Wireless IoT networking method based on star flash technology

Abstract

This invention relates to the field of wireless Internet of Things (IoT) technology, and provides a wireless IoT networking method based on StarFlash technology. The method includes: collecting a network topology set of a target IoT area using a StarFlash networking control device; encoding and generating identifier groups for IoT devices to be connected; optimizing the signal coverage of the network topology set to generate an optimized set, and then stitching them together to form a regional network topology map; generating an original network area model using a pre-trained network information generation model; acquiring model information of IoT devices and filling the original model with the encoded identifier groups to obtain the target network area model; controlling the networking devices based on this model to complete the networking, collecting target network data and performing connection verification, and achieving closed-loop management by marking devices that fail the verification. This invention can improve networking efficiency, enhance signal coverage, and ensure the reliability of device connections.

Description

Technical Field

[0001] This invention relates to the field of wireless Internet of Things (IoT) technology, and more specifically, to a wireless IoT networking method based on StarFlash technology. Background Technology

[0002] With the rapid development of IoT technology, wireless IoT networking technology has been widely used in many fields such as smart homes, smart cities, and industrial automation. Traditional wireless IoT networking technologies, such as Wi-Fi and Bluetooth, while meeting the communication needs between devices to a certain extent, have gradually revealed many problems when facing large-scale, complex IoT device networking environments. For example, Wi-Fi is prone to signal congestion when multiple devices are connected, leading to increased transmission latency; Bluetooth has limited transmission distance and a limited number of connected devices. These technologies mostly adopt star or mesh topologies, which, while enabling device connectivity in simple scenarios, lack flexibility in complex IoT environments. They are difficult to dynamically adjust according to device distribution and communication needs, thus affecting the overall performance and stability of the IoT system.

[0003] In the process of implementing the embodiments of the present invention, the inventors discovered that the prior art has at least the following problems or defects: When a large number of devices are connected, the existing wireless IoT networking technology has a fixed network topology and cannot be dynamically optimized according to the actual distribution of devices and communication needs, resulting in blind spots in signal coverage and low communication efficiency between devices; at the same time, the device connection verification after the network is completed is not perfect, and it is impossible to detect and solve the problems in device connection in a timely manner, which affects the reliability and stability of the IoT system. Summary of the Invention

[0004] This invention provides a wireless IoT networking method based on star-flash technology, comprising:

[0005] The network deployment data of the target IoT area is collected through the StarFlash networking control device. The network deployment data includes: the network topology set of the corresponding target IoT area.

[0006] Each IoT device to be connected within the target IoT area is encoded to generate an IoT device code identifier, resulting in an IoT device code identifier group;

[0007] An optimized network topology set is generated from the network topology set, where the optimized network topology is the structure after signal coverage optimization of multiple network topologies;

[0008] By splicing together the various optimized network topologies in the optimized network topology set, a regional network topology map is obtained.

[0009] Based on the regional network topology map and the pre-trained network information generation model, generate the original network region model corresponding to the target IoT region;

[0010] Obtain the IoT device model information corresponding to each IoT device code identifier in the IoT device code identifier group to obtain the IoT device model information group, wherein the IoT device model information includes: IoT device model, access location information, and communication parameter information;

[0011] Based on the IoT device model information group and the IoT device code identification group, the original network area model is filled to obtain the target network area model;

[0012] Based on the target network area model, control the associated networking devices to perform networking processing for each IoT device to be connected;

[0013] In response to the determination that the networking of each IoT device is completed, the StarSpark networking control device collects the network deployment data of the target IoT area after the networking is completed, as the target network data, and performs connection verification on each IoT device that has completed networking based on the target network data.

[0014] Furthermore, based on the regional network topology map and the pre-trained network information generation model, an original network region model corresponding to the target IoT region is generated, including:

[0015] Based on the regional network topology map and the network information generation model, an initial network regional model and network regional model information are generated. The network regional model information includes: a set of network regional model element information, and the network regional model element information includes: network element type, network element feature vector, and network element position information. The network element position information represents the position of the network regional model element corresponding to the network regional model element information in the initial network regional model.

[0016] For each network region model element in the network region model element information set, perform the following processing steps: Based on the network element type and network element feature vector included in the network region model element information, determine whether there is a network region model element in the pre-set network region model element library that matches the network region model element information;

[0017] In response to the existence of a matching network region model element in the network region model element library, the network region model element that matches the network region model element information is determined as a candidate network region model element.

[0018] Based on the elements of each candidate network region model, the initial network region model is updated to generate an updated initial network region model, which serves as the original network region model.

[0019] Furthermore, based on the regional network topology map and the network information generation model, an initial network region model and network region model information are generated, including:

[0020] Based on the encoder included in the network information generation model, the topology map encoding features corresponding to the regional network topology map are generated.

[0021] The network information generation model includes a decoder, which generates topology graph decoding features corresponding to topology graph encoding features.

[0022] The topology graph extraction features corresponding to the topology graph decoding features are generated by using the feature extraction model included in the network information generation model.

[0023] By using the threshold adjustment network included in the network information generation model, adjustment parameters corresponding to the regional network topology map are generated.

[0024] By adjusting parameters and extracting features from the topology map, an initial network region model and network region model information are generated.

[0025] Furthermore, based on the IoT device model information group and the IoT device code identifier group, the original network area model is filled in to obtain the target network area model, including:

[0026] For each IoT device model information in the IoT device model information group, perform the following population steps: populate the IoT device model included in the IoT device model information into the access location of the corresponding access location information in the original network area model;

[0027] Determine the IoT device code identifier corresponding to the IoT device model information, and use it as the target IoT device code identifier;

[0028] Mark the target IoT device code identifier on the IoT device model corresponding to the IoT device model information in the original network area model;

[0029] The original network region model that has been filled in is identified as the target network region model.

[0030] Furthermore, an optimized network topology set is generated using the network topology set, including:

[0031] For each network topology in the network topology set, perform the following optimization steps: calculate the signal coverage parameters of the network topology, where the signal coverage parameters include signal strength distribution values ​​and signal coverage blind zone area values; adjust the network topology parameters of the network topology based on the signal coverage parameters, where the network topology parameters include node distribution density values ​​and inter-node connection weight values; determine the network topology corresponding to the adjusted network topology parameters as the optimized network topology; combine the various optimized network topologies into an optimized network topology set.

