SPONTANEOUS EDGE APPLICATION DEPLOYMENT AND PRICING METHOD BASED ON INCENTIVE MECHANISM

Abstract

Disclosed in the present invention is a spontaneous edge application deployment and pricing method based on an incentive mechanism. The method comprises the following steps: building an edge end application oriented spontaneous deployment system architecture; then proposing an incentive mechanism aiming at spontaneous edge application deployment and prizing; solving the spontaneous edge application deployment and prizing problem based on a backward induction method, thereby obtaining an optimal deployment solution of an edge server and an optimal prizing strategy of an application provider.

Claims

1. A spontaneous edge application deployment and pricing method based on an incentive mechanism, characterized by comprising the following steps: 1) building an edge application oriented spontaneous deployment system architecture; 2) proposing an incentive mechanism aiming at spontaneous edge application deployment and prizing; and 3) obtaining the spontaneous edge application deployment and prizing method.

2. The spontaneous edge application deployment and pricing method based on the incentive mechanism according to claim 1, characterized in that the edge application oriented spontaneous deployment system architecture in step 1) comprises three portions: (1) deployment monitoring, wherein the deployment monitoring is used for collecting real-time information from a user application deployment request; (2) deployment planning, wherein a core of a deployment system is calculated and is used for planning a spontaneous edge application deployment and prizing method according to related given information; and (3) a deployment engine, wherein current application deployment operation is executed according to the obtained spontaneous deployment and prizing method.

3. The spontaneous edge application deployment and pricing method based on the incentive mechanism according to claim 2, characterized in that the incentive mechanism aiming at spontaneous edge application deployment and prizing in step 2) is defined as follows: a ratio of capacity of each edge server to obtain deployment right is defined as ${\beta }_i\left(x_m^i,x_{-m}^i\right)=\frac{x_m^i}{{.Math.}_{k\in ℳ}{x}_k^i},$ wherein: (1) x.sub.m.sup.i is a deployment willingness of an edge server m to service i, and x.sub.. . . m.sup.i is a deployment willingness matrix of all edge servers except the edge server m; and (2) the deployment willingness x.sub.m.sup.i is related to an intention value of a current round of deployment right competition, and processing capacity and a memory space of the edge server m, which is as follows:
x.sub.m.sup.i=γ.sub.m.sup.i(ξu.sub.m.sup.i+εf.sub.m)  (1) wherein ξ and ε represent weighting parameters, u.sub.m.sup.i is a memory space left when the edge server m deploys service i, f.sub.m represents a clock frequency of the edge server m, and γ.sub.m.sup.i represents a given intention value, aiming a charge of an application provider, of deployment right competition when the edge server m deploys service i.

4. The deployment and pricing method according to claim 3, characterized in that an objective function of the spontaneous edge application deployment and prizing method in step 3) is: profit U.sup.i.sub.A obtained by the application provider in each round of service deployment: wherein, $\begin{array}{cc}{U}_A^i=\underset{p_m}{\max }\underset{m\in ℳ}{.Math.}{p}_m^i{x}_m^i& \left(2\right)\end{array}$ 1) the profit U.sup.i.sub.A obtained by the application provider in each round of service deployment comes from expense needing to be paid to the provider by the edge server after winning the deployment right; where p.sub.m.sup.i represents pricing proposed by the application provider to the edge server m obtaining the deployment right in the ith round of deployment right competition; profit U.sup.is obtained by the edge server in each round of service deployment: $\begin{array}{cc}{U}_S^i=\underset{x_m,i\in \mathrm{��},n\in ℳ}{\max }\left\{\left(R+{\mathrm{rs}}_i\right)\frac{x_m^i}{{.Math.}_{k\in ℳ}{x}_k^i}\mathrm{QoS}-p_m^i{x}_m^i-c_m^i\right\}& \left(3\right)\end{array}$ 2) the profit U.sup.is obtained by the edge server in each round of service deployment consists of reward available for the server after winning the deployment right, expense paid to the application provider by the server and spending of service operation of the server; where R represents fixed reward earning available for the edge server after succeeding in obtaining the deployment right; rs.sub.i represents part of variable award value; r is a given variable award factor; s.sub.i is a memory size of an ith service; and c.sup.i.sub.m is source energy consumption cost of the edge server m when deploying and operating service i, a value thereof is related to a unit memory cost of the server and a unit calculation cost of a CPU, and a calculation equation thereof is as follows:
c.sub.m.sup.i=a.sub.ms.sub.i+b.sub.mo.sub.s.sub.i  (4)
b.sub.m=ψf.sub.m  (5) wherein a.sub.m represents the unit memory cost of the edge server m; b.sub.m represents the unit calculation cost of a CPU of the edge server m and is relates to a clock frequency of the server, and ψ is a weighting coefficient; and o.sub.si represents a clock number consumed by operating service i; oS, a quality of service, is a function related to service transmission time and execution time, the lower the quality of service is, the lower the reward obtained by the edge server is, the calculation is as follows:
QoS=f(b.sub.m.sub.tran.sup.i,b.sub.m.sub.exec.sup.i)  (6) wherein, b.sub.m.sub.tran.sup.i is a transmission time of data needed by edge server m before deploying service i; b.sub.m.sub.exec.sup.i is an executed time of service i on the edge server m; and f is a function for normalization processing of b.sub.m.sub.tran.sup.i and b.sub.m.sub.exec.sup.i.

