High capacity vehicle clustering method and system based on cyber-physical dual network

Abstract

The application discloses a high-capacity vehicle clustering method and system based on an information-physical dual network, which comprises the following steps: firstly, constructing a physical network topology graph corresponding to the current network of vehicles, and calculating a vehicle mobility correlation index; then, constructing a physical network topology graph corresponding to the current network of vehicles, and calculating a link transmission quality; further, superimposing the physical network and the information network by weighted summation, calculating a correlation density under a comprehensive network, and further establishing an optimal clustering model based on the correlation density; finally, using a matrix decomposition algorithm to approximately solve, and obtaining a final clustering result. In the calculation process, the channel fading under the LOS and NLOS environments is considered, and a channel link capacity calculation method is given through a matrix mother function. The method can reflect the stability of the physical topology, and can also reflect the link transmission quality of the whole channel, so that the VC construction with a more efficient transmission structure can be realized.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, specifically relating to a high-capacity vehicle clustering method and system based on cyber-physical dual networks. Background Technology

[0002] To improve traffic efficiency and user driving experience in Intelligent Transportation Systems (ITS), vehicle-to-everything (V2X) networks are required to support ultra-high bandwidth and intensive computing services, including advanced driver assistance systems (ADAS), safe autonomous driving services, and immersive applications. Therefore, Mobile Edge Computing (MEC) technology has been introduced into V2X. To enable nearby vehicles to collaborate on communication and computing resources to support more services, MEC incorporates a vehicular cloud network architecture. Therefore, how to cluster vehicles to form a high-capacity vehicular cloud (VC) structure to effectively improve the performance of various services through vehicle-to-vehicle collaborative computing and transmission is a problem that needs to be studied.

[0003] To date, scholars both domestically and internationally have proposed various clustering methods for vehicular networks (V2X), including vehicle clustering based on evolutionary game theory, vehicle clustering using node similarity merging strategies and network incremental analysis, and dynamic clustering based on link connectivity prediction. However, current clustering research is based on link connectivity in V2X. Unlike existing work, the VC structure in MEC requires higher transmission capacity to support the various ultra-high bandwidth and intensive computing services in V2X. Therefore, clustering for VC will face the following two challenges: First, in wireless transmission networks, channel link quality is not only affected by random characteristics, but also significantly impacted by inter-vehicle interference as the number of intelligent connected vehicles increases. Second, in physical networks, the vehicle topology determines the length of channel links and the distribution of obstacles, leading to the coexistence of line-of-sight and non-line-of-sight channels; and their dynamic changes affect channel conditions and interference distribution, exacerbating the instability and complexity of channel transmission. In summary, existing research methods are not suitable for V2X network clustering, which requires an efficient transmission structure.

