Ad-hoc network multipath routing simulation method for cross-platform parameter interaction

Abstract

The invention relates to the technical field of routing simulation, in particular to an ad hoc network multipath routing simulation method for cross-platform parameter interaction. By constructing a weight simulation platform and a routing simulation platform, weight calculation and a routing algorithm can be independently developed on different platforms, and data synchronization is realized through socket-based communication, so that the problem of test limitation caused by platform closure in ad hoc network routing algorithm development is solved; according to the method, a multi-path routing algorithm with different weight parameters can be cooperatively developed and tested by different platforms, congestion diffusion and stability mutation are captured in advance by utilizing a space-time differential operator through space-time differential characteristics and a Monte Carlo Dropout graph convolution prediction mechanism, and network oscillation caused by prediction errors is avoided by dynamically fusing observation values and prediction values; the technical problems that at present, it is difficult to directly associate a weight routing algorithm and a multipath routing algorithm independently developed by different platforms to achieve decentralized development, so that time is saved, and cost is reduced are solved.

Description

Technical Field

[0001] This invention relates to the field of routing simulation technology, and in particular to a cross-platform parameter interaction method for simulating multipath routing in self-organizing networks. Background Technology

[0002] Routing algorithms for ad hoc networks play a crucial role in the communication performance of these networks. To this end, numerous researchers have devoted significant effort to research, proposing a multitude of routing algorithms with varying weights and diverse multipath routing techniques. The design and evaluation of these algorithms and techniques are core elements in building efficient networks. Currently, various simulation platforms have been developed to conduct detailed tests on the routing performance of ad hoc networks. However, existing weighted routing algorithms and multipath routing algorithms can only be tested and correlated as a whole on a single simulation platform, making it difficult to directly correlate weighted routing algorithms and multipath routing algorithms developed independently on different platforms to achieve decentralized development. Therefore, for some complex ad hoc networks, this significantly increases development time and cost. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a cross-platform parameter interaction self-organizing network multipath routing simulation method, which solves the current technical problem of being unable to directly associate weighted routing algorithms and multipath routing algorithms independently developed on different platforms to achieve decentralized development, thereby saving time and reducing costs.

[0004] To address the aforementioned technical problems, this invention provides the following technical solution: a cross-platform parameter interaction self-organizing network multipath routing simulation method, which specifically includes the following steps: S1. Set up several weight simulation platforms with independently developed weight calculation methods and several routing simulation platforms with independently developed multipath routing algorithms; S2. Create socket communication programs independently on the weight simulation platform and the routing simulation platform, and run either the server side or the server side on the weight simulation platform and the routing simulation platform. S3. The routing simulation platform calculates the link data required for weighting and transmits the link data to each weighting simulation platform through a socket communication program. S4. The weight simulation platform calculates the weight of each available route based on the weight calculation method of the local machine and the received link data, selects the optimal route, and transmits the corresponding relay node IP address to the route simulation platform through the socket communication program. S5. The routing simulation platform selects routes and communicates based on the IP address of the relay node of the received optimal route.

[0005] Preferably, in step S1, the link data includes the data packets transmitted per unit time, the total number of packets sent up to the current time, the total lifetime of the local next-hop node within the local coverage area, and the routing sequence number of the local next-hop node.

[0006] Preferably, the total number of packets sent up to the current moment is updated according to the following method: The formula for calculating the total number of packets sent up to the current moment, for each data packet sent by a node, is: The formula for calculating the total number of packets sent up to the current moment, for each data packet received by the node, is: In the above formula, and These represent the total number of packets sent up to the current time, before and after the update, respectively.

[0007] Preferably, in step S4, the weight of each available route includes a congestion weight and a stability weight, and the formula for calculating the congestion weight is as follows: The formula for calculating the stability weight is: In the above formula, and These represent congestion weight and stability weight, respectively, and pcktsps represents the number of data packets transmitted per unit time. Indicates the first parameter. This indicates the second parameter, which is equal to the maximum number of packets that can be sent divided by 10. The maximum number of packets that can be sent is the maximum number of data packets that a node can send when it is fully charged. The third parameter is represented by pckts, which represents the total number of packets sent up to the current time, Nblt, which represents the total lifetime of the next-hop node within the local coverage area, and seqno, which is the routing sequence number of the next-hop node.

[0008] Preferably, in step S4, the method for selecting the optimal route is as follows: choose The route with the lowest congestion level is the route with the best congestion level. choose The largest route is the optimal stability route.