5. The spontaneous edge application deployment and pricing method based on the incentive mechanism according to claim 1, characterized in that a process for application deployment and pricing on the basis of the method is as follows: 1) request stage, wherein a deployment request of any application from a user is received by the application provider; 2) competitive stage, wherein the edge server is made to compete for the deployment right of each service of the required application, and a Nash equilibrium between the service pricing and a deployment intention value of the edge server are found on the basis of a Stackelberg game model; 3) deployment stage, wherein the service is deployed; and 4) repeating steps 2)-3) until after all services of the application are deployed.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0041] FIG. 1 is an edge application oriented spontaneous deployment system architecture;

[0042] FIG. 2 is an incentive mechanism aiming at spontaneous edge application deployment and prizing;

[0043] FIG. 3 is a relation between service average optimal price and a service number;

[0044] FIG. 4 is a relation between a deployment willingness total value of an edge server and the service number;

[0045] FIG. 5 is a relation between the service average optimal price and the dependent data number;

[0046] FIG. 6 is a relation between a deployment willingness total value of all edge servers and the dependent data number;

[0047] FIG. 7 is a relation between a profit average value of the edge server and a residue memory space in a single round of service deployment;

[0048] FIG. 8 is a relation between a total profit value of an application provider and the residue memory space.

DESCRIPTION OF THE EMBODIMENTS

[0049] The present invention will be further elaborated hereafter in conjunction with accompanying drawings and the specific embodiments.

[0050] A spontaneous edge application deployment and pricing method based on an incentive mechanism includes the following steps:

[0051] 1) building an edge application oriented spontaneous deployment system architecture;

[0052] 2) proposing an incentive mechanism aiming at spontaneous edge application deployment and prizing; and

[0053] 3) obtaining the spontaneous edge application deployment and prizing method.

[0054] As shown in FIG. 1, the edge application oriented spontaneous deployment system architecture in the present invention mainly includes three portions of:

[0055] (1) deployment monitoring, wherein the deployment monitoring is used for collecting real-time information from a user application deployment request;

[0056] (2) deployment planning, wherein a core of a deployment system is calculated and is used for planning a spontaneous edge application deployment and prizing method according to related given information; and

[0057] (3) a deployment engine, wherein current application deployment operation is executed according to the obtained spontaneous deployment and prizing strategy.

[0058] In a traditional edge application deployment system without any intermediary agent, the edge server may not be willing to become the deployment side because a large amount of resource space and computing power consumption are needed for data storage and service execution, and the edge server without considering the incentive rewards may submit an incorrect result in a great probability, and postpones the deployment process at will. For solving the problem, we provide an incentive mechanism aiming at spontaneous edge application deployment and prizing.

[0059] Capacity proportion of each edge server to obtain deployment right is defined as

[00004]${\beta }_i\left(x_m^i,x_{-m}^i\right)=\frac{x_m^i}{{.Math.}_{k\in ℳ}{x}_k^i},$

wherein:

[0060] (1) x.sub.m.sup.i is a deployment willingness of an edge server m to service i, and x.sub.. . . m.sup.i is a deployment willingness matrix of all edge servers except the edge server m; and

[0061] (2) the deployment willingness x.sub.m.sup.i is related to an intention value of a current round of deployment right competition, and processing capacity and a memory space of the edge server m, which is as follows:

x.sub.m.sup.i=γ.sub.m.sup.i(ξu.sub.m.sup.i+εf.sub.m)  (1)

wherein ξ and ε represent weighting parameters, the value thereof is 0-1, u.sub.m.sup.i is a memory space left when the edge server m deploys service i, f.sub.m represents a clock frequency of edge server m, and γ.sub.m.sup.i represents a given intention value, aiming a charge of an application provider, of deployment right competition when the edge server m deploys service i.

