A multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling

Abstract

The application discloses a kind of multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling, comprising the following steps: S1. routing discovery and networking are carried out, and mesh network is formed;S2. channel modeling is carried out using interpretable neural network, for predicting channel information, and judging link deterioration and node movement, using routing protocol to carry out fast convergence, and using potential path to route maintenance for current data transmission;S3. network controller is used to broadcast and collect each node state information, according to network state dynamics and service particularity, introduce routing selection process, use node state and service characteristics to select the best route and secondary route.The application can adaptively model and predict based on historical channel data by constructing self-learning channel model, update channel state in real time, improve the accuracy and intelligent level of route maintenance, and can also comprehensively analyze channel quality, optimize network performance.

Description

Technical Field

[0001] The present invention relates to the field of mobile communication technology, and in particular to a multi-hop multi-frequency wireless mesh network adaptive routing method, device and storage medium based on channel modeling. Background Art

[0002] With the rapid development of the Internet and wireless communication technologies in the digital age, the scale and complexity of networks continue to increase. Especially in application scenarios such as the Internet of Things, smart cities, and mobile ad hoc networks, the requirements for network performance are becoming more stringent.

[0003] Traditional network architectures often find it difficult to adapt quickly to dynamically changing environments, resulting in inefficient resource allocation and insufficient network reliability. Therefore, software-defined networking (SDN) is often used in existing technologies as an innovative network architecture. By separating the control plane from the data plane, it achieves higher flexibility and programmability, and can effectively support dynamic resource allocation and automated management.

[0004] However, node mobility and frequent link changes often make it difficult to meet the required network management and performance monitoring requirements. To address this issue, the existing technology further proposes a BATMAN protocol that uses transmission quality (TQ) values ​​to quantify link status, effectively reflecting the dynamic behavior of nodes in the network and link quality fluctuations.

[0005] However, even if existing technologies use transmission quality (TQ) values ​​to quantify link status, they cannot effectively predict and preset node path selection within the existing network performance evaluation model framework. In the event of a link failure, it is also not convenient to seamlessly complete route switching and ensure network stability.

[0006] In addition, although traditional deep learning models such as deep neural networks (DNN), convolutional neural networks (CNN) and recurrent neural networks (RNN) perform well in processing complex tasks, they are often regarded as "black boxes" because their internal decision-making processes are difficult for humans to intuitively understand. Therefore, a neural network model that improves the transparency and understandability of the model's decision-making process has been proposed in the prior art, which is usually called an interpretable neural network model (XNN).

[0007] At the same time, with the development of new technologies and applications in 5G and beyond, a new trend has emerged in wireless channels, including higher frequencies, larger antenna arrays, and more complex wireless communication schemes. At this time, the wireless channels in the channel modeling process have brought huge challenges to traditional channel modeling. Therefore, there is an urgent need for an effective and accurate channel modeling method to discover basic channel characteristics.

[0008] To this end, this application proposes a multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling to solve the above technical problems. Summary of the Invention

[0009] The main purpose of the present invention is to provide a multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling. In view of the complexity of the multi-hop multi-frequency wireless mesh network, the optimal routing path and frequency combination are dynamically selected to maximize the satisfaction of business needs. At the same time, by combining multi-frequency switching based on Wifi6 with a new routing mechanism, the adaptability and performance of the network in high-interference environments are improved to solve the technical problems raised in the background technology.

[0010] The present invention adopts the following technical solutions to solve the above technical problems:

[0011] A multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling includes the following steps:

[0012] S1. Use the Batman protocol for route discovery and networking to build a mesh network;

[0013] S2. Use interpretable neural networks for channel modeling to predict channel information and determine link degradation and node mobility. Use routing protocols for rapid convergence and use potential paths to maintain routing for current data transmission.

[0014] S3. Using the SDN network controller, the node reports the service type and transmission request to the SDN network controller. The SDN network controller broadcasts and collects the status information of each node, introduces a routing selection process based on the dynamic network status and service characteristics, and uses the node status and service characteristics to select the best route and secondary route.

[0015] Preferably, the specific operation process of step S1 includes:

[0016] S11. All nodes periodically generate and send OGM messages. These messages contain key information, including the unique identifier of the sending node, sequence number, hop count, and lifetime. The purpose of OGM messages is to inform other nodes in the network of the existence of the source node and provide a basis for the route discovery process. OGM messages are broadcast to surrounding nodes. The propagation range is controlled by the lifetime (TTL) field. When the TTL is reduced to 0, the OGM message is discarded, limiting its maximum propagation range and avoiding network overload.