[0009] Preferably, before the weight simulation platform calculates the weight of each available route based on the local weight calculation method and the received link data, it predicts and compensates for the link status. The specific steps are as follows: S401. Calculate the spatial and temporal differential eigenvalues ​​of each node. The calculation formula is as follows: In the above formula, and Let represent the spatial and temporal differential eigenvalues ​​of node i, respectively, and N represent the number of neighboring nodes of node i. and Let represent the data packets transmitted by node i and its neighbor node j within a unit of time, respectively. This represents the Euclidean distance between node i and node j. and These represent the total number of packets sent up to the previous time and up to the time before that time, respectively. S402. Construct a four-dimensional feature vector containing time differential feature values, spatial differential feature values, data packets transmitted within a bit time, and the total number of packets sent up to the current time. S403. Perform several sampling predictions using a Monte Carlo Dropout graph convolutional network, and output the prediction mean PredMean and standard deviation PredStd. S404. Compensate the data for each link and use the compensated link data as the link data at the current moment. The calculation formula is as follows: In the above formula, and These represent the parameter values ​​of the link data before and after compensation, respectively. PredMean represents the predicted mean, and StdNorm represents the standard deviation standardized to the interval [0, 1]. This represents the fusion weighting coefficient.

[0010] Preferably, in step S404, the formula for calculating the standard deviation StdNorm standardized to the [0, 1] interval is: In the above formula, and These represent the maximum and minimum standard deviations of the training set, respectively.

[0011] Preferably, the Monte Carlo Dropout graphical convolutional network uses a Chebyshev polynomial kernel function as the spatial convolutional layer, with a polynomial order of 3. It also uses a one-dimensional causal convolution with a kernel size of 3 as the temporal convolutional layer. The activation function of the Monte Carlo Dropout graphical convolutional network is... .

[0012] Preferably, step S5 specifically includes the following steps: S51. Construct a path selection decision function, the function expression of which is: In the above formula, Representing quantum decision states, This represents the overall weight of path k. Describes the quantum state of path k. This represents the return coefficient, and M represents the number of available paths; S52. Perform space-time block coding on the data packets to be sent. The expression is: In the above formula, This represents the spatiotemporal block encoding of the i-th data packet. Indicates the timestamp of the data packet's generation. Indicates life cycle, This represents the coordinates of the source node of the i-th data packet. This indicates the type of the i-th data packet, where 0 represents low latency, 1 represents high bandwidth, and H represents the Hadamard transformation matrix. This represents the feature vector of the i-th data packet; S53. Calculate the path allocation probability of data packets being assigned to each path. The calculation formula is: In the above formula, The spacetime curvature of path k is represented. and Let (u, v) represent the mean and standard deviation of the current grid curvature, respectively, and (u, v) represent the communication link on path k, where u represents the sending node and v represents the receiving node. The time acceleration representing the survival time of neighbors. For a second-order differential operator in the time dimension, This represents the lifetime of the neighboring nodes of the link (u, v). The spatial gradient representing the packet sending rate, The first-order differential operator representing the spatial dimension, This represents the number of packets sent per unit time along the link (u, v).

[0013] S54. Based on the path allocation probability, use the roulette wheel algorithm to select a communication path for the data packets.

[0014] By employing the above technical solution, the present invention provides a cross-platform parameter interaction self-organizing network multipath routing simulation method, which has at least the following beneficial effects: 1. This invention constructs a cross-platform weight simulation platform and routing simulation platform collaborative mechanism, allowing weight calculation and routing algorithms to be developed independently on different professional platforms. Data synchronization is achieved through socket-based dual-channel communication, solving the testing limitations caused by platform closure in the development of self-organizing network routing algorithms. This enables different platforms to collaboratively develop and test multipath routing algorithms with different weight parameters.

[0015] 2. This invention utilizes spatiotemporal differential features and Monte Carlo Dropout graph convolution prediction mechanism to capture congestion diffusion and stability mutations in advance using spatiotemporal differential operators. By dynamically fusing observed and predicted values, it avoids network oscillations caused by prediction errors and improves the accuracy of link data in real time.

[0016] 3. This invention distinguishes video streams from file streams through spatiotemporal block coding and greatly reduces the probability of high-instability path allocation through curvature penalty terms, thereby significantly improving the splitting efficiency and transmission stability of different types of data packets. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of the self-organizing network multipath routing simulation method of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This will allow for a full understanding of how the present application uses technical means to solve technical problems and achieve technical effects, and to facilitate its implementation.

