Power distribution network edge-edge cooperative task unloading method, device, equipment and medium

Abstract

The invention relates to an edge-edge cooperative task unloading method, device and equipment for a power distribution network and a medium, and the method comprises the steps: determining a utility function of a remote edge server based on the earnings obtained through the calculation service provided by the remote edge server and the data storage cost; based on earnings obtained by task calculation of the local edge server, payment and cooperative calculation of service cost and data storage cost are carried out to determine a utility function of the local edge server; constructing a Stackelberg game model by taking maximization of a remote edge server utility function and a local edge server utility function as a target; solving the Stackelberg game model by adopting a reverse induction method to obtain the optimal task unloading proportion of the local edge server and the optimal unit price of the remote edge server for providing the calculation service; and performing task unloading by adopting the optimal task unloading proportion of the local edge server and the optimal unit price of the calculation service provided by the remote edge server. The task processing efficiency of the power distribution network can be improved.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology for distribution networks, and in particular to a method, device, equipment and medium for edge-to-edge collaborative task unloading in distribution networks. Background Art

[0002] With the rapid development of the Power Internet of Things (Power IoT), the application of a large number of intelligent distribution terminals has dramatically increased power data collection and information interaction. At the same time, the distribution network has also derived diversified and large-scale computing business needs such as distributed power supply regulation and load demand side management. On the one hand, massive power data increases the complexity of distribution network operation and control. On the other hand, it also brings huge computing pressure to the edge server (ES) of the distribution network, and even affects the normal decision-making of the distribution network control system. Edge collaborative computing is one of the effective technologies to improve task computing performance. Through a reasonable edge collaborative computing architecture and collaborative strategy, the efficiency of task execution can be effectively improved and the quality of experience (QoE) of user experience can be improved. To this end, studying the computing resource scheduling strategy based on edge collaborative computing is one of the key issues to solve the insufficient computing power of the distribution network and provide low-latency and efficient computing services.

[0003] If edge-cloud collaborative task computing is adopted, although cloud servers have powerful computing capabilities and edge-cloud collaboration can give full play to the advantages of their respective system resources, it also has disadvantages such as extended latency, high communication bandwidth and data security requirements, and high construction and use costs. If edge-edge collaborative task computing is adopted, there are currently few research methods, especially in terms of its collaborative mechanism, which needs further exploration. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a distribution network edge-to-edge collaborative task unloading method, device, equipment and medium, which can improve the processing efficiency of distribution network tasks.

[0005] The technical solution adopted by the present invention to solve the technical problem is: to provide a distribution network edge-to-edge collaborative task unloading method, comprising the following steps:

[0006] Determine the utility function of the remote edge server based on the revenue obtained from providing computing services and the data storage cost of the remote edge server;

[0007] Determine the utility function of the local edge server based on the revenue obtained by performing task computation on the local edge server, the cost of paying the edge-to-edge collaborative computing service, and the cost of data storage;

[0008] Constructing a Stackelberg game model with the goal of maximizing the utility function of the remote edge server and the utility function of the local edge server;

[0009] The Stackelberg game model is solved by using the reverse induction method to obtain the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server;

[0010] The task offloading is performed using the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server.

[0011] The remote edge server utility function is expressed as: in, is the utility function of the remote edge server, p is the unit price of computing services provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete the task, The cache cost that is used by the task during the time it is queued on the remote edge server.

[0012] The cache cost occupied by the task during the queuing period on the remote edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the remote edge server.

[0013] The local edge server utility function is expressed as: in, is the utility function of the local edge server, γ is the conversion coefficient from task computing delay to economic benefit, t total,i is the execution time of the local edge server without edge collaboration, is the execution time of the task in the local edge server, is the execution time of the task on the remote edge server, is the cache cost occupied by the task during the queuing period of the local edge server, p is the unit price of the computing service provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete a task.

[0014] The cache cost occupied by the task during the queuing period of the local edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the local edge server.

[0015] The execution time of the local edge server without edge collaboration is expressed as: The execution time of the task on the local edge server is expressed as: The execution time of the task on the remote edge server is expressed as: in, is the queuing delay of the task in the local edge server, expressed as: is the queued task volume of the local edge server, The computing power of the local edge server; is the computational delay of the task in the local edge server, expressed as: is the computational delay of the task in the remote edge server, expressed as: in, The computing power of the remote edge server; is the queuing delay of the task in the remote edge server, expressed as: is the queued task volume of the remote edge server; is the communication delay of task offloading to the remote edge server, expressed as: R n is the transmission rate offloaded to the remote edge server, expressed as: R n =w·log2(1+P n ·h n / (N0·w)), w is the signal transmission bandwidth, P n is the side-to-side signal transmission power, h n is the channel gain, and N0 is the Gaussian white noise spectral density.