[0017] S12. After receiving the OGM message, the neighbor node checks the validity of the message. First, the validity of the message is checked, especially by the OGM sequence number to determine whether it is a duplicate message. If it is a non-duplicate message, that is, the sequence number is received for the first time, it indicates that the message is new. The node forwards it to other neighbor nodes and increments the hop count field in the message to indicate that the number of hops on the path has increased.

[0018] S13. All nodes gradually build and update their own routing tables based on the OGM messages received from neighboring nodes.

[0019] Preferably, the specific operation process in step S2 includes:

[0020] S21 receives the channel parameter information of the node through the SDN network controller as a channel quality indicator;

[0021] S22. Use an interpretable neural network for self-learning channel modeling. Build a channel environment model by analyzing channel parameters and node locations. Input historical channel parameter data into the trained channel environment model to perform channel prediction.

[0022] S23. Using an interpretable neural network model, the channel changes and network connectivity status between nodes are predicted based on location information and channel information. On this basis, path presetting is completed according to different business types and using the WD algorithm. This is used to achieve seamless link switching when a link fails, realizing rapid signal routing and seamless signal connection.

[0023] Preferably, the specific construction process of the channel environment model in step S22 includes:

[0024] S221. Preprocess the channel training data by MinMaxScaler normalization, transform the original channel data to the same dimension and map it to the given [0,1];

[0025] S222. Initialize the model weights and parameters, and use an interpretable neural network to train the preprocessed data. During the training process, define the loss function as mean squared error and add an L1 penalty to control model complexity and reduce overfitting. At the same time, use the Adam optimizer to determine the adaptive learning rate. Iterate the training until the specified number of iterations is reached or the specified threshold error is met.

[0026] Set the parameters that need to be learned and updated in the XNN model, including: {α, β j ,h j ,w j}, j = 1, ..., k, where k is the update threshold, α is the initial learning rate of the model, and β j is the update decay rate, hj is a subnetwork, w j is the network weight;

[0027] The loss function (loss LOSS) of the XNN model is defined as the root mean square error, and L1 penalty is added to control model complexity and reduce overfitting.

[0028]

[0029] where n is the mini-batch size, k is the number of subnetworks, λ1, λ2 are hyperparameters that control the strength of the regularization penalty, and is the predicted value of the XNN model, y i is the true value, ω j is the projection weight vector, and β is the subnetwork weight vector.

[0030] In addition to w j , the remaining parameters are updated by gradient descent θ, and the adaptive learning rate of the gradient descent update iteration is determined by Adam Optimizer. For each iteration t+1=t, we have:

[0031]

[0032] where △ θ represents the loss gradient of the parameter θ, and η t Indicates a learning rate that can be automatically adjusted using the ADAM optimizer.

[0033] The validation error is defined as the mean squared error between the predicted value and the true value in the validation set. The training is iterated until the maximum number of iterations is reached or the threshold error (for example, the threshold error ε = 5dB) is met. After fixing the projection layer and projection matrix, λ1 = λ2 = 0 and fine-tuning the residual network using the Adam optimizer.

[0034] S223. After the training is completed, a channel environment model based on an interpretable neural network is obtained.

[0035] Preferably, a noise source module is further provided in the channel prediction process in step S22 to simulate errors caused by the environment or an imperfect measurement system. The final prediction value of the channel prediction is: where ε i is the Gaussian noise error, is the true value, is the predicted value.

[0036] Preferably, the specific operation process of step S23 includes:

[0037] S231. Based on the channel prediction results of the channel environment model, a utility matrix is ​​constructed for the influencing factors of different business types, and the WD routing algorithm is used to achieve path prediction and complete path presetting;

[0038] S232 sets the channel quality threshold according to business needs to determine whether the channel quality meets the threshold requirements. If the current path channel quality drops below the threshold requirement, the path switching is triggered and the path is switched to the preset path.

[0039] S233 after switching, according to the set channel quality threshold, real-time verification of the performance of the new path, the path channel continues to monitor the network status after the stability of the path channel, if the path channel is unstable, fall back to the previous path or continue to look for a new path;

[0040] S234. Summarize and record the path switching log information and feed it back to the channel environment model as training data.