[0032] Furthermore, the optimized network topologies in the optimized network topology set are spliced ​​together to obtain a regional network topology map, including:

[0033] Obtain the set of network topology nodes and the set of network topology edges for each optimized network topology; determine the overlapping regions of nodes between each optimized network topology based on the position coordinates of each set of network topology nodes; perform node matching processing on the set of network topology nodes of each optimized network topology based on the overlapping regions to obtain a set of matched node pairs; perform edge fusion processing on the set of network topology edges of each optimized network topology based on the set of matched node pairs to generate a spliced ​​edge set; generate a regional network topology map based on the set of matched node pairs and the spliced ​​edge set.

[0034] Furthermore, after filling the IoT device model information, including the IoT device model, into the access location of the corresponding access location information in the original network area model, it also includes:

[0035] Obtain communication parameter information from the IoT device model information;

[0036] Based on the communication parameter information, adjust the communication parameters of the corresponding IoT device models in the original network area model. The communication parameters include transmission power parameters and communication frequency band parameters.

[0037] Update the IoT device model corresponding to the adjusted communication parameters into the original network area model.

[0038] Furthermore, based on the target network data, connection verification is performed on each IoT device that has completed the network, including:

[0039] Based on the target network data, determine the connection verification parameters for each IoT device. These parameters include signal transmission delay and data transmission success rate.

[0040] Determine whether the connection verification parameters meet the preset connection verification conditions;

[0041] In response to the connection verification parameters meeting the preset connection verification conditions, the corresponding IoT device is determined to have passed the connection verification.

[0042] In response to the connection verification parameters not meeting the preset connection verification conditions, IoT devices that fail the connection verification are marked in the target network area model.

[0043] Furthermore, the encoder included in the network information generation model generates topology map encoding features corresponding to the regional network topology map, including:

[0044] Convert the regional network topology graph into a topology graph matrix;

[0045] The topology graph matrix is ​​convolutionally encoded by an encoder to generate topology graph convolutional features.

[0046] The convolutional features of the topological graph are processed by fully connected encoding to generate coded features of the topological graph.

[0047] Further, determining whether a pre-set network region model element library contains a network region model element that matches the network region model element information includes:

[0048] Calculate the similarity value between the feature vector of a network element in the network region model element information and the feature vector of each network region model element in the network region model element library;

[0049] Determine whether the similarity value exceeds a preset similarity threshold;

[0050] In response to the existence of a similarity value exceeding a preset similarity threshold, it is determined that there is a network region model element in the network region model element library that matches the network region model element information.

[0051] The embodiments of the present invention have at least the following beneficial effects:

[0052] 1. The network topology set of the target IoT area can be collected in real time through the StarFlash networking control device, and the network topology can be dynamically adjusted by the signal coverage optimization algorithm to optimize the node distribution density and connection weight parameters, effectively solving the problems of uneven signal strength distribution and coverage blind spots in traditional wireless networking, and significantly improving the integrity and stability of network coverage.

[0053] 2. It can automatically construct an initial network area model based on a pre-trained network information generation model, and achieve accurate access of IoT devices and automated allocation of network resources through intelligent matching of IoT device coding identifier groups and device model information, which greatly reduces the workload and error rate of manual configuration and improves networking efficiency and accuracy.

[0054] 3. After the network is completed, multi-dimensional connection verification of each IoT device can be performed by collecting target network data, including signal transmission delay detection and data transmission success rate analysis. Devices that fail the verification can be automatically identified and marked, which effectively solves the problem of difficult fault location in traditional networking methods and ensures the long-term stable operation of the entire IoT network. Attached Figure Description

[0055] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein:

[0056] Figure 1 This is a flowchart illustrating a wireless IoT networking method based on star-flash technology provided in an embodiment of the present invention. Detailed Implementation

[0057] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make the invention more thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

[0058] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0059] It should be noted that the number of any elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0060] The following is for reference. Figure 1 , Figure 1 This is a flowchart illustrating a wireless IoT networking method based on star-flash technology according to an embodiment of the present invention. Figure 1 As shown, a wireless IoT networking method based on star-flash technology includes:

[0061] S1. Collect network deployment data of the target IoT area through the Star Flash Network Control Device. The network deployment data includes: the network topology set of the corresponding target IoT area.

[0062] S2. Encode each IoT device to be connected within the target IoT area to generate an IoT device code identifier, thus obtaining an IoT device code identifier group;

[0063] S3. Generate an optimized network topology set from the network topology set, where the optimized network topology is the structure after signal coverage optimization of multiple network topologies;

[0064] S4. Perform network splicing on each optimized network topology in the optimized network topology set to obtain a regional network topology map;

[0065] S5. Generate the original network region model corresponding to the target IoT region based on the regional network topology map and the pre-trained network information generation model.

[0066] S6. Obtain the IoT device model information corresponding to each IoT device code identifier in the IoT device code identifier group to obtain the IoT device model information group, wherein the IoT device model information includes: IoT device model, access location information, and communication parameter information;

[0067] S7. Fill in the original network area model with the IoT device model information group and the IoT device code identification group to obtain the target network area model;

[0068] S8. Based on the target network area model, control the associated networking devices to perform networking processing on each IoT device to be connected;

[0069] S9. In response to determining that the networking of each IoT device is completed, control the StarShine networking control device to collect the network deployment data of the target IoT area after the networking is completed, as the target network data, and perform connection verification on each IoT device that has completed networking based on the target network data.

[0070] It should be noted that this invention collects network deployment data of the target IoT area through a StarSignal networking control device. This network deployment data includes a set of network topologies for the target IoT area. The StarSignal networking control device is a device used to manage and control IoT networking; it can collect various information related to network deployment. The set of network topologies refers to the collection of all possible network topologies within the target IoT area, describing the connection relationships and layout of devices in the network. Each IoT device to be connected within the target IoT area is encoded to generate an IoT device coded identifier, resulting in an IoT device coded identifier group. The IoT device coded identifier is a unique identifier for each IoT device, used to distinguish different devices during the networking process. An optimized network topology set is generated from the network topology set. Optimized network topology refers to the structure after optimizing the signal coverage of multiple network topologies, aiming to improve network signal coverage and communication efficiency.

