the device comprises: a collection module 100 for collecting state data, user behavior data and environment data of the FTTR network of the family in real time; a dynamic selection module 200 for constructing a user moving trajectory prediction model, inputting the user behavior data and the environment data, outputting the moving probability distribution of the user in a certain future time, and predicting the family area that the user will enter soon according to the moving probability distribution, and dynamically selecting the optimal wireless access point in the area; an adjustment and optimization module 300 for constructing a fiber and wireless cooperative optimization model, taking the state data and the selected optimal wireless access point as the model input, dynamically adjusting the fiber path resources, and adaptively optimizing the wireless parameters of the selected optimal wireless access point according to the adjustment result of the fiber path resources; a dynamic adjustment module 400 for feeding back the wireless parameters of the adaptively optimized optimal wireless access point and the adjustment result of the fiber path resources to the data collection stage, so as to dynamically adjust the collection of the state data, the user behavior data and the environment data, so as to realize the self-optimization of the FTTR network.

[0084] In the following, the present application will be exemplarily described by combining specific data.

[0085] In a FTTR network system deployed in a 120 square meter family residence, the house contains 4 main areas: living room (area A), kitchen (area B), master bedroom (area C) and secondary bedroom (area D), and each area is provided with a wireless access point (AP). When the user is active at home, the terminal device (such as a mobile phone, AR glasses) generates continuous trajectory data, and the system collects the historical trajectory sequence, the room layout graph (binarized into a 100x100 binary matrix), and the environment dynamic data (such as room temperature 26°C, temporary obstacles in the kitchen, etc.).

[0086] The system predicts the user's activity path in the next 60 seconds through the trained multi-modal trajectory prediction model. The current time t is 10:30:00, and the user has just moved from the living room to the kitchen. After inputting the following information: historical trajectory sequence: [(85, 60), (87, 62), (89, 64), (91, 66)] (units: pixels); environment state: the kitchen area obstacle is marked as 1, and the others are 0; the current temperature is normalized to 0.72, and the new obstacle position coordinates are (95, 68); the prediction target is which room area (A / B / C / D) the user will appear in at the next time.

[0087] The model output prediction probability is shown in Table 1: Table 1

[0088] The true label is B (kitchen), i.e. .

[0089] Considering the spatial weight at this position , the time step weight , set , then:

[0090] By bringing in the loss function calculation, we can get:

[0091] By dynamically adjusting the loss weight, the system pays more attention to the prediction accuracy of key areas (such as the kitchen), and optimizes the wireless access point configuration accordingly. The system then increases the transmission power of the kitchen AP from 12 dBm to 16 dBm, switches to the least interference channel 6, and reduces the signal priority in the living room area, achieving dynamic scheduling of network resources.

[0092] Through the above process, it is not difficult to find that based on the scheme described in the present application, not only the area that the user may enter is predicted in advance, but also the wireless parameter configuration is optimized in real time based on the prediction result, realizing the coordinated improvement of network performance and service experience. Compared with the static strategy, the overall bandwidth utilization rate of the system is improved by about 15%, and the user access delay is reduced by an average of 20ms, verifying the practicality and intelligence of the technical scheme in real scenarios.

[0093] Based on the above embodiment, with reference to Figure 4 , the computer-readable storage medium of the exemplary embodiment of the present application is described, please refer to Figure 4The computer readable storage medium shown is an optical disc 40, and a computer program (i.e., a program product) is stored on the optical disc 40, which, when executed by a processor, implements each step described in the above method embodiments, for example, collecting state data, user behavior data and environment data of a home FTTR network in real time; constructing a user movement trajectory prediction model, inputting user behavior data and environment data, outputting a movement probability distribution of the user within a certain time in the future, and predicting a home area that the user will enter according to the movement probability distribution, and dynamically selecting an optimal wireless access point in the area; constructing a fiber and wireless cooperative optimization model, taking the state data and the selected optimal wireless access point as model inputs, to dynamically adjust fiber path resources, and adaptively optimize wireless parameters of the selected optimal wireless access point according to the adjustment result of the fiber path resources; and feeding back the wireless parameters of the adaptively optimized optimal wireless access point and the adjustment result of the fiber path resources to the data collection stage, to dynamically adjust the collection of the state data, user behavior data and environment data, to realize self-optimization of the FTTR network. The specific implementation methods of each step are not repeated here.

[0094] It should be noted that the computer readable storage medium includes, but is not limited to, a phase change memory (PRAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), other types of random access memory (RAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, or other optical or magnetic storage medium, which will not be repeated here.

[0095] On the basis of the above embodiments, the present application further provides an electronic device, which will be described below with reference to Figure 5 An electronic device for file download according to an exemplary embodiment of the present application is described.

[0096] Figure 5 A block diagram of an exemplary electronic device 50 suitable for implementing an embodiment of the present application is shown, which can be a computer system or a cloud server. Figure 5 The electronic device 50 shown is only an example, and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0097] As shown in Figure 5 The electronic device 50 includes, but is not limited to, one or more processors or processing units 501, a system memory 502, and a bus 503 connecting different system components (including the system memory 502 and the processing unit 501).