[0041] Preferably, the batman protocol is used to perform routing maintenance during the execution of step S2, and the specific maintenance process includes:

[0042] Each node periodically sends source announcement data to neighboring nodes according to the Batman protocol and updates the node routing table according to the network status. If a neighboring node loses connection or does not receive source announcement data, the neighbor is removed from the node routing table to prevent the link status from deteriorating and affecting data transmission.

[0043] Preferably, the specific operation process of step S3 includes:

[0044] S31 business connection node, request transmission, the node reports the business type and transmission request to the SDN network controller;

[0045] S32. After sending the broadcast, the SDN network controller uses the batctl and iwinfo commands to obtain the underlying parameters of the batman protocol. It collects and uploads multiple parameter information for each node in the 5.8 GHz and 2.4 GHz bands, including RSSI, throughput, and TQ. It then constructs a maximum channel capacity matrix, a throughput matrix, and a packet loss rate matrix. These matrices are multiplied by the influencing factors of each parameter and added together. The influencing factors can be adjusted for different service requests. Finally, a channel matrix containing information about each node is constructed.

[0046] (1) Voice services have high requirements for stability. The impact factor of packet loss rate can be increased to give this service transmission priority.

[0047] (2) Video services have higher throughput requirements. The throughput impact factor can be increased to meet the higher throughput requirements of this service.

[0048] S33. Compare each element in the 2.4G and 5.8G channel information matrices to obtain the channel matrix, and select the optimal transmission frequency band based on the size of the matrix element value. The channel matrix uses the WD algorithm to find the optimal path and secondary path;

[0049] S34. The network controller sends the routing table to each node and updates the routing table of each node.

[0050] Preferably, the information collection process in step S32 includes:

[0051] S321. Use the iw dev mesh0 station dump command to obtain the mesh node network data of the mesh network;

[0052] S322. Observe the received signal strength RSSI, the transmit bit rate tx_bitrate, the receive bit rate rx_bitrate and the signal noise strength N0, and calculate the throughput parameter T x , the calculation formula is:

[0053] S323. Calculate the maximum channel capacity C, the calculation formula is: Where B is the channel bandwidth;

[0054] S324. Use the batctlo command to obtain the TQ value parameter. The larger the TQ value, the lower the packet loss rate of the link.

[0055] S325. Normalization processing parameters C, T x and TQ value, and obtain the processed parameter C r 、T r and P r , the calculation formula is:

[0056]

[0057] Among them, C max It is expressed as the theoretical maximum value of the channel capacity, T max It represents the theoretical maximum value of the channel throughput parameter.

[0058] Preferably, in the step S33, the WD algorithm (wide & deep algorithm) is run by the SDN network controller to obtain the shortest path and the second shortest path from the source node to the destination node, and the corresponding IP addresses are collected. According to the source IP, destination IP, port number, frequency band selection, and frequency point selection, a strategy is issued to the nodes on each path. The WD algorithm steps used include:

[0059] S331 creates a graph representation using a channel information matrix, sets the initial distance of the starting node to 0, and sets the initial distance of all other nodes to infinity;

[0060] S332. Create a minimum heap priority queue to store each node and its distance to the starting node, initially containing only the starting node;

[0061] S333. Use the D algorithm (Deep Learning) to traverse and calculate the shortest path from the source node to the destination node. After finding the shortest path, reset the distance information of all nodes, set the distance of the starting node to 0, and set the distance of other nodes to infinity, run the Dijkstra algorithm again, and ensure that the edges of the shortest path are not repeated.

[0062] S334. Repeat the D algorithm traversal process, add a check in the D algorithm traversal process to ensure that the same edge as the shortest path found is not selected, and stop the algorithm when a path to the target node is found and the path is not the shortest path.

[0063] S335 outputs the shortest path and the second shortest path from the start node to the target node. If there is no second shortest path, the shortest path and None are output.

[0064] Preferably, the transmission communication between the service connection node and the network controller in the step S3 is realized through the RYU network controller on a server and the OpenvSwitch on the node, and the Openflow protocol is used to realize information interaction.

[0065] Preferably, when the nodes are in the routing maintenance phase, each node uploads channel information and location information to the SDN network controller every 2 seconds to ensure network stability.

