A Differentiated Service Transmission Method Based on a Double-Layer Satellite Heterogeneous Network

Abstract

The present invention relates to a differential service transmission method based on a double-layer satellite heterogeneous network, belonging to the field of satellite communication technology. The method includes: S1: obtaining the position sets of users, base stations and satellites by using a satellite positioning system and ephemeris information; meanwhile, constructing a service model of the users, and forming a candidate service node set and a candidate service path set by using user request information, position information and node resources; S2: aiming at maximizing the average service experience quality of users, jointly considering user service quality level selection, service path selection, and service node computing resource and power resource allocation, and establishing a service quality level adaptive optimization problem; S3: solving the problem by using the variable relaxation method, the successive convex approximation method and the alternating direction multiplier method; S4: converting the solved linear solution into a non-linear solution by using the marginal utility function. The present invention can improve the service experience quality of users and the throughput of information transmission, and enhance the reliability of information transmission.

Description

Technical Field

[0001] The present invention belongs to the technical field of satellite communication, and relates to a differential service transmission method based on a two-layer satellite heterogeneous network. Background Art

[0002] Satellite networks have advantages that cannot be compared with traditional terrestrial networks, such as global coverage, resistance to natural disasters, low latency, and high bandwidth. Therefore, satellite networks play an important role in communication. With the development of multimedia technology, future households will have an increasing demand for bandwidth. For example, in the case of video services, most current flat TV installations will be ultra-high definition (UHD), and the bit rate of UHD video is about 15 - 18 Mbps. This will pose a huge challenge to the next-generation mobile network. At the same time, many developing terrestrial communication technologies have been further improved for use in satellite networks. Utilizing the advantages of satellite networks can effectively improve the service transmission rate and the user's service quality experience. Currently, the satellites launched into space mainly consist of LEO satellites and GEO satellites. And with the continuous increase in the number of satellites, the frequency band, as a non-renewable resource, will severely restrict the future development of the air-ground integrated network. Therefore, different satellite constellations must operate in the same frequency band and share spectrum resources. However, sharing the spectrum between LEO satellites and GEO satellites will cause co-channel interference. The International Telecommunication Union Radio Regulations stipulate that NGEO satellites can share the same frequency band with GEO satellites on the condition that they do not cause unacceptable interference to GEO satellites. Therefore, designing a service transmission method based on a two-layer satellite heterogeneous network is an inevitable development result.

[0003] However, the current research focus is mainly on terrestrial service transmission mechanisms, and there is relatively little research on satellite-ground joint differential service transmission mechanisms. The mechanism that only considers terrestrial propagation is easily restricted by the terrestrial environment when deploying backhaul equipment, and terrestrial communication resources are difficult to meet the growing user needs. Therefore, by deploying a differential service transmission mechanism in a two-layer satellite heterogeneous network, the user's service quality experience can be improved. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a differential service transmission method based on a two-layer satellite heterogeneous network. This method combines power allocation, computing resource allocation, path selection, and service level selection algorithms, and can improve the user service experience quality and the throughput of information transmission, and enhance the reliability of information transmission.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] A differential service transmission method based on a two-layer satellite heterogeneous network specifically includes the following steps:

[0007] S1: Obtain the position sets of nodes such as users, base stations, and satellites using the satellite positioning system and ephemeris information; at the same time, construct the service model of the user, and use the user request information, location information, and node resources to form a candidate service node set and a candidate service path set;

[0008] S2: Based on the service model constructed in step S1, and considering the co-frequency interference limitation of the two-layer satellite at the same time, with the goal of maximizing the average service experience quality of the user, jointly consider the selection of user service quality level, service path selection, and the allocation of computing resources and power resources of service nodes, and establish a service quality level adaptive optimization problem model;

[0009] S3: Use the variable relaxation method and the continuous convex approximation method to transform the constructed optimization problem into a convex problem, and then use the alternating direction multiplier method to solve to obtain the user service quality level selection and service path selection results, and obtain the corresponding resource allocation plan;

[0010] S4: Use the marginal utility function to convert the solved linear solution into a non-linear solution.

[0011] Furthermore, the physical nodes of the network include: users, low-earth orbit satellites, geostationary satellites, base stations, cloud servers, and SDN controllers. Among them, users receive downlink data and select the service quality level. Satellites, base stations, and cloud servers are responsible for the transmission of downlink data, the collection and upload of channel state information, resource allocation control, and acting as relay links to transmit signaling and data. The SDN controller is responsible for aggregating the global channel state information, executing the proposed joint power allocation, computing resource allocation, path selection, and service level selection algorithms under the two-layer satellite network and feeding back the results to the corresponding nodes.