[0004] To achieve clustering in vehicle-to-everything (V2X) networks, the patent application "Method, Apparatus, Device and Storage Medium for Vehicle Clustering in V2X" (application date: September 30, 2020, application number: CN202011058759.7, publication number: CN112243213B) filed by Beijing University of Aeronautics and Astronautics discloses a method for dividing vehicles into clusters using the average mobility of the target vehicle and the data packets sent by neighboring vehicles within a preset range; the patent application "A NOMA-based V2X Broadcast Communication Resource Allocation Method" (application date: July 7, 2021, application number: CN20211) filed by Southeast University discloses a method for dividing vehicles into clusters using the average mobility of the target vehicle and the data packets sent by neighboring vehicles within a preset range. Patent application 0767610.4 (Publication No. CN113490275) discloses a method for classifying vehicles into clusters using the geographical location of V2V users; Southwest University for Nationalities' patent application "Artificial Fish Swarm Clustering Method, Storage Medium and Device in Vehicle Networking" (Application Date: June 7, 2020, Application No.: CN202010509269.8, Publication No.: CN112004208B) discloses a method for vehicle networking clustering using the fitness value F of the artificial fish swarm algorithm; Beijing University of Aeronautics and Astronautics' patent application "Clustering Method, Device and Equipment Suitable for High Dynamic Large-Scale Vehicle Networking" (Application Date: 2020-20 ... classifying vehicles into clusters using the geographical location of V2V users; Southwest University for Nationalities' patent application "Artificial Fish Swarm Clustering Method, Storage Medium and Device in Vehicle Networking" (Application Date: 2020-2020, Application No.: CN202010509269.8, Publication No.: CN112004208B) discloses a method for vehicle networking clustering using the fitness value F of the artificial fish swarm algorithm; Beijing University of Aeronautics and Astronautics' patent application "Clustering Method, Device and Equipment Suitable for High Dynamic Large-Scale Vehicle Networking" (Application Date: 2020-2020, Application No.: CN113490275) discloses a method for classifying vehicles into clusters using the geographical location of V2V users; Southwest University for Nationalities' patent application "Artificial Fish Swarm Clustering Method, Storage Medium and Device in On September 25, 2019, application number CN201910912214.9, publication number CN110662182B disclosed a vehicle clustering method that uses inter-node communication messages to establish a neighbor table and then update the cluster head; Nanjing University of Posts and Telecommunications' patent application "A vehicle clustering method based on link reliability and stability" (application date: October 29, 2019, application number: CN201911036660.4, patent number: CN110855563B) disclosed a vehicle clustering method based on link reliability and stability; Huazhong University of Science and Technology's patent application "Different security The patent application "A Distance-Based Genetic Clustering Method for Fog Nodes in Vehicle-to-Everything (V2X) under Different Security Constraints" (application date: March 26, 2018, application number: CN201810251973.0, publication number: CN108616568B) filed by Nanjing University of Aeronautics and Astronautics discloses a distance-based clustering method for fog nodes in V2X.

[0005] The above methods address the impact of a single network and do not consider the randomness of links within the network or the impact of interference on link connectivity and overall transmission capacity. Therefore, they cannot guarantee that their clustering results will achieve efficient network transmission performance. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a high-capacity vehicle clustering method and system based on cyber-physical dual networks in the Internet of Vehicles (IoV). This method addresses the problems of existing technologies not taking into account the coupling effect between the physical network and the information network in the IoV network, and not considering the random characteristics of links in the network and the impact of interference on the connectivity of links and the overall transmission capacity.

[0007] To achieve the above objectives, the present invention employs the following technical solution:

[0008] A high-capacity vehicle clustering method based on cyber-physical dual networks includes the following steps:

[0009] A vehicle physical network topology graph is constructed, which is a weighted graph constructed based on mobility relationships during vehicle operation. The mobility correlation index between vehicles is calculated through the physical network topology graph.

[0010] A vehicle information network topology map is constructed, which is a weighted graph based on the transmission link quality between vehicles. The transmission quality between vehicle links is calculated through the information network topology map.

[0011] By overlaying the vehicle physical network topology and the vehicle information network topology, a cyber-physical dual network is obtained. Based on the mobility correlation index between vehicles and the link transmission quality between vehicles, the correlation density of two vehicles in the information dual network is established. The weights of mobility correlation and link transmission quality are adjusted by coefficients in the correlation density, and a model for maximizing the correlation density is established.

[0012] By approximating the model that maximizes the correlation density using a matrix factorization algorithm, the clustering results of vehicles are obtained, and vehicles with greater correlation are grouped into one cluster.

[0013] A further improvement of the present invention is that:

[0014] Preferably, the mobility correlation index between vehicles The calculation formula is:

[0015]

[0016] in, The distance correlation index is calculated using the following formula:

[0017]

[0018] for in For similarity in speed magnitude, For vehicle v iWith vehicle v j Similarity in velocity direction between them;

[0019] The calculation formulas are as follows:

[0020]

[0021]

[0022] Preferably, the link transmission quality between the vehicles is... for:

[0023]

[0024] Among them, C ij C ji and C d It is a channel and and a safe distance from interference d The expected value of the corresponding channel transmission capacity.

[0025] Preferred, C d The expression is as follows:

[0026]

[0027] Preferably, channel C ij The calculation formula is:

[0028]

[0029] in,

[0030]

[0031] Preferably, the correlation density is:

[0032]

[0033] in, Let k represent the vertex set of the c-th VC in the vehicle set, where k is the number of VCs. express The number of vehicles in the area.