[0071] Specifically, each network topology in the network topology set describes the connection method and layout between devices, such as star topology, mesh topology, or hybrid topology. The IoT device identification group contains the identification codes of all IoT devices to be connected; these codes can be numbers, letters, or combinations thereof, used to uniquely identify each device. The optimized network topology set is obtained by optimizing signal coverage based on the original network topology set using algorithms. The algorithm considers parameters such as signal strength distribution and signal coverage blind spot area to ensure that the optimized network topology better meets the communication needs of IoT devices. The optimized network topologies in the set are then stitched together to obtain a regional network topology map. Network stitching refers to connecting multiple optimized network topologies to form a complete regional network topology map. This process requires considering the connection points and connection methods between the various network topologies to ensure the connectivity and stability of the entire regional network. Based on the regional network topology map and a pre-trained network information generation model, an original network region model corresponding to the target IoT region is generated. A network information generation model is a pre-trained model that can generate a corresponding network region model based on an input regional network topology map. The input parameters of the model include information such as the node positions and connection relationships of the network topology, and the output is the original network region model, which contains the basic structure of the network and the preliminary layout of the devices.

[0072] Preferably, when generating the optimized network topology set, a genetic algorithm-based optimization method can be used. The genetic algorithm optimizes the network topology by simulating the process of natural selection. First, a fitness function is defined, which takes the signal strength distribution value and the signal coverage blind zone area as parameters to evaluate the quality of the network topology. Then, the population is initialized, with each individual representing a network topology. Through crossover, mutation, and other operations, the process iterates continuously to obtain the optimized network topology set. When generating the original network region model, the network information generation model can employ a Generative Adversarial Network (GAN) from deep learning. The generator generates a network region model based on the regional network topology map, and the discriminator evaluates the generated model. Through continuous training, the generator can generate network region models that more closely resemble actual needs. The generator's input parameters include the node coordinates and connection weights of the network topology, and the output is the original network region model containing network element types, network element feature vectors, and network element location information. When populating the original network area model, the IoT device model can be accurately placed in the corresponding position in the model based on the access location information in the IoT device model information, and the communication parameters of the device, such as transmission power and communication frequency band, can be adjusted according to the communication parameter information to ensure the communication performance of the device after networking.

[0073] In some embodiments, an original network region model corresponding to the target IoT region is generated based on the regional network topology map and a pre-trained network information generation model, including:

[0074] Based on the regional network topology map and the network information generation model, an initial network regional model and network regional model information are generated. The network regional model information includes: a set of network regional model element information, and the network regional model element information includes: network element type, network element feature vector, and network element position information. The network element position information represents the position of the network regional model element corresponding to the network regional model element information in the initial network regional model.

[0075] For each network region model element in the network region model element information set, perform the following processing steps: Based on the network element type and network element feature vector included in the network region model element information, determine whether there is a network region model element in the pre-set network region model element library that matches the network region model element information;

[0076] In response to the existence of a matching network region model element in the network region model element library, the network region model element that matches the network region model element information is determined as a candidate network region model element.

[0077] Based on the elements of each candidate network region model, the initial network region model is updated to generate an updated initial network region model, which serves as the original network region model.

[0078] It should be noted that this invention generates the original network region model for the corresponding target IoT region based on the regional network topology diagram and a pre-trained network information generation model. The regional network topology diagram is a graphical representation of the connection relationships and layout of network devices within the target IoT region, intuitively reflecting the network's structural characteristics. The network information generation model is a trained model capable of generating a corresponding network region model based on the input network topology information. This model includes the basic network structure and preliminary device layout information. When generating the initial network region model and network region model information, a set of network region model element information is obtained. Each network region model element contains key data such as network element type, network element feature vector, and network element location information. This data describes each element and its attributes within the network region model. For each network region model element in the set of network region model element information, a matching process is performed to determine if a matching network region model element exists, thereby generating the original network region model.

[0079] Specifically, each network region model element in the network region model element information set can be a different type of network device, such as a router, sensor node, or gateway, each performing different functions within the network. The network element feature vector is a multi-dimensional vector used to describe various attributes of the network element, such as communication capabilities, processing power, and energy consumption. These parameters are quantified using feature vectors. The network element location information represents the specific position of the network element within the network region model, typically represented by coordinates in a coordinate system. For example, in a two-dimensional plane, (x, y) coordinates can be used to determine the device's position. The network region model element library is a pre-set database storing various possible network region model elements and their feature vectors, used for matching and selection during network region model generation. The matching process involves comparing the feature vectors in the network region model element information with the feature vectors of elements in the library. When a matching element is found, it is designated as a candidate network region model element. The initial network region model is then updated based on these candidate elements to generate the final original network region model.

[0080] Preferably, the network information generation model can be constructed using a deep neural network model based on machine learning. The input to this model is a matrix representation of the regional network topology, where the rows and columns of the matrix correspond to the nodes in the network and the connections between them, respectively. The element values ​​in the matrix represent information such as connection weights or distances between nodes. The model training process involves learning from a large number of network topology samples and adjusting the weights and bias parameters of the neural network to enable the model to accurately generate network regional model information. When matching network regional model element information, cosine similarity can be used to calculate the similarity between feature vectors. Specifically, the dot product of two feature vectors is divided by the product of their magnitudes; the closer the value is to 1, the more similar the two vectors are. When the similarity value exceeds a preset similarity threshold, a matching network regional model element is considered to have been found. Furthermore, to improve the model's accuracy and robustness, regularization techniques, such as L2 regularization, can be introduced during training. By adding the sum of squared weights as a penalty term to the loss function, overfitting is prevented, thereby improving the model's generalization ability in practical applications.

[0081] In some embodiments, an initial network region model and network region model information are generated based on the regional network topology map and the network information generation model, including:

[0082] Based on the encoder included in the network information generation model, the topology map encoding features corresponding to the regional network topology map are generated.

[0083] The network information generation model includes a decoder, which generates topology graph decoding features corresponding to topology graph encoding features.

[0084] The topology graph extraction features corresponding to the topology graph decoding features are generated by using the feature extraction model included in the network information generation model.

[0085] By using the threshold adjustment network included in the network information generation model, adjustment parameters corresponding to the regional network topology map are generated.