[0019] To address the current technical challenge of directly linking independently developed weighted routing algorithms and multipath routing algorithms across different platforms to achieve decentralized development and thus save time and reduce costs, this invention proposes a cross-platform parameter interaction-based self-organizing network multipath routing simulation method. This method supports collaborative development and testing of multipath routing algorithms based on different weight parameters across different platforms. The cross-platform development simultaneously provides professional model debugging capabilities and a user-friendly visual development environment, offering both code-level tracing and testing, and facilitating a more intuitive understanding of the simulation system and environment for users. The specific process of this simulation method is as follows: S1. Set up several independent weight simulation platforms for developing weight calculation methods and several independent routing simulation platforms for developing multipath routing algorithms. Weight calculation requires complex mathematical models (such as differential equations and neural networks), which are suitable for development on the Matlab / Python platform. Multipath routing algorithms require network protocol stack support, which are suitable for development on the NS-3 / OMNeT++ platform. Assume there are four different simulation platforms, where platforms A and B develop weight calculation methods a and b, and platforms C and D develop multipath routing algorithms c and d. Link data includes the number of data packets transmitted per unit time, the total number of packets sent up to the current time, the total lifetime of the local next-hop node within the local coverage area, and the routing sequence number of the local next-hop node. The total number of packets sent up to the current time is updated according to the following method: The formula for calculating the total number of packets sent up to the current moment, for each data packet sent by a node, is: The formula for calculating the total number of packets sent up to the current moment, for each data packet received by the node, is: In the above formula, and These represent the total number of packets sent up to the current time, before and after the update, respectively.

[0020] S2. Create socket communication programs independently on the weight simulation platform and the routing simulation platform. To enable communication, each platform needs a dedicated communication port (e.g., UDP 5000-5020 for the weight platform, UDP 5021-5040 for the routing platform) to lay the foundation for subsequent socket communication. The choice of socket communication is based on two points: 1. TCP protocol ensures the reliability of weight parameter transmission; 2. UDP protocol is used for routing command transmission, meeting low latency requirements. Then, choose to run either the server-side or the routing simulation platform. For example, run the client on the weight simulation platform and the server-side on the routing simulation platform. The positions of the client and server-side can be interchanged. S3, the routing simulation platform can use a multi-threaded Socket service architecture to calculate the link data required for weights, and transmit the link data to each weight simulation platform through a socket communication program; S4. The weight simulation platform calculates the weight of each available route based on the local weight calculation method and the received link data. When the destination node receives a route request, it returns a response RREP. When the relay node receives this RREP, it calculates its own weight and compares it with the weight of the RREP. If the current node's weight is larger, the RREP weight is updated. When the RREP reaches the source node, the source node knows the worst-case scenario for each node on each path. Before calculating the weight, the link data is first predicted and compensated. The specific steps are as follows: S401. Calculate the spatial differential characteristic value of each node to quantify the direction of congestion propagation, and the temporal differential characteristic value to capture the acceleration of stability changes. The calculation formula is as follows: In the above formula, and Let represent the spatial and temporal differential eigenvalues ​​of node i, respectively, and N represent the number of neighboring nodes of node i. and Let represent the data packets transmitted by node i and its neighbor node j within a unit of time, respectively. This represents the Euclidean distance between node i and node j. and These represent the total number of packets sent up to the previous time and up to the time before that time, respectively. S402. Spatial differential eigenvalues ​​and temporal differential eigenvalues ​​need to be fused with the original data to fully describe the network state. Therefore, a four-dimensional feature vector is constructed that includes temporal differential eigenvalues, spatial differential eigenvalues, data packets transmitted within a bit time, and the total number of packets sent up to the current time. S403. The four-dimensional feature vector is used as input to the Monte Carlo Dropout graph convolutional network. Several sampling predictions are then performed, and the predicted mean (PredMean) and standard deviation (PredStd) are output. The Monte Carlo Dropout graph convolutional network uses a Chebyshev polynomial kernel function as the spatial convolutional layer to capture the influence of topological neighbors. The polynomial order of the Chebyshev polynomial kernel function is 3. A one-dimensional causal convolution with a kernel size of 3 is used as the temporal convolutional layer to handle historical state dependencies. The activation function of the Monte Carlo Dropout graph convolutional network is... .

[0021] S404. Next, the predicted results are fused with the actual values ​​to compensate for the data of each link. The compensated link data is then used as the link data at the current moment. The calculation formula is as follows: In the above formula, and represents the parameter values ​​of the link data before and after compensation, respectively. PredMean represents the predicted mean, StdNorm represents the standard deviation standardized to the interval [0, 1], 10 is an experimental optimization value, Yongkai controls the transition steepness, and 0.5 is the confidence threshold obtained from the experiment. Indicates the fusion weighting coefficient. and These represent the maximum and minimum standard deviations of the training set, respectively.