[0016] The reverse induction method is used to solve the Stackelberg game model to obtain the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server. Specifically, under the condition that the task offloading ratio is fixed, the optimal task offloading ratio of the local edge server is first determined according to the price of the computing service provided by the remote edge server, and then the optimal unit price of the computing service provided by the remote edge server is obtained according to the optimal task offloading ratio of the local edge server.

[0017] The technical solution adopted by the present invention to solve the technical problem is: to provide a distribution network edge-to-edge collaborative task unloading device, comprising:

[0018] A first determination module is used to determine a utility function of a remote edge server based on the revenue obtained by providing computing services and data storage costs of the remote edge server;

[0019] A second determination module is used to determine the utility function of the local edge server based on the revenue obtained by the local edge server from performing task computation, the cost of paying for the edge-to-edge collaborative computing service, and the cost of data storage;

[0020] A construction module, configured to construct a Stackelberg game model with the goal of maximizing the utility function of the remote edge server and the utility function of the local edge server;

[0021] A solution module, used to solve the Stackelberg game model by using a reverse induction method to obtain an optimal task offloading ratio of a local edge server and an optimal unit price of computing services provided by a remote edge server;

[0022] The task offloading module is used to offload tasks by using the optimal task offloading ratio of the local edge server and the optimal unit price of computing services provided by the remote edge server.

[0023] The utility function of the remote edge server determined by the first determination module is expressed as: in, is the utility function of the remote edge server, p is the unit price of computing services provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete the task, The cache cost that is used by the task during the time it is queued on the remote edge server.

[0024] The cache cost occupied by the task during the queuing period on the remote edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the remote edge server.

[0025] The local edge server utility function determined by the second determination module is expressed as: in, is the utility function of the local edge server, γ is the conversion coefficient from task computing delay to economic benefit, t total,i is the execution time of the local edge server without edge collaboration, is the execution time of the task in the local edge server, is the execution time of the task on the remote edge server, is the cache cost occupied by the task during the queuing period of the local edge server, p is the unit price of the computing service provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete a task.

[0026] The cache cost occupied by the task during the queuing period of the local edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the local edge server.

[0027] The execution time of the local edge server without edge collaboration is expressed as: The execution time of the task on the local edge server is expressed as: The execution time of the task on the remote edge server is expressed as: in, is the queuing delay of the task in the local edge server, expressed as: is the queued task volume of the local edge server, The computing power of the local edge server; is the computational delay of the task in the local edge server, expressed as: is the computational delay of the task in the remote edge server, expressed as: in, The computing power of the remote edge server; is the queuing delay of the task in the remote edge server, expressed as: is the queued task volume of the remote edge server; is the communication delay of task offloading to the remote edge server, expressed as: R n is the transmission rate offloaded to the remote edge server, expressed as: R n =w·log2(1+P n ·h n / (N0·w)), w is the signal transmission bandwidth, P n is the side-to-side signal transmission power, h n is the channel gain, and N0 is the Gaussian white noise spectral density.

[0028] The solution module includes:

[0029] The task optimal offloading ratio determination unit is used to determine the task optimal offloading ratio of the local edge server according to the price of computing services provided by the remote edge server under the condition that the task offloading ratio is fixed;

[0030] The optimal unit price determination unit is used to obtain the optimal unit price of the computing service provided by the remote edge server according to the optimal task offloading ratio of the local edge server.

[0031] The technical solution adopted by the present invention to solve its technical problem is: to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned distribution network edge-to-edge collaborative task unloading method when executing the computer program.

[0032] The technical solution adopted by the present invention to solve its technical problem is: providing a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned distribution network edge-to-edge collaborative task unloading method are implemented.

[0033] Beneficial Effects

[0034] Due to the adoption of the above-mentioned technical scheme, compared with the prior art, the present invention has the following advantages and positive effects: the present invention comprehensively considers the execution time and execution cost of the task, establishes a Stackelberg game model for task unloading between different edge servers of the distribution network, and gives a calculation method for its optimal game result through the proof of Nash equilibrium existence. Finally, according to the maximization of the interests of the local edge server, the target edge server for unloading and the task unloading ratio are selected to realize the edge-to-edge collaborative computing of the distribution network computing tasks. This method improves the computing efficiency of local business while increasing the benefits of other edge servers, achieving a win-win situation for all parties, solving the problem of unloading computing tasks in the distribution network, and improving the processing efficiency of distribution network tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the edge-cloud collaborative architecture of the distribution network;