[0086] By adjusting parameters and extracting features from the topology map, an initial network region model and network region model information are generated.

[0087] It should be noted that this invention employs a collaborative approach involving multiple components of the network information generation model when generating the initial network region model and network region model information. The network information generation model is a machine learning model that processes the regional network topology map through multiple modules, including an encoder, decoder, feature extraction model, and threshold adjustment network, to generate the initial network region model and network region model information. The encoder converts the regional network topology map into a compact feature representation, i.e., topology map encoded features; the decoder decodes the encoded features into a more easily processed form, i.e., topology map decoded features; the feature extraction model further extracts key information from the decoded features to generate topology map extracted features; finally, the threshold adjustment network generates adjustment parameters based on these features to generate the initial network region model and network region model information. This process involves the collaborative work of multiple steps and modules to ensure that the generated model accurately reflects the network topology and device information.

[0088] Specifically, the encoder part of the network information generation model converts the regional network topology map into a topology map matrix, and then performs convolutional coding on this matrix to generate topology map convolutional features. Convolutional coding is a common image processing technique used to extract local features from images, suitable for processing spatially structured data like network topology maps. Fully connected coding further compresses and encodes the convolutional features, generating more compact topology map encoded features. The decoder part decodes the topology map encoded features into topology map decoded features. This process can be seen as the reverse of the encoding process, aiming to restore the encoded features to a form closer to the original data while retaining the key information extracted during encoding. The feature extraction model further extracts key information from the decoded features to generate topology map extracted features, which can more accurately describe the structure and properties of the network topology map. The threshold adjustment network generates adjustment parameters based on these features. These parameters are used to fine-tune the generated initial network region model to better adapt to the actual network environment.

[0089] Preferably, the network information generation model can be constructed using a deep learning framework, such as TensorFlow or PyTorch. The model input is a matrix representation of the regional network topology, where rows and columns correspond to nodes in the network and the connections between them, respectively, and element values ​​represent connection weights or distances between nodes. In the encoder part, a multi-layer convolutional neural network (CNN) can be used to extract local features from the topology. The kernel size and stride of each convolutional layer can be adjusted according to the size and complexity of the network topology. For example, for a complex network topology with a large number of nodes, a larger kernel and a smaller stride can be used to extract more detailed features. Fully connected encoding can be implemented using a multi-layer perceptron (MLP). By adjusting the number of layers and neurons in each layer of the MLP, the dimensionality and complexity of the encoded features can be controlled. The decoder part can employ deconvolution operations to gradually restore the encoded features to a form close to the original data. The feature extraction model can use an attention mechanism to extract the most critical parts describing the network topology by calculating the importance weights of the features. The threshold adjustment network can dynamically generate adjustment parameters based on extracted features. These parameters can include node weight adjustments and connection strength corrections to optimize the generated initial network region model. During training, mean squared error (MSE) can be used as the loss function, and the model parameters can be adjusted through backpropagation to enable the model to better fit the training data and generate an accurate network region model.

[0090] In some embodiments, the original network area model is filled in according to the IoT device model information group and the IoT device code identifier group to obtain the target network area model, including:

[0091] For each IoT device model information in the IoT device model information group, perform the following population steps: populate the IoT device model included in the IoT device model information into the access location of the corresponding access location information in the original network area model;

[0092] Determine the IoT device code identifier corresponding to the IoT device model information, and use it as the target IoT device code identifier;

[0093] Mark the target IoT device code identifier on the IoT device model corresponding to the IoT device model information in the original network area model;

[0094] The original network region model that has been filled in is identified as the target network region model.

[0095] It should be noted that this invention generates the target network area model by filling the original network area model with IoT device model information. The IoT device model information includes the IoT device model, access location information, and communication parameter information. This information describes the characteristics of the IoT devices and their location and communication capabilities within the network. Specifically, the IoT device model refers to the functional and structural description of the IoT devices; the access location information indicates the specific location of the IoT devices within the network; and the communication parameter information relates to the devices' communication capabilities, such as transmission power and communication frequency bands. By filling this information into the original network area model, a more specific and detailed network area model, namely the target network area model, can be generated. This process ensures that the network area model accurately reflects the actual layout and communication characteristics of the IoT devices.

[0096] Specifically, each IoT device model in the IoT device model information group contains the IoT device model, access location information, and communication parameter information. The IoT device model can be a model of different types of devices such as sensors, actuators, and gateways, each with its specific functions and interfaces. Access location information is typically a coordinate value indicating the device's specific location in the network; for example, (x, y) coordinates can be used to represent the device's location on a two-dimensional plane. Communication parameter information includes transmission power parameters and communication frequency band parameters, which determine the device's communication capabilities and range. During the filling process, the IoT device model is first placed into its corresponding access location in the original network area model. Then, the IoT device code identifier corresponding to the IoT device model information is determined and marked on the model. Finally, the filled model is determined as the target network area model. This process involves parsing and processing the IoT device model information, as well as updating and marking the original network area model.

[0097] Preferably, the processing steps for filling in IoT device model information can be further refined. First, a template-based matching method can be used for filling in the IoT device model. A set of templates is predefined based on the type and function of the IoT device model, with each template corresponding to a device type. During filling, the appropriate template is selected based on the device type in the IoT device model information and placed at the corresponding access location in the original network area model. The access location information can be processed through coordinate transformation to ensure the device model is placed in the correct position. For communication parameter information, the communication capabilities of the device model can be adjusted according to the type and range of the communication parameters. For example, for transmission power parameters, the transmission power value can be adjusted according to the actual needs of the device to ensure that the device can communicate within the required range. Adjusting the communication frequency band parameters requires considering the device's communication protocol and frequency band allocation to ensure that the device can communicate on the appropriate frequency band. Furthermore, to improve the accuracy and efficiency of filling, automated tools, such as scripting languages ​​or graphical user interface tools, can be introduced during the filling process. These tools can quickly parse the IoT device model information and fill it into the original network area model.

[0098] In some embodiments, generating an optimized network topology set from the network topology set includes:

[0099] For each network topology in the network topology set, perform the following optimization steps: calculate the signal coverage parameters of the network topology, where the signal coverage parameters include signal strength distribution values ​​and signal coverage blind zone area values; adjust the network topology parameters of the network topology based on the signal coverage parameters, where the network topology parameters include node distribution density values ​​and inter-node connection weight values; determine the network topology corresponding to the adjusted network topology parameters as the optimized network topology; combine the various optimized network topologies into an optimized network topology set.