[0098] Electronic device 50 typically includes a variety of computer system readable media. These media can be any available media that is located either internally or externally to electronic device 50, including both volatile and nonvolatile media, removable and non-removable media.

[0099] System memory 502 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 5021 and / or cache memory 5022. Electronic device 50 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, ROM 5023 can be used to read-only memory, such as a magnetic disk drive (not shown in Figure 5 FIG. 5), although the drives represented in FIG. 5 are merely examples and are not intended to limit the scope of possible storage media types that can be used in electronic device 50. Further, it should be appreciated by those skilled in the art that other types of computer system storage media that can be used in the operation of electronic device 50 can comprise, but are not limited to, magnetic cassettes, DVDs, tapes, cartridges, MEMs, optical media such as CD-ROMs, and / or solid state media such as flash memory, among others. It is considered that these types of media can be nonvolatile storage media. Accordingly, the disk drives illustrated in FIG. 5 represent the disk storage devices in a tangible, computer system storage medium configuration employed in electronic device 50. Figure 5 Although not shown in FIG. 5, it is contemplated that some embodiments of electronic device 50 can employ other types of data storage media as appropriate, while the same are contemplated to be within the spirit and scope of the present application. In this regard, various implementations can present a number of components, elements, and / or modules. These are illustrated as a collection of distinct components, elements and / or modules in order to more particularly emphasize their make-up. However, it should be understood that such components, elements and / or modules can be combined into more

[0100] Program / utility 5025, having a set (at least one) of program modules 5024, can be stored in, for example, system memory 502 and implemented or accessed by electronic device 50. It is appreciated that each of the program modules 5024 includes, and / or is implemented with, one or more programs and / or program modules, which are configured to carry out the functions of embodiments of the application. Program / utility 5025 can also include, for example, device drivers, operating system binaries, applications, and / or other code.

[0101] Electronic device 50 can also communicate with one or more external devices 504 such as a keyboard or pointing device, a display 505, etc.; through Input / Output (I / O) interface(s) 505. Furthermore, electronic device 50 can communicate with one or more networks such as a local area network (LAN), a general wide area network (WAN), and / or a public network such as the Internet; through network adapter 506. As Figure 5 illustrated, network adapter 506 communicates with the other components of electronic device 50 via bus 503. It should be appreciated that although not shown, other hardware and / or software components that are ​ well known in the art can be utilized in conjunction with electronic device 50. For example, hardware and / or software components such as a keyboard and / or mouse and / or other input device 509 can be utilized with electronic device 50.

[0102] The processing unit 501 performs various function applications and data processing by running programs stored in the system memory 502, such as collecting state data, user behavior data and environment data of the FTTR network in real time; constructing a user movement trajectory prediction model, inputting user behavior data and environment data, outputting the movement probability distribution of the user in a certain future time, and predicting the home area that the user will enter according to the movement probability distribution, and dynamically selecting the optimal wireless access point in the area; constructing a fiber and wireless cooperative optimization model, taking the state data and the selected optimal wireless access point as the model input, dynamically adjusting the fiber path resources, and adaptively optimizing the wireless parameters of the selected optimal wireless access point according to the adjustment result of the fiber path resources; feeding back the wireless parameters of the adaptively optimized optimal wireless access point and the adjustment result of the fiber path resources to the data collection stage to dynamically adjust the collection of state data, user behavior data and environment data, so as to realize the self-optimization of the FTTR network. The specific implementation of each step will not be repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the file concurrent download device are mentioned in the foregoing detailed description, such division is only exemplary and not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into a plurality of units / modules.

[0103] In the description of the present application, it should be noted that the terms "first", "second", "third" are only for the purpose of description and cannot be understood or implied as indicating or implying relative importance.

[0104] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0105] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some communication interfaces, devices or units, and can be electrical, mechanical or other forms.

[0106] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0107] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present alone, or two or more units can be integrated in one unit.

[0108] If the functions are realized in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present application or the part of the present application that essentially contributes to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for executing all or part of the steps of the method described in each embodiment of the present application by a computer device (which can be a personal computer, a cloud server, or a network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, and various program code storage media.

[0109] The above embodiments are only for illustrating the technical concept and characteristics of the present application, and the purpose is to enable those skilled in the art to understand the content of the present application and to implement it, and cannot limit the protection scope of the present application. Any equivalent changes or modifications made according to the spirit and essence of the present application shall be covered within the protection scope of the present application.

Claims

1. A self-optimization method for home FTTR networks, characterized in that, The method includes: Real-time collection of status data, user behavior data, and environmental data from home FTTR networks; A user movement trajectory prediction model is constructed. User behavior data and environmental data are input, and the probability distribution of user movement within a certain period of time is output. Based on the probability distribution of movement, the home area that the user is about to enter is predicted, and the optimal wireless access point in the area is dynamically selected. A fiber optic and wireless collaborative optimization model is constructed, using the aforementioned state data and the selected optimal wireless access point as model inputs, to dynamically adjust fiber optic path resources, and to adaptively optimize the wireless parameters of the selected optimal wireless access point based on the adjustment results of fiber optic path resources. The adaptively optimized wireless parameters and fiber optic path resources of the optimal wireless access point are fed back to the data acquisition stage to dynamically adjust the collection of status data, user behavior data, and environmental data, thereby achieving self-optimization of the FTTR network.

