Task-driven MAC Protocol Reconstruction Method, Device, Equipment and Medium

Abstract

The present application relates to a task-driven MAC protocol reconstruction method, apparatus, device and medium. The method includes: on the one hand, considering the differences in service characteristics brought by node task heterogeneity, the network load is refined into three-dimensional features, and the constructed network load model can fully reflect the performance of the MAC protocol in different network environments. On the other hand, a closed-loop software-defined distributed subnet node system architecture is constructed. This architecture provides a solution for distributed nodes to cooperate to complete the reconstruction process. The reconstruction process considers different task requirements to solve the problem of balancing protocol performance, constructs a data set and uses a classification model to realize the reconstruction decision of different protocols. Finally, the optimal MAC protocol based on task requirements and taking into account different performance dimensions can be reconstructed, improving the adaptability and reliability of the MAC protocol to dynamic network environments and task requirements.

Description

Technical Field

[0001] This application relates to the field of network communication technologies, and particularly to a task-driven MAC protocol reconstruction method, apparatus, device, and medium. Background Art

[0002] A wireless ad hoc network is a distributed network that does not rely on fixed infrastructure. Its nodes can move freely and automatically configure connections, featuring characteristics such as being centerless, self-organizing, and having a dynamic topology. It is suitable for scenarios such as emergency response and disaster relief, and research on ad hoc networks has received extensive attention in recent years.

[0003] The multiple access control (MAC) protocol plays a crucial role in the process of network formation and information transmission, directly affecting the communication performance of the network. Generally speaking, MAC protocols applicable to wireless ad hoc networks can be divided into contention-based and scheduling-based types. Contention-based MAC protocols obtain transmission opportunities by seizing the channel, while scheduling-based protocols orderly access the channel by dividing the channel. The essential difference between the two types of protocols lies in their different channel utilization rates under different network conditions. Usually, the Poisson distribution is used to model the service arrival rate, and it is considered that using contention-based protocols in low-load environments and scheduling-based protocols in high-load environments can give full play to the respective advantages of the two types of protocols.

[0004] However, due to the essential differences in the access mechanisms of MAC protocols, there is no single protocol that can meet all network environments and requirements. In practical applications, the dynamic changes in the network environment and the time-varying nature of task requirements pose challenges to a single fixed MAC protocol. Therefore, an intelligent MAC protocol reconstruction algorithm is very necessary for a dynamic network environment. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a task-driven MAC protocol reconstruction method, apparatus, device, and medium. By re-modeling the network load, a network load model based on task characteristics is constructed. On this basis, the MAC protocol is reconstructed, and a closed-loop distributed node system architecture is designed to reconstruct the optimal MAC protocol that meets task requirements and takes into account different performance dimensions.

[0006] A task-driven MAC protocol reconstruction method, the method comprising:

[0007] Considering the multi-domain distributed wireless ad hoc network scenario, constructing a network load model through multi-dimensional feature division and extraction of network load, Gaussian random process modeling, and feature fusion;

[0008] Based on the network load model, a software-defined distributed subnet node system architecture is constructed, which consists of a task plane, a data plane, and a joint control plane;

[0009] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames into the data plane to form queues. The data plane queues include the service data frames generated by node tasks, and also include the control data frames generated by feature interaction during the MAC protocol reconstruction process and channel reservation and preemption behaviors in the MAC protocol process;

[0010] The joint control plane includes an observation plane, a knowledge plane, and a control plane; among them, the observation plane is used to observe and statistically analyze the services of each node itself, sequentially obtain the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, and predict the service characteristics of each node in the next reconstruction period through Kalman filtering. Then, the service characteristics of all single-hop reachable nodes obtained through broadcast interaction are fused to obtain the network characteristics in the next reconstruction period and transmitted to the knowledge plane;

[0011] The knowledge plane is used to carry a pre-trained classification model, and according to the network characteristics input by the observation plane and the task requirements input by the task plane, output the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane;

[0012] The control plane is used to carry a MAC protocol component library, and judge whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If it is the same, the reconstruction process is not executed; if it is different, search for components and adjust the protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction period.

[0013] In one embodiment, considering the multi-domain distributed wireless ad hoc network scenario, through multi-dimensional feature partitioning extraction, Gaussian random process modeling, and feature fusion of network loads, a network load model is constructed, including:

[0014] Considering the multi-domain distributed wireless ad hoc network scenario, partition and extract the multi-dimensional features of network loads, including network node scale, node activity level, and node service characteristics;

[0015] Use a multi-dimensional Gaussian random process with the same dimension to model the service characteristics of each node in the network, and use the weighted fusion algorithm of multi-dimensional Gaussian components to perform weighted fusion on the service characteristics of all nodes in the network to obtain the global network service characteristics;

[0016] Build a network load model based on the network node scale, the global network activity level obtained by averaging based on the activity levels of all nodes, and the global network service characteristics.

[0017] In one embodiment, a multi-dimensional Gaussian random process with the same dimension is used to model the service characteristics of each node in the network, including:

[0018] Considering that the tasks executed by each node in the network are different, a multi-dimensional Gaussian random process with the same dimension is used to model the service characteristics of each node in the network, expressed as , where is a set composed of all sample functions of the service characteristics of a single node at t time, , is the i th sample function, d is the number of sample functions, that is, the dimension number of the Gaussian random process is the same as the number of sample functions, is the period;

[0019] is a random process used to describe the change of the service length generated by the node over time. The -dimensional normal probability density function of this process is expressed as:

[0020] ;

[0021] where is the expectation value, represents a random variable of any sample function at time; is the expectation value, represents a random variable of any sample function at another time; is the variance of the random variable ; is the variance of the random variable ; represents the expectation calculation; is the determinant of the normalized covariance matrix ; represents the th service length generated by the node, represents the service characteristic random variable of the task mapped to d dimensions; represents any d times; and both represent the dimension;

[0022] Further considering that the tasks of each node in the network are different and the service types are relatively fixed, assuming that the service characteristics of each node are wide-sense stationary within a limited period of time, the service characteristics of each node in the network are modeled as a D-dimensional stationary Gaussian process, and the service characteristics of the stationary Gaussian process satisfy the following constraints:

[0023] ;

[0024] ;

[0025] where, is the expected value of; is an expected value constant, used to represent that the expected value of the stationary Gaussian process does not change with time; represents the value of the autocorrelation function at time and ; is the expected value of the product of and; is the time difference, that is ; is the autocorrelation function, which only depends on the time difference .

[0026] In one embodiment, the service characteristics of each node after modeling are a D-dimensional Gaussian distribution with the service length as a random vector at any moment. The service lengths of each dimension of node are described as a D-dimensional random vector , and the probability density function of its service characteristics is expressed as: ;

[0027] ;

[0028] where, is t the probability density function of the service characteristics of node at time; represents the mean vector of the service lengths of different types of node , represents the mean of the th service length of node i , , the superscript T represents the transpose; is the covariance matrix, used to represent the correlation between the mean vectors of the service lengths of different types of node , represents the determinant of the covariance matrix ;

[0029] By means of the -dimensional Gaussian distribution to describe the service characteristics of different nodes, the set of probability density functions of the service characteristics of each node in the network is expressed as ; where is the scale of the network nodes, and .

[0030] In one embodiment, the weighted fusion algorithm of multi-dimensional Gaussian components is used to perform weighted fusion on the service characteristics of all nodes in the network to obtain the global network service characteristics;

[0031] Using the -dimensional Gaussian component weighted fusion algorithm, considering that each node in the network satisfies the independent and identically distributed condition, the goal is to fuse the service characteristics of all nodes in the network to obtain a -dimensional Gaussian distribution of the global network service characteristics;

[0032] Assume that the Gaussian component of the service characteristic of node is , and the activity level of node is , and the activity level represents the probability that node generates services within a unit time; by normalizing , the weight of the service characteristic of each node is obtained, which is expressed as:

[0033] ;

[0034] where satisfies and ;

[0035] Based on the weight , the service characteristics of all nodes in the network are weighted and fused to obtain the global network service characteristics , and conforms to -dimensional Gaussian distribution, that is ; where, the mean vector of the global network service characteristics and the covariance matrix are respectively expressed as:

[0036] ;

[0037] ;

[0038] where , where represents the th mean in i , and ; Superscript T denotes transpose.

[0039] In one embodiment, a network load model is constructed by modeling based on the network node scale, the global network activity level obtained by averaging based on the activity levels of all nodes, and the global network service characteristics, including:

[0040] Based on the network node scale , the global network activity level obtained by averaging based on the activity levels of all nodes and the mean vector of the global network service characteristics are used for modeling to construct a network load model, where t the expectation of the network load at time is expressed as:

[0041] ;

[0042] where is the time-varying network load, is the expected value of; the global network activity level , representing the average probability of generating services by all nodes in the network per unit time; is the activity level of node ; is the activity level of node at the i th mean of the service length.

[0043] In one embodiment, the observation plane observes and statistically analyzes the services of each node itself, and obtains the service characteristics of each node's current reconstruction period through maximum likelihood estimation, including:

[0044] The observation plane observes and statistically analyzes the number of data generations by each node in the network during the current observation period and the data length observation data set through the task and service mapping between the observation plane and the task plane; where the reconstruction period contains time slots, and the observation period contains time slots, and satisfies i.e., ;

[0045] According to the statistical result of the number of data generations, the activity level of node n is calculated as , and the data length observation data set is used as an observation sample for service characteristic estimation and prediction;

[0046] Adopt Model the service characteristics of nodes with a D-dimensional stationary Gaussian random process, and reduce the D-dimensional Gaussian distribution of the service characteristics of nodes n to a one-dimensional Gaussian distribution for description. Then, according to the prior information that the data frame length variable at the sampling moment follows a Gaussian distribution, obtain the mean n and standard deviation of the service characteristics of nodes through maximum likelihood estimation; specifically, let , for the data set , its likelihood function is the product of the probability density functions of all observed values, expressed as:

[0047] ; n Model the service characteristics of nodes n of the service characteristics of nodes n to a one-dimensional Gaussian distribution for description, and then, according to the prior information that the data frame length variable at the sampling moment follows a Gaussian distribution, obtain the mean n and standard deviation of the service characteristics of nodes through maximum likelihood estimation; n mean and standard deviation ; Specifically, let For the data set , its likelihood function is the product of the probability density functions of all observed values, expressed as:

[0047] ;

[0048] where represents the data length generated at the th time within the observation period i , and ;

[0049] Take the logarithm of this likelihood function to obtain the log-likelihood function , and take the partial derivatives of with respect to and respectively and set them equal to zero, and solve to obtain and which are respectively expressed as:

[0050] ;

[0051] ;

[0052] Based on the above solutions and determine that the service characteristics of nodes n in the current reconstruction period follow ; where , is the scale of the network nodes.