[0066] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0067] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0068] As can be seen from the above technical solution, the present invention provides a multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling. Compared with the existing technology, the present invention has the following advantages:

[0069] 1. The application can build a more comprehensive and accurate network performance evaluation model by considering the maximum channel capacity, packet loss rate and throughput as new network performance evaluation parameters, and using the TQ value of the BATMAN protocol, which provides an important basis for improving the reliability, efficiency and intelligent management of the network.

[0070] 2. The application introduces a channel modeling and prediction mechanism based on an interpretable neural network on the basis of traditional network routing maintenance, and through the integration of network performance evaluation parameters and the learning ability of the interpretable neural network, the system can predict and preset the path selection of the node to build a more comprehensive and intelligent network performance evaluation model, improve the adaptability and intelligent level of the network, and seamlessly complete the routing switching when a link fault occurs, ensuring the stability of the network.

[0071] 3. In the routing maintenance step, when a node in the network fails or disappears from the network, the routing protocol quickly converges to maintain the connectivity of the network, and the channel prediction and preset are used to ensure the stability of the network, which is convenient to use.

[0072] 4. The application adopts a software-defined manner, and the controller uniformly allocates routing to realize the function of on-demand routing according to the service type, and also runs in the data link layer to transfer routing information through Ethernet frames, which reduces the data packet processing delay.

[0073] 5. In the process of building a mesh network, the application receives OGM messages, self-checks and updates, and forwards incrementally to gradually build and update the routing table of the node, so that OGM can traverse the entire network, and each node only needs to know the neighbor nodes directly connected thereto, and does not need to know the topology of the entire network, which facilitates the rapid assembly and processing of network data.

[0074] 6. The application can avoid excessive switching by introducing a channel quality threshold in the path switching judgment, and can flexibly adjust the path selection according to different traffic types to ensure network load balancing and reduce network fluctuations caused by switching.

[0075] 7. This invention constructs a self-learning channel model and performs adaptive modeling and prediction based on historical channel data. It can update the channel status in real time, improve the accuracy and intelligence level of route maintenance, and dynamically adjust the model parameter selection according to different business needs. In combination with factors such as throughput, packet loss rate, and signal strength, it can comprehensively analyze channel quality and optimize network performance.

[0076] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become easy to understand through the following description. Of course, it is not necessary to achieve all of the above-mentioned advantages simultaneously in order to implement any product of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings:

[0078] Figure 1 This is a schematic diagram of the effect of using the model of the method of the present invention;

[0079] Figure 2 This is a schematic diagram of the protocol routing discovery process of the present invention;

[0080] Figure 3 This is a schematic diagram of the protocol routing maintenance process of the present invention;

[0081] Figure 4 A schematic diagram of the routing process of the present invention;

[0082] Figure 5 Schematic diagram of the experimental comparison result curve of the method of the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. In the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0084] In the embodiment, see Figures 1 to 5 .

[0085] The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling provided by the embodiment of the application adds an SDN network controller and a neural network model in a traditional batman protocol-based Mesh network based on an interpretable neural network.

[0086] The SDN network controller collects channel information between nodes, dynamically adjusts routing strategies for different service requirements from the aspects of maximum channel capacity, packet loss rate, throughput and the like between nodes, and realizes finding of different optimal routes required; an effect diagram is shown in the figure Figure 1 The improved routing algorithm adds secondary route finding and route prediction and presetting, thereby better guaranteeing data transmission of services in the route maintenance stage. The combination of network dynamics and service dynamics is realized, and the stability of the mobile network and the transmission requirements of the services are guaranteed.

[0087] The multi-hop multi-frequency wireless mesh network adaptive routing method is composed of a route discovery process, a route maintenance process and a route use process, wherein:

[0088] (1) The route discovery process is shown in the figure Figure 2 and includes the following steps:

[0089] Step S1: The route discovery step, the steps of network formation by using the batman protocol are as follows:

[0090] Step S1.1: Each node periodically generates and sends an OGM (Originator Message), and the message contains the unique identifier of the sending node, the sequence number, the hop count (Hop Count) and the time-to-live (Time-To-Live, TTL) and other key information. The purpose of the OGM message is to let other nodes in the network know the existence of the source node and provide a basis for the route discovery process. The OGM message is sent to the surrounding nodes by broadcast, and the propagation range is controlled by the TTL field. When TTL is reduced to 0, the OGM message will be discarded, limiting the maximum range of its propagation, and avoiding network overload.