2. The self-optimization method for a home FTTR network according to claim 1, characterized in that, The user movement trajectory prediction model includes: The multimodal feature encoder, environmental constraint encoding module, and trajectory prediction module are connected in sequence, among which... The multimodal feature encoder is used to fuse user behavior data and environmental perception information to extract a joint representation with spatial structure and temporal dynamic features; The environmental constraint encoding module is used to model the physical constraints of room layout and environmental conditions on user behavior in order to form a trainable global constraint vector. The trajectory prediction module is used to predict the user's future movement trajectory based on the combined features of environment and behavior.

3. The self-optimization method for a home FTTR network according to claim 2, characterized in that, Multimodal feature encoders include: The input layer, spatiotemporal feature extraction layer, and association layer are connected sequentially, among which, The input layer is used to receive and process user behavior data and environmental data; The spatiotemporal feature extraction layer is used to extract spatial and temporal features from user behavior data and environmental data, respectively. The association layer is used to dynamically associate spatial and temporal features to generate trajectory features with spatial awareness capabilities.

4. The self-optimization method for a home FTTR network according to claim 2, characterized in that, The environmental constraint coding module includes: The constraint input layer, static constraint mapping branch, dynamic constraint mapping branch, constraint fusion mechanism, and output layer are as follows: The constraint input layer is used to input environmental data and perform formatting and feature preprocessing on the environmental data; The static constraint mapping branch is used to perform spatial modeling on the static data in the formatted and characterized preprocessed environmental data to obtain static vectors. The dynamic constraint mapping branch is used to encode the dynamic data in the formatted and characterized preprocessed environmental data to obtain dynamic high-dimensional feature vectors. The constraint fusion mechanism is used to weightedly fuse static vectors and dynamic high-dimensional feature vectors to obtain fused features; The output layer is used to output fused features.

5. A self-optimization method for a home FTTR network according to claim 4, characterized in that, The trajectory prediction module includes: The feature alignment and fusion layer, the spatiotemporal encoder, and the prediction head are connected sequentially, where, The feature alignment and fusion layer is used to perform dimensional alignment and adaptive fusion of spatially perceived trajectory features and fused features to obtain fused features; The spatiotemporal encoder is used to perform global dependency modeling and local feature extraction on the fused features, and outputs an encoded feature sequence. The prediction head is used to predict the spatial location distribution of users at future times based on encoded feature sequences.

6. The self-optimization method for a home FTTR network according to claim 1, characterized in that, The user movement trajectory prediction model is trained through the following steps: Acquire historical user behavior data and environmental data, construct multimodal training samples, and divide them into training set and validation set according to the ratio; Set the training parameters and train the model using the training set until the cross-entropy loss function converges. The trained model is validated using a validation set. During the validation process, if the model's test metrics meet the threshold, the model is validated; otherwise, the training parameters are adjusted and the model is retrained until it is validated.

7. A self-optimization method for a home FTTR network according to claim 1, characterized in that, The fiber optic and wireless collaborative optimization model includes: Fiber optic resource dynamic allocator and wireless parameter adaptive optimizer, among which, The fiber optic resource dynamic allocator is used to analyze fiber optic path status, user demand, and topology information in real time to dynamically adjust fiber optic resource allocation. The wireless parameter adaptive optimizer is used to adjust the transmit power, channel allocation, and beamforming parameters of the wireless access point in real time based on the fiber optic resource allocation results.

8. A home FTTR network self-optimization device, characterized in that, The device includes: The data acquisition module is used to collect real-time status data, user behavior data, and environmental data of the home FTTR network. The dynamic selection module is used to build a user movement trajectory prediction model. It takes user behavior data and environmental data as input, outputs the user's movement probability distribution over a certain period of time, and predicts the home area that the user is about to enter based on the movement probability distribution, and dynamically selects the optimal wireless access point in that area. The adjustment and optimization module is used to construct a fiber optic and wireless collaborative optimization model. The state data and the selected optimal wireless access point are used as model inputs to dynamically adjust the fiber optic path resources and adaptively optimize the wireless parameters of the selected optimal wireless access point based on the adjustment results of the fiber optic path resources. The dynamic adjustment module is used to feed back the adjustment results of the wireless parameters and fiber optic path resources of the optimal wireless access point after adaptive optimization to the data acquisition stage, so as to dynamically adjust the collection of status data, user behavior data and environmental data to achieve self-optimization of the FTTR network.

9. A storage medium, characterized in that, It includes instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, The electronic device includes: Memory, processor, and computer programs stored in memory and executable on the processor, wherein, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

Images