[0036] Figure 2 is a flow chart of a distribution network edge-to-edge collaborative task unloading method according to a first embodiment of the present invention;

[0037] Figure 3 is a flow chart of an algorithm for solving the Stackelberg game model in the first embodiment of the present invention;

[0038] Figure 4 is a diagram of optimal game results under different servers in an embodiment of the present invention;

[0039] Figure 5 is a graph of edge-edge collaborative computing results under different γ in an embodiment of the present invention;

[0040] Figure 6 is a diagram of task offloading results under different strategies in an embodiment of the present invention;

[0041] Figure 7 This is a graph of edge-to-edge collaboration results under different local ESs in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0043] The edge-cloud collaborative architecture of the distribution network is as follows: Figure 1 As shown in the figure, taking the improved IEEE 33 node as an example, the distribution network is equipped with multiple ESs, each serving the task computing in its own area. The ESs are interconnected through the communication network.

[0044] In the distribution network, multiple edge servers are generally deployed. When a computing task is generated, the local ES can either perform local computing or offload the task to other ES for computing. Therefore, it is key to rationally utilize the computing resources in the network to reduce the computing pressure and improve the quality of task execution. The Stackelberg game model formulates the optimal task offloading strategy based on the information and interests of both parties in the game, thereby realizing the distributed decision-making of the system. This enables the strategy to adapt to the characteristics and changes of different nodes and improves the flexibility of task offloading. Secondly, its Nash equilibrium solution is the balance point where the overall interests of both parties in the game are maximized, ensuring the balance of interests of both parties in the game.

[0045] In this research scenario, the key issue of distribution network task unloading under edge-edge collaborative computing is the task unloading ratio and unloading cost. To this end, the first embodiment of the present invention relates to a distribution network edge-edge collaborative task unloading method, which comprehensively considers the execution time and execution cost of the task, establishes a task unloading Stackelberg game model between different ESs in the distribution network, and gives the calculation method of its optimal game result through the proof of Nash equilibrium existence. Finally, according to the maximization of the interests of the local ES, the unloading target ES and the task unloading ratio are selected to realize the edge-edge collaborative computing of the distribution network computing tasks.

[0046] like Figure 1 As shown, a distribution network edge-to-edge collaborative task unloading method of this embodiment specifically includes the following steps:

[0047] Step 1: Determine the utility function of the remote edge server based on the revenue obtained by the remote edge server from providing computing services and the data storage cost.

[0048] Step 2: Determine the utility function of the local edge server based on the revenue obtained by the local edge server from performing task computations, the cost of paying for edge-to-edge collaborative computing services, and the cost of data storage.

[0049] In the distribution network system, multiple distribution substations are equipped with an edge server, and different edge servers collaborate to improve the computing power of the entire distribution network. Suppose the edge server set in the distribution network is S = {S1, S2, S3, · · · S i ,···}. Each edge server in S is represented as Among them, p n For edge server S n Unit price of providing task offloading service (￥ / (cycle·s)), f n For edge server S n Computing power (cycle / s), Q n For edge server S n The number of queued tasks (cycle). For edge server S n The unit price (￥ / (bit·s)) for providing cache for task data. The task set generated by the distribution area is J = {J1, J2, J3, ···J i ,···}. Each task in J is represented as Among them, D i For Task J i The data size (bit), G i The number of CPU cycles required to complete the task. The maximum delay allowed for the task (s).

[0050] Task J i Computation latency between local ES and remote ES and Respectively expressed as:

[0051]

[0052]

[0053] Among them, λ is the task J i The offload ratio on the remote ES. and They are the computing capabilities of the local and remote ES respectively.

[0054] Task J i The queuing delay of waiting for calculation in the local ES and remote ES and They are:

[0055]

[0056]

[0057] in, and The number of queued tasks for local and remote ES respectively.

[0058] According to Shannon's theorem, task J i Offload to other edge servers S n The transmission rate (bit / s) is:

[0059] R n =w·log2(1+P n ·h n / (N0·w)) (5)

[0060] Where w is the signal transmission bandwidth, P n is the side-to-side signal transmission power, h n is the channel gain, and N0 is the Gaussian white noise spectral density.

[0061] Therefore, Task J i The communication time for offloading to other ES is:

[0062]

[0063] Since the amount of result data returned after task execution is very small, its transmission time is generally ignored.