[0053] In one embodiment, the observation surface predicts the service characteristics of each node in the next reconstruction period through Kalman filtering, and then fuses the service characteristics of all single-hop reachable nodes obtained through broadcast interaction to obtain the network characteristics in the next reconstruction period and transmit them to the knowledge surface, including:

[0054] Set that the true service characteristics of nodes in the th reconstruction period follow , and obtain that the service characteristics follow ; Among them, is the node at the true mean of the service characteristics in the th reconstruction period, and it is set that is the node at the th reconstruction period of the observed value of the service characteristics, representing the average data length, and it is set that is the true mean of the noisy measurement; is the node at the th reconstruction period of the variance of the service characteristics, reflecting the degree of data frame dispersion;

[0055] The goal of the Kalman filter is based on the historical observed values and and its corresponding variance and , predict that the service characteristics of the node at the th reconstruction period follow , and then weight and fuse the prediction results of the th reconstruction period with the service characteristics of all single-hop reachable nodes obtained by broadcast interaction to obtain the global network service characteristics of the th reconstruction period. Finally, combine the network node scale, the global network activity degree obtained by averaging the activity degrees of all nodes in the network, and the global network service characteristics into the network characteristics of the th reconstruction period, and transfer the network characteristics to the knowledge surface;

[0056] Among them, in the first two reconstruction periods when the network is initially established, the Kalman filter prediction process is not carried out. The new nodes added at the end of the reconstruction period do not participate in the algorithm process of the observation surface, and the observation surface only performs the processes of observation, maximum likelihood estimation, and feature interaction fusion in the first two reconstruction periods.

[0057] In one embodiment, the Kalman filter estimates the state through two-step iteration of prediction and update. First, obtain the prior estimates of the mean and variance of the service characteristics in the th reconstruction period through the prediction step. The predictions of the mean and variance are respectively expressed as:

[0058] ;

[0059] ;

[0060] Among them, is the The prior estimate of the mean of the business characteristics for a reconstruction cycle is the posterior estimate of the mean of the business characteristics for the -th reconstruction cycle; is the prior estimate of the variance of the business characteristics for the -th reconstruction cycle, is the posterior estimate of the variance of the business characteristics for the -th reconstruction cycle;

[0061] Then, based on the update step of the posterior estimate, the updated Kalman gain is obtained, and the expression is:

[0062] ;

[0063] According to the Kalman gain , the posterior estimates of the mean and the posterior estimate of the variance of the business characteristics for the -th reconstruction cycle are updated respectively, and are expressed as:

[0064] ;

[0065] ;

[0066] Thus, at the next iteration, the predicted distribution of the business characteristics of node in the -th reconstruction cycle is , and respectively represent the predicted values of the mean and the variance of the business characteristics of node in the -th reconstruction cycle; among them, the prior estimate of the mean of the business characteristics of node in the -th reconstruction cycle is predicted as , which is obtained by prediction through the exponential smoothing method and is expressed as ; among them, is the preset smoothing exponent, represents the -th predicted value of the variance of the business characteristics.

[0067] In one embodiment, the knowledge plane is equipped with a pre-trained classification model, and according to the network features input by the observation plane and the task requirements input by the task plane, the optimal MAC protocol reconstruction strategy for the next reconstruction cycle is output to the control plane, including:

[0068] First, based on the performance parameters of the MAX protocol in three dimensions: throughput, packet loss rate, and average network delay, the task requirements are modeled. This includes quantifying the priorities of task requirements using effectiveness requirements, reliability requirements, and timeliness requirements, and defining effectiveness requirement parameters , reliability requirement parameters , and timeliness requirement parameters as the weights of throughput , packet loss rate , and average network delay respectively. Among them, the performance parameters of the three dimensions are normalized so that , , , and it satisfies ;

[0069] Then, the normalized performance parameters and task requirement parameters are weighted and summed as the scoring evaluation index for different MAC protocols. The score value is expressed as:

[0070] ;

[0071] Among them, the subscript h represents the number of alternative MAC protocols, and the superscript T represents the transpose;

[0072] Finally, based on the pre-trained XGBoost classification model carried, with network features including the mean and variance of network node scale, global network activity level, and global network service characteristics as classification features, and task requirements including effectiveness requirement parameters , reliability requirement parameters , and timeliness requirement parameters as evaluation features, a data set feature is formed to classify alternative MAC protocols, and the alternative MAC protocol with the highest score value is selected as the best MAC protocol that best balances the network features and task requirements in the next reconstruction cycle, expressed as:

[0073] .

[0074] A task-driven MAC protocol reconstruction device, the device includes:

[0075] A load modeling module, which is used to consider the multi-domain distributed wireless ad hoc network scenario, and constructs a network load model through multi-dimensional feature division extraction of network load, Gaussian random process modeling, and feature fusion;

[0076] A protocol reconstruction module, which is used to construct a software-defined distributed subnet node system architecture based on a network load model. This architecture consists of a task plane, a data plane, and a joint control plane;

[0077] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames into the data plane to form a queue. The data plane queue contains the service data frames generated by node tasks, and also contains the control data frames generated by feature interaction during the MAC protocol reconstruction and channel reservation and preemption behaviors in the MAC protocol process;

[0078] The joint control plane includes an observation plane, a knowledge plane, and a control plane; among them, the observation plane is used to observe and statistically analyze the services of each node itself, obtain the service characteristics of each node in the current reconstruction period through maximum likelihood estimation in sequence, and predict the service characteristics of each node in the next reconstruction period through Kalman filtering. Then, the service characteristics of all single-hop reachable nodes obtained through broadcast interaction are fused to obtain the network characteristics in the next reconstruction period and transmitted to the knowledge plane;

[0079] The knowledge plane is used to carry a pre-trained classification model, and output the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane according to the network characteristics input by the observation plane and the task requirements input by the task plane;

[0080] The control plane is used to carry a MAC protocol component library, and judge whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If they are the same, the reconstruction process is not executed; if they are different, search for components and adjust the protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction period.

[0081] A computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0082] Considering the multi-domain distributed wireless ad hoc network scenario, a network load model is constructed through multi-dimensional feature division and extraction of network load, Gaussian random process modeling, and feature fusion;

[0083] Based on the network load model, a software-defined distributed subnet node system architecture is constructed. This architecture consists of a task plane, a data plane, and a joint control plane;

[0084] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames into the data plane to form a queue. The data plane queue contains the service data frames generated by node tasks, and also contains the control data frames generated by feature interaction during the MAC protocol reconstruction and channel reservation and preemption behaviors in the MAC protocol process;

[0085] The joint control plane includes an observation plane, a knowledge plane, and a control plane. Among them, the observation plane is used to observe and statistically analyze the services of each node itself. It sequentially obtains the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, and predicts the service characteristics of each node in the next reconstruction period through Kalman filtering. Then, it fuses the service characteristics of all single-hop reachable nodes obtained through broadcast interaction to obtain the network characteristics in the next reconstruction period and transmits them to the knowledge plane.

[0086] The knowledge plane is used to carry a pre-trained classification model, and according to the network characteristics input by the observation plane and the task requirements input by the task plane, it outputs the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane.

[0087] The control plane is used to carry a MAC protocol component library, and it judges whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If they are the same, the reconstruction process is not executed. If they are different, it searches for components and adjusts the protocol parameters to complete the MAC protocol component-level reconstruction and enters the loop of the next reconstruction period.

[0088] A computer-readable storage medium stores a computer program thereon. When the computer program is executed by a processor, the following steps are implemented:

[0089] Considering the scenario of a multi-domain distributed wireless ad hoc network, a network load model is constructed through multi-dimensional feature partitioning and extraction of network load, Gaussian random process modeling, and feature fusion.

[0090] Based on the network load model, a software-defined distributed subnet node system architecture is constructed. This architecture consists of a task plane, a data plane, and a joint control plane.

[0091] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames into the data plane to form a queue. The data plane queue contains the service data frames generated by node tasks, and also contains the control data frames generated by feature interaction during the MAC protocol reconstruction process and channel reservation and contention behaviors in the MAC protocol process.

[0092] The joint control plane includes an observation plane, a knowledge plane, and a control plane. Among them, the observation plane is used to observe and statistically analyze the services of each node itself. It sequentially obtains the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, and predicts the service characteristics of each node in the next reconstruction period through Kalman filtering. Then, it fuses the service characteristics of all single-hop reachable nodes obtained through broadcast interaction to obtain the network characteristics in the next reconstruction period and transmits them to the knowledge plane.

[0093] The knowledge plane is used to carry a pre-trained classification model, and according to the network characteristics input by the observation plane and the task requirements input by the task plane, it outputs the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane.

[0094] The control plane is used to carry the MAC protocol component library, and determine whether the reconstructed strategy output by the knowledge plane is the same as the MAC protocol of the current reconstruction cycle. If they are the same, the reconstruction process is not executed; if they are different, search for components and adjust protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction cycle.