[0100] It should be noted that, when generating the optimized network topology set, this invention performs a series of optimization steps for each network topology in the set. These steps include calculating signal coverage parameters, adjusting network topology parameters, and ultimately determining the optimized network topology. Signal coverage parameters are used to evaluate the signal coverage effect of the network topology, including signal strength distribution values ​​and signal coverage blind spot area values. Network topology parameters describe the characteristics of the network topology, such as node distribution density values ​​and inter-node connection weight values. By adjusting these parameters, the network topology can be optimized to better adapt to the signal coverage requirements of the target IoT area, thereby improving the overall performance and reliability of the network.

[0101] Specifically, the signal strength distribution value in the signal coverage parameters refers to the distribution of signal strength at various locations within the network topology. It is typically represented by a two-dimensional array, where each element corresponds to a location in the network, and its value represents the signal strength at that location. The signal coverage blind zone area value refers to the total area of ​​regions in the network where the signal strength is below a certain threshold; the smaller this value, the better the network's signal coverage. The node distribution density value in the network topology parameters refers to the number of nodes per unit area, affecting signal propagation and coverage range. The inter-node connection weight value represents the connection strength or communication cost between nodes, usually related to the distance between nodes and communication quality. During optimization, the signal coverage parameters for each network topology are first calculated. Then, the network topology parameters are adjusted based on these parameters. For example, the node distribution density value is increased to reduce the signal coverage blind zone area value, or the inter-node connection weight value is adjusted to optimize the signal propagation path. The network topology corresponding to the adjusted network topology parameters is determined as the optimized network topology, and these optimized structures are combined into an optimized network topology set.

[0102] Preferably, a physical model-based calculation method can be used when calculating signal coverage parameters. For example, signal strength distribution values ​​can be calculated by simulating the signal propagation process in space, considering factors such as signal attenuation, reflection, and refraction. Specific steps include: first, calculating the signal coverage range of each node based on its location and transmission power; then, obtaining the signal strength distribution map of the entire network by superimposing the signal coverage ranges of each node. The signal coverage blind zone area can be determined by analyzing the signal strength distribution map, identifying areas with signal strength below a threshold, and calculating the total area of ​​these areas. When adjusting network topology parameters, an optimization algorithm-based method, such as a genetic algorithm or simulated annealing algorithm, can be used. Taking a genetic algorithm as an example, first, a fitness function is defined, which takes the signal coverage parameters as input and outputs the degree of superiority or inferiority of the network topology; then, the population is initialized, with each individual representing a configuration of network topology parameters; through selection, crossover, and mutation operations, the population is gradually optimized, ultimately obtaining the optimal network topology parameter configuration. These parameter configurations can include the specific location of the node, the node's transmission power, etc. By adjusting these parameters, the area of ​​the signal coverage blind zone can be effectively reduced, and the uniformity of signal strength distribution can be improved.

[0103] In some embodiments, network splicing is performed on each optimized network topology in the optimized network topology set to obtain a regional network topology map, including:

[0104] Obtain the set of network topology nodes and the set of network topology edges for each optimized network topology; determine the overlapping regions of nodes between each optimized network topology based on the position coordinates of each set of network topology nodes; perform node matching processing on the set of network topology nodes of each optimized network topology based on the overlapping regions to obtain a set of matched node pairs; perform edge fusion processing on the set of network topology edges of each optimized network topology based on the set of matched node pairs to generate a spliced ​​edge set; generate a regional network topology map based on the set of matched node pairs and the spliced ​​edge set.

[0105] It should be noted that the main purpose of this invention in stitching together various optimized network topologies from a set of optimized network topologies is to generate a complete regional network topology map. This process involves obtaining the node set and edge set of each optimized network topology, determining overlapping node regions, and performing node matching and edge fusion processing. A node set refers to the collection of all nodes in the network, and an edge set refers to the set of connections between nodes. Overlapping node regions refer to areas where nodes in the same position exist between different optimized network topologies; these regions are the key parts of the stitching. Through node matching and edge fusion processing, multiple optimized network topologies can be connected into a complete regional network topology map, thereby achieving network coverage of the entire target IoT area.

[0106] Specifically, each node in the network topology node set has its own coordinates, which determine its specific location in the network space. Each edge in the network topology edge set represents the connection between two nodes, usually represented by node pairs, such as (node ​​A, node B) indicating an edge between nodes A and B. The determination of overlapping regions is achieved by comparing the coordinates of nodes in different optimized network topologies; when two or more nodes have the same coordinates, these nodes constitute an overlapping region. Node matching involves matching nodes with the same position in different optimized network topologies to generate a set of matched node pairs. Edge fusion, based on the matched node pairs, merges the edges from different optimized network topologies to generate a spliced ​​edge set. Each edge in the spliced ​​edge set represents the connection between the merged nodes, and finally, a regional network topology map is generated based on the matched node pairs and the spliced ​​edge set.

[0107] Preferably, a spatial partitioning method can be used to determine the overlapping area of ​​nodes. First, the target IoT region is divided into multiple small grid cells, each with a unique identifier and coordinate range. Then, the coordinates of the nodes in each optimized network topology are mapped to these grid cells. By checking whether the grid cells mapped to nodes in different optimized network topologies are the same, the overlapping area of ​​nodes is determined. This method can improve the efficiency and accuracy of determining the overlapping area of ​​nodes. In node matching, a matching weight can be assigned to each pair of matched nodes. This weight can be calculated based on factors such as node type and communication capability to reflect the similarity and matching degree between nodes. In edge fusion, for multiple edges connecting the same pair of matched nodes, a weighted average can be calculated based on the edge weights, such as communication quality and transmission delay, to generate new spliced ​​edges, thereby optimizing the connectivity of the entire regional network topology. Furthermore, to ensure the connectivity and stability of the generated regional network topology, a connectivity check mechanism can be introduced during the splicing process. Graph theory algorithms, such as depth-first search or breadth-first search, can be used to verify the network connectivity and adjust the splicing strategy as necessary.