[0064] In summary, Task J i Execution time on local and remote ES and They are:

[0065]

[0066]

[0067] With the cooperation of the side by side, task J i The final execution time is:

[0068]

[0069] Since the task data to be processed needs to occupy valuable server cache resources, the cost of data occupying the cache needs to be considered in edge-to-edge collaborative computing. i Cache cost occupied during local and remote ES queueing and They are:

[0070]

[0071]

[0072] In edge-edge collaborative computing, the goal of offloading tasks to the local ES is toi Part of the task is offloaded to other edge servers to reduce the task execution time. At the same time, the ES needs to pay the corresponding edge collaborative computing service costs, and the income of the local ES The difference between the benefit obtained from task computing latency and the cost of paying for edge-to-edge collaborative computing services and data storage costs is:

[0073]

[0074] Where γ is the conversion coefficient of task computing latency to economic benefits. p is the unit price of computing services provided by the remote ES (￥ / cycle). total,i It is the execution time of local ES without edge collaboration, that is:

[0075]

[0076] In edge-to-edge collaborative computing, the goal of remote ES to perform task offloading is to provide computing services and obtain revenue through reasonable pricing. The revenue of remote ES is It includes the difference between the revenue from providing computing services and the cost of data storage, namely:

[0077]

[0078] In this embodiment, the profit of the local ES is used as the local edge server utility function, and the profit of the remote ES is used as the remote edge server utility function.

[0079] Step 3: constructing a Stackelberg game model with the goal of maximizing the utility function of the remote edge server and the utility function of the local edge server.

[0080] The task offloading of edge-to-edge collaborative computing in the distribution network system is modeled as a Stackelberg game model. Among them, M is the leader, that is, the remote ES party; N is the follower, that is, the local ES party; p is the policy set of the target edge server, that is, the unit price of providing computing services; λ is the policy set of the local edge server, that is, the offloading ratio of the local edge server; and are the utility functions of the leader and follower, respectively.

[0081] In the Stackelberg game model, the goal of the leader (remote ES) is to maximize its own benefits by providing computing services, that is:

[0082]

[0083] The goal of the follower (local EC) is to maximize the benefits of reducing task execution time by offloading tasks to remote ECs, namely:

[0084]

[0085] Step 4: Use the reverse induction method to solve the Stackelberg game model to obtain the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server.

[0086] First, it is proved that the game model established above has a Nash equilibrium, and on this basis, a calculation method for the optimal game result is given. The calculation method is as follows: first, under the condition of a fixed task offloading ratio, the optimal task offloading ratio of the local edge server is determined according to the price of the computing service provided by the remote edge server, and then the optimal unit price of the computing service provided by the remote edge server is obtained according to the optimal task offloading ratio of the local edge server.

[0087] Specifically, when the two parties in the game reach a Nash equilibrium, each participant reaches its maximum value, and unilaterally changing the strategy of a participant will not increase its benefits, that is:

[0088] A non-cooperative game has at least one Nash equilibrium, namely

[0089] (1) The strategy space X(λ) is a non-empty bounded closed convex set;

[0090] (2) Utility function is continuously quasiconcave in X.

[0091] Obviously, for all local ES, the task offloading strategy space X is a bounded non-empty closed convex set in Euclidean space. It is known that bounded closed sets in Euclidean space are compact, and the utility function It is continuous in its strategy space. The following proves that the utility function satisfies the characteristics of a concave function.

[0092] First, Right now: We can obtain λ a , the following two cases are discussed:

[0093] (1) When 0≤λ≤λ a When The utility function of the follower's local ES is The first and second order partial derivatives of λ are calculated as follows:

[0094]

[0095]

[0096] Since γ>0, Therefore, the utility function is strictly concave, and the Stackelberg game model constructed in this implementation mode has a Nash equilibrium solution.

[0097] Let (17) equal to 0 to find the extreme point. So this point is the utility function The maximum value point of is the optimal unloading ratio λ b (p):

[0098]

[0099] because Therefore, the value range corresponding to p can be obtained at the same time. That is, setting λ equal to 0 can obtain the maximum value p corresponding to p max :

[0100]

[0101] Similarly, let λ be equal to λ a The minimum value p corresponding to p can be obtained min .

[0102] Furthermore, for λ b (p) The first-order derivative is as follows:

[0103]

[0104] In practical problems, the storage cost of data is much smaller than the computation cost, so therefore λ b (p) is a monotonically decreasing function of p. Therefore, when p>p max , λ=0.

[0105] (2) When λ a ≤λ≤1, we can see The utility function of the follower's local ES is The first-order and second-order partial derivatives are calculated as follows:

[0106]

[0107]

[0108] Similar to case (1), since γ>0, Utility Function is strictly concave and has a Nash equilibrium solution.