[0012] Furthermore, in step S1, the service model refers to general services, including traffic-intensive tasks, computing-intensive tasks, and other types of tasks; for a certain type of service, let the service quality that the user can choose be q ∈ {1,..., Q}, and the link data rate required for each service quality is v q , and the computing resources required are c q , that is, the user service can be expressed as (q, v q , c q ).

[0013] Due to the difficulty of deploying wired backhaul links and the significant advantages of the current satellite network, a satellite network is added as a wireless backhaul network, and modern communication technologies (edge computing, caching, software-defined network) are introduced. Users can first request services from edge nodes. If the edge nodes cannot meet the user's needs, they can obtain the resources of the cloud server through the satellite backhaul link, thereby further improving the service transmission performance.

[0014] Furthermore, in step S2, the established self-adaptive optimization problem model for service quality level is expressed as:

[0015] P1:

[0016] s.t.C2.1:

[0017] C2.2:

[0018] C2.3:

[0019] C2.4:

[0020] C2.5:

[0021] C2.6:

[0022] C2.7:

[0023] C2.8:

[0024] Among them, C2.1 means that each user can only select one service quality level, C2.2 means that the sum of the service flows of each user should be equal to the data rate required by the user, C2.3 means the rate limit provided by the node for the user, C2.4 means the relationship between the node computing resource allocation and the service level, C2.5 means the downlink interference limit of the LEO satellite on the GEO satellite, C2.6 means the link capacity limit, C2.7 means the computing resource limit, and C2.8 means the power resource limit; among them, {z qm}, {y mn}, and {p l} are the elements of sets Z, Y, R, and P respectively. z qm represents the service quality level q selected by user m, y mn represents whether service node n can provide computing services for user m, represents the service transmission rate of path ; represents the number of users, s q represents the revenue of each service quality level q; represents the set of all users, and Q represents the highest service quality level; represents the k-th path from service node n to user m, represents the set of all paths of user m, and v qm represents the rate required for user m to select a service with quality level q; Denotes the set of paths from service node n to user m, v Qm Denotes the rate required for user m to select the service with the highest quality level, z Qm Denotes that user m selects the highest quality level, Denotes the set of all service nodes; EPFD m Denotes the interference received by the geostationary satellite ground station, EPFD th Denotes the interference threshold of the geostationary satellite, Denotes the geostationary ground station; Denotes the link Whether it has passed through link l, Denotes all paths in the network, Denotes the maximum capacity of link l; h mn Denotes whether service node n can provide services for user m, c m Denotes the computing resources required by user m, C n Denotes the computing resources owned by the node; p l Denotes the power allocated to link l, P n max Denotes the maximum power that node n can provide, Denotes the set of all links from node n to the next-hop node.

[0025] The proposed problem model aims to help users select appropriate service quality levels, while invoking the resources of the network and nodes to support users' selections, and helping the network maximize the utility of service quality levels.

[0026] Furthermore, step S3 specifically includes the following steps:

[0027] S31: The network control plane initializes parameters;

[0028] S32: Relax the binary discrete user association variable to the continuous interval of [0,1];

[0029] S33: The controller estimates the initial point of continuous relaxation and continuous convex approximation method iteration according to the global channel state information;

[0030] S34: Use the continuous convex approximation method to convert the non-convex link capacity constraint condition into a linear constraint condition;

[0031] S35: Solve the service quality level selection sub-problem on the user side based on the updated parameters;

[0032] S36: Solve the computing resource allocation sub-problem on the service node side based on the updated parameters;

[0033] S37: Solve the service path selection sub-problem on the wireless network side based on the updated parameters;

[0034] S38: Solve the link power allocation sub - problem on the radio network side based on the updated parameters;

[0035] S39: Update the Lagrange multipliers;

[0036] S39: If it has not converged or has not reached the maximum number of iterations, return to step S35; otherwise, continue;

[0037] S310: Determine whether it has converged. If it has not converged, it is determined that the problem has no solution; if it has converged, the problem has a feasible solution.