[0108] In some embodiments, after filling the IoT device model information, which includes the IoT device model, into the access location of the corresponding access location information in the original network area model, the method further includes:

[0109] Obtain communication parameter information from the IoT device model information;

[0110] Based on the communication parameter information, adjust the communication parameters of the corresponding IoT device models in the original network area model. The communication parameters include transmission power parameters and communication frequency band parameters.

[0111] Update the IoT device model corresponding to the adjusted communication parameters into the original network area model.

[0112] It should be noted that after filling the corresponding access locations in the original network area model with the IoT device model information, this invention further obtains the communication parameter information from the IoT device model information and adjusts the communication parameters of the corresponding IoT device models in the original network area model based on this communication parameter information. The purpose of this process is to ensure that the communication performance of IoT devices reaches its optimal level after networking. The communication parameter information includes transmission power parameters and communication frequency band parameters, which directly affect the communication quality and coverage between devices. By adjusting these parameters, the communication capabilities of the devices can be optimized, improving the performance and stability of the entire IoT system.

[0113] Specifically, the transmission power parameter in the communication parameter information refers to the power of the signal transmitted by the IoT device during communication. This parameter determines the signal propagation distance and strength. Higher transmission power can increase the signal coverage, but may also lead to increased energy consumption. The communication frequency band parameter refers to the wireless frequency band used by the IoT device. Different frequency bands have different propagation characteristics and anti-interference capabilities. For example, lower frequency bands have a longer propagation distance but narrower bandwidth, while higher frequency bands have a wider bandwidth but a shorter propagation distance and are more susceptible to interference. When adjusting communication parameters, these parameters need to be set according to the specific application scenario and performance requirements of the IoT device. For example, in scenarios requiring long-distance communication, the transmission power can be appropriately increased; in scenarios requiring high data transmission rates, a higher frequency band can be selected. In addition, the adjustment of communication parameters also needs to consider the compatibility between devices and the overall network performance to ensure that all devices can work collaboratively under optimized communication conditions.

[0114] Preferably, a dynamic adjustment method based on performance evaluation can be adopted when adjusting communication parameters. First, a set of performance indicators, such as signal transmission delay, data transmission success rate, and energy consumption, are defined to evaluate the communication performance of IoT devices under the current communication parameters. Then, performance data of the devices under different communication parameter configurations is collected through simulation or actual testing. Based on this data, a performance evaluation model is established, which can predict the device performance under different communication parameter configurations. In practical applications, the communication parameters of the IoT devices are dynamically adjusted according to the prediction results of the performance evaluation model to achieve optimal communication performance. For example, if a device is found to have high signal transmission delay, its transmission power can be appropriately increased; if there is significant communication interference in a certain frequency band, it can be switched to another frequency band. Furthermore, to improve the efficiency and accuracy of adjustments, machine learning algorithms, such as reinforcement learning algorithms, can be introduced to automatically adjust communication parameters through continuous learning and optimization to adapt to the complex IoT environment.

[0115] In some embodiments, connection verification is performed on each IoT device that has completed its network deployment based on target network data, including:

[0116] Based on the target network data, determine the connection verification parameters for each IoT device. These parameters include signal transmission delay and data transmission success rate.

[0117] Determine whether the connection verification parameters meet the preset connection verification conditions;

[0118] In response to the connection verification parameters meeting the preset connection verification conditions, the corresponding IoT device is determined to have passed the connection verification.

[0119] In response to the connection verification parameters not meeting the preset connection verification conditions, IoT devices that fail the connection verification are marked in the target network area model.

[0120] It should be noted that when verifying the connectivity of IoT devices after network formation, this invention primarily relies on target network data to determine the connectivity verification parameters for each device, and uses these parameters to determine whether the device passes verification. The connectivity verification parameters include signal transmission delay and data transmission success rate, which are used to evaluate the communication performance of IoT devices after network formation. The signal transmission delay reflects the time required for a signal to travel from the sender to the receiver, while the data transmission success rate indicates the proportion of data that successfully reaches the receiver during transmission. These parameters effectively evaluate the connection quality between IoT devices, ensuring network stability and reliability.

[0121] Specifically, signal transmission delay refers to the time required for a signal to be sent from one IoT device to another and received, usually measured in milliseconds (ms). Data transmission success rate refers to the ratio of successfully transmitted data packets to the total number of transmitted data packets within a certain time period, usually expressed as a percentage. During connection verification, connection verification parameters for each IoT device are first calculated based on the target network data. Preset connection verification conditions are thresholds set according to actual application requirements; for example, signal transmission delay should not exceed a certain maximum delay value, and data transmission success rate should not be lower than a certain minimum success rate value. If the connection verification parameters of an IoT device meet these preset conditions, the device is considered to have passed the connection verification; otherwise, the device fails the verification and is marked in the target network area model for subsequent troubleshooting and repair.

[0122] Preferably, a timestamp-based method can be used to calculate the signal transmission delay value. At the sending end, a timestamp is added to each transmitted data packet to record the transmission time; at the receiving end, the time of the received data packet is recorded, and the difference between the received and transmitted timestamps is calculated to obtain the signal transmission delay value. To improve measurement accuracy, multiple measurements can be taken and the average value calculated. When evaluating data transmission success rate, a series of data packets can be sent within a certain time period, and the number of successfully received data packets can be counted to calculate the success rate. Furthermore, to more comprehensively evaluate the connection quality of IoT devices, other parameters such as signal strength and packet loss rate can be introduced. When marking IoT devices that fail connection verification in the target network area model, different colors or symbols can be used to distinguish the device status; for example, red is used to mark devices that fail verification, and green is used to mark devices that pass verification. This provides a visual representation of the connection status of devices in the network, facilitating management and maintenance by network administrators.

[0123] In some embodiments, the encoder included in the network information generation model generates topology map encoding features corresponding to the regional network topology map, including:

[0124] Convert the regional network topology graph into a topology graph matrix;

[0125] The topology graph matrix is ​​convolutionally encoded by an encoder to generate topology graph convolutional features.

[0126] The convolutional features of the topological graph are processed by fully connected encoding to generate coded features of the topological graph.