[0091] Step S1.2: After receiving the OGM message, the neighbor node first checks the validity of the message, especially by the sequence number of the OGM to determine whether it is a repeated message. If the message is new (i.e. the sequence number is the first received), the node will forward it to other neighbor nodes, and will appropriately increment the hop count field in the message to represent the increase of the hop count of the path.

[0092] Step S1.3: Each node gradually builds and updates its routing table based on the OGM messages it receives from its neighbors. This incremental forwarding of messages enables the OGM to traverse the entire network, while each node only needs to know its directly connected neighbors and does not need to understand the entire network topology.

[0093] (2) The routing maintenance process is as follows Figure 3 As shown, the following steps are included:

[0094] Step S2: Route maintenance step, autonomously determine whether the link is deteriorating or the node is moving, use the routing protocol for rapid convergence, and use potential paths to maintain current data transmission. The specific steps are as follows:

[0095] Step S2.1: The first step is to rely on the routing maintenance process of the BATMAN protocol itself. The BATMAN protocol will periodically send "source announcements" and update the routing table according to the network conditions. If a neighboring node loses connection (for example, the source announcement is not received on time), the neighbor will be removed from the routing table. This is to prevent the deterioration of the link status from affecting data transmission. In addition, in order to prevent the movement of nodes in the mobile ad hoc network from affecting the overall network conditions and to reduce the frequency of uploading data too quickly to cause data redundancy, a logarithmic distance path loss model is added to construct an expression for the signal strength RSSI and the distance between nodes. When the node reaches the mobility threshold, it is uploaded to the SDN network controller and the channel information is collected again to achieve optimal path planning.

[0096] Step S2.2: The SDN network controller collects channel information from each node and uses XNN (interpretable neural network) for self-learning channel modeling. It builds a channel environment model by analyzing the maximum channel capacity, throughput, packet loss rate channel parameters and node location. The channel parameters (throughput, packet loss rate, maximum channel transmission rate, node location) are input into the trained XNN model for channel prediction.

[0097] Step S2.3: Analyze the prediction results of XNN, construct a utility matrix based on the influencing factors of different business types, and use the WD routing algorithm to achieve path prediction and complete path presetting.

[0098] Step S2.4: Determine whether the current path meets the service requirements. Set a threshold based on the requirements. If the channel quality deteriorates (such as excessive packet loss rate or low throughput), a path switch is triggered, switching to the preset path.

[0099] Step S2.5: After the switch, verify the performance of the new path in real time to check whether it meets requirements such as throughput, latency, and packet loss rate. If stable, continue to monitor the network status; if unstable, fall back to the previous path or continue to search for a new path.

[0100] Step S2.6: Summarize and record the path switching log information and feed it back to the XNN model as training data to further improve the accuracy of the model.

[0101] New network performance evaluation parameters are now considered. In network performance evaluation, maximum channel capacity, packet loss rate, and throughput are key parameters affecting network quality: maximum channel capacity determines the theoretical transmission limit of the network, packet loss rate reflects the reliability of data transmission, and throughput indicates the actual data transmission efficiency. By comprehensively considering these parameters and utilizing the TQ value of the BATMAN protocol, a more comprehensive and accurate network performance evaluation model can be constructed, providing an important basis for improving network reliability, efficiency, and intelligent management.

[0102] At the same time, to further enhance the adaptability and intelligence of the network, this application also innovatively introduces a channel modeling and prediction mechanism based on an interpretable neural network (XNN) on the basis of traditional network routing maintenance. By integrating these parameters and combining the learning capabilities of XNN, the system can predict and preset node path selection to build a more comprehensive and intelligent network performance evaluation model. In the event of a link failure, routing switching can be completed seamlessly to ensure network stability.

[0103] (3) Routing usage process Figure 4 As shown in the figure, using the SDN network controller, broadcasting requirements collect and report node status information, introducing a routing selection process based on the dynamic nature of the network status and the specificity of the service, and selecting the best route and secondary route based on the node status and service characteristics. Specifically, the following steps are included:

[0104] Step A1: The service connects to the node and requests transmission. The node reports the service type and transmission request to the SDN controller.