[0034] Preferably, the correlation density is:

[0035]

[0036] Here, λ is used to adjust the weights of mobility association and link transmission quality.

[0037] Preferably, the model for maximizing the correlation density is:

[0038]

[0039] Preferably, the specific process of solving the model using the matrix factorization algorithm is as follows:

[0040] Use x ·c Let represent the c-th column vector of matrix X. Using the symbol ' to denote the transpose of matrix X, we derive:

[0041]

[0042] The equivalent transformation from correlation density to matrix trace yields the following formula:

[0043]

[0044]

[0045] in, The c-th column vector in is have

[0046] Association density is expressed as:

[0047]

[0048] in

[0049] W PT =λW P +(1-λ)W T Substituting into equation (28), the symmetric nonnegative matrix factorization (SNMF) based on cyber-physical dual networks can approximately solve the maximum trace optimization problem, where the binary constraint is relaxed to a continuous interval [0,1].

[0050] To discrete solutions The continuous variables in the array are approximately assigned a value of 1 for the largest element in each row and 0 for all other elements. The column containing the element assigned a value of 1 represents the cluster to which that vehicle belongs.

[0051] A high-capacity vehicle clustering system based on cyber-physical dual networks includes:

[0052] A physical network unit is used to construct a vehicle physical network topology map, which is a weighted graph constructed based on mobility relationships during vehicle operation. The mobility correlation index between vehicles is calculated through the physical network topology map.

[0053] An information network unit is used to construct a vehicle information network topology map, which is a weighted graph constructed based on the transmission link quality between vehicles. The transmission quality between vehicle links is calculated through the information network topology map.

[0054] Cyber-physical dual network unit is used to overlay the vehicle physical network topology map and the vehicle information network topology map to obtain a cyber-physical dual network; the association density of two vehicles in the cyber-physical dual network is established based on the mobility association index between vehicles and the link transmission quality between vehicles. The association density is adjusted by coefficients to adjust the weights of mobility association and link transmission quality, and a model for maximizing the association density is established.

[0055] The solution unit is used to approximate the model that maximizes the correlation density through matrix factorization algorithm, obtain the clustering results of vehicles, and group vehicles with greater correlation into one cluster.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] This invention discloses a high-capacity vehicle clustering method based on cyber-physical dual networks. The specific steps of the clustering method are as follows: (1) Constructing a physical network topology map corresponding to the vehicle in the current network and calculating the vehicle mobility correlation index; (2) Constructing a physical network topology map corresponding to the vehicle in the current network and calculating the link transmission quality; (3) Superimposing the physical network and the information network by weighted summation, calculating the correlation density under the integrated network, and further establishing an optimal clustering model based on the correlation density; (4) Approximating the solution using a matrix factorization algorithm to obtain the final clustering result. This invention considers channel fading under LOS and NLOS environments during the calculation process and provides a method for calculating channel link capacity through a moment generating function. This method can reflect the stability of the physical topology and the link transmission quality of the entire channel, thereby enabling the construction of VCs with more efficient transmission structures. Simulation results show that this invention can obtain clustering results with higher network transmission performance in mobile vehicle networks.

[0058] This invention also discloses a high-capacity vehicle clustering system based on cyber-physical dual networks. The system includes physical network units, information network units, cyber-physical dual network units, and a solution unit. The system first establishes physical network units and information network units respectively. Taking advantage of the fact that the boundary conditions of the two units are the same, the two units are merged to obtain a cyber-physical dual network unit. A clustering model is established in the cyber-physical dual network unit. After solving the clustering model, the vehicle clustering results are obtained. Attached Figure Description

[0059] Figure 1 This is a flowchart of the present invention;

[0060] Figure 2 This is a graph showing the variation of association density with the number of vehicles under different cluster numbers according to the present invention;

[0061] Figure 3This is a graph showing the variation of correlation density with vehicle speed under different hyperparameters according to the present invention;

[0062] Figure 4 This is a graph showing the variation of channel link capacity with vehicle speed under different cluster numbers according to the present invention. Detailed Implementation

[0063] The present invention will now be described in further detail with reference to the accompanying drawings:

[0064] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. The terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, unless otherwise explicitly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection or a detachable connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal connection of two elements. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0065] The network topology of vehicles in the current environment is constructed, the corresponding weights under different topologies are calculated, the vehicle association density is further calculated, and finally the vehicle integrated network is clustered by matrix factorization to output the clustering results. The specific implementation steps are as follows: (1) Construct the vehicle physical network topology and calculate the vehicle mobility association index; (2) Construct the vehicle information network topology and calculate the link transmission quality; (3) Calculate the vehicle association density under the integrated network and establish an optimal clustering model; (4) Use the matrix factorization algorithm to approximate the solution and output the clustering results.

[0066] The specific implementation steps are as follows:

[0067] Step 1: Construct the vehicle physical network topology diagram The physical network is a weighted graph constructed based on the mobility relationships of vehicles during road travel; in the topology graph... Let ε be the set of vertices for vehicles, where each vertex corresponds to a vehicle in the network; P ={(v i ,v j )} represents the edge set corresponding to the vehicle's physical network, where each edge Defined as vehicle v i With vehicle v jThere is a direct impact on driving between them; This is the weighted adjacency matrix corresponding to the vehicle in the physical network, where each element in the set... Defined as vehicle v i With vehicle v j Mobility correlation index between them;

[0068] Calculate the mobility correlation index between vehicles Assume all vehicles are equipped with a Global Positioning System (GPS), which provides information about vehicle movement, including location, speed, and direction (with right as positive). For each vehicle v i Its speed is expressed as z i The direction of the velocity is θ. i , represents the angle between the vehicle's direction of travel and the positive direction; d ij Indicates vehicle v i With vehicle v j The relative distance between them. For each element in the weighted adjacency matrix of a vehicle in the physical network topology graph. Defined as in and These are the distance correlation index and the movement similarity correlation index for vehicles, respectively. The specific calculation method is as follows:

[0069] Considering the impact between following vehicles during vehicle operation, use This indicates a predefined effective distance range, meaning the distance between vehicles is less than... At that time, it was considered that there was a driving influence between the vehicles. Furthermore, to ensure safety, the distance between the two vehicles should not be less than [amount missing]. d Therefore, the vehicle distance correlation index is defined. The calculation is as follows:

[0070]

[0071] Define the vehicle movement similarity association index for in For similarity in speed magnitude, For vehicle v i With vehicle v j The similarity of their velocity directions. Their calculation methods are as follows:

[0072]

[0073]

[0074] Step 2, Construct the vehicle information network topology The information network is constructed as a weighted graph based on the quality of the transmission links between vehicles; the information network has the same set of vehicle vertices as the physical network. ε T ={ <v i ,v j <} represents the edge set corresponding to the vehicle information network, where each edge... Defined as vehicle v i With vehicle v j There is a direct transmission link connection between them; This is the weighted adjacency matrix corresponding to the vehicle in the information network, where each element in the set... Defined as vehicle v i With vehicle v j Normalized link transmission quality between them.

[0075] Calculate the transmission quality of the link between vehicles Link transmission quality is defined as the weight of the information network topology; the communication range between vehicles is defined as R. If vehicle v i With vehicle v j If the distance between them does not exceed R, then a transmission link is considered to exist between them. The transmission quality of this link depends on path loss, channel fading, and interference. To represent a bidirectional transmission channel... and The average quality will determine the link transmission quality. Defined as:

[0076]

[0077] Among them, C ij C ji and C d It is a channel and and a safe distance from interference d The expected value of the corresponding channel transmission capacity.

[0078] Step 2.1. Due to path loss, channel fading, and interference, the channel... The received signal-to-interference-plus-noise ratio (SINR) is expressed as:

[0079]

[0080] Among them, |h ij | represents the magnitude of the channel fading coefficient, and α represents the path loss exponent. It is a link The set of interfering nodes. P i and P k These are the launching vehicles v i and interfering vehicle v kThe power. Typically, N0 represents the power spectral density of additive white Gaussian noise. Where i is the transmitting vehicle, j is the receiving vehicle, k is the jamming vehicle, and d... ij It is the distance between i and j, d kj It is the distance between vehicles k and j.