[0095] The above MAC protocol reconstruction method, device, equipment and medium based on task driving have the following beneficial effects:

[0096] 1. Considering the differences in service characteristics brought by node task heterogeneity, the network load is refined and split into three-dimensional characteristics. Compared with traditional network load modeling methods, the network load model constructed in this application can fully reflect the performance of the MAC protocol in different network environments.

[0097] 2. Construct a closed-loop software-defined distributed subnet node system architecture. This architecture gives a complete system design and MAC protocol reconstruction algorithm process, provides a solution for distributed nodes to cooperate to complete the reconstruction process. The reconstruction process considers different task requirements to solve the problem of balancing protocol performance, constructs a data set and uses the XGBoost classification model to realize the reconstruction decision of different protocols. Finally, the best MAC protocol based on task requirements and taking into account different performance dimensions can be reconstructed, improving the adaptability and reliability of the MAC protocol to dynamic network environments and task requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 It is a schematic diagram of a multi-domain distributed wireless ad-hoc network scenario in an embodiment;

[0099] Figure 2 It is a schematic diagram of a software-defined distributed subnet node system architecture in an embodiment;

[0100] Figure 3 It is a schematic diagram of the process of each plane in the software-defined distributed subnet node system architecture in an embodiment;

[0101] Figure 4 It is a schematic diagram of the performance difference comparison of the 0.6-CSMA protocol under different network characteristic parameters in an embodiment; among them, Figure 4 (a) is the throughput performance comparison curve of the 0.6-CSMA protocol under different network characteristic parameters, Figure 4 (b) is the packet loss rate performance comparison curve of the 0.6-CSMA protocol under different network characteristic parameters, Figure 4 (c) is the network average delay performance comparison curve of the 0.6-CSMA protocol under different network characteristic parameters;

[0102] Figure 5 Schematic diagram for comparing the performance differences of the SOTDMA protocol under different network characteristic parameters in an embodiment; wherein, Figure 5 (a) is the throughput performance comparison curve of the SOTDMA protocol under different network characteristic parameters, Figure 5 (b) is the packet loss rate performance comparison curve of the SOTDMA protocol under different network characteristic parameters, Figure 5 (c) is the network average delay performance comparison curve of the SOTDMA protocol under different network characteristic parameters;

[0103] Figure 6 Schematic diagram of the throughput performance curve of the reconstruction protocol under the network scale variable in an embodiment;

[0104] Figure 7 Schematic diagram of the packet loss rate performance curve of the reconstruction protocol under the network scale variable in an embodiment;

[0105] Figure 8 Schematic diagram of the network average delay performance curve of the reconstruction protocol under the network scale variable in an embodiment;

[0106] Figure 9 Schematic diagram of the throughput performance curve of the reconstruction protocol under the activity level variable in an embodiment;

[0107] Figure 10 Schematic diagram of the packet loss rate performance curve of the reconstruction protocol under the activity level variable in an embodiment;

[0108] Figure 11 Schematic diagram of the delay performance curve of the reconstruction protocol under the activity level variable in an embodiment;

[0109] Figure 12 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0110] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0111] The multi-domain distributed wireless ad-hoc network scenario is as Figure 1As shown in the figure, the network adopts a distributed structure, which means that there is no central node for centralized control. The functions of the nodes in the network are diverse. The nodes are interconnected and cooperate to execute cluster tasks. The mapping of nodes with different functions and applications to communication services is also different. The upper-layer communication services cover heterogeneous modal data such as instructions, texts, audio, images, and video information, and the characteristics reflected in the MAC layer are different data lengths and priorities. Different services also have different requirements for throughput, effectiveness, and timeliness, which pose challenges to the existing single and fixed MAC protocol.

[0112] Specifically, in order to highlight the performance analysis of the MAC protocol and make the analysis general, the following settings are made for the characteristics of the network system:

[0113] (1) Nodes in the network are one-hop reachable and the wireless channel is ideal;

[0114] (2) Nodes in the network operate in the same frequency band, the transmission rate is fixed, and the time-slot channel is considered;

[0115] (3) Data frame collision is the only reason for receiving errors;

[0116] (4) If a data frame collides, all data frames are damaged.

[0117] According to the above settings, the transmission rate of the nodes in the time-slot channel is fixed, which means that the MAC protocol transmits data with fixed-length data frames. Then, the service data may be split into multiple data frames, which means occupying multiple channel time slots. Even if the length is less than one data frame, one time slot is still occupied. To improve the performance of the contention-based MAC protocol, the TXOP (Transmission Opportunity) transmission mechanism is introduced. After a node obtains the transmission opportunity, it is allowed to transmit multiple frames of data, that is, multiple frames of continuous transmission are realized through a single channel contention, thereby reducing the contention times and improving the channel utilization rate. In this scenario, it is considered that after a node obtains the transmission opportunity through contention, it sends at least one complete piece of data (multiple data frames) to ensure the integrity of the upper-layer service data.

[0118] The heterogeneity of network characteristics and node communication services is reflected in three aspects, namely the scale of network nodes, the activity level of nodes, and the length of service data. The number of nodes in the network reflects the size of the network node scale. The activity level of nodes is reflected in the generation probability of node communication services. The length of service data is reflected in the occupation duration of channel time slots.

[0119] Classical models in queuing theory, such as the M / M / 1 (single-server queuing) model, usually assume that the arrival process is a Poisson process, the service time is exponentially distributed, and the system is in a steady state. However, in an actual communication network, the arrival process may not follow a Poisson distribution. Especially when the load is high, the arrival of data often exhibits burstiness and correlation, resulting in the model underestimating queue buildup and delay, thus causing deviations in model prediction. In a single-channel scenario, the transmission of large files may require a longer service time, blocking short data packets of other users and leading to greater delay fluctuations. Simple models in queuing theory may not be able to handle this heterogeneous service time distribution, especially when the load intensity is high, this effect will be more significant.

[0120] Based on the above analysis, three-dimensional parameters of network heterogeneity are introduced, which can precisely describe the essence of network load changes and can also mathematically model the burstiness and correlation of data at different nodes. Let's qualitatively analyze these three dimensions. For example, considering a high-load network scenario, traditional queuing theory-based modeling analysis suggests that a reservation-based or polling-based MAC protocol should be used in high-load situations. In fact, any of these dimension parameters will increase the network load. When the network load increases only due to an increase in the number of network nodes and the node activity level and data length remain at a low level, using a reservation-based MAC protocol is not the best choice.

[0121] It should be noted that different MAC protocols have their own unique performance advantages, and there is no MAC protocol that has the best performance in multiple dimensions. Throughput (efficiency), packet loss rate (reliability), and delay (fairness, timeliness) are the basic indicators for measuring protocol performance. Different application scenarios and tasks have different performance requirements for the ad hoc network cluster. It is necessary to change the performance requirements according to different task needs and achieve the adaptation of the task and the balance of performance through protocol reconstruction.

[0122] In a distributed network scenario, each node independently makes a decision to obtain a reconstructed strategy. Therefore, it is necessary to design a node system architecture with a closed-loop of observation, decision-making, and control. For time-varying network characteristics and task requirements, global information is obtained through observation, distributed fusion, and data prediction, and the most suitable MAC protocol is adapted and the protocol parameters are optimized to overall optimize the three performance indicators of throughput, packet loss rate, and delay in the entire ad hoc network.

[0123] Based on the above analysis, in one embodiment of the present application, a task-driven MAC protocol reconstruction method is provided, including the following steps:

[0124] Step 1: Considering a multi-domain distributed wireless ad hoc network scenario, a network load model is constructed through multi-dimensional feature extraction, Gaussian random process modeling, and feature fusion of network load.

[0125] Among them, considering the complex multi-domain distributed wireless ad-hoc network scenario, the functions of different nodes in the network are diverse, and there are also significant differences in the corresponding communication service characteristics. If the service characteristics are simply described by equal-length data packets, it is only applicable to nodes of the same service type. For the scenario of different service characteristics caused by diverse node functions in heterogeneous networks, the load cannot fully represent the network strength characteristics. Therefore, this application designs a new method for constructing a network load model, including:

[0126] Step 1.1: Considering the multi-domain distributed wireless ad-hoc network scenario, extract multi-dimensional features of the network load, including the network node scale, node activity level, and node service characteristics.

[0127] Step 1.2: Use a multi-dimensional Gaussian random process with the same dimension to model the service characteristics of each node in the network, and use the weighted fusion algorithm of multi-dimensional Gaussian components to perform weighted fusion on the service characteristics of all nodes in the network to obtain the global network service characteristics.

[0128] Specifically, first consider that the tasks executed by each node in the network are different, and use a multi-dimensional Gaussian random process with the same dimension to model the service characteristics of each node in the network, expressed as , where is the set composed of all sample functions of the service characteristics of a single node at t time, , is the i th sample function, d is the number of sample functions, that is, the dimension number of the Gaussian random process is the same as the number of sample functions, is the period.

[0129] is used to describe the random process of the service length (number of data frames) generated by the node changing with time. The dimensional normal probability density function of this process is expressed as:

[0130] ;

[0131] Among them, is the expectation value of, represents the random variable of any sample function at time; is the expectation value of, represents the random variable of any sample function at another time; is the random variable Variance of; is the random variable Variance of; Denotes the expectation calculation; is the normalized covariance matrix Determinant of; Denotes the th service length generated by the node, Denotes the service feature random variable of the task mapped to d dimensions; Denotes any d moments; and both denote the dimension.

[0132] Further considering that the tasks of each node in the network are different and the service types are relatively fixed, it is assumed that the service features of each node are generally stationary within a finite time period. At the same time, the service features of the data transmitted in the communication process reflected by different tasks should follow dimensional Gaussian distribution. Therefore, the service features of each node in the network are modeled as dimensional stationary Gaussian process. The service features of the stationary Gaussian process satisfy the following constraints:

[0133] ;

[0134] ;

[0135] where, is Expected value of; Is the expected value constant, used to represent that the expected value of the stationary Gaussian process does not change with time; Denotes the value of the autocorrelation function at time and ; is and Expected value of the product of; Is the time difference, i.e., ; Is the autocorrelation function, which only depends on the time difference .

