CN115361392A - Control method, system and storage medium of blockchain-based computing power network - Google Patents

Abstract

The application relates to a control method, a system and a computer readable storage medium of a block chain-based computational power network. The control method of the computational power network based on the block chain models the interactive relation between each computational power provider in the computational power network into an optimization problem which takes a revenue function with resource utilization rate as a parameter as an optimization target and takes operation cost and task waiting and allocating time as constraint conditions, and the optimization problem is optimized and solved to obtain the optimal matching between a required task and the computational power resources, so that the maximization target of the computational power resource utilization rate in the computational power network is realized.

Description

Control method, system and storage medium of blockchain-based computing power network

technical field

The present application relates to the technical field of blockchain-based computing power network, and more specifically, to a control method, system and computer-readable storage medium for a blockchain-based computing power network.

Background technique

As a new type of network architecture, computing power network, as shown in Figure 1, its core idea is to connect distributed computing power with the network through the joint participation of multiple computing power providers and multiple computing power demanders, and integrate the ubiquitous and dynamic Distributed computing, network, storage and other computing resources, and intelligently schedule and optimally allocate them. At present, the computing power network mainly includes two modes: cloud-edge-device collaboration and computing-network integration. With the gradual transformation of services, data, and content from centralized to coordinated distribution, the ubiquitous computing integrating computing and network has also put forward new requirements for the development of computing and network integration technology. Therefore, in order to break the status quo of isolated islands of computing power resources, balance the distribution of computing power resources, and leverage the advantages of distributed computing power resource pooling, research on computing power resource scheduling among multiple computing power providers under the computing power network architecture has become fundamental and cutting-edge. key scientific questions.

The sharing and allocation of distributed computing power resources in the computing power network is essentially the negotiation and cooperation among distributed nodes, but the lack of trust among multiple computing power providers hinders the sharing of resources. Blockchain is a multi-party co-construction, sharing and co-management technology. Its consensus mechanism is used to ensure that nodes reach a consensus on the validity and consistency of data and information in a distributed system, so it has been widely used in computing power networks. . In a computing power network that has reached a trust consensus, through optimal scheduling and precise allocation of computing power resources, competition for shared computing power resources can be reduced and the overall performance of the computing power network can be improved. Therefore, how to realize the bilateral matching of multi-demand tasks and multi-computing resources in the computing power network is of great significance for realizing the distributed consensus of computing power resources and improving the overall utilization of computing power resources in the computing power network.

Contents of the invention

The technical problem to be solved in this application is to provide a blockchain-based computing power network control method that can realize bilateral matching of demand tasks-computing power resources to maximize the utilization of computing power resources in view of the above-mentioned defects of the existing technology , systems and computer readable storage media.

In order to solve its technical problems, the present application proposes a blockchain-based computing power network control method in the first aspect. The method includes: modeling the interaction relationship between computing power providers in the computing power network as follows: It is an optimization problem that takes the resource utilization rate as a parameter, the profit function as the optimization objective, and the operating cost and task waiting time as the constraints. The optimization problem is solved to obtain the optimal balance between the required tasks and the computing power resources. Matching, where the overall revenue function of each computing power provider in the computing power network is expressed as:

The constraints are:

Subject to Time _i ≥ time _i , i∈(1,n),

为运营成本涉及的电费，

为运营成本涉及的环境处理费，Cost_j为常数，指算力提供方Service_j,j∈{1,2,3,…}的成本预算，n为算力需求方的待分配任务个数，Time_i为算力需求方能接受的最长等待被分配时间，time_i为算力需求方的实际等待时间。Among them, Reward _j is the total income of Service _j , j ∈ {1,2,3,…} of the computing power provider, and QoE is the satisfaction measure index of the computing power demander.

Electricity charges involved in operating costs,

is the environmental processing fee involved in the operating cost, Cost _j is a constant, and refers to the cost budget of the computing power provider Service _j , j∈{1,2,3,…}, n is the number of tasks to be allocated by the computing power demander, Time _i is the longest waiting time that the computing power demander can accept to be allocated, and time _i is the actual waiting time of the computing power demander.

According to an embodiment of the method described in the first aspect of the present application, the satisfaction measure index QoE of the computing power demander is calculated by the following formula:

Among them, s represents the number of services provided by the computing power demand side.