[0109] Similarly, since the data storage cost is much smaller than the computation cost, the sum of the second and third terms in equation (22) is thereby Utility Function In λ a ≤λ≤1 is monotonically decreasing, In λ a Similarly, the corresponding p at this time satisfies p<p min .

[0110] In summary, for different main ES computing service unit prices p, the optimal corresponding strategy λ of the local ES is * for:

[0111]

[0112] In the complete information game, the master ES obtains its optimal pricing strategy based on the optimal corresponding strategy of the slave ES. * = 0, that is, no task is unloaded from ES, then p * = 0. If λ * =λ a , the main ES utility function Find the first-order partial derivative with respect to p:

[0113]

[0114] because It is monotonically increasing with respect to p. Therefore p * =p min .

[0115] If * =λ b (p), Find the first-order partial derivative with respect to p:

[0116]

[0117] Provable As shown in formula (27).

[0118]

[0119] Derivative (27), the first term is The second term is a linear function of p, and its value is also less than zero.

[0120] Furthermore, we can set equation (26) equal to 0 to obtain the utility function: The maximum point of If we obtain p min ≤p0≤p max , this point is the optimal price p * ; If p0≥p max ,because Therefore, p * =pmax ; Similarly, if p0≤p min , then p * =p max .

[0121] In summary, the optimal price for the master server is:

[0122]

[0123] Step 5: Unloading tasks using the optimal task offloading ratio of the local edge server and the optimal unit price of computing services provided by the remote edge server.

[0124] Based on the above game model and solution method, when the local ES obtains a computing task, it first sends a task offloading request to different remote ES. Then, different remote ESs respectively form the Stackelberg game with the local ES. After obtaining the game results of different remote ESs, the local ES selects the remote ES with the best utility function to offload the task. Since the relationship between the parameters p and λ of the two game parties is segmented, a decreasing sequence is used to determine the initial p value, and multiple rounds of games are performed to obtain the optimal task offloading solution. The specific process is as follows: Figure 2 As shown, finally, the task is unloaded using the optimal task unloading solution obtained.

[0125] In order to verify the effectiveness of the method of this embodiment, the improved IEEE 33-node distribution network system is used as the test system for verification. Assume that the computing tasks of all substations in the distribution network system are completed by four edge servers ES1-ES4. Figure 1 The simulation is implemented by Matlab, and the edge server parameters are shown in Table 1.

[0126] Table 1 Edge server parameters

[0127]

[0128] Distribution network computing tasks can generally be divided into three categories according to real-time requirements. The first category is tasks with high real-time requirements, such as distribution network topology analysis, power grid dynamic stability calculation, fault monitoring, etc. The data and calculation amount of this type of task are relatively small. The second category is conventional status monitoring tasks, such as substation electrical data collection, data processing, status monitoring, etc. Although the single-point calculation amount of this type of task is small, the scale is large. The data and calculation amount of this type of task is medium, and the real-time requirements are also general. The third type of task is some dispatching business, such as intraday dispatching, load forecasting, new energy grid dispatching, etc. This type of task has the lowest real-time requirements, but its calculation load is also the largest. The types of distribution network computing tasks and parameter settings are shown in Table 2.

[0129] Table 2 Calculation task parameters

[0130]

[0131]

[0132] Taking edge server ES0 as the local ES, 100 different types of tasks are randomly generated according to the type of distribution network computing tasks. Among them, there are 40 tasks with high real-time performance, 40 tasks with medium real-time performance, and 20 tasks with low real-time performance. The optimal task offloading strategy is as follows: Figure 4 shown.

[0133] Table 3 Optimal game results under different servers

[0134]

[0135] Depend on Figure 4 As shown in Table 3, different remote ESs can effectively improve the execution efficiency of tasks under the edge-edge collaboration based on Stackelberg game. Compared with the computation without edge-edge collaboration, the utility value of the local ES under the optimal result is increased by 75.1%. At the same time, the average execution time of the task is reduced by 32.7%. Compared with the task offloading of a single ES, the optimal utility value of ES0 under the multi-ES game is further improved. In addition, Figure 4 As can be seen from (c), ES0 rarely chooses ES1 for task offloading. In this embodiment, although ES1 has stronger computing power, it is not the best choice for ES0 for different types of power services, especially when executing services with lower real-time requirements (tasks 41 to 100), its cost-effectiveness is lower. The average task execution time under the optimal result is slightly slower than the execution time under single ES task offloading, but it has been greatly improved compared to edge-to-edge collaborative computing.