[0136] Without loss of generality, the service features of each node after modeling are dimensional Gaussian distribution with the service length as the random vector at any moment. The service lengths of each dimension of node are described as dimensional random vector , and the probability density function of its service features is expressed as:

[0137] ;

[0138] Among them, is t the probability density function of the service characteristics at the time node; represents the mean vector of the lengths of different types of services at node , represents the mean of the th i service length at node , and the superscript T represents the transpose; is the covariance matrix, which is used to represent the correlation between the mean vectors of the lengths of different types of services at node , represents the determinant of the covariance matrix .

[0139] After describing the service characteristics of different nodes through the -dimensional Gaussian distribution of each node, the set of probability density functions of the service characteristics of each node in the network is expressed as ; among them, is the scale of the network nodes, and .

[0140] To obtain the global network service characteristics at a certain moment, it is necessary to fuse the service characteristics of all nodes in the network. Specifically, using the -dimensional Gaussian component weighted fusion algorithm, considering that each node in the network satisfies the independent and identically distributed condition, the goal is to fuse the service characteristics of all nodes in the network to obtain a -dimensional Gaussian distribution of the global network service characteristics.

[0141] Assume that the Gaussian component of the service characteristics of node is , and the activity level of node is , and the activity level represents the probability that node generates services within a unit time; by normalizing , the weights of the service characteristics of each node are obtained, which is expressed as:

[0142] ;

[0143] Among them, satisfies and .

[0144] Based on the weights , the service characteristics of all nodes in the network are weighted and fused to obtain the global network service characteristics , and Conform to a Gaussian distribution, that is ; among them, the mean vector of the global network service characteristics and the covariance matrix are respectively expressed as:

[0145] ;

[0146] ;

[0147] Among them, , among which, represents the i th mean in , and T the superscript

[0148] represents the transpose.

[0149] Step 1.3: Model according to the network node scale, the global network activity degree obtained by averaging based on the activity degrees of all nodes, and the global network service characteristics, and construct a network load model. Specifically, according to the network node scale , the global network activity degree obtained by averaging based on the activity degrees of all nodes , and the mean vector t of the global network service characteristics, a network load model is constructed, where

[0150] ;

[0151] Among them, is the time-varying network load, is the expected value of; the global network activity degree , which represents the average probability of generating services by all nodes in the network per unit time; is the activity degree of node ; is the activity degree of node the i th mean of the service length.

[0152] Through the above modeling method, the network load is fused with the service characteristics of each node, making the description of the network load more refined and enabling more accurate network characteristics to be obtained. Compared with the traditional load-based MAC protocol handover algorithm, this task-characteristic-based network load model provides more favorable support for the task-driven MAC protocol reconstruction algorithm.

[0153] Step 2: Based on the network load model, construct the software-defined distributed subnet node system architecture. This architecture is as shown in Figure 2 Figure Figure 2 , and it consists of a task plane, a data plane, and a joint control plane. Each different functional plane is separated and equipped with different algorithms and protocols, and is implemented through a full-stack software-defined approach. All nodes in the network are developed based on this architecture, and each node exists independently and performs different tasks in the network to form a distributed heterogeneous network.

[0154] Moreover, since all nodes in the network are single-hop reachable, the characteristic results of the observation plane fusion are the same. The knowledge plane is equipped with the same set of pre-trained intelligent algorithm classification models, and the reconstruction strategies obtained by inputting into the knowledge plane are also the same, which can ensure the consistency of the reconstruction strategies within the subnet. Guided by the software-defined distributed subnet node system architecture, the closed-loop control problem of node observation, decision-making, and action in a pure distributed network can be solved.

[0155] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames into the data plane to form a queue. Cluster tasks are the upper-layer tasks jointly completed by all nodes in the network, such as tasks like silent approach, on-the-fly reorganization, and joint action. Node tasks are the specific actions of heterogeneous nodes, such as tasks like high-altitude reconnaissance, continuous surveillance, environmental perception, cargo transportation, and user access.

[0156] The data plane queue contains the service data frames generated by node tasks, and also contains control data frames generated by feature interaction during the MAC protocol reconstruction process and channel reservation and contention behaviors in the MAC protocol process.

[0157] The joint control plane includes an observation plane, a knowledge plane, and a control plane; among them, the observation plane is used to observe and statistically analyze the services of each node itself, sequentially obtain the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, and predict the service characteristics of each node in the next reconstruction period through Kalman filtering, and then perform feature fusion on the service characteristics of all single-hop reachable nodes obtained through broadcast interaction to obtain the network characteristics in the next reconstruction period and transmit them to the knowledge plane. The knowledge plane is used to carry the pre-trained classification model, and according to the network characteristics input by the observation plane and the task requirements input by the task plane, output the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane. The control plane is used to carry the MAC protocol component library, judge whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If it is the same, the reconstruction process is not executed; if it is different, search for components and adjust the protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction period.

[0158] Based on the above settings of the network system characteristics, considering the time-slot channel, assume a reconstruction period contains time slots, reconstruction period should be less than the task execution period All nodes in the network complete clock synchronization, network feature statistics, and service feature observation and estimation within a reconstruction period, and complete feature interaction and fusion. Combining the task requirements and network features, the best MAC protocol is decided, and the MAC protocol reconstruction process is completed within the first few time slots of the next reconstruction period. Suppose an observation period contains time slots and satisfies That is The processes of each plane in the software-defined distributed sub-network node system architecture are as Figure 3 shown

[0159] Through Figure 3 the process shown, it can be seen that the node clusters in the network jointly execute a certain type of task. Distributed nodes need to know the network features of the task cluster in the next reconstruction period through three steps: service statistics, observation prediction, and interaction and fusion. The obtained network features and the current task requirements are transmitted to the knowledge plane to decide the best MAC protocol for the next reconstruction period. During this process, the data plane continuously transmits data according to the MAC protocol of this reconstruction period. After obtaining the reconstruction strategy, if it is the same as the protocol of this reconstruction period, there is no need to execute the reconstruction process; if it is different, the nodes in the cluster synchronously execute the protocol reconstruction. After completion, the system clock synchronization is completed through GPS (Global Positioning System) timing, and the cycle of the next reconstruction period begins.

[0160] In Figure 2 the software-defined distributed sub-network node system architecture shown, due to the distributed characteristics of the wireless ad-hoc network, it is difficult to centrally obtain the global information of the network through a single node. Therefore, other network features in the above network load model except the network node scale cannot be directly obtained. To obtain accurate service features in the above model, each node needs to statistically analyze, fit, and predict the parameters of its own node through its own observation plane. The algorithm process executed by the observation plane includes the following steps:

[0161] Feature estimation part: The observation plane observes and statistically analyzes the number of data generations within the current observation period and the data length observation data set of each node in the network through the task and service mapping between the observation plane and the task plane; calculates the activity degree of node n as , and uses the data length observation data set as the observation sample for service feature estimation and prediction.

[0162] According to the above network load model construction process, using Model the service characteristics of nodes with a d-dimensional stationary Gaussian random process. To simplify the computational complexity of the observation surface, further reduce the d-dimensional Gaussian distribution of the service characteristics of nodes n to a one-dimensional Gaussian distribution for description. Then, according to the prior information that the data frame length variable at the sampling moment follows a Gaussian distribution, obtain the mean and standard deviation of the service characteristics of nodes n through maximum likelihood estimation; specifically, let , for the data set , its likelihood function is the product of the probability density functions of all observed values, expressed as:

[0163] ; n where represents the data length generated at the i-th time within the observation period , and . n of the d-dimensional Gaussian distribution of service characteristics is reduced to a one-dimensional Gaussian distribution for description, and then based on the prior information that the data frame length variable at the sampling moment follows a Gaussian distribution, the mean and standard deviation of the service characteristics of nodes n are obtained through maximum likelihood estimation; n mean and standard deviation ; Specifically, let , for the data set , its likelihood function is the product of the probability density functions of all observed values, expressed as: For the data set , its likelihood function is the product of the probability density functions of all observed values, expressed as:

[0163] ;

[0164] where, represents the data length generated at the i-th time within the observation period and i is the i-th time, and .

[0165] Take the logarithm of this likelihood function to obtain the log-likelihood function , expressed as:

[0166] .

[0167] To find the and that maximize the log-likelihood function, take the partial derivatives of with respect to and respectively and set them equal to zero, expressed as:

[0168] ;

[0169] ;

[0170] Solve to obtain and which are respectively expressed as:

[0171] ;

[0172] ;

[0173] Based on the above solutions for and , determine that the service characteristics of nodes n in the current reconstruction period follow n ; where, ; where, , It is the scale of network nodes. However, the resulting outcome is the service characteristics of this reconstruction cycle and cannot directly provide guidance for the control behavior of the next reconstruction cycle. Therefore, based on the previous reconstruction cycle and the estimation results of this reconstruction cycle, the Kalman filtering method is used to predict the service characteristics of the next reconstruction cycle.

[0174] Feature prediction part: After estimating the service characteristics of each node in this reconstruction cycle, to ensure the rigor of the reconstruction logic, the service characteristics of the next reconstruction cycle are predicted to guide the protocol reconstruction strategy. Considering the autocorrelation of heterogeneous node communication services, each node first uses the Kalman filtering algorithm for prediction, and then based on the prediction results and the interaction and fusion of the characteristics of other nodes in the network, the global network characteristics are obtained.

[0175] Specifically, it is set that the true service characteristics of node in the th reconstruction cycle follow , and through observation and maximum likelihood estimation, the service characteristics are obtained to follow ; where is the true mean of the service characteristics of node in the th reconstruction cycle; is the observed value of the service characteristics of node in the th reconstruction cycle, representing the average data length; is the variance of the service characteristics of node in the th reconstruction cycle, reflecting the degree of data frame dispersion.