[0105] Step A1.1: The RYU network controller on a host communicates with the OpenvSwitch on the node, using the OpenFlow protocol to exchange information. After receiving a transmission request, the RYU network controller broadcasts a collection message to collect the status information of all nodes.

[0106] Step A1.2: When the node is in the routing maintenance phase, it sends channel information irregularly. Only when the routing link deteriorates or the node moves, the channel information is re-uploaded to the router, which can effectively reduce the redundant data transmission burden.

[0107] Step A2: After the SDN sends the broadcast, it collects and uploads the maximum channel capacity, throughput, and packet loss rate information for each node in the 5.8GHz and 2.4GHz frequency bands. It then constructs a maximum channel capacity matrix, a throughput matrix, and a packet loss rate matrix, weighting them using impact factors. Different impact factors can be adjusted for different service requests, ultimately constructing a channel matrix containing information about each node. The specific steps are as follows:

[0108] Step A2.1: In a mesh network built with the Batman protocol, you can use the iwdev mesh0 stationdump command to obtain network data from the mesh nodes in the network structure. The main observations are the received signal strength RSSI, the transmit bit rate tx_bitrate, the receive bit rate rx_bitrate, and the signal noise intensity N0, and calculate the throughput parameter T x , the calculation formula is:

[0109] The maximum channel capacity C is calculated using the Shannon formula: Where B is the channel bandwidth;

[0110] The batctlo command can be used to obtain the TQ value parameter. The larger the TQ value, the lower the packet loss rate of the link.

[0111] Normalization processing parameters C, T x and TQ value, and obtain the processed parameter C r 、T r and P r , the calculation formula is:

[0112]

[0113] Among them, C max It is expressed as the theoretical maximum value of the channel capacity, T max It represents the theoretical maximum value of the channel throughput parameter.

[0114] Step A2.2: Construct matrices based on the information obtained Multiply the influence factors of each parameter and add them together. Then compare each element in the 2.4G and 5.8G channel information matrices, and finally obtain the channel matrix This matrix effectively displays channel information between nodes and selects frequency bands based on element values ​​to achieve optimal transmission. Voice services require high stability, so the packet loss rate impact factor can be increased to prioritize transmission. Video services require high throughput, so the throughput impact factor can be increased to meet these high throughput requirements.

[0115] Step A3: The channel matrix uses the improved WD algorithm to select the optimal path and find the secondary path. The SDN controller then sends the routing table to each node and updates the routing table of each node. The specific steps are as follows:

[0116] A3.1: Using the WD algorithm, create a graph representation using a channel information matrix. Set the initial distance to 0 for the starting node and to infinity for all other nodes. Create a priority queue (minimum heap) to store each node and its distance to the starting node, initially containing only the starting node. Use the Dijkstra algorithm to traverse and calculate the shortest path from the source node to the destination node. After finding the shortest path, reset the distance information for all nodes, setting the distance to 0 for the starting node and infinity for all other nodes. Run the Dijkstra algorithm again, but this time ensure that edges in the shortest path are not repeated. Repeat the Dijkstra algorithm process, but add a check to ensure that the same edges as those in the shortest path already found are not selected. Stop the algorithm when a path to the destination node is found that is not the shortest path. Output the shortest path and the next shortest path from the starting node to the destination node. If no next shortest path is found, output the shortest path and None.

[0117] In summary, this application adopts a software-defined approach, with the controller uniformly allocating routes, realizing the function of on-demand routing based on business types; in addition, channel modeling based on an interpretable neural network model (XNN) is used to achieve channel quality prediction and path presetting, and complete seamless routing switching when a failure occurs. By collecting network status information, node weights are adaptively changed, and appropriate routes are selected based on node performance conditions; this study runs at the data link layer and transmits routing information through Ethernet frames, reducing data packet processing delays. Compared with routing strategies based on the shortest path, this study considers multiple channel parameters and adaptively selects routes and frequencies.

[0118] Based on the above method, this application conducted an experimental comparison and found that the throughput and packet loss rate of the algorithm based on the present invention were significantly improved. The experimental data are as follows:

[0119]

[0120] The experiment uses 3 or 4 nodes for transmission test. The experimental scenarios are divided into L-type (NLOS) and linear (LOS). UDP is used to test the improvement of packet loss rate, and TCP is used to test the improvement of throughput. Figure 5, by observing the selection of the route, whether the route decision is different from that of the traditional batman, can be used to show the effectiveness of the application applied to the wireless communication scheme to improve the adaptability and performance of the network in a high interference environment, and the throughput and packet loss rate are obviously improved.