[0022] Then, the weight of each available route can be calculated based on the compensated link data. The weight of each available route includes a congestion weight and a stability weight. The formula for calculating the congestion weight is as follows: The formula for calculating the stability weight is: In the above formula, and These represent congestion weight and stability weight, respectively, and pcktsps represents the number of data packets transmitted per unit time. Indicates the first parameter. This indicates the second parameter, which is equal to the maximum number of packets that can be sent divided by 10. The maximum number of packets that can be sent is the maximum number of data packets that a node can send when it is fully charged. The third parameter is represented by pckts, which represents the total number of packets sent up to the current time, Nblt, which represents the total lifetime of the next-hop node within the local coverage area, and seqno, which is the routing sequence number of the next-hop node.

[0023] Then, the optimal route can be selected based on the calculated congestion weight and stability weight, and the corresponding relay node IP address can be transmitted to the routing simulation platform via a socket communication program. The method for selecting the optimal route is as follows: Select The route with the lowest congestion is the optimal route; therefore, it should be selected. The largest route is the optimal stability route.

[0024] S5. The routing simulation platform selects routes and communicates based on the IP address of the relay node of the received optimal route. Compared with the traditional classic greedy algorithm for path selection efficiency, this invention provides a more efficient method, which specifically includes the following steps: S51. Construct a path selection decision function, the function expression of which is: In the above formula, Representing quantum decision states, This represents the overall weight of path k. Reflects the amplitude of path k. Describes the quantum state of path k. This represents the return coefficient, and M represents the number of available paths. It is a congestion penalty item. The larger the value, the lower the weight. S52. Perform space-time block coding on the data packets to be sent. The expression is: In the above formula, This represents the spatiotemporal block encoding of the i-th data packet. Indicates the timestamp of the data packet's generation. Indicates life cycle, This represents the coordinates of the source node of the i-th data packet. This indicates the type of the i-th data packet, where 0 represents low latency, 1 represents high bandwidth, and H represents the Hadamard transformation matrix. This represents the feature vector of the i-th data packet; S53. Calculate the path allocation probability of data packets being assigned to each path. The calculation formula is: In the above formula, The spacetime curvature of path k is represented. and Let (u, v) represent the mean and standard deviation of the current grid curvature, respectively, and (u, v) represent the communication link on path k, where u represents the sending node and v represents the receiving node. The time acceleration representing the survival time of neighbors. For a second-order differential operator in the time dimension, This represents the lifetime of the neighboring nodes of the link (u, v). The spatial gradient representing the packet sending rate, The first-order differential operator representing the spatial dimension. This represents the number of packets sent per unit time along the link (u, v).

[0025] S54. Based on the path allocation probability, use the roulette wheel algorithm to select a communication path for the data packets.

[0026] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0027] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Since the above embodiments are substantially similar to the method embodiments, their descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0028] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A cross-platform parameter interaction simulation method for self-organizing network multipath routing, characterized in that, The simulation method specifically includes the following steps: S1. Set up several weight simulation platforms with independently developed weight calculation methods and several routing simulation platforms with independently developed multipath routing algorithms; S2. Create socket communication programs independently on the weight simulation platform and the routing simulation platform, and run either the server side or the server side on the weight simulation platform and the routing simulation platform. S3. The routing simulation platform calculates the link data required for weighting and transmits the link data to each weighting simulation platform through a socket communication program. S4. The weight simulation platform calculates the weight of each available route based on the weight calculation method of the local machine and the received link data, selects the optimal route, and transmits the corresponding relay node IP address to the route simulation platform through the socket communication program. S5. The routing simulation platform selects routes and communicates based on the IP address of the relay node of the received optimal route.

2. The self-organizing network multipath routing simulation method according to claim 1, characterized in that, In step S1, the link data includes the data packets transmitted per unit time, the total number of packets sent up to the current time, the total lifetime of the local next-hop node within the local coverage area, and the routing sequence number of the local next-hop node.

3. The self-organizing network multipath routing simulation method according to claim 2, characterized in that, The total number of packets sent up to the current moment is updated according to the following method: The formula for calculating the total number of packets sent up to the current moment, for each data packet sent by a node, is: The formula for calculating the total number of packets sent up to the current moment, for each data packet received by the node, is: In the above formula, and These represent the total number of packets sent up to the current time, before and after the update, respectively.