[0176] Among them, it is set that the change of follows a random walk process, and the state equation is expressed as:

[0177] ;

[0178] where is the process noise variance, estimated by the variance of the change of the historical data mean, expressed as , is the number of reconstructions.

[0179] And, it is set that is the noisy measurement of the true mean , and the observation equation is:

[0180] ;

[0181] where the process noise and the measurement noise are independent of each other and are both zero-mean Gaussian white noise.

[0182] The goal of the Kalman filter is based on historical observations and their corresponding variances and , predict the service characteristics of the node at the th reconstruction period, and the service characteristics follow . Then, the prediction results of the th reconstruction period are weighted and fused with the service characteristics of all single-hop reachable nodes obtained through broadcast interaction to obtain the global network service characteristics of the th reconstruction period. Finally, the network node scale, the global network activity level obtained by averaging the activity levels of all nodes in the network, and the global network service characteristics are combined into the network characteristics of the th reconstruction period, and the network characteristics are passed to the knowledge surface.

[0183] Specifically, the Kalman filter estimates the state through two-step iteration of prediction and update. First, the prior estimates of the mean and variance of the service characteristics of the th reconstruction period are obtained through the prediction step. The predictions of the mean and variance are respectively expressed as:

[0184] ;

[0185] ;

[0186] where is the prior estimate of the mean of the service characteristics of the th reconstruction period, is the posterior estimate of the mean of the service characteristics of the th reconstruction period; is the prior estimate of the variance of the service characteristics of the th reconstruction period, is the posterior estimate of the variance of the service characteristics of the th reconstruction period; is the process noise variance.

[0187] Then, the updated Kalman gain is obtained based on the posterior estimate in the update step, and the expression is:

[0188] ;

[0189] According to the Kalman gain , the posterior estimate of the mean of the service characteristics of the th reconstruction period and the posterior estimate of the variance are updated respectively, and the expression is:

[0190] ;

[0191] ;

[0192] so as to obtain the node at the next iteration At the th reconstruction period, the predicted distribution of the service characteristics is , and respectively represent the predicted values of the mean and variance of the service characteristics of the node at the th reconstruction period; among them, the prior estimate of the mean of the service characteristics of the node at the th reconstruction period is predicted to be , predicted by the exponential smoothing method and expressed as ; among them, is a preset smoothing exponent, represents the predicted value of the variance of the service characteristics at the th reconstruction period.

[0193] It should be noted that within the first two reconstruction periods when the network is initially established, the Kalman filter prediction process is not performed. The new nodes added at the end of the reconstruction period do not participate in the algorithm process of the observation surface, and the observation surface only performs the processes of observation, maximum likelihood estimation, and feature interaction fusion within the first two reconstruction periods. By introducing the Kalman filter algorithm, the present application can not only predict the service characteristics of each node in the next reconstruction period, but also eliminate the observation and estimation errors caused by the randomness of the data generated by the nodes to a certain extent, making the mapping of the node tasks to the characteristics of the communication service more accurate.

[0194] Figure 2 The algorithm processes executed in the knowledge surface in include: Different self-organizing network MAC protocols have different performance in different network characteristics. The specific performance evaluation indicators can be divided into throughput, packet loss rate, and network average delay, reflecting the advantages and disadvantages of a certain MAC protocol in three different dimensions. The throughput

[0195] is defined as the number of successfully transmitted data frames per unit time and can be expressed as:

[0196] Among them, represents the number of successfully received data frames transmitted within the observation period , represents the total number of time slots in the observation period .

[0197] Packet loss rate It is defined as the ratio of the sum of data frames damaged during transmission and discarded due to failure to transmit in time within a unit time to the total number of data frames generated, and can be expressed as:

[0198] ;

[0199] Among them, represents the observation period The number of data frames damaged during transmission by the node within represents the observation period The number of data frames discarded by the node due to failure to transmit in time or backlogged in the queue at the end of the period within represents the observation period The total number of data frames generated by all nodes in the network within

[0200] Network average delay It is defined as the average waiting time of all data frames generated in the network in the queue. Due to the TXOP mechanism, the delay statistical method can be simplified. It only needs to count the waiting time of each node in the send-pending state and the number of times the send process is activated, and average the average delay of all nodes in the network to obtain the network average delay, which can be expressed as:

[0201] ;

[0202] Among them, is the observation period The waiting time of node in the send-pending state within is the number of times the send process is activated. Suppose there are h alternative MAC protocols, and the performance of different protocols under different network characteristics forms a vector, which can be expressed as , , .

[0203] To achieve the unbiasedness of the evaluation of the three performance indicators, normalization is performed on them. The definitions of throughput and packet loss rate have achieved normalization, so only the network average delay needs to be normalized. Using the local information normalization method, the network average delay performance of the alternative MAC protocols obtained under different network characteristics conditions is normalized to obtain:

[0204] ;

[0205] Among them, From this, the normalized network average delay can be obtained.

[0206] First, based on the performance parameters of the MAX protocol in three dimensions: throughput, packet loss rate, and average network delay, the task requirements are modeled, including quantifying the priorities of task requirements using effectiveness requirements, reliability requirements, and timeliness requirements, and defining effectiveness requirement parameters , reliability requirement parameters and timeliness requirement parameters as the weights of throughput , packet loss rate and average network delay respectively.

[0207] Among them, the performance parameters in the three dimensions are normalized so that , , , and it satisfies .

[0208] Then, the normalized performance parameters and task requirement parameters are weighted and summed to be used as the scoring evaluation index for different MAC protocols. The score value is expressed as:

[0209] ;

[0210] where the subscript h represents the number of alternative MAC protocols, and the superscript T represents the transpose.

[0211] Finally, based on the pre-trained XGBoost classification model carried, using network features including the mean and variance of network node scale, global network activity level, and global network service characteristics as classification features, and using task requirements including effectiveness requirement parameters , reliability requirement parameters and timeliness requirement parameters as evaluation features, a dataset feature is formed to classify alternative MAC protocols, and the alternative MAC protocol with the highest score value is selected as the best MAC protocol that best balances the network features and task requirements in the next reconstruction cycle, which is expressed as:

[0212] .

[0213] Among them, the dataset feature settings of the XGBoost classification model are shown in Table 1.

[0214] Table 1 Dataset feature settings:

[0215]

[0216] The scale of network nodes and task requirements are randomly generated within the value range and constraints. At the same time, in order to fully eliminate the probability error of data generated within the cycle and the randomness of node transmission conflicts, each different network feature is reconstructed with a step length of 200 times, and the average of throughput, packet loss rate and average network delay is obtained as a complete data.

[0217] XGBoost (eXtreme Gradient Boosting, XGB for short) is an efficient implementation of the GBDT (gradient boosting decision tree) algorithm, which optimizes algorithm efficiency, model performance and scalability. XGBoost introduces regularization terms to prevent overfitting, and uses efficient tree splitting algorithms and column sampling techniques to perform well on large-scale data sets. The idea of ​​the algorithm is to continuously add trees and continuously perform feature splitting to grow a tree. Each time a tree is added, a new function is learned to fit the residual of the last prediction.

[0218] The objective function of XGBoost consists of two parts: training loss and regularization term. The objective function is defined as follows:

[0219] ;

[0220] in, is the training loss term, represents the loss function; is the regularization term, representing the complexity of the tree.

[0221] Since the application scenario is to deal with multi-classification problems and hope to obtain the probability estimate of each category, it is more appropriate to use the logistic regression loss function as the loss function. If there are more noise or outliers in the data set, the logistic regression loss function can also provide better robustness. The expression of the logistic loss function is:

[0222] ;

[0223] in It is i Samples Since XGBoost is an additive model, the prediction score is the cumulative sum of the scores of each tree, expressed as:

[0224] ;

[0225] in represents the sum of the trees, For the The prediction function of the tree.

[0226] Assume The tree model trained after the -th iteration is . The prediction result of sample

[0227] can be expressed as:

[0228] where is the prediction result of the first trees. Substituting it into the objective function, we get:

[0229] .

[0230] Using the XGBoost algorithm, according to the mapping relationship between task requirements and protocol performance, an XGB classification model is trained by constructing a dataset through simulation. As an algorithm model of knowledge, it is carried on each distributed node to provide policy support for the intelligent reconstruction of distributed nodes in scenarios with different network characteristics and task requirements.

[0231] Furthermore, the method proposed in this paper optimizes the performance in terms of three dimensions: throughput, packet loss rate, and network average delay. It only focuses on the channel access mechanisms of different MAC protocols and ignores the considerations of existing MAC protocols for performance dimensions such as priority, security, and low energy consumption. The functions for optimizing other dimensions of the protocol can be added on this basis.

[0232] Figure 2 The control plane shown in

[0233] summarizes different access and transmission mechanisms based on the existing MAC protocol for the distributed wireless ad hoc network application scenario, obtaining two types of transmission mechanisms, namely contention-based and scheduling-based MAC protocols, as well as a MAC protocol component library composed of six different types of MAC protocols, as shown in Table 2.

[0234]

[0235] Based on the above setting of the time slot channel and the equal length of the data frame, the protocol will only compete for and occupy the channel at the beginning of the time slot, and will also occupy a complete time slot when the data volume is less than the frame length. At the same time, following the TXOP transmission mechanism, when the contention-based protocol and the polling-based protocol obtain the transmission opportunity, they will continuously occupy the channel until the entire data is sent, ensuring the integrity of the data.

[0236] As the simplest MAC protocol, the S-ALOHA protocol directly sends data via broadcast as long as communication traffic is generated, and it is applicable under the conditions of a small number of nodes, low activity level, and short service mean. However, due to the lack of listening to the channel occupancy status in the protocol mechanism, any time slot may be occupied by the data frames of other nodes, leading to conflicts and the destruction of data frames. This situation is particularly obvious in the transmission of long data, and its reliability and throughput are difficult to guarantee.