[0081] link The instantaneous capacity is expressed as C( <v i ,v j >), its expression is:

[0082] C( <v i ,v j >)=Blog2(1+SINR ij (6)

[0083] Where B is the channel bandwidth, SINR ij C represents the signal-to-interference-plus-noise ratio (SIR) of the link. <v i ,v j The random variable in >) is the corresponding channel fading coefficient.

[0084] in, For a safe distance without interference d The expected value of transmission capacity C under the following conditions d Its expression is as follows:

[0085]

[0086] Among them, h d Here, P represents the channel coefficient, and P represents the vehicle's transmit power.

[0087] The expected value C of the transmission capacity in equation (7) d The specific calculation method is as follows:

[0088] Step 2.1.1. Convert the logarithmic function to an integral function when r > 0:

[0089]

[0090] Where r is the parameter and t is the integrand.

[0091] Will Substituting into equation (8), the transmission capacity C( d This can be represented as:

[0092]

[0093] Among them, C( d The random variable in ) is the channel coefficient h d .

[0094] Will Substituting into the above equation, the average capacity can be expressed as:

[0095]

[0096] Step 2.1.2. Based on the concept of the Moment Generating Function (MGF), that is: Take P T, d =P|h d | 2 d -α Then there is C can be d Represented as:

[0097]

[0098] Step 2.1.3. Based on the Nakagami-m channel probability density function (PDF):

[0099]

[0100] Where m is the shape parameter representing the degree of fading, Ω is the expansion parameter, and Γ(·) represents the gamma function.

[0101] P T, d Further expressed as:

[0102]

[0103] Then the moment generating function (MGF)M T, d (s) can be expressed as:

[0104]

[0105] M T, d (s) Substitute into C d In the middle, we can obtain C. d expression:

[0106]

[0107] Step 2.2. In channel C ij There are two related channels, including the corresponding target channel. and interference channels Then for C ij The calculation, as analyzed above, yields the following results:

[0108]

[0109] Define target channel MGF is M T,<i,j> (s), then the target channel The joint MGF of the cumulative interference channel is represented as follows: We can obtain the following formula:

[0110]

[0111] Therefore, channel C can be obtained. ij The expression is as follows:

[0112]

[0113] in,

[0114]

[0115] Step 3: Establish an optimal clustering model based on vehicle association density; define V C It is a vehicle collection A subset of [the network]. Considering that the nodes and edges corresponding to the vehicle physical network and the vehicle information network are the same, only the weights of the two networks are different, by superimposing the edge weights in the graph according to a certain ratio, a cyber-physical dual network can be obtained. The VCs in the cyber-physical dual network are represented as [the following]. in and Let represent the set of vertices containing the c-th vertex in the vehicle set VC, where k is the number of vertices in VC. In graph theory, In the physical network topology diagram G P And information network topology diagram G T The internal correlation coefficients are respectively expressed as: and G P and G T The correlation densities are respectively expressed as and The calculation formula is as follows:

[0116]

[0117]

[0118] in express The number of vehicles in the system. To demonstrate overall stability and capacity, the transmission strength of the VC is defined as the correlation density, as shown below:

[0119]

[0120] λ is used to adjust the weights of mobility association and link transmission quality.

[0121] definition Let x be the indicator matrix of VC. ic Instruction vehicle v i Does it belong to the c-th cloud? Right now:

[0122]

[0123] according to By definition, the sum of all elements in each row of X equals 1, that is: ∑ c x ic =1. Therefore, the strength optimization VC construction problem based on cyber-physical dual networks can be expressed as the following model P:

[0124]

[0125] Step 4: Approximate the solution using matrix factorization algorithm; use x ·c Let represent the c-th column vector of matrix X. The symbol ' is used to denote the transpose of matrix X. It can be derived that:

[0126]

[0127] The equivalent transformation from correlation density to matrix trace can be derived as follows:

[0128]

[0129]

[0130] in, The c-th column vector in is have Association density can be expressed as:

[0131]

[0132] in

[0133] W PT =λW P +(1-λ)W T Substituting into equation (29) above, the symmetric nonnegative matrix factorization (SNMF) based on cyber-physical dual networks can approximately solve the maximum trace optimization problem, where the binary constraints are relaxed to a continuous interval [0,1]. Then, to discretize the solution... In this context, continuous variables are assigned approximately 1 for the largest element in each row and 0 for all other elements. The column containing an element assigned 1 represents the cluster to which that vehicle belongs; elements with a higher correlation are more likely to be grouped into the same cluster.