According to an embodiment of the control method for a blockchain-based computing power network described in the first aspect of the present application, the total revenue Reward of the computing power provider Service _j , j∈{1,2,3,...} _j is represented as:

Among them, Rev _i is the computing power resource required by the computing power demander, TR _j is the sum of resources of the computing power provider Service _j , j∈{1,2,3,…}, Com _i ={1,0} means The completion of the task.

According to an embodiment of the method for controlling a blockchain-based computing power network described in the first aspect of the present application, the method further includes: based on the contribution value of each computing power provider participating in the consensus process in the consensus process to Each computing power provider distributes the income of the corresponding weight. Among them, the contribution value of a computing power provider Service _j , j∈{1,2,3,…} to a task i is:

Among them, Rev _i is the computing power resource required by the computing power demander, and Resource _j is the available resources of the computing power provider Service _j , j∈{1,2,3,…} current cooperative members.

According to an embodiment of the method for controlling a computing power network based on blockchain described in the first aspect of the present application, the method further includes: providing service j to the computing power provider Service _j ,j∈{1, 2,3,…}, and use the average contribution value of the task as the basis for dynamically updating the credit value of the computing power provider, where the average contribution value of the task is calculated as follows:

According to an embodiment of the method for controlling a blockchain-based computing power network described in the first aspect of the present application, the method further includes: dynamically updating each computing power according to the participation of each computing power provider in the computing power network in the consensus process Based on the real-time credit value of the provider, each computing power provider is divided into the main control layer, the coordination layer and the peripheral layer based on the level of the credit value, and a new master node is selected from the main control layer. The coordination layer is composed of nodes dynamically selected and marked by the main control layer based on the credit value, and the peripheral layer is composed of nodes other than the main control layer and the coordination layer.

According to an embodiment of the control method of the blockchain-based computing power network described in the first aspect of the application, the credit value of node l (l≥1) in the k-th (k≥1) round of consensus process is defined as Reputation _{l, k} , the multiple weighting formula of credit value Reputation _{l, k} is:

Reputation _l,k ＝αA _l,k +βB _l,k +γC _l,k +ηD _l,k +μE _l,k +Reputation _l,0 ,

Among them, A _l represents the computing power, B _l represents the memory capacity, C _l represents the bandwidth level, D _l represents the online stability, E _l represents the trust degree of interactive scoring, Reputation _{l, 0} represents the initial credit value, α, β, γ, η , μ represent the weights of these five dimensions respectively.

In order to solve its technical problems, the present application proposes a block chain-based control system of computing power network in the second aspect, the system includes a processor and a memory, the memory stores a computer program, and the computer program is controlled by the processor During execution, the control method of the computing power network based on the block chain as described above is realized.

In order to solve its technical problems, the present application proposes a computer-readable storage medium in the third aspect, which stores a computer program, and when the computer program is executed by a processor, the control method of the computing power network based on blockchain as described above is realized. .

Implementing the control method, system and computer-readable storage medium of the blockchain-based computing power network of the present application has the following beneficial effects:

(1) This application is based on the cooperative game theory, and considers the available computing power in the computing power network, the computing power required for the required task, and the cost of task completion and other related factors to study the matching problem between the required task and the computing power resource. A coalition formation algorithm based on cooperative game is proposed. By analyzing, solving and optimizing the target revenue function under the equilibrium strategy, the goal of maximizing the utilization of computing power resources in the computing power network is realized.

(2) This application further addresses the distributed cooperation and resource sharing requirements among computing power providers, and proposes a hierarchical distributed credit consensus mechanism, which improves the efficiency of distributed consensus in the computing power network and reduces the waste of computing power resources.

Description of drawings

The application will be further described below in conjunction with the accompanying drawings and embodiments, in the accompanying drawings:

Figure 1 is a schematic diagram of a computing power network architecture in the prior art;

Fig. 2 is a logical structure diagram of a control system of a blockchain-based computing power network according to an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not intended to limit the present application. Moreover, the embodiments in the present application and the features in the embodiments can be combined with each other under the condition of no conflict.