[0136] We further compared the key parameters in the game model: the impact of the computational delay conversion coefficient γ on the edge-to-edge collaborative computing results. Figure 5 .Depend on Figure 5 It can be seen that under different γ, the results of edge-to-edge coordination only affect the economic aspects of both parties in the game. From equations (17), (22), (26), etc., it can be seen that the utility functions of both the master and the slave and the optimal unloading unit price p have a complex functional relationship with γ, but within the range of the parameter γ, the results are almost linear with the change of γ. However, under different changes in γ, the performance indicators such as the unloading ratio λ and the task execution time are almost the same, which also verifies the effectiveness of the proposed game method from another perspective.

[0137] In order to compare the performance of task offloading under different methods, two other common strategies are used for task offloading:

[0138] (1) Random task offloading (RTO): Randomly select the target ES and randomly set the offloading ratio to offload tasks. The offloading unit price is based on the historical average price.

[0139] (2) Greedy task offloading (GTO): The offloading ratio is set in a decreasing sequence, and the target ES is selected based on the minimum task execution time. The offloading unit price is based on the historical average price.

[0140] The uninstall results under different strategies are as follows Figure 6 shown.

[0141] Table 4 Comparison of uninstallation results under different strategies

[0142] server Main utility function (￥) From the utility function (￥) Execution time(s) STO 2.556 3.556 0.111 GTO 3.012 2.618 0.101 RTO 5.202 2.394 0.138 Local execution 0 2.031 0.165

[0143] Depend on Figure 6 As shown in Table 4, in terms of task execution utility value, the proposed method is 35.8% and 48.4% higher than the GTO and RTO methods, and 75.2% higher than local execution alone. In terms of task execution latency, the STO method (i.e., the method of this embodiment) is 19.8% and 32.9% higher than the RTO algorithm and local execution, respectively, and is only 9.3% lower than the GTO algorithm, which is primarily aimed at improving execution time. The results show that the proposed STO method has better overall performance.

[0144] Furthermore, the performance of the proposed method in executing different types of tasks is also different. Figure 6 It can be seen that compared with the other two algorithms, the utility value of local ES is improved more under the proposed algorithm in terms of high and medium real-time tasks, which account for most of the distribution network computing tasks, and the advantage is more obvious. RTO has a greater performance improvement in the execution of low real-time tasks. The results show that the proposed algorithm has a wider range of applications.

[0145] The above are all based on the same edge server ES0 requesting task offloading. In reality, different edge servers in the distribution network should generate computing tasks. To this end, based on the above 100 tasks, randomly use them as local ES and perform edge-edge collaborative computing through the proposed task offloading method. The results are as follows: Figure 7 shown.

[0146] Depend on Figure 7It can be seen that under the edge-to-edge collaboration of multiple edge servers, the computing tasks generated by the entire distribution network system can be executed more efficiently. In terms of task calculation time, the average latency of all tasks is reduced by 26.2% compared to local execution alone. At the same time, since each edge server collaboratively executes computing tasks in a master-slave game mode, the average payment cost of task offloading is greatly reduced. This also reflects the principle of "common but differentiated responsibilities" of edge-to-edge collaborative computing between edge servers in the distribution network.

[0147] A second embodiment of the present invention relates to a distribution network edge-to-edge collaborative task unloading device, comprising:

[0148] A first determination module is used to determine a utility function of a remote edge server based on the revenue obtained by providing computing services and data storage costs of the remote edge server;

[0149] A second determination module is used to determine the utility function of the local edge server based on the revenue obtained by the local edge server from performing task computation, the cost of paying for the edge-to-edge collaborative computing service, and the cost of data storage;

[0150] A construction module, configured to construct a Stackelberg game model with the goal of maximizing the utility function of the remote edge server and the utility function of the local edge server;

[0151] A solution module, used to solve the Stackelberg game model by using a reverse induction method to obtain an optimal task offloading ratio of a local edge server and an optimal unit price of computing services provided by a remote edge server;

[0152] The task offloading module is used to offload tasks by using the optimal task offloading ratio of the local edge server and the optimal unit price of computing services provided by the remote edge server.

[0153] The utility function of the remote edge server determined by the first determination module is expressed as: in, is the utility function of the remote edge server, p is the unit price of computing services provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete the task, The cache cost that is used by the task during the time it is queued on the remote edge server.

[0154] The cache cost occupied by the task during the queuing period on the remote edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the remote edge server.

[0155] The local edge server utility function determined by the second determination module is expressed as: in, is the utility function of the local edge server, γ is the conversion coefficient from task computing delay to economic benefit, t total,i is the execution time of the local edge server without edge collaboration, is the execution time of the task in the local edge server, is the execution time of the task on the remote edge server, is the cache cost occupied by the task during the queuing period of the local edge server, p is the unit price of the computing service provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete a task.