[0134] Output the clustering results.

[0135] To implement the above-mentioned units, the present invention is implemented through the following system, specifically including:

[0136] A physical network unit is used to construct a vehicle physical network topology map, which is a weighted graph constructed based on mobility relationships during vehicle operation. The mobility correlation index between vehicles is calculated through the physical network topology map.

[0137] An information network unit is used to construct a vehicle information network topology map, which is a weighted graph constructed based on the transmission link quality between vehicles. The transmission quality between vehicle links is calculated through the information network topology map.

[0138] Cyber-physical dual network unit is used to overlay the vehicle physical network topology map and the vehicle information network topology map to obtain a cyber-physical dual network; the association density of two vehicles in the cyber-physical dual network is established based on the mobility association index between vehicles and the link transmission quality between vehicles. The association density is adjusted by coefficients to adjust the weights of mobility association and link transmission quality, and a model for maximizing the association density is established.

[0139] The solution unit is used to approximate the model that maximizes the correlation density through matrix factorization algorithm, obtain the clustering results of vehicles, and group vehicles with greater correlation into one cluster.

[0140] Example

[0141] The following simulation examples further illustrate this:

[0142] I. Simulation conditions:

[0143] In the vehicle-to-everything (V2X) network, the road is 1000m long, with two lanes in one direction and a lane width of 3m. The safe distance between vehicles is... d The maximum connectivity distance of the vehicle physical network is 20m. The path loss is 100m, the path loss exponent α is 2, and the vehicle's transmit power P i The power is 0.1W, the vehicle information network communication range R is 300m, and the Gaussian white noise power is 10. -6 w, the Nakagami-m channel shape parameter m and spread parameter Ω are 3 and 1 respectively, and the minimum vehicle speed z min The speed is 20 km / h.

[0144] II. Simulation Content and Results:

[0145] Figure 2 The correlation density of the proposed VC-DCP algorithm with other algorithms in terms of vehicle number is described. Figure 3 The correlation density of the proposed VC-DCP algorithm is compared with that of other algorithms in terms of vehicle maximum speed. Figure 4 The link capacity comparison between the proposed VC-DCP algorithm and other algorithms in terms of maximum vehicle speed is described. From Figure 2 It can be seen that the correlation density of all different algorithms increases with the increase in the number of vehicles. This is because the link transmission and inter-vehicle influence increase with the increase in the number of vehicles. Furthermore, the correlation density of the proposed VC-DCP algorithm is higher than that of the LC-SNMF and k-means algorithms because the proposed algorithm comprehensively considers the inter-vehicle influence in the physical network and the link capacity in the information transmission network, and then uses a dual network model to represent the coupling between the dynamic physical network and the information transmission network. Moreover, this algorithm considers fading characteristics and interference from a more detailed perspective, which is beneficial for forming an efficient transmission structure. Figure 3 It can be seen that the association density of all algorithms decreases as the maximum vehicle speed increases. This is because as the maximum vehicle speed increases, the mobility similarity association index between vehicles decreases, leading to a decrease in association density. Furthermore, the value of association density and its changing trend increase with increasing λ, meaning that in the integrated network, the higher the proportion of physical mobility factors, the higher the association density. This is because, due to the fading of transmission links, the weights in the physical network are higher than those in the general information transmission network. A comparison of the average normalized link capacity and outage probability of the proposed VC-DCP algorithm with other algorithms is shown below. Figure 1 As shown. From Figure 4 It can be seen that the average normalized link capacity decreases as the maximum vehicle speed increases. This is because the initial link capacity changes more significantly with increasing vehicle speed, and may even decrease due to increased physical distance. Furthermore, the proposed VC-DCP algorithm outperforms the comparative algorithms on average because comprehensively considering the stability of the physical topology and the link transmission quality in the transmission network can improve the stability and efficiency of link transmission.