[0237] The CSMA (Carrier Sense Multiple Access) protocol adds the function of carrier sensing on the basis of the S-ALOHA protocol. After generating traffic, it first joins the queue and only sends the data of this node after detecting that the channel is idle. The CSMA protocol guarantees the reliability of data transmission, but there may still be conflicts when multiple nodes occupy the same time slot simultaneously when the channel is idle. Therefore, this transmission mechanism only guarantees certain performance under the condition of low network activity.

[0238] P The p-persistent CSMA protocol, as a variant, sends data with a probability after detecting that the channel is idle and backs off with a probability of . This mechanism essentially weakens the network activity by reducing the competition probability, and its performance has a limited improvement compared with the 1-persistent CSMA protocol.

[0239] The 802.11 DCF protocol and the fixed backoff window CSMA / CA (Carrier Sense Multiple Access / Collision Avoidance) protocol add the function of channel reservation on the basis of CSMA and can solve the problem of hidden / exposed terminals, but this consideration is not required in the scenario where the network is reachable in one hop. The essential difference in their mechanisms lies in the non-persistent type of backoff algorithm. After each conflict occurs, the node will adjust the contention window and reselect the backoff duration. In the 802.11 standard, the DCF (Distributed Coordination Function) protocol considers dynamic window backoff and adopts the binary backoff algorithm. The node will adjust the contention window size according to the number of retransmissions as follows:

[0240] ;

[0241] where is the contention window size at the th retransmission, is the minimum window value, is the maximum window value; , is the maximum number of retransmissions. After determining the contention window size, the node randomly selects an integer value within this range as the backoff duration , and after waiting for After a certain number of time slots, try to send the data frame again. Similarly, a fixed backoff window means that the range of the backoff duration is fixed, and the contention window size does not change after a collision. Instead, a random integer value is selected again as the backoff duration. 。

[0242] The polling protocol realizes the access scheduling of nodes by sequentially querying each node or token passing, and is often used in wired networks and wireless networks with a central node. In the distributed node scenario, the design of the polling protocol is somewhat different, but the transmission mechanism of polling is completely retained. When reconstructing the protocol, count the number of network nodes ,from node 1 to node Pass the token in order and occupy the channel for transmission in turn. The TXOP mechanism is also followed during transmission. After a node completes the transmission of the complete data, it passes the token to the next node.

[0243] SOTDMA is a MAC protocol designed specifically for dynamic and distributed wireless communication environments. Its transmission mechanism is similar to that of the traditional TDMA (Time Division Multiple Access) protocol, but it can update the time slot allocation scheme according to the dynamic changes of the topology to achieve efficient utilization of the channel. In this scenario, the update period of the time slot allocation of this protocol is set to the reconstruction period to reflect the performance characteristics of this transmission mechanism. The difference between the two scheduling protocols of SOTDMA and polling is that the nodes of TDMA take turns to send only one data frame of data, and the complete data is transmitted through multiple rounds of cycling.

[0244] In specific implementation, to verify the guiding significance of the network load model constructed in this application for MAC protocol reconstruction, further simulation experiments were carried out for verification. Comparing with the network load modeling method based on Poisson distribution of the queuing theory model, the performance of the contention-based protocol and the scheduling-based protocol was simulated respectively, and the influence of the two models on the MAC protocol performance was obtained.

[0245] The p-CSMA protocol was selected for simulation of the contention-based protocol, and the probability . The simulation time was stepped at 5000 time slots per cycle, and the throughput, packet loss rate, and network average delay performance of 10 cycles were counted. Three networks with different characteristics were set, and the network characteristic parameters are shown in Table 3.

[0246] Table 3 Setting of network characteristic parameters for simulation of contention-based protocol:

[0247]

[0248] Among them, the characteristic variance is randomly generated during the simulation process to reflect the error of model parameter estimation and simulate the heterogeneity of service data generated by network nodes. The variance value range of the global network service characteristics is set to . The network load is calculated through the above parameters as , Then the expectation of the Poisson distribution is 0.8, and the probability density function is:

[0249] ;

[0250] The above formula represents that 0.8 data frames arrive at the channel per unit time (time slot). Although the above three network characteristic parameters are different, they have the same load.

[0251] Figure 4 The throughput, packet loss rate, and network average delay performance curves of the contention-based 0.6-CSMA protocol are given for three different network characteristics respectively. Although the three networks have the same network load, it can be seen from the simulation results that there are significant differences in the protocol performance. Under the same activity level, Network 1 has a large number of nodes and short service data lengths, with lower latency, but poor throughput and packet loss rate performance; Network 3 has fewer nodes and long service data lengths, with better throughput and packet loss rate performance, but longer latency; the performance of Network 2 is between the two. The reason is that the increase in the number of nodes leads to an increase in the probability of collision when accessing the channel, and the competition for time slots results in the collision and loss of data frames. The transmission of short frames also increases the number of times network nodes compete for the channel, exacerbating the situation of collisions. However, this also brings an increase in the opportunity for nodes to access the channel, reducing the average waiting latency. Vice versa.

[0252] The scheduling protocol selects the SOTDMA protocol for simulation. Similarly, three networks with different characteristics are set, and the fused characteristic parameters are shown in Table 4.

[0253] Table 4 Settings of the simulation network characteristic parameters for the scheduling protocol:

[0254]

[0255] Similarly, the variance value range of the global network service characteristics is set to . Through the above parameter calculations, the network load is approximately obtained as , Then the expectation of the Poisson distribution is 1, and the probability density function , and the three networks with different characteristics still have the same load.

[0256] Figure 5The throughput, packet loss rate, and network average delay performance curves of the scheduling-based SOTDMA protocol in three different characteristic parameter networks are given respectively. The three networks also have the same network load characteristics, and the protocol performance differences in the simulation results are still obvious. Network 1 has many nodes, but the nodes are inactive and the probability of generating traffic is extremely low. Once traffic is generated, it will accumulate a long amount of data. Network 3 has fewer nodes and shorter service frame lengths, but the network nodes are active and the probability of generating traffic per unit time is relatively large. The characteristics of Network 2 are between the two. Based on the analysis of the access mechanism of the scheduling class, the performance of the SOTDMA protocol in Network 3 should be better than that in Network 1, which is also verified in the simulation results. The throughput and packet loss rate performance of Network 2 and Network 3 gradually improve with the increase of the simulation time, showing a trend of converging to the performance of Network 3. This is because at the beginning of the simulation, the activity level of Network 3 is extremely low, resulting in a large difference in the service queues of each node. As time goes by, the randomness of traffic generation is gradually eliminated, and the difference in the service queues of each node tends to be flat. Due to the inherent characteristics of SOTDMA, the delay performance is highly positively correlated with the number of network nodes.

[0257] Through the simulation of contention-based protocols and scheduling-based protocols in different characteristic networks, the performance of different protocols under the same load conditions is obtained.

[0258] From Figure 4 and Figure 5 it can be seen that the same network load of the three networks means the same load parameters of the Poisson distribution. This application splits the network load into three-dimensional characteristics. Although the characteristics are different but the network loads are the same, there are significant differences in performance, which means that the traditional load modeling method based on the Poisson distribution has certain limitations in analyzing the performance of the MAC protocol, reflected in the lack of fine description of network characteristics and the inability to fully describe the burstiness and self-correlation of traffic. In some studies, using the parameter of network load as the standard for protocol reconstruction is actually not perfect. Different protocols have essential differences in the access mechanism and are highly sensitive to different network characteristics. Therefore, it is necessary to refine the modeling of the characteristic parameters that affect the protocol performance for different network environments.

[0259] The simulation proves that the network load model based on task characteristics constructed in this application is effective and sufficient in analyzing the performance of the MAC protocol.

[0260] For the construction of the simulation verification dataset and the training results of the XGBoost algorithm, the dataset parameters and examples are set as shown in Table 5. Three types of indicators, namely network features, task requirements, and labels in the complete dataset, are retained as the training set to train the XGBoost classification model. Among the generated 30,000 pieces of data, 70% of the dataset is divided as the training set and the other 30% as the test set, ensuring that the test set and the training set have no overlap and are consistently distributed. After training and parameter adjustment using the grid search method, the best model parameters for classification performance are obtained as shown in Table 6. The parameters not specified in the table are default values.

[0261] Table 5 Dataset Parameters and Examples:

[0262]

[0263] Table 6 Parameter Settings of the XGBoost Classification Model:

[0264]

[0265] After setting the above parameters and completing the training, the accuracy of the classification model is evaluated. 5-fold cross-validation is used to reduce the randomness of the evaluation results, and the classification accuracy results of the training set and the test set are obtained, as shown in Table 7.

[0266] Table 7 Classification Accuracy Results of the XGBoost Classification Model:

[0267]

[0268] Based on the trained XGBoost classification model, the performance of the reconstructed protocol under different network features is verified. Taking the network scale, activity level, and mean business feature as variables respectively, the effect of protocol reconstruction is simulated. The simulation parameter settings for protocol reconstruction verification are shown in Table 8, where is the reconstruction step size.