4. The self-organizing network multipath routing simulation method according to claim 1, characterized in that, In step S4, the weight of each available route includes a congestion weight and a stability weight. The formula for calculating the congestion weight is as follows: The formula for calculating the stability weight is: In the above formula, and These represent congestion weight and stability weight, respectively, and pcktsps represents the number of data packets transmitted per unit time. Indicates the first parameter. This indicates the second parameter, which is equal to the maximum number of packets that can be sent divided by 10. The maximum number of packets that can be sent is the maximum number of data packets that a node can send when it is fully charged. The third parameter is represented by pckts, which represents the total number of packets sent up to the current time, Nblt, which represents the total lifetime of the next-hop node within the local coverage area, and seqno, which is the routing sequence number of the next-hop node.

5. The self-organizing network multipath routing simulation method according to claim 4, characterized in that, In step S4, the method for selecting the optimal route is as follows: choose The route with the lowest congestion level is the route with the best congestion level. choose The largest route is the optimal stability route.

6. The self-organizing network multipath routing simulation method according to claim 2, characterized in that, Before the weight simulation platform calculates the weight of each available route based on the local weight calculation method and the received link data, it predicts and compensates for the link status. The specific steps are as follows: S401. Calculate the spatial and temporal differential eigenvalues ​​of each node. The calculation formula is as follows: In the above formula, and Let represent the spatial and temporal differential eigenvalues ​​of node i, respectively, and N represent the number of neighboring nodes of node i. and Let represent the data packets transmitted by node i and its neighbor node j within a unit of time, respectively. This represents the Euclidean distance between node i and node j. and These represent the total number of packets sent up to the previous time step and up to the time step two time steps ago, respectively. S402. Construct a four-dimensional feature vector containing time differential feature values, spatial differential feature values, data packets transmitted within a bit time, and the total number of packets sent up to the current time. S403. Perform several sampling predictions using a Monte Carlo Dropout graph convolutional network, and output the prediction mean PredMean and standard deviation PredStd. S404. Compensate the data for each link and use the compensated link data as the link data at the current moment. The calculation formula is as follows: In the above formula, and These represent the parameter values ​​of the link data before and after compensation, respectively. PredMean represents the predicted mean, and StdNorm represents the standard deviation standardized to the interval [0, 1]. This represents the fusion weighting coefficient.

7. The self-organizing network multipath routing simulation method according to claim 6, characterized in that, In step S404, the formula for calculating the standard deviation StdNorm standardized to the [0,1] interval is: In the above formula, and These represent the maximum and minimum standard deviations of the training set, respectively.

8. The self-organizing network multipath routing simulation method according to claim 6, characterized in that, The Monte Carlo Dropout Graph Convolutional Network uses a Chebyshev polynomial kernel function as the spatial convolutional layer, with a polynomial order of 3. It uses a one-dimensional causal convolution with a kernel size of 3 as the temporal convolutional layer. The activation function of the Monte Carlo Dropout Graph Convolutional Network is... .

9. The self-organizing network multipath routing simulation method according to claim 1, characterized in that, Step S5 specifically includes the following steps: S51. Construct a path selection decision function, the function expression of which is: In the above formula, Representing quantum decision states, This represents the overall weight of path k. Let z represent the quantum state of path k, z represent the annealing coefficient, and M represent the number of available paths; S52. Perform space-time block coding on the data packets to be sent. The expression is: In the above formula, This represents the spatiotemporal block encoding of the i-th data packet. Indicates the timestamp of the data packet's generation. Indicates life cycle, This represents the coordinates of the source node of the i-th data packet. This indicates the type of the i-th data packet, where 0 represents low latency, 1 represents high bandwidth, and H represents the Hadamard transformation matrix. This represents the feature vector of the i-th data packet; S53. Calculate the path allocation probability of data packets being assigned to each path. The calculation formula is: In the above formula, The spacetime curvature of path k is represented. and Let (u, v) represent the mean and standard deviation of the current grid curvature, respectively, and (u, v) represent the communication link on path k, where u represents the sending node and v represents the receiving node. The time acceleration representing the survival time of neighbors. For the second-order differential operator in the time dimension, This represents the lifetime of the neighboring nodes of the link (u, v). The spatial gradient representing the packet sending rate, The first-order differential operator representing the spatial dimension. This represents the number of packets sent per unit time along the link (u, v); S54. Based on the path allocation probability, use the roulette wheel algorithm to select a communication path for the data packets.

Images