[0146] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-capacity vehicle clustering method based on cyber-physical dual networks, characterized in that, Includes the following steps: A vehicle physical network topology graph is constructed, which is a weighted graph constructed based on mobility relationships during vehicle operation. The mobility correlation index between vehicles is calculated through the physical network topology graph. A vehicle information network topology map is constructed, which is a weighted graph based on the transmission link quality between vehicles. The transmission quality between vehicle links is calculated through the information network topology map. By overlaying the vehicle physical network topology diagram and the vehicle information network topology diagram, a cyber-physical dual network is obtained. The association density of two vehicles in an information dual network is established based on the mobility association index between vehicles and the link transmission quality between vehicles. The weights of mobility association and link transmission quality are adjusted by coefficients in the association density, and a model for maximizing the association density is established. The model that maximizes the correlation density is approximately solved by matrix factorization algorithm to obtain the clustering results of vehicles, and vehicles with greater correlation are grouped into one cluster. The mobility correlation index between vehicles The calculation formula is: in, The distance correlation index is calculated using the following formula: （2） for ,in For similarity in speed magnitude, For vehicles With vehicles Similarity in velocity direction between them; The calculation formulas are as follows: （3） （4） The link transmission quality between the vehicles for: （5） in, , and It is a channel and and a safe distance from interference The expected value of the corresponding channel transmission capacity; The correlation density is: （21） in, Indicates the first vehicle in the set. The vertex set of a VC It refers to the quantity in VC. express The number of vehicles in the area; The specific process of solving the model using the matrix factorization algorithm is as follows: use Representation matrix The Column vectors, using symbols Representation matrix The transpose of is derived as follows: （24） The equivalent transformation from correlation density to matrix trace yields the following formula: （25） （26） in, The first in Column vector is ,have ; Association density is expressed as: （27） in ; Will Substituting into equation (28), the symmetric nonnegative matrix factorization (SNMF) based on cyber-physical dual networks can approximately solve the maximum trace optimization problem, where the binary constraint is relaxed to a continuous interval. ; To discrete solution The continuous variables in the array are approximately assigned a value of 1 for the largest element in each row and 0 for all other elements; the column containing the element assigned a value of 1 is the cluster to which the vehicle belongs.

2. The high-capacity vehicle clustering method based on cyber-physical dual networks according to claim 1, characterized in that, The expression is as follows: （15）。 3. The high-capacity vehicle clustering method based on cyber-physical dual networks according to claim 2, characterized in that, Channel The calculation formula is: （18） in, （20） 。 4. The high-capacity vehicle clustering method based on cyber-physical dual networks according to claim 1, characterized in that, The correlation density is: （22） in, Weights used to adjust mobility associations and link transmission quality.

5. The high-capacity vehicle clustering method based on cyber-physical dual networks according to claim 4, characterized in that, The model for maximizing the correlation density is: P: （23）。 6. A high-capacity vehicle clustering system based on a cyber-physical dual network for implementing the high-capacity vehicle clustering method of claim 1, characterized in that, include: A physical network unit is used to construct a vehicle physical network topology map, which is a weighted graph constructed based on mobility relationships during vehicle operation. The mobility correlation index between vehicles is calculated through the physical network topology map. An information network unit is used to construct a vehicle information network topology map, which is a weighted graph constructed based on the transmission link quality between vehicles. The transmission quality between vehicle links is calculated through the information network topology map. Cyber-physical dual network unit is used to overlay the vehicle physical network topology map and the vehicle information network topology map to obtain a cyber-physical dual network; the association density of two vehicles in the cyber-physical dual network is established based on the mobility association index between vehicles and the link transmission quality between vehicles. The association density is adjusted by coefficients to adjust the weights of mobility association and link transmission quality, and a model for maximizing the association density is established. The solution unit is used to approximate the model that maximizes the correlation density through matrix factorization algorithm, obtain the clustering results of vehicles, and group vehicles with greater correlation into one cluster.

Images