[0269] Table 8 Reconstruction Simulation Parameter Settings:

[0270]

[0271] Figures 6 to 8 Schematic diagrams of the throughput, packet loss rate, and network average delay performance curves of the reconstructed protocol are given respectively under the variable of network scale. The initial value of the network node scale is set to , and the range is . The performance curves of the reconstructed protocol are represented by dashed lines, and the solid lines represent the theoretical performance of different protocols. It can be seen from Figures 6 to 8 that as the number of network nodes increases, a total of four reconstruction processes are experienced, respectively at , , and Reconfiguration of the MAC protocol occurred. The decision-making protocol types include p-persistent CSMA, fixed-window backoff CSMA / CA protocol, and SOTDMA protocol. Each protocol is the best MAC protocol under the set network characteristics and demand parameters. According to the setting of the task demand parameters, it can be known that the network has a high demand for throughput. Figure 6 From the throughput performance curve shown in

[0272] Figures 9 to 11 it can be observed that the throughput of the reconfiguration protocol has always remained near the curve of the best MAC protocol. The packet loss rate and delay performance of the reconfiguration protocol also remain in a good range. respectively give the schematic diagrams of the throughput, packet loss rate, and network average delay performance curves of the reconfiguration protocol under the activity variable. The initial value of the activity is set to , and the range is Figures 9 to 11 . It is not difficult to see from and that as the number of network nodes increases, a total of two reconfiguration processes have occurred. Reconfiguration of the MAC protocol occurred at

[0273] respectively. The decision-making protocol types include p-persistent CSMA and SOTDMA protocol. Since the task has high requirements for timeliness, p-persistent CSMA can provide nodes with sufficient opportunities to compete for the channel, and the corresponding waiting delay is the smallest, but the throughput and packet loss rate performance are not the best among all reconfigurable protocols.

[0274] From the above simulation experiment results, it can be seen that the task-driven MAC protocol reconfiguration method proposed in this application can take into account the performance of different dimensions of the protocol and the requirements of different tasks, and select the best MAC protocol under the conditional constraints.

[0274] In one embodiment, a task-driven MAC protocol reconfiguration device is provided, including:

[0275] A load modeling module, which is used to consider the multi-domain distributed wireless ad hoc network scenario, and construct a network load model through multi-dimensional feature division and extraction of network load, Gaussian random process modeling, and feature fusion;

[0276] A protocol reconfiguration module, which is used to construct a software-defined distributed subnet node system architecture on the basis of the network load model. This architecture consists of a task plane, a data plane, and a joint control plane;

[0277] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks into the data plane in the form of data frames to form a queue. The data plane queue contains the service data frames generated by node tasks, and also contains the control data frames generated by feature interaction during the MAC protocol reconfiguration process and channel reservation and contention behaviors in the MAC protocol process;

[0278] The joint control plane includes an observation plane, a knowledge plane, and a control plane. Among them, the observation plane is used to observe and statistically analyze the services of each node itself, obtain the service characteristics of each node in the current reconstruction period through maximum likelihood estimation in sequence, and predict the service characteristics of each node in the next reconstruction period through Kalman filtering. Then, the service characteristics of all single-hop reachable nodes obtained through broadcast interaction are subjected to feature fusion to obtain the network characteristics in the next reconstruction period and transmit them to the knowledge plane;

[0279] The knowledge plane is used to carry a pre-trained classification model, and according to the network characteristics input by the observation plane and the task requirements input by the task plane, output the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane;

[0280] The control plane is used to carry a MAC protocol component library, and determine whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If they are the same, the reconstruction process is not executed; if they are different, search for components and adjust protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction period.

[0281] For the specific limitations of the task-driven MAC protocol reconstruction device, reference can be made to the limitations of the task-driven MAC protocol reconstruction method in the above text, which will not be elaborated here. Each module in the above task-driven MAC protocol reconstruction device can be implemented in whole or in part through software, hardware, and their combinations. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0282] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 12 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a task-driven MAC protocol reconstruction method.

[0283] Those skilled in the art can understand, Figure 12The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0284] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0285] Considering the multi-domain distributed wireless ad hoc network scenario, through multi-dimensional feature partitioning and extraction of network load, Gaussian random process modeling, and feature fusion, a network load model is constructed;

[0286] Based on the network load model, a software-defined distributed subnet node system architecture is constructed, which consists of a task plane, a data plane, and a joint control plane;

[0287] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks into the data plane in the form of data frames to form a queue. The data plane queue contains the service data frames generated by node tasks, and also contains the control data frames generated by feature interaction during the MAC protocol reconstruction process and channel reservation and preemption behaviors in the MAC protocol process;

[0288] The joint control plane includes an observation plane, a knowledge plane, and a control plane; among them, the observation plane is used to observe and statistically analyze the services of each node itself, sequentially obtain the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, and predict the service characteristics of each node in the next reconstruction period through Kalman filtering, and then perform feature fusion on the service characteristics of all single-hop reachable nodes obtained through broadcast interaction to obtain the network characteristics in the next reconstruction period and transmit them to the knowledge plane;

[0289] The knowledge plane is used to carry a pre-trained classification model, and according to the network characteristics input by the observation plane and the task requirements input by the task plane, output the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane;

[0290] The control plane is used to carry a MAC protocol component library, judge whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If it is the same, the reconstruction process is not executed; if it is different, search for components and adjust the protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction period.

[0291] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0292] Considering the scenario of a multi-domain distributed wireless ad-hoc network, a network load model is constructed through the extraction of multi-dimensional features of network load, Gaussian random process modeling, and feature fusion.

[0293] Based on the network load model, a software-defined distributed subnet node system architecture is constructed, which consists of a task plane, a data plane, and a joint control plane.

[0294] Among them, the task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks into the data plane in the form of data frames to form a queue. The data plane queue contains the service data frames generated by node tasks, and also contains the control data frames generated by feature interaction during the MAC protocol reconstruction process and channel reservation and contention behaviors in the MAC protocol process.

[0295] The joint control plane includes an observation plane, a knowledge plane, and a control plane. Among them, the observation plane is used to observe and statistically analyze the services of each node itself, and sequentially obtain the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, and predict the service characteristics of each node in the next reconstruction period through Kalman filtering. Then, the service characteristics of all single-hop reachable nodes obtained through broadcast interaction are fused to obtain the network characteristics in the next reconstruction period and transmitted to the knowledge plane.

[0296] The knowledge plane is used to carry a pre-trained classification model, and output the best MAC protocol reconstruction strategy in the next reconstruction period to the control plane according to the network characteristics input by the observation plane and the task requirements input by the task plane.

[0297] The control plane is used to carry a MAC protocol component library, and judge whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol in the current reconstruction period. If it is the same, the reconstruction process is not executed. If it is different, search for components and adjust the protocol parameters to complete the MAC protocol component-level reconstruction, and enter the loop of the next reconstruction period.

[0298] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0299] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0300] The above-described embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A task-driven MAC protocol reconstruction method, characterized in that: The method comprises: Considering the multi-domain distributed wireless self-organizing network scenario, a network load model is constructed through multi-dimensional feature extraction, Gaussian random process modeling and feature fusion of network load. Based on the network load model, a software-defined distributed subnet node system architecture is constructed, which consists of a task plane, a data plane, and a joint control plane; The task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames to form a queue on the data plane. The data plane queue includes the service data frames generated by the node tasks, and also includes the control data frames generated by the feature interaction during the MAC protocol reconstruction process and the channel reservation and preemption behavior in the MAC protocol process. The joint control plane includes an observation plane, a knowledge plane and a control plane; wherein the observation plane is used to observe and count the services of each node, obtain the service characteristics of each node in the current reconstruction period by maximum likelihood estimation, and obtain the service characteristics of each node in the next reconstruction period by Kalman filtering prediction, and then perform feature fusion on the service characteristics of all single-hop reachable nodes obtained by broadcast interaction to obtain the network characteristics of the next reconstruction period and pass them to the knowledge plane; The knowledge plane is used to carry the pre-trained classification model, and output the best MAC protocol reconstruction strategy for the next reconstruction cycle to the control plane according to the network features input by the observation plane and the task requirements input by the task plane; specifically, the knowledge plane first models the task requirements based on the performance parameters of the MAC protocol in three dimensions: throughput, packet loss rate, and average network delay; and then, based on the pre-trained XGBoost classification model carried, the task requirements are used as evaluation features; The control plane is used to carry the MAC protocol component library to determine whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol of the current reconstruction cycle. If they are the same, the reconstruction process is not performed; if they are not the same, the components are searched and the protocol parameters are adjusted to complete the MAC protocol component-level reconstruction, and enter the cycle of the next reconstruction cycle.

2. The method according to claim 1, characterized in that Considering the multi-domain distributed wireless self-organizing network scenario, the network load model is constructed through multi-dimensional feature partitioning and extraction of network load, Gaussian random process modeling and feature fusion, including: Considering the multi-domain distributed wireless self-organizing network scenario, the multi-dimensional characteristics of the network load are divided and extracted, including the network node scale, node activity level and node business characteristics; The service characteristics of each node in the network are modeled using a multi-dimensional Gaussian random process with the same dimension, and the service characteristics of all nodes in the network are weighted and fused using a weighted fusion algorithm of multi-dimensional Gaussian components to obtain the global network service characteristics. The network load model is constructed based on the scale of network nodes, the global network activity obtained by averaging the activity levels of all nodes, and the global network service characteristics.

3. The method according to claim 2, characterized in that A multi-dimensional Gaussian random process with the same dimension is used to model the service characteristics of each node in the network, including: Considering that each node in the network performs different tasks, we use The Gaussian random process models the business characteristics of each node in the network, which can be expressed as ,in, Reason t The set of all sample functions of the business characteristics of a single node at a certain moment, , For the i A sample function, d is the number of sample functions, that is, the number of dimensions of the Gaussian random process is consistent with the number of sample functions, For the cycle; It is used to describe the random process in which the length of the business generated by the node changes over time. The normal probability density function is expressed as: ； in, for The expected value of Represents any sample function exist Random variables at time; for The expected value of Represents any sample function In another Random variables at time; is a random variable The variance of is a random variable The variance of represents the expected calculation; is the normalized covariance matrix The determinant of ; Indicates the node generated Business length, Indicates that the task is mapped to d Business characteristic random variables of dimensions; Indicates any d a moment; and Both represent dimension; Further considering that the tasks of each node in the network are different and the business types are relatively fixed, the business characteristics of each node are set to be wide-sense stationary within a limited period of time, and the business characteristics of each node in the network are modeled as The business characteristics of a stationary Gaussian process satisfy the following constraints: ； ； in, for Expected value; is the expected value constant, which is used to indicate that the expected value of a stationary Gaussian process does not change with time; Represents the autocorrelation function at time and The value at for and The expected value of the product of is the time difference, that is ; is the autocorrelation function, which depends only on the time difference .