[0156] The cache cost occupied by the task during the queuing period of the local edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the local edge server.

[0157] The execution time of the local edge server without edge collaboration is expressed as: The execution time of the task on the local edge server is expressed as: The execution time of the task on the remote edge server is expressed as: in, is the queuing delay of the task in the local edge server, expressed as: is the queued task volume of the local edge server, The computing power of the local edge server; is the computational delay of the task in the local edge server, expressed as: is the computational delay of the task in the remote edge server, expressed as: in, The computing power of the remote edge server; is the queuing delay of the task in the remote edge server, expressed as: is the queued task volume of the remote edge server; is the communication delay of task offloading to the remote edge server, expressed as: R n is the transmission rate offloaded to the remote edge server, expressed as: R n =w·log2(1+P n·h n / (N0·w)), w is the signal transmission bandwidth, P n is the side-to-side signal transmission power, h n is the channel gain, and N0 is the Gaussian white noise spectral density.

[0158] The solution module includes:

[0159] The task optimal offloading ratio determination unit is used to determine the task optimal offloading ratio of the local edge server according to the price of computing services provided by the remote edge server under the condition that the task offloading ratio is fixed;

[0160] The optimal unit price determination unit is used to obtain the optimal unit price of the computing service provided by the remote edge server according to the optimal task offloading ratio of the local edge server.

[0161] It is not difficult to find that the present invention comprehensively considers the execution time and execution cost of the task, establishes a Stackelberg game model for task offloading between different edge servers in the distribution network, and gives a calculation method for the optimal game result through the proof of the existence of Nash equilibrium. Finally, according to the maximization of the interests of the local edge server, the target edge server for offloading and the task offloading ratio are selected to realize the edge-to-edge collaborative computing of the distribution network computing tasks. This method improves the local computing efficiency while increasing the benefits of other servers. Multiple parties share the solution to the problem of offloading computing tasks in the distribution network and improve the processing efficiency of distribution network tasks.

[0162] A third embodiment of the present invention relates to an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned distribution network edge-to-edge collaborative task unloading method when executing the computer program.

[0163] A fourth embodiment of the present invention relates to a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the above-mentioned distribution network edge-to-edge collaborative task offloading method are implemented.

[0164] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0165] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction method, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0168] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A distribution network edge-to-edge collaborative task unloading method, characterized in that: The following steps are involved: Determine the utility function of the remote edge server based on the revenue obtained from providing computing services and the data storage cost of the remote edge server; Determine the utility function of the local edge server based on the revenue obtained by performing task computation on the local edge server, the cost of paying the edge-to-edge collaborative computing service, and the cost of data storage; Constructing a Stackelberg game model with the goal of maximizing the utility function of the remote edge server and the utility function of the local edge server; The Stackelberg game model is solved by using the reverse induction method to obtain the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server; The task offloading is performed using the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server.

2. The distribution network edge-to-edge collaborative task unloading method according to claim 1, characterized in that: The remote edge server utility function is expressed as: in, is the utility function of the remote edge server, p is the unit price of computing services provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete the task, The cache cost that is used by the task during the time it is queued on the remote edge server.

3. The distribution network edge-to-edge collaborative task unloading method according to claim 2, characterized in that: The cache cost occupied by the task during the queuing period on the remote edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the remote edge server.

4. The distribution network edge-to-edge collaborative task unloading method according to claim 1, characterized in that: The local edge server utility function is expressed as: in, is the utility function of the local edge server, γ is the conversion coefficient from task computing delay to economic benefit, t total,i is the execution time of the local edge server without edge collaboration, is the execution time of the task in the local edge server, is the execution time of the task on the remote edge server, is the cache cost occupied by the task during the queuing period of the local edge server, p is the unit price of the computing service provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete a task.

5. The distribution network edge-to-edge collaborative task unloading method according to claim 4, characterized in that: The cache cost occupied by the task during the queuing period of the local edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the local edge server.

6. The distribution network edge-to-edge collaborative task unloading method according to claim 4, characterized in that: The execution time of the local edge server without edge collaboration is expressed as: The execution time of the task on the local edge server is expressed as: The execution time of the task on the remote edge server is expressed as: in, is the queuing delay of the task in the local edge server, expressed as: is the queued task volume of the local edge server, The computing power of the local edge server; is the computational delay of the task in the local edge server, expressed as: is the computational delay of the task in the remote edge server, expressed as: in, The computing power of the remote edge server; is the queuing delay of the task in the remote edge server, expressed as: is the queued task volume of the remote edge server; is the communication delay of task offloading to the remote edge server, expressed as: R n is the transmission rate offloaded to the remote edge server, expressed as: R n =w·log2(1+P n ·h n / (N0·w)), w is the signal transmission bandwidth, P n is the side-to-side signal transmission power, h n is the channel gain, and N0 is the Gaussian white noise spectral density.