[0062] FIG. 2 shows a flow schematic diagram of the spontaneous edge application deployment and pricing method. After the user sends an application deployment request, a responder analyses service and data elements in an application and input service information needing deployment into a work network in a digraph sequence, then all edge servers start to compete for the deployment right and the operation right of the service, and the winning capacity of the edge servers are measured according to their memory spaces, clock frequency and the deployment intention values. For making the mechanism effectively solve the spontaneous deployment problem of the application, the present invention introduces the Stackelberg game module with profit maximization on the basis of the mechanism. After the application deployment request generates, the application provider provides the edge server side with a bidding of each round of service deployment right, the edge server side decides this round of deployment willingness according to the provided bidding, and then a strategy space and a profit function of the two game sides are further constructed.

[0063] Profit U.sup.i.sub.A obtained by the application provider in each round of service deployment:

[00005]$\begin{array}{cc}{U}_A^i=\underset{p_m}{\max }\underset{m\in ℳ}{.Math.}{p}_m^i{x}_m^i& \left(2\right)\end{array}$

[0064] 1) The profit U.sup.i.sub.A obtained by the application provider in each round of service deployment comes from expense needing to be paid to the provider by the edge server after winning the deployment right.

[0065] p.sub.m.sup.i represents pricing proposed by the application provider to the edge server m obtaining the deployment right in the ith round of deployment right competition;

[0066] profit U.sup.is obtained by the edge server in each round of service deployment:

[00006]$\begin{array}{cc}{U}_S^i=\underset{x_m,i\in \mathrm{��},n\in ℳ}{\max }\left\{\left(R+{\mathrm{rs}}_i\right)\frac{x_m^i}{{.Math.}_{k\in ℳ}{x}_k^i}\mathrm{QoS}-p_m^i{x}_m^i-c_m^i\right\}& \left(3\right)\end{array}$

[0067] 2) The profit U.sup.is obtained by the edge server in each round of service deployment consists of reward available for the server after winning the deployment right, expense paid to the edge service provider by the server and spending of service operation of the server.

[0068] R represents fixed reward earning available for the edge server after succeeding in obtaining the deployment right; rs.sub.i represents part of variable award value; r is a given variable award factor; s.sub.i is a memory size of an ith service; and c.sup.i.sub.m is source energy consumption cost of the edge server m when deploying and operating service i, a value thereof is related to a unit memory cost of the server and a unit calculation cost of a CPU, and a calculation equation thereof is as follows:

c.sub.m.sup.i=a.sub.ms.sub.i+b.sub.mo.sub.s.sub.i  (4)

b.sub.m=ψf.sub.m  (5)

[0069] wherein a.sub.m represents the unit memory cost of the edge server m; b.sub.m represents the unit calculation cost of a CPU of the edge server m and is relates to a clock frequency of the server, and ψ is a weighting coefficient; and o.sub.si represents a clock number consumed by operating service i.

[0070] oS, a quality of service, is a function related to service transmission time and execution time, the lower the quality of service is, the lower the reward obtained by the edge server is, the calculation is as follows:

QoS=f(b.sub.m.sub.tran.sup.i,b.sub.m.sub.exec.sup.i)  (6)

[0071] wherein, b.sub.m.sub.tran.sup.i is a transmission time of data needed by edge server m before deploying service i; b.sub.m.sub.exec.sup.i is an executed time of service i on the edge server m; and f is a function for normalization processing of b.sub.m.sub.tran.sup.i an b.sub.m.sub.exec.sup.i.