4. The method according to claim 3, characterized in that The business characteristics of each node after modeling are random vectors with business length at any time. Dimensional Gaussian distribution, nodes The business length of each dimension is described as dimensional random vector , the probability density function of its business characteristics is expressed as: ； in, for t Time Node The probability density function of the business characteristics; Representation Node The mean vector of different types of business lengths, Representation Node No. i The mean of the business length, , superscript T represents transpose; is the covariance matrix, which is used to represent the node The correlation between the mean vectors of different types of business lengths, Represents the covariance matrix The determinant of ; Through each node After using the Gaussian distribution to describe the service characteristics of different nodes, the probability density function set of the service characteristics of each node in the network is expressed as ;in, is the network node size, and .

5. The method according to claim 4, characterized in that Using the weighted fusion algorithm of multi-dimensional Gaussian components, the service characteristics of all nodes in the network are weighted and fused to obtain the global network service characteristics; use The weighted fusion algorithm of Gaussian components considers that the nodes in the network meet the independent and identically distributed conditions. The goal is to fuse the business characteristics of all nodes in the network to obtain a Global network service characteristics of Gaussian distribution; Assume Node The Gaussian component of the business characteristics is ,node The activity level is , the activity level indicates the node The probability of generating business in a unit time; Perform normalization to obtain the weight of each node’s business characteristics , expressed as: ； in, satisfy and ; Based on weight Perform weighted fusion on the service features of all nodes in the network to obtain the global network service features ,and conform to Gaussian distribution, that is ; Among them, the mean vector of global network service characteristics With the covariance matrix Respectively expressed as: ； ； in, ,in, express The i means, and ; Superscript T Indicates transpose.

6. The method according to claim 5, characterized in that The network load model is constructed based on the scale of network nodes, the global network activity obtained by taking the average of the activity levels of all nodes, and the global network service characteristics, and includes: According to the scale of network nodes , the global network activity level is obtained by taking the average activity level of all nodes And the mean vector of global network service characteristics Modeling is performed to construct a network load model, where: t The expected network load at a given moment is expressed as: ； in, is the time-varying network load, for Expected value of global network activity , represents the average probability of all nodes in the network generating business per unit time; For Node activity level; For Node No. i The mean length of the business.

7. The method according to claim 1, characterized in that The observation plane observes and counts the services of each node, and obtains the service characteristics of each node in the current reconstruction period through maximum likelihood estimation, including: The observation plane maps tasks and services with the task plane to observe and count the nodes in the network in the current observation period. Number of data generated within Observation dataset with data length ; Among them, the reconstruction cycle Include time slots, observation period Include time slots, and satisfy Right now ; Calculate the nodes based on the statistical results of the number of times data is generated n The activity level is , and the data length observation data set Use it as an observation sample to estimate and predict business characteristics; use dimensional stationary Gaussian random process for nodes n The business characteristics of the node are modeled and the n Business characteristics The dimensional Gaussian distribution is reduced to a one-dimensional Gaussian distribution description, and then the node is obtained by maximum likelihood estimation based on the prior information that the data frame length variable at the sampling time obeys the Gaussian distribution. n Mean of business characteristics and standard deviation Specifically, let , for the dataset , its likelihood function is the product of the probability density functions of all observations, expressed as: ； in, Represents the observation period Neidi i The length of the data generated, and ; Taking the logarithm of the likelihood function, we get the log-likelihood function , and About and Find the partial derivative and set it equal to zero, and the solution is and Respectively expressed as: ； ； Based on the above solution, we get and , determine the node n The business characteristics in the current reconstruction cycle are subject to ;in, , is the network node size.

8. The method according to claim 7, characterized in that The observation plane predicts the service features of each node in the next reconstruction cycle through Kalman filtering, and then fuses the service features of all single-hop reachable nodes obtained through broadcast interaction to obtain the network features of the next reconstruction cycle and pass them to the knowledge plane, including: Set Nodes within the reconstruction cycle The real business characteristics of , through observation and maximum likelihood estimation, the business characteristics are obtained ;in, For Node In the The true mean of the business characteristics of the reconstruction period, and set The change of follows a random walk process; For Node In the The observed values ​​of the business characteristics of the reconstruction period represent the average data length and set The true mean Noisy measurement of For Node In the The variance of the service characteristics of the reconstruction period reflects the degree of discreteness of the data frame; The goal of Kalman filtering is to and and its corresponding variance and , prediction node In the The business characteristics of the reconstruction cycle are subject to , and then The prediction results of the first reconstruction cycle are weightedly integrated with the service characteristics of all single-hop reachable nodes obtained by broadcast interaction to obtain the Finally, the network node scale, the global network activity obtained by averaging the activity of all nodes in the network, and the global network service characteristics are combined into the first The network features of the reconstruction cycle are transmitted to the knowledge plane; Among them, the Kalman filter prediction process is not performed in the first two reconstruction cycles when the network is initially established, the new nodes added to the network at the end of the reconstruction cycle do not participate in the algorithm process of the observation surface, and the observation surface only performs the process of observation, maximum likelihood estimation and feature interactive fusion in the first two reconstruction cycles.

9. The method according to claim 8, characterized in that The Kalman filter estimates the state through two steps of prediction and update. First, the prediction step obtains the The prior estimates of the mean and variance of the business characteristics of the reconstruction period are expressed as follows: ； ； in, For the A priori estimate of the mean of the business characteristics of the reconstruction period, For the The posterior estimate of the mean of the business characteristics of the reconstruction period; For the The variance of the business characteristics of the reconstruction cycle A priori estimate of For the A posterior estimate of the variance of the business characteristics of the reconstruction period; is the process noise variance; Then, the updated Kalman gain is obtained based on the update step of the posterior estimate , the expression is ； According to the Kalman gain Respectively The posterior estimate of the mean of the service characteristics of the reconstruction period and the posterior estimate of the variance Update, expressed as: ； ； So that in the next iteration, the node In the The predicted distribution of business characteristics of a reconstruction cycle is , and Respectively represent nodes In the The predicted values ​​of the mean and variance of the business characteristics of the reconstruction period; among them, the node In the A priori estimate of the mean of the service characteristics of the reconstruction period Prediction , The forecast is obtained by exponential smoothing method and is expressed as ;in, is the preset smoothing index, Indicates The predicted value of the variance of the business characteristics of the reconstruction period.

10. The method according to claim 1, characterized in that The knowledge plane carries a pre-trained classification model, and outputs the best MAC protocol reconstruction strategy for the next reconstruction cycle to the control plane according to the network features input by the observation plane and the task requirements input by the task plane, including: First, the knowledge aspect is to model the task requirements based on the performance parameters of the MAC protocol in three dimensions: throughput, packet loss rate, and average network delay. This includes using effectiveness requirements, reliability requirements, and time sensitivity requirements to quantify the importance of task requirements, and defining effectiveness requirement parameters. , reliability requirement parameters and time-sensitive demand parameters As throughput , Packet loss rate and average network delay The performance parameters of the three dimensions are normalized so that , , , and satisfies ; Then, the normalized performance parameters and task requirement parameters are weighted and summed as the score evaluation index of different MAC protocols. It is expressed as: ； Among them, the subscript h Indicates the number of alternative MAC protocols, the superscript T represents transpose; Finally, the knowledge aspect is based on the pre-trained XGBoost classification model, with network features including network node scale, global network activity, and the mean and variance of global network business features as classification features, and validity requirement parameters as validation parameters. , reliability requirement parameters and time-sensitive demand parameters The task requirements are used as evaluation features to form data set features to classify candidate MAC protocols, and the candidate MAC protocol with the highest score is selected as the best MAC protocol that takes into account the network characteristics and task requirements of the next reconstruction cycle, expressed as: 。 11. A task-driven MAC protocol reconstruction device, characterized in that: The device comprises: The load modeling module is used to consider the multi-domain distributed wireless self-organizing network scenario. The network load model is constructed through multi-dimensional feature partitioning and extraction of network load, Gaussian random process modeling and feature fusion. A protocol reconstruction module, used to construct a software-defined distributed subnet node system architecture based on the network load model, the architecture consisting of a task plane, a data plane and a joint control plane; The task plane is used to cache the communication services generated during the execution of node tasks and cluster tasks in the form of data frames to form a queue on the data plane. The data plane queue includes the service data frames generated by the node tasks, and also includes the control data frames generated by the feature interaction during the MAC protocol reconstruction process and the channel reservation and preemption behavior in the MAC protocol process. The joint control plane includes an observation plane, a knowledge plane and a control plane; wherein the observation plane is used to observe and count the services of each node, obtain the service characteristics of each node in the current reconstruction period by maximum likelihood estimation, and obtain the service characteristics of each node in the next reconstruction period by Kalman filtering prediction, and then perform feature fusion on the service characteristics of all single-hop reachable nodes obtained by broadcast interaction to obtain the network characteristics of the next reconstruction period and pass them to the knowledge plane; The knowledge plane is used to carry the pre-trained classification model, and output the best MAC protocol reconstruction strategy for the next reconstruction cycle to the control plane according to the network features input by the observation plane and the task requirements input by the task plane; specifically, the knowledge plane first models the task requirements based on the performance parameters of the MAC protocol in three dimensions: throughput, packet loss rate, and average network delay; and then, based on the pre-trained XGBoost classification model carried, the task requirements are used as evaluation features; The control plane is used to carry the MAC protocol component library to determine whether the reconstruction strategy output by the knowledge plane is the same as the MAC protocol of the current reconstruction cycle. If they are the same, the reconstruction process is not performed; if they are not the same, the components are searched and the protocol parameters are adjusted to complete the MAC protocol component-level reconstruction, and enter the cycle of the next reconstruction cycle.

12. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 10 are implemented.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images