[0038] Furthermore, in step S35, the obtained user - side service quality level selection sub - problem is as follows, and this problem can be separated from the users and independently selected by each user:

[0039] P2:

[0040]

[0041] s.t.C3.1:

[0042] where α mn and β mn are introduced auxiliary variables, whose role is to convert the C2.3 and C2.4 inequalities in the original problem into equalities. μ m , λ mn and ω mn are the Lagrange multipliers for constraints C2.2, C2.3, and C2.4. ρ1, ρ2, and ρ3 are penalty parameters, and it is required that the penalty parameters are greater than zero. Since the objective function is a convex function and the constraints are all linear constraints, this problem is a convex problem and can be solved using standard convex optimization solution methods such as simulation with the CVX toolbox in MATLAB.

[0043] Furthermore, in step S36, the calculation resource allocation optimization sub - problem on the service node side is as follows, which is independently solved by each service node:

[0044] P3:

[0045] s.t.C3.2:

[0046] In this problem, the user service level selection, path selection, and other parameters are given in advance, and only the calculation resource allocation index of the service node is a variable. Therefore, this problem is a convex optimization problem, and the optimal calculation resource allocation situation can be directly solved by using simulation with the CVX toolbox in MATLAB.

[0047] Further, in step S37, the service path selection sub-problem on the radio network side is as follows:

[0048] P4:

[0049]

[0050] where χ l is an introduced auxiliary variable, whose role is to convert the C2.3 and C2.4 inequalities in the original problem into equalities. is the lower bound obtained by the sequential convex approximation method, ν l is the Lagrange multiplier of constraint C2.6, ρ4 is the penalty parameter, and it is required that the penalty parameter is greater than zero. The objective function is a quadratic function, which is a common convex function. By using the CVX toolbox to input variables and the objective function, the optimal service transmission path selection situation can be obtained.

[0051] Further, in step S38, the link power allocation sub-problem on the radio network side is as follows:

[0052] P5:

[0053] s.t.C3.3:

[0054] C3.4:

[0055] In this problem, there are only link power allocation variables, and the objective function is a convex function and the constraints are linear constraints. The CVX toolbox can be used to input variables, the objective function, and the constraints to solve the optimal link power resource allocation situation.

[0056] However, the current binary discrete variables are relaxed to continuous values, and the solution obtained based on the alternating direction method of multipliers is the linear solution of the relaxed problem. For convenient application in practice, therefore, the present invention must restore the user service quality level selection variables and service node computing resource allocation variables to binary values.

[0057] Further, step S4 specifically includes: restoring the obtained linear solution to a discrete solution through the marginal utility function. First, obtain the marginal utility of each user for different service quality levels and each service node for different users' computing resource allocations; each user selects the service quality level with the maximum marginal utility, and each service node provides computing services for several users with the maximum marginal utility.

[0058] The beneficial effects of the present invention are as follows: The present invention combines power allocation, computing resource allocation, path selection, and service level selection algorithms, which can improve the user service experience quality and the throughput of information transmission, and enhance the reliability of information transmission.

[0059] Other advantages, objectives and features of the present invention will be described to some extent in the following specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be learned from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the following specification. Brief Description of the Drawings

[0060] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be described in detail preferably with reference to the accompanying drawings, where:

[0061] Figure 1 It is a flowchart of the differential service transmission method based on the double-layer heterogeneous network of the present invention;

[0062] Figure 2 It is a flowchart of the resource allocation, path selection and service quality level selection algorithm based on the alternating direction method of multipliers. Detailed Embodiments

[0063] The following uses specific specific examples to illustrate the embodiments of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0064] Please refer to Figures 1 to 2 , the present invention provides a differential service transmission method based on a double-layer heterogeneous network. In this method, the physical nodes of the network include: users, low-earth orbit satellites, geostationary orbit satellites, base stations, cloud servers, and SDN controllers. Among them, users receive downlink data and select service quality levels. Satellites, base stations, and cloud servers are responsible for sending downlink data, collecting and uploading channel state information, resource allocation control, and acting as relay links to transmit signaling and data. The SDN controller is responsible for aggregating global channel state information, executing the proposed joint power allocation, computing resource allocation, path selection, and service level selection algorithms under the double-layer satellite network and feeding back the results to the corresponding nodes.

[0065] As Figure 1 shown, the method specifically includes the following steps:

[0066] S1: Obtain the position sets of nodes such as users, base stations, and satellites using the satellite positioning system and ephemeris information. At the same time, construct the user's service model, and use the user request information, location information, and node resources to form a candidate service node set and a candidate service path set.

[0067] The defined service model refers to general services, including traffic-intensive tasks, computing-intensive tasks, and other types of tasks. For a certain type of service, assume that the service quality q that the user can choose belongs to {1,...,Q}, and each service quality corresponds to the required link data rate v q , and the required computing resource is c q . That is, the user service can be expressed as (q, v q , c q ).