[0121] In yet another aspect, the application also discloses a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to make the processor execute the steps of the above method.

[0122] In yet another aspect, the application also discloses a computer device, which comprises a memory and a processor, and the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the above method.

[0123] In yet another embodiment of the application, a computer program product containing instructions is also provided, which, when running on a computer, causes the computer to execute the channel modeling based multi-hop multi-frequency wireless mesh adaptive routing method in any of the above embodiments.

[0124] It can be understood that the system provided by the embodiments of the application corresponds to the method provided by the embodiments of the application, and the explanation, examples and beneficial effects of the related content can refer to the corresponding part in the above method.

[0125] The application also provides an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus,

[0126] The memory is used to store a computer program.

[0127] The processor is used to execute the program stored on the memory, and realize the channel modeling based multi-hop multi-frequency wireless mesh adaptive routing method.

[0128] The communication bus mentioned in the above electronic device can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0129] The communication interface is used for communication between the above electronic device and other devices.

[0130] The memory can include a random access memory (RAM) and can also include a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory can also be at least one storage device located away from the aforementioned processor.

[0131] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0132] It should also be noted that electronic devices also include terminal devices, which can also be referred to as terminals, user equipment, mobile stations, mobile terminals, etc. Terminal devices can be mobile phones, smart TVs, wearable devices, tablet computers, computers with wireless transceiver functions, virtual reality terminal devices, augmented reality terminal devices, wireless terminals in industrial control, wireless terminals in unmanned driving, wireless terminals in remote surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, etc. The embodiments of this application do not limit the specific technologies and specific device forms used by terminal devices.

[0133] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium, or a semiconductor medium (e.g., a solid-state drive).

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0135] In addition, it needs to be explained that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement condition, etc. between components in a certain posture, and if the certain posture changes, the directional indications will also change accordingly.

[0136] In addition, if the embodiments of the present application involve descriptions such as "first", "second", etc., the descriptions of "first", "second", etc. are only for description purposes and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include at least one of the features. In addition, the meaning of "and / or" appearing throughout the text includes three parallel schemes. Taking "A and / or B" as an example, it includes A scheme, or B scheme, or A and B scheme. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the realization of ordinary skilled in the art, when the combination of technical solutions appears contradictory or cannot be realized, it should be considered that the combination of technical solutions does not exist, nor is it within the scope of protection required by the present application.

Claims

1. A multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling, characterized in that: The following steps are involved: S1. Use the Batman protocol for route discovery and networking to build a mesh network; S2. Use interpretable neural networks for channel modeling to predict channel information and determine link degradation and node mobility. Use routing protocols for rapid convergence and use potential paths to maintain routing for current data transmission. S3. The node reports the service type and transmission request to the network controller. The network controller broadcasts and collects the status information of each node. Based on the dynamic network status and service characteristics, a routing selection process is introduced to select the optimal route and secondary routes based on the node status and service characteristics. The specific operation process in the S2 step includes: S21 receives the channel parameter information of the node through the network controller as a channel quality indicator; S22. Use an interpretable neural network for self-learning channel modeling. Build a channel environment model by analyzing channel parameters and node locations. Input historical channel parameter data into the trained channel environment model to perform channel prediction. S23. According to different business types and the use of WD algorithm to complete the path preset, for when a link failure occurs, to achieve seamless link switching; The specific construction process of the channel environment model in step S22 includes: S221. Preprocess the channel training data by normalization, transform the original channel data to the same dimension and map it to the given [0, 1]; S222. Initialize the model weights and parameters, and use an interpretable neural network to train the preprocessed data. During the training process, define the loss function as mean squared error and add an L1 penalty to control model complexity and reduce overfitting. At the same time, use the Adam optimizer to determine the adaptive learning rate. Iterate the training until the specified number of iterations is reached or the specified threshold error is met. S223. After the training is completed, a channel environment model based on an interpretable neural network is obtained; The specific operation process of step S23 includes: S231. Based on the channel prediction results of the channel environment model, a utility matrix is ​​constructed for the influencing factors of different business types, and the WD routing algorithm is used to achieve path prediction and complete path presetting; S232 sets the channel quality threshold according to business needs to determine whether the channel quality meets the threshold requirements. If the current path channel quality drops below the threshold requirement, the path switching is triggered and the path is switched to the preset path. S233. After switching, the performance of the new path is verified in real time according to the channel quality threshold. After the path channel is stable, the network status is continuously monitored. If the path channel is unstable, the path is returned to the previous path or a new path is searched for. S234. Summarize and record the path switching log information and feed it back to the channel environment model as training data.