[0072] For the spontaneous edge application deployment problem, the present invention uses the Stackelberg game model to optimize the target. The spontaneous edge application deployment problem is mapped to a double-target model, after the deployment request generates, the application provider analyses the application into several service and data elements and provides the pricing p.sub.m.sup.i needing to be paid by the edge server in each round of deployment right competition, and then the edge server determines the corresponding strategy space and obtains the capacity proportion of the deployment right available for the edge server in the competition process through a calculation equation (1), wherein the larger the capacity proportion is, the larger the probability obtaining the deployment right is. Finally, the strategy space and the profit function of the two game sides are further constructed. The incentive mechanism mainly includes four stages of:

[0073] 1) Initialization

[0074] In the initialization stage, it is necessary to determine the application digraph needing to be deployed, the source energy consumption cost c.sup.i.sub.m of the edge server m when deploying and operating service i, the variable award factor r and the fixed reward earning R.

[0075] 2) Incentive Operation

[0076] By means of the incentive operation, the edge server actively competes for the deployment right of service for obtaining the profit. The capacity proportion of the edge server to obtain the deployment right is calculated according to the following equation:

[00007]$\begin{array}{cc}{\beta }_i\left(x_m^i,x_{-m}^i\right)=\frac{x_m^i}{{.Math.}_{k\in ℳ}{x}_k^i}& \left(8\right)\end{array}$

[0077] 3) Game Operation

[0078] On the basis of the provided equations (2) and (7), meanwhile considering the influence of the quality of service on the user experience during deployment, the equation (7) is subjected to secondary derivation, and it is certified that the equation satisfies strictly concave function features.

[00008]$\begin{array}{cc}\frac{\partial U_s}{\partial x_m^i}=\left(R+{\mathrm{rs}}_i\right)\frac{\partial {\beta }_m^i}{\partial x_m^i}\mathrm{QoS}-p_m^i& \left(9\right)\\ \frac{\partial U_s}{{\partial }^2{x}_m^i}=\left(R+{\mathrm{rs}}_i\right)\frac{{\partial }^2{\beta }_m^i}{{\partial }^2{x}_m^i}\mathrm{QoS}<0& \left(10\right)\\ \mathrm{Since}\frac{{\partial }^2{\beta }_m^i}{{\partial }^2{x}_m^i}=-2\frac{{.Math.}_{m\ne k}{x}_k^i}{{\left({.Math.}_{k\in ℳ}{x}_k^i\right)}^2}<0& \left(11\right)\end{array}$

[0079] Thus, it may be known from equations (9), (10) and (11) that a Nash equilibrium point exist in the game model.

[0080] After optimal X*.sup.i is solved, the equation (7) is subjected to derivation to obtain a maximum value, the optimal solution P*.sup.i is determined, the optimal price does have a closed mode, and the optimal price of a single service is closely linked to the prices of other services, that is, when the price of one service changes, other services also need to update the corresponding expense, and accordingly, the optimization process of the price need to be solved by means of an iteration manner.

[0081] 4) Deployment Operation

[0082] The Nash equilibrium point is solved between the strategy space and the profit of the application provider and the edge server, and then the service may be deployed.

[0083] The process for application deployment and pricing by using the method of the present invention is as follows:

[0084] 1) request stage, wherein a deployment request of any application from a user is received by the application provider;

[0085] 2) competitive stage, wherein the edge server is made to compete for the deployment right of each service of the required application, and a Nash equilibrium between the service pricing and a deployment intention value of the edge server are found on the basis of a Stackelberg game model;

[0086] 3) deployment stage, wherein the service is deployed; and

[0087] 4) repeating steps 2)-3) until after all services of the application are deployed.

[0088] 5) Simulation Result

[0089] Data results of the method (SELL) of the present invention and conventional methods (genetic algorithm GA, particle swarm optimization PSO, and Hill Climbing algorithm) are compared. Comparing FIGS. 3 and 4, it may be known that the method of the present invention may obtain an optimal deployment solution of the edge server and an optimal pricing strategy of the application provider, so as to make the application provider and the edge server obtain the highest profits under the condition of low delay. Comparing FIGS. 5 and 6, it may be known that along with the dependent data number changes, the effect of the method of the present invention is hardly influenced, and compared with the other three methods, the method more easily obtains the optimal deployment solution of the edge server and the optimal pricing strategy of the application provider, so as to make the application provider and the edge server obtain the highest profits under the condition of low delay. Comparing FIGS. 7 and 8, it is known that the change of the residue memory space in the edge server may influence the profits of the application provider and the edge server, and by means of comparison and analysis, the weighting parameter of the residue memory space is set to be 0.8.