7. The distribution network edge-to-edge collaborative task unloading method according to claim 1, characterized in that: The reverse induction method is used to solve the Stackelberg game model to obtain the optimal task offloading ratio of the local edge server and the optimal unit price of the computing service provided by the remote edge server. Specifically, under the condition that the task offloading ratio is fixed, the optimal task offloading ratio of the local edge server is first determined according to the price of the computing service provided by the remote edge server, and then the optimal unit price of the computing service provided by the remote edge server is obtained according to the optimal task offloading ratio of the local edge server.

8. A distribution network edge-to-edge collaborative task unloading device, characterized in that: include: A first determination module is used to determine a utility function of a remote edge server based on the revenue obtained by providing computing services and data storage costs of the remote edge server; A second determination module is used to determine the utility function of the local edge server based on the revenue obtained by the local edge server from performing task computation, the cost of paying for the edge-to-edge collaborative computing service, and the cost of data storage; A construction module, configured to construct a Stackelberg game model with the goal of maximizing the utility function of the remote edge server and the utility function of the local edge server; A solution module, used to solve the Stackelberg game model by using a reverse induction method to obtain an optimal task offloading ratio of a local edge server and an optimal unit price of computing services provided by a remote edge server; The task offloading module is used to offload tasks by adopting the optimal task offloading ratio of the local edge server and the optimal unit price of computing services provided by the remote edge server.

9. The distribution network edge-to-edge collaborative task unloading device according to claim 8, characterized in that: The utility function of the remote edge server determined by the first determination module is expressed as: in, is the utility function of the remote edge server, p is the unit price of computing services provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete the task, The cache cost that is used by the task during the time it is queued on the remote edge server.

10. The distribution network edge-to-edge collaborative task unloading device according to claim 9, characterized in that: The cache cost occupied by the task during the queuing period on the remote edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the remote edge server.

11. The distribution network edge-to-edge collaborative task unloading device according to claim 8, characterized in that: The local edge server utility function determined by the second determination module is expressed as: in, is the utility function of the local edge server, γ is the conversion coefficient from task computing delay to economic benefit, t total,i is the execution time of the local edge server without edge collaboration, is the execution time of the task in the local edge server, is the execution time of the task on the remote edge server, is the cache cost occupied by the task during the queuing period of the local edge server, p is the unit price of the computing service provided by the remote edge server, λ is the offloading ratio of the local edge server, G i The number of CPU cycles required to complete a task.

12. The distribution network edge-to-edge collaborative task unloading device according to claim 11, characterized in that: The cache cost occupied by the task during the queuing period of the local edge server is expressed as: Among them, D i is the data volume of the task, The unit price for the edge server to provide cache for task data, The computing power of the local edge server.

13. The distribution network edge-to-edge collaborative task unloading device according to claim 11, characterized in that: The execution time of the local edge server without edge collaboration is expressed as: The execution time of the task on the local edge server is expressed as: The execution time of the task on the remote edge server is expressed as: in, is the queuing delay of the task in the local edge server, expressed as: is the queued task volume of the local edge server, The computing power of the local edge server; is the computational delay of the task in the local edge server, expressed as: is the computational delay of the task in the remote edge server, expressed as: in, The computing power of the remote edge server; is the queuing delay of the task in the remote edge server, expressed as: is the queued task volume of the remote edge server; is the communication delay of task offloading to the remote edge server, expressed as: R n is the transmission rate offloaded to the remote edge server, expressed as: R n =w·log2(1+P n ·h n / (N0·w)), w is the signal transmission bandwidth, P n is the side-to-side signal transmission power, h n is the channel gain, and N0 is the Gaussian white noise spectral density.

14. The distribution network edge-to-edge collaborative task unloading device according to claim 8, characterized in that: The solution module includes: The task optimal offloading ratio determination unit is used to determine the task optimal offloading ratio of the local edge server according to the price of computing services provided by the remote edge server under the condition that the task offloading ratio is fixed; The optimal unit price determination unit is used to obtain the optimal unit price of the computing service provided by the remote edge server according to the optimal task offloading ratio of the local edge server.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the distribution network edge-to-edge collaborative task unloading method as described in any one of claims 1-7 are implemented.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the distribution network edge-to-edge collaborative task unloading method as described in any one of claims 1-7 are implemented.

Images