2. The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling according to claim 1, characterized in that: The specific operation process of step S1 includes: S11. All nodes generate and send OGM messages regularly; S12 neighbor node receives the OGM message after checking the validity of the message to determine whether it is a duplicate message, if it is a non-duplicate message, it is forwarded to other neighbor nodes, and according to the message hop number field is incremented to indicate that the path hop count increases; S13. All nodes gradually build and update their own routing tables based on the OGM messages received from neighboring nodes.

3. The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling according to claim 1, wherein: During the execution of step S2, the batman protocol is used to perform routing maintenance, and the specific maintenance process includes: Each node sends source announcement data to neighboring nodes regularly according to the Batman protocol and updates the node routing table according to the network status. If a neighboring node loses connection or does not receive source announcement data, the neighbor is removed from the node routing table.

4. The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling according to claim 1, wherein: The specific operation process of the S3 step includes: S31 business connection node reports the business type and transmission request to the network controller; S32. After sending the broadcast, the network controller uses the batctl and iwinfo commands to obtain the underlying parameters of the batman protocol. It collects and uploads multiple parameter information for each node in the 5.8 GHz and 2.4 GHz bands, including RSSI, throughput, and TQ. It then constructs a maximum channel capacity matrix, a throughput matrix, and a packet loss rate matrix, respectively. These matrices are multiplied by the influencing factors of each parameter and then added together. S33. Compare each element in the 2.4G and 5.8G channel information matrices to obtain the channel matrix, and select the optimal transmission frequency band based on the size of the matrix element value. The channel matrix uses the WD algorithm to find the optimal path and secondary path; S34. The network controller sends the routing table to each node and updates the routing table of each node.

5. The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling according to claim 4, characterized in that: The information collection process in step S32 includes: S321. Get mesh node network data of mesh network; S322. Observe the received signal strength RSSI, transmit bit rate tx_bitrate, receive bit rate rx_bitrate and signal noise strength , calculate the throughput parameters , the calculation formula is: ; S323. Calculate the maximum channel capacity , the calculation formula is: in is the channel bandwidth; S324. Use the batctlo command to obtain the TQ value parameter. The larger the TQ value, the lower the packet loss rate of the link. S325. Normalization processing parameters 、 and , get the processed parameters 、 and , the calculation formula is: in, Expressed as the theoretical maximum channel capacity, Expressed as the theoretical maximum value of the channel throughput parameter.

6. The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling according to claim 4, characterized in that: In step S33, the WD algorithm is run through the SDN network controller to obtain the shortest path and the second shortest path from the source node to the destination node, and the corresponding IP addresses are collected. According to the source IP, destination IP, port number, frequency band selection, and frequency point selection, a policy is issued to the nodes on each path. The WD algorithm steps used include: S331 creates a graph representation using a channel information matrix, sets the initial distance of the starting node to 0, and sets the initial distance of all other nodes to infinity; S332. Create a minimum heap priority queue to store each node and its distance to the starting node, initially containing only the starting node; S333. Use the Dijkstra algorithm to traverse and calculate the shortest path from the source node to the destination node. After finding the shortest path, reset the distance information of all nodes, set the distance of the starting node to 0, and set the distance of other nodes to infinity. Run the Dijkstra algorithm again, and ensure that the edges of the shortest path are not repeated. S334. Repeat the D algorithm traversal process, adding a check to ensure that the same edge as the shortest path found is not selected during the D algorithm traversal process. When a path to the target node is found and the path is not the shortest path, stop the algorithm. S335 outputs the shortest path and the second shortest path from the start node to the target node. If there is no second shortest path, the shortest path and None are output.

7. The multi-hop multi-frequency wireless mesh network adaptive routing method based on channel modeling according to claim 1, wherein: In the step S3, the transmission communication between the service connection node and the network controller is realized through the RYU network controller on a server and the OpenvSwitch on the node, and the Openflow protocol is used to realize information interaction.

Images