[0127] It should be noted that, in generating the topology graph encoding features corresponding to the regional network topology graph through the encoder of the network information generation model, this invention mainly involves converting the regional network topology graph into a topology graph matrix, and then performing convolutional coding and fully connected coding on this matrix. The purpose of this process is to transform the complex network topology structure into a compact feature representation for subsequent processing and analysis. The topology graph matrix is ​​a mathematical representation used to describe the connection relationships between nodes in the network, while convolutional coding and fully connected coding are commonly used feature extraction methods in deep learning. These methods can extract key features of the network topology graph, providing a foundation for generating the initial network region model and network region model information.

[0128] Specifically, the topology graph matrix is ​​a two-dimensional array where rows and columns represent nodes in the network, and element values ​​represent connections between nodes. For example, a value of 1 indicates a connection between two nodes, while 0 indicates no connection. Convolutional coding is a technique based on Convolutional Neural Networks (CNNs) that extracts local features by sliding convolutional kernels across the matrix. Fully connected coding further compresses and encodes these convolutional features, generating a more compact feature representation. In this process, parameters such as the kernel size, stride, and number of neurons in the fully connected layers can be set according to the complexity of the network topology and the needs of feature extraction. For example, for a complex network topology with a large number of nodes, a larger kernel and a smaller stride can be used to extract more detailed local features, while increasing the number of neurons in the fully connected layers ensures feature richness.

[0129] Preferably, deep learning frameworks such as TensorFlow or PyTorch can be used when constructing the encoder of the network information generation model. First, the input layer of the network is defined, with the input being a matrix representation of the regional network topology. Next, multiple convolutional layers are added, with the kernel size and stride of each layer adjusted according to the size and complexity of the topology. For example, the first convolutional layer can use a 3x3 kernel and a stride of 1 to extract basic local features. Then, pooling layers are added to reduce the dimensionality of the features and improve the computational efficiency of the model. Afterward, fully connected layers are added, and the dimensionality of the encoded features is controlled by adjusting the number of neurons. During training, mean squared error (MSE) can be used as the loss function, and the model parameters are adjusted through backpropagation to ensure the model accurately generates the topology-encoded features. Furthermore, to improve the model's generalization ability, regularization techniques, such as L2 regularization, can be introduced during training. This prevents overfitting by adding the sum of squared weights as a penalty term to the loss function.

[0130] In some embodiments, determining whether a network region model element exists in a pre-set network region model element library that matches the network region model element information includes:

[0131] Calculate the similarity value between the feature vector of a network element in the network region model element information and the feature vector of each network region model element in the network region model element library;

[0132] Determine whether the similarity value exceeds a preset similarity threshold;

[0133] In response to the existence of a similarity value exceeding a preset similarity threshold, it is determined that there is a network region model element in the network region model element library that matches the network region model element information.

[0134] It should be noted that, in determining whether a matching network region model element exists in the network region model element library, this invention primarily achieves this by calculating the similarity value between the feature vector of the network element in the network region model element information and the feature vector of each network region model element in the network region model element library. The purpose of this process is to find the known network region model element most similar to the current network region model element information, thereby providing a basis for subsequent model updates and optimizations. The similarity value is calculated by comparing the closeness of two feature vectors, typically using methods such as cosine similarity. When the similarity value exceeds a preset similarity threshold, a matching network region model element can be considered to have been found.

[0135] Specifically, the network element feature vector in the network region model element information is a multi-dimensional vector used to describe various attributes of the network element, such as device type, communication capability, and processing capability. The network region model element library is a pre-set database that stores various possible network region model elements and their feature vectors. The similarity value is usually calculated by dividing the dot product of two feature vectors by their modulus product; this method is called cosine similarity. The preset similarity threshold is a value set according to actual application requirements to determine whether two feature vectors are sufficiently similar. For example, if the similarity threshold is set to 0.8, then when the calculated similarity value is greater than or equal to 0.8, a matching network region model element can be considered to have been found.

[0136] Preferably, when calculating similarity values, a machine learning-based approach can be used to optimize the representation of feature vectors and the similarity calculation process. First, the feature vectors in the network region model element library are preprocessed by using dimensionality reduction techniques such as Principal Component Analysis (PCA) to reduce the dimensionality of the feature vectors while retaining the main feature information. Then, a deep learning model, such as an autoencoder, is used to further encode and learn the feature vectors, enabling the model to automatically extract key information from them. When calculating similarity, cosine similarity or other more complex similarity metrics, such as Euclidean distance or Mahalanobis distance, can be used. Furthermore, to improve matching accuracy, more feature vector samples can be introduced into the network region model element library, and similar feature vectors can be grouped using clustering algorithms, thus finding the closest feature vector more quickly during the matching process. In practical applications, the preset similarity threshold can be dynamically adjusted according to different network topologies and device types to adapt to different matching needs.

[0137] The above embodiments of the present invention have the following beneficial effects:

[0138] 1. The network topology set of the target IoT area can be collected in real time through the StarFlash networking control device, and the network topology can be dynamically adjusted by the signal coverage optimization algorithm to optimize the node distribution density and connection weight parameters, effectively solving the problems of uneven signal strength distribution and coverage blind spots in traditional wireless networking, and significantly improving the integrity and stability of network coverage.

[0139] 2. It can automatically construct an initial network area model based on a pre-trained network information generation model, and achieve accurate access of IoT devices and automated allocation of network resources through intelligent matching of IoT device coding identifier groups and device model information, which greatly reduces the workload and error rate of manual configuration and improves networking efficiency and accuracy.

[0140] 3. After the network is completed, multi-dimensional connection verification of each IoT device can be performed by collecting target network data, including signal transmission delay detection and data transmission success rate analysis. Devices that fail the verification can be automatically identified and marked, which effectively solves the problem of difficult fault location in traditional networking methods and ensures the long-term stable operation of the entire IoT network.