[0068] The service transmission path can be divided according to different types of service nodes. Specifically, base stations, satellites, and cloud servers can all be used as service nodes, and different service transmission paths can be further formed according to different relay nodes. According to the user's service requirements, the distribution of node service resources, and whether the nodes are reachable, the user's candidate path set can be constructed

[0069] S2: Based on the service model constructed in step S1, and considering the co-frequency interference limitation of the double-layer satellite, with the goal of maximizing the average service experience quality of the user, jointly consider the selection of user service quality levels, service path selection, and the allocation of computing resources and power resources of service nodes, and establish a service quality level adaptive optimization problem model.

[0070] Use the Equivalent Power Fluw Density (EPFD) as the interference threshold. The EPFD is specified by the International Telecommunication Union, and its definition is the sum of the power flux densities generated by all transmitting radio stations on the Earth's surface or at the receiving radio stations of the geostationary satellite system in the geostationary orbit within the non-geostationary satellite system range, and consider the off-axis discrimination that may point to the reference receiving antenna. The specific expression of EPFD is as follows.

[0071]

[0072] Among them, p i represents the transmission power of the LEO satellite, G t (θ i ) is the antenna gain of the LEO satellite transmitting antenna in the off-axis angle θ i direction, is the antenna gain of the GEO satellite ground station receiving antenna in the off-axis angle direction, d iis the distance from the LEO satellite to the GEO satellite ground station, G rmax is the maximum antenna gain at the receiving end. Considering that the link capacity is determined by bandwidth, power, and interference, the link capacity is limited under resource constraints. Therefore, the maximum capacity R of the link l max is written as:

[0073]

[0074] where the transmit power p l is a continuous variable, G l is the channel gain, is the noise power. Thus, the service quality adaptation problem model is established as:

[0075] P1:

[0076] s.t. C2.1:

[0077] C2.2:

[0078] C2.3:

[0079] C2.4:

[0080] C2.5:

[0081] C2.6:

[0082] C2.7:

[0083] C2.8:

[0084] where C2.1 means that each user can only select one service quality level, C2.2 means that the sum of the traffic flows of each user should be equal to the data rate required by the user, C2.3 means the rate limit provided by the node for the user, C2.4 means the relationship between the node computing resource allocation and the service level, C2.5 means the downlink interference limit of the LEO satellite to the GEO satellite, C2.6 means the link capacity limit, C2.7 means the computing resource limit, C2.8 means the power resource limit; where, {z qm}, {y mn}, and {p l} are the elements of sets Z, Y, R, and P respectively, z qm represents the service quality level q selected by user m, y mnIndicates whether service node n can provide computing services for user m. Indicates a path of the business transmission rate. Indicates the number of users, s q Indicates the revenue for each quality of service level q; Indicates the set of all users, and Q represents the highest quality of service level; Indicates the k-th path from service node n to user m, Indicates the set of all paths of user m, v qm Indicates the rate required for user m to select a service with quality level q; Indicates the set of paths from service node n to user m, v Qm Indicates the rate required for user m to select the service with the highest quality level, z Qm Indicates that user m selects the highest quality level, Indicates the set of all service nodes; EPFD m Indicates the interference received by the geostationary satellite ground station, EPFD th Indicates the interference threshold of the geostationary satellite, Indicates the geostationary ground station; Indicates a link Whether it has passed through link l, Indicates all paths in the network, Indicates the maximum capacity of link l; h mn Indicates whether service node n can provide services for user m, c m Indicates the computing resources required by user m, C n Indicates the computing resources owned by the node; p l Indicates the power allocated to link l, P n max Indicates the maximum power that node n can provide, Indicates the set of all links from node n to the next-hop node.

[0085] S3: Since the quality of service adaptive problem model in step S1 is a mixed-integer non-linear programming (MINLP), and this problem is usually considered NP-hard. Therefore, first use the variable relaxation method and the continuous convex approximation method to transform the constructed optimization problem into a convex problem, and then use the alternating direction multiplier method to solve to obtain the user quality of service level selection and service path selection results, and obtain the corresponding resource allocation scheme. The specific solution process is shown in Figure 2 .

[0086] For the integer variables {z qm , y mn} ∈ {0, 1}, relax them into continuous variables {z qm , ymn} ∈ [0, 1]. Then, considering that l max the expression of R is not a convex function, it is rewritten using the continuous convex approximation method as follows:

[0087]

[0088] where is a matrix stacked by all the approximation coefficients a l and b l according to their indices.