The blockchain-based computing power network solves the problem of reaching a consensus among computing power providers. Under the guarantee of distributed computing power resource sharing and distribution, in order to maximize the utilization of widely distributed computing power resources in the computing power network, It is necessary to fully consider the bilateral supply and demand relationship between demand tasks and computing power resources. And because different tasks have different requirements, such as low-latency response tasks or high computing resource tasks, and the resources of a single computing power provider are limited, it cannot cope with multiple intensive service requests at the same time, so the computing power provider still Cooperation with other computing power providers is required to achieve resource sharing and collaborative work. Therefore, based on the cooperative game theory, this application proposes a blockchain-based computing power network control method, modeling the interaction relationship between computing power providers in the computing power network as a model with resource utilization as a parameter The profit function is an optimization problem with the optimization objective and the constraint conditions of operating cost and task waiting time for allocation. The optimization problem is solved to obtain the optimal match between the required tasks and the computing power resources.

Specifically, for resource scheduling and computing power sharing in the computing power network, each computing power provider needs to provide idle computing power to work together to complete required tasks and maximize the overall computing power network performance. That is, each computing power provider has the same objective function and forms a cooperative relationship with each other. Therefore, the control method of the block chain-based computing power network in this application chooses a cooperative game for each computing power provider. Model the relationship between them. The specific definition of the cooperative game is as follows:

1) Participants: various computing power providers;

2) Strategy: whether to join a new alliance or whether to leave the current alliance;

3) Profit function: a function that takes the utilization rate of the overall computing resource as a parameter.

和环境处理费

等，于是各方的收益情况取决于运营成本和自身算力资源的使用情况。At the same time, in order to carry out the bilateral matching of tasks and resources, the satisfaction of the computing power demander with regard to the completion of the task must be considered. The control method of the blockchain-based computing power network of this application formalizes the satisfaction of the computing power demander The amount of time spent waiting for a task to be assigned. And because each computing power provider Service _j ,j∈{1,2,3,…} needs to consider its own construction cost and operating cost, where the construction cost is constant, and the operating cost involves electricity

and environmental treatment fees

etc., so the income of all parties depends on the operating costs and the use of their own computing power resources.

At present, a computing power demander has n tasks to be allocated. The longest waiting time for Task _i , i∈(1,n) to be assigned is Time _i , and the actual waiting time is time _i . The required computing power The resource is Rev _i , the sum of the resources of the computing power provider Service _j , j∈{1,2,3,…} is TR _j , according to the completion of the task Com _i ={1,0}, the computing power provider’s The total revenue Reward _j can be expressed as:

In addition, the available resource information of each computing power provider is independent of each other, and only when a computing power provider decides to cooperate with other computing power providers, the information will be shared with each other. The number X _j of tasks currently undertaken by each computing power provider is public information. Therefore, the computing power provider Service _j , j∈{1,2,3,…} can infer other computing power based on the Bayesian formula based on the available resource Resource _j of the current cooperative members and the task commitment of each computing power provider. The probability of the resource availability of the force provider, namely:

Among them, X={X ₁ ,X ₂ ,X ₃ ,…} represents the set of the number of tasks undertaken by all computing power providers, and the probability inference of the above formula (1) is also believed in game theory. P(X|Resource _j ) and P(Resource _j ) can be directly calculated based on the data shared after the cooperation. In order to avoid underflow caused by the possible small data set, the above formula (1) is changed to the following formula :

Finally, the overall revenue function of each computing power provider in the computing power network can be expressed as:

The constraints are:

Subject to Time _i ≥ time _i , i∈(1,n) (4)

Among them, Cost _j is a constant, which refers to the cost budget of Service _j , j ∈ {1,2,3,…} of the computing power provider, n is the number of tasks to be assigned by the computing power demander, and QoE (Quality of Experience) is Satisfaction indicator of computing power demand side. Specifically, QoE is calculated by the following formula:

Among them, s represents the number of services provided by the computing power demand side.