[0141] Furthermore, the storage medium in the embodiments of this application stores program instructions capable of implementing all the above methods. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0142] The above description is merely a selection of preferred embodiments of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention as described in the embodiments is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A wireless IoT networking method based on star-flash technology, characterized in that, include: The network deployment data of the target IoT area is collected through the Star Flash networking control device. The network deployment data includes the network topology set of the corresponding target IoT area. Each IoT device to be connected within the target IoT area is encoded to generate an IoT device code identifier, resulting in an IoT device code identifier group; An optimized network topology set is generated from the network topology set, where the optimized network topology is the structure after signal coverage optimization of multiple network topologies; By stitching together the various optimized network topologies in the optimized network topology set, a regional network topology map is obtained. Based on the regional network topology map and the pre-trained network information generation model, generate the original network region model corresponding to the target IoT region; The IoT device model information group is obtained by acquiring the IoT device model information corresponding to each IoT device code identifier in the IoT device code identifier group. The IoT device model information includes: IoT device model, access location information, and communication parameter information. Based on the IoT device model information group and the IoT device code identification group, the original network area model is filled to obtain the target network area model; Based on the target network area model, control the associated networking devices to perform networking processing for each IoT device to be connected; In response to the determination that the networking of each IoT device is completed, the StarSpark networking control device collects the network deployment data of the target IoT area after the networking is completed, as the target network data, and performs connection verification on each IoT device that has completed networking based on the target network data.

2. The wireless IoT networking method based on star-flash technology according to claim 1, characterized in that, Based on the regional network topology map and a pre-trained network information generation model, an original network region model corresponding to the target IoT region is generated, including: Based on the regional network topology map and the network information generation model, an initial network regional model and network regional model information are generated. The network regional model information includes: a set of network regional model element information, and the network regional model element information includes: network element type, network element feature vector, and network element position information. The network element position information represents the position of the network regional model element corresponding to the network regional model element information in the initial network regional model. For each network region model element in the network region model element information set, perform the following processing steps: Based on the network element type and network element feature vector included in the network region model element information, determine whether there is a network region model element in the pre-set network region model element library that matches the network region model element information; In response to the existence of a matching network region model element in the network region model element library, the network region model element that matches the network region model element information is determined as a candidate network region model element. Based on the elements of each candidate network region model, the initial network region model is updated to generate an updated initial network region model, which serves as the original network region model.

3. The wireless IoT networking method based on star-flash technology according to claim 2, characterized in that, Based on the regional network topology map and network information generation model, an initial network region model and network region model information are generated, including: Based on the encoder included in the network information generation model, the topology map encoding features corresponding to the regional network topology map are generated. The network information generation model includes a decoder, which generates topology graph decoding features corresponding to topology graph encoding features. The topology graph extraction features corresponding to the topology graph decoding features are generated by using the feature extraction model included in the network information generation model. By using the threshold adjustment network included in the network information generation model, adjustment parameters corresponding to the regional network topology map are generated. By adjusting parameters and extracting features from the topology map, an initial network region model and network region model information are generated.

4. The wireless IoT networking method based on star-flash technology according to claim 2, characterized in that, Based on the IoT device model information group and the IoT device code identifier group, the original network area model is filled in to obtain the target network area model, including: For each IoT device model information in the IoT device model information group, perform the following population steps: populate the IoT device model included in the IoT device model information into the access location of the corresponding access location information in the original network area model; Determine the IoT device code identifier corresponding to the IoT device model information, and use it as the target IoT device code identifier; Mark the target IoT device code identifier on the IoT device model corresponding to the IoT device model information in the original network area model; The original network region model that has been filled in is identified as the target network region model.

5. The wireless IoT networking method based on star flash technology according to claim 1, characterized in that, Generate an optimized network topology set from the network topology set, including: For each network topology in the network topology set, perform the following optimization steps: calculate the signal coverage parameters of the network topology, where the signal coverage parameters include signal strength distribution values ​​and signal coverage blind zone area values; adjust the network topology parameters of the network topology based on the signal coverage parameters, where the network topology parameters include node distribution density values ​​and inter-node connection weight values; determine the network topology corresponding to the adjusted network topology parameters as the optimized network topology; combine the various optimized network topologies into an optimized network topology set.

6. The wireless IoT networking method based on star-flash technology according to claim 1, characterized in that, By stitching together the various optimized network topologies in the optimized network topology set, a regional network topology map is obtained, including: Obtain the set of network topology nodes and the set of network topology edges for each optimized network topology; determine the overlapping regions of nodes between each optimized network topology based on the position coordinates of each set of network topology nodes; perform node matching processing on the set of network topology nodes of each optimized network topology based on the overlapping regions to obtain a set of matched node pairs; perform edge fusion processing on the set of network topology edges of each optimized network topology based on the set of matched node pairs to generate a spliced ​​edge set; generate a regional network topology map based on the set of matched node pairs and the spliced ​​edge set.

7. The wireless IoT networking method based on star-flash technology according to claim 4, characterized in that, After filling the IoT device model information, including the IoT device model, into the access location of the corresponding access location information in the original network area model, it also includes: Obtain communication parameter information from the IoT device model information; Based on the communication parameter information, adjust the communication parameters of the corresponding IoT device models in the original network area model. The communication parameters include transmission power parameters and communication frequency band parameters. Update the IoT device model corresponding to the adjusted communication parameters into the original network area model.

8. The wireless IoT networking method based on star flash technology according to claim 1, characterized in that, Based on the target network data, perform connection verification on each IoT device that has completed the network, including: Based on the target network data, determine the connection verification parameters for each IoT device. These parameters include signal transmission delay and data transmission success rate. Determine whether the connection verification parameters meet the preset connection verification conditions; In response to the connection verification parameters meeting the preset connection verification conditions, the corresponding IoT device is determined to have passed the connection verification. In response to the connection verification parameters not meeting the preset connection verification conditions, IoT devices that fail the connection verification are marked in the target network area model.

9. The wireless IoT networking method based on star-flash technology according to claim 3, characterized in that, The network information generation model, including its encoder, generates topology map encoding features corresponding to the regional network topology map, including: Convert the regional network topology graph into a topology graph matrix; The topology graph matrix is ​​convolutionally encoded by an encoder to generate topology graph convolutional features. The convolutional features of the topological graph are processed by fully connected encoding to generate coded features of the topological graph.

10. The wireless IoT networking method based on star-flash technology according to claim 2, characterized in that, Determine whether a pre-set network region model element library exists that matches the network region model element information, including: Calculate the similarity value between the feature vector of a network element in the network region model element information and the feature vector of each network region model element in the network region model element library; Determine whether the similarity value exceeds a preset similarity threshold; In response to the existence of a similarity value exceeding a preset similarity threshold, it is determined that there is a network region model element in the network region model element library that matches the network region model element information.

Images