[0089] Then, the alternating direction method of multipliers is used to solve the relaxed problem. First, the augmented Lagrangian function form of the problem is obtained:

[0090]

[0091] where, α mn β mn and χ l are the introduced auxiliary variables, whose role is to convert the C2.3, C2.4, and C2.6 inequalities in the original problem into equalities. μ m λ mn ω mn and ν l are the Lagrange multipliers for the constraints C2.2, C2.3, C2.4, and C2.6, and ρ1, ρ2, ρ3, and ρ4 are the penalty parameters.

[0092] Then, according to the augmented Lagrangian function, the problem is divided into the service quality level selection problem on the user side, the computing resource allocation problem on the service node side, the path selection problem on the wireless network side, and the link power allocation problem:

[0093] The service quality level selection optimization problem on the user side can be separated from the users and independently selected by each user according to the resource allocation situation:

[0094] P2:

[0095] s.t. C3.1:

[0096] Since the objective function is a convex function and the constraints are all linear constraints, this problem is a convex problem, and a standard convex optimization solution method, including simulation using the CVX toolbox in MATLAB, is used to solve it.

[0097] The computing resource allocation optimization problem on the service node side can be separated from the service nodes and independently solved by each service node according to the requests of the users and the network conditions:

[0098] P3:

[0099] s.t.C3.2:

[0100] In this problem, the user service level selection, path selection, and other parameters are given in advance. Only the computing resource allocation metric of the service node is a variable. Therefore, this problem is a convex optimization problem, and the optimal computing resource allocation can be directly obtained by simulating using the CVX toolbox in MATLAB.

[0101] The service path selection problem solved by the SDN controller of the wireless network is:

[0102] P4:

[0103] The objective function is a quadratic function, which is a common convex function. The optimal service transmission path selection can be obtained by inputting variables and the objective function using the CVX toolbox.

[0104] The power resource allocation problem solved by the SDN controller of the wireless network is:

[0105] P5:

[0106] s.t.C3.3:

[0107] C3.4:

[0108] In this problem, there are only link power allocation variables, and the objective function is a convex function and the constraints are linear constraints. The optimal link power resource allocation can be solved by inputting variables, the objective function, and the constraints using the CVX toolbox.

[0109] The SDN controller passes the Lagrange multipliers to the network nodes and users to help them solve the corresponding problems. After optimizing the local variables, the network nodes feedback them to the SDN controller to help it adjust the decision, and update the Lagrange multipliers in each iteration based on the following formula:

[0110]

[0111]

[0112]

[0113]

[0114] S4: However, the current binary discrete variables are relaxed to continuous values, and the obtained solution is the linear solution of the relaxed problem. Therefore, for convenient application in practice, we must restore them to binary values. Thus, the marginal utility function is used to convert the solved linear solution into a non-linear solution. First, obtain the marginal utility of each user for different service quality levels and each service node for the computing resource allocation to different users, that is

[0115] For z qm , each user m selects the service quality level q′ with the largest marginal benefit . For y mn , assume y′ mn ∈[0,1] is the solution achieved by solving the relaxed problem. We can calculate the number of computing services that can be provided on node n, that is where means taking the largest integer value less than x. For each node n, the Γ with a larger marginal benefit can be selected from n users

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the present technical solution, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A differential service transmission method based on a two-layer satellite heterogeneous network, characterized in that, The method specifically includes the following steps: S1: Obtain the position sets of the user, base station, and satellite using the satellite positioning system and ephemeris information; meanwhile, construct the service model of the user, and use the user request information, location information, and node resources to form a candidate service node set and a candidate service path set; S2: Based on the service model constructed in step S1, and considering the co-frequency interference limitation of the double-layer satellite, with the goal of maximizing the average service experience quality of the user, jointly consider the selection of user service quality levels, service path selection, and the allocation of computing resources and power resources of service nodes, and establish an adaptive optimization problem model for service quality levels; S3: Use the variable relaxation method and the successive convex approximation method to transform the constructed optimization problem into a convex problem, and then use the alternating direction multiplier method to solve to obtain the results of user service quality level selection and service path selection, and obtain the corresponding resource allocation plan; S4: Use the marginal utility function to convert the solved linear solution into a non-linear solution.

2. The differential service transmission method according to claim 1, characterized in that In step S1, the service model refers to general services, including traffic-intensive tasks, computing-intensive tasks, and other types of tasks. For a certain type of service, assume that the service quality options available to the user are q ∈ {1,..., Q}, and the required link data rate for each service quality is v q , and the required computing resource is c q , that is, the user service is represented as (q, v q , c q ).

3. The differential service transmission method according to claim 2, wherein In step S2, the established adaptive optimization problem model for service quality levels is expressed as: Among them, C2.1 means that each user can only select one service quality level, C2.2 means that the sum of the traffic flows of each user should be equal to the data rate required by the user, C2.3 means the rate limit provided by the node for the user, C2.4 means the relationship between the node's computing resource allocation and the service level, C2.5 means the downlink interference limit of the LEO satellite to the GEO satellite, C2.6 means the link capacity limit, C2.7 means the computing resource limit, and C2.8 means the power resource limit; among them, {z qm}, {y mn}, and {p l} are the elements of sets Z, Y, R, and P respectively, z qm represents the service quality level q selected by user m, y mn represents whether service node n can provide computing services for user m, represents the service transmission rate of path , represents the number of users, s q represents the revenue of each service quality level q; represents the set of all users, and Q represents the highest service quality level; represents the k-th path from service node n to user m, represents the set of all paths of user m, v qm represents the rate required by user m to select a service with quality level q; represents the set of paths from service node n to user m, v Qm represents the rate required by user m to select the service with the highest quality level, z Qm represents that user m selects the highest quality level, represents the set of all service nodes; EPFD m represents the interference received by the GEO satellite ground station, EPFD th represents the interference threshold of the GEO satellite, represents the GEO ground station; represents the link whether it passes through link l, represents all paths in the network, represents the maximum capacity of link l; h mn represents whether service node n can provide services for user m, c m represents the computing resources required by user m, C n represents the computing resources owned by the node; p l represents the power allocated to link l, P n max represents the maximum power that node n can provide, represents the set of all links from node n to the next-hop node.

4. The differential service transmission method according to claim 3, wherein Step S3 specifically includes the following steps: S31: Initialize the parameters on the network control plane; S32: Relax the binary discrete user association variable to the continuous interval of [0,1]; S33: The controller estimates the initial point of the successive relaxation and successive convex approximation method iterations according to the global channel state information; S34: Use the successive convex approximation method to convert the non-convex link capacity constraint condition into a linear constraint condition; S35: Solve the sub-problem of user-side service quality level selection based on the updated parameters; S36: Solve the sub-problem of computing resource allocation on the service node side based on the updated parameters; S37: Solve the sub-problem of service path selection on the wireless network side based on the updated parameters; S38: Solve the sub-problem of link power allocation on the wireless network side based on the updated parameters; S39: Update the Lagrange multiplier; S39: If it has not converged or has not reached the maximum number of iterations, return to step S35, otherwise continue; S310: Judge whether it has converged. If it has not converged, it is determined that the problem has no solution. If it has converged, the problem has a feasible solution.

5. The differential service transmission method according to claim 4, wherein In step S35, the obtained sub-problem of user-side service quality level selection is as follows, and this problem can be separated from the user and independently selected by each user: Among them, α mn and β mn are introduced auxiliary variables, whose role is to convert the C2.3 and C2.4 inequalities in the original problem into equalities. μ m , λ mn and ω mn are the Lagrange multipliers that constrain C2.2, C2.3 and C2.4, and ρ1, ρ2 and ρ3 are penalty parameters.

6. The differential service transmission method according to claim 5, wherein In step S36, the optimization sub-problem of computing resource allocation on the service node side is as follows, and it is independently solved by each service node:

7. The differential service transmission method according to claim 6, wherein In step S37, the sub-problem of service path selection on the wireless network side is: where, χ l is an introduced auxiliary variable, whose role is to convert the C2.6 inequality in the original problem into an equation, is the lower bound obtained by the sequential convex approximation method, ν l is the Lagrange multiplier for the constraint C2.6, and ρ4 is the penalty parameter.

8. The differentiated service transmission method according to claim 7, wherein In step S38, the sub-problem of link power allocation on the wireless network side is:

9. The differential service transmission method according to claim 1, wherein Step S4 specifically includes: restoring the obtained linear solution to a discrete solution through the marginal utility function. First, obtain the marginal utility of each user for different service quality levels and the computing resource allocation of each service node for different users; each user selects the service quality level with the maximum marginal utility, and each service node provides computing services for several users with the maximum marginal utility.

Images