The above optimization objective formula (3) requires the resource utilization rate of the computing power provider and the satisfaction degree of the computing power demander for the completion of the task to be the largest, and also requires the cost of the computing power provider to be minimized. Constraint (4) requires that the computing power demander's waiting time for each task cannot exceed the corresponding waiting upper bound, and constraint (5) requires that the computing power provider's electricity and environmental costs cannot exceed the cost budget Cost _j . Combined with the optimization problem composed of the optimization objective formula (3) and the constraint condition formula (4)-(5), it is theoretically proved that the cooperative alliance formation algorithm can converge to a stable equilibrium point, and reach a suboptimal solution with a faster convergence speed . According to the computing power demand of the demand task, the idle computing power of the computing power provider, the waiting delay, etc., after the optimization goal finds the equilibrium point, it is considered that the task and the demand are the best match, and the corresponding computing power provider is used to complete The corresponding demand tasks can achieve the goal of maximizing the utilization of computing power resources in the computing power network.

Further, the blockchain-based computing power network control method of the present application also distributes corresponding weighted income to each computing power provider based on the contribution value of each computing power provider participating in the consensus process in the consensus process. In order to measure the contribution of a computing power provider Service _j , j∈{1,2,3,…} to a task i, this application proposes the following contribution measurement indicators to complete the final revenue distribution:

为算力提供方Service_j,j∈{1,2,3,…}对于一个任务i的贡献值，Rev_i为算力需求方所需要的算力资源，Resource_j为算力提供方Service_j,j∈{1,2,3,…}当前合作成员的可用资源。in,

is the contribution value of the computing power provider Service _j , j∈{1,2,3,…} to a task i, Rev _i is the computing power resource required by the computing power demander, and Resource _j is the computing power provider Service _j , j ∈ {1,2,3,…} the available resources of the current cooperative member.

Further, the control method of the blockchain-based computing power network of this application also calculates the average contribution value of the task of the computing power provider Service _j , j∈{1,2,3,...} at a certain period T as follows :

The average contribution value of this task will be used as a basis for dynamically updating the credit value of computing power providers to encourage each computing power provider to actively participate in the collaborative allocation of computing power resources within a certain period.

Further, according to the control method of the blockchain-based computing power network according to the above-mentioned embodiments of the present application, in order to meet the needs of distributed computing power providers in the computing power network to cooperate and share computing power resources, valid blocks are generated based on consensus nodes and uploaded to the chain. A hierarchical distributed credit consensus model is proposed to promote trust cooperation and resource sharing among computing power providers.

Specifically, the blockchain-based computing power network control method of this application dynamically updates the real-time credit value of each computing power provider according to the participation of each computing power provider in the computing power network in the consensus process. The computing power provider is divided into three layers, namely the master control layer, the coordination layer and the peripheral layer. Among them, the master control layer is composed of trusted high-performance computing nodes arranged in advance, and new master nodes are selected from the master control layer first to reduce the consumption of selection. The coordination layer is composed of nodes dynamically selected and marked by the main control layer based on the credit value. The peripheral layer is composed of nodes other than the main control layer and the coordination layer. Due to the low trust value, these nodes do not participate in the consensus process, but they can By increasing the trust value, it is marked into the coordination layer by the master control layer.

Each incoming new computing power provider will get a unique ID and initial credit value, and the credit value will be accumulated and counted along with the consensus behavior of the computing power provider and other computing power providers. The return of positive credit can motivate computing power providers to participate in the consensus behavior to a higher degree, while negative credit can reduce the weight of the member in the consensus. Define the credit value of node l (l ≥ 1) in the k (k ≥ 1) round of consensus process as Reputation _{l, k} , then the multiple weighted formula for calculating credit value Reputation _{l, k} is:

Reputation _l,k ＝αA _l,k +βB _l,k +γC _l,k +ηD _l,k +μE _l,k +Reputation _l,0 , (9)

Among them, A _l represents the computing power, B _l represents the memory capacity, C _l represents the bandwidth level, D _l represents the online stability, E _l represents the trust degree of interactive scoring, Reputation _{l, 0} represents the initial credit value, α, β, γ, η , μ represent the weights of these five dimensions respectively. The first four items of the expression reflect the objective processing capability of the node, and the fifth item reflects the forwarding and communication completed during the distributed interaction process of the node. According to the level of the credit value calculated by the above formula (9), the application can divide the nodes into three layers: the main control layer, the coordination layer, and the peripheral layer.

As the number of computing power providers increases, the interactive information will increase exponentially, the performance of the consensus algorithm will decline, and the scalability will be limited to a certain extent, resulting in a large waste of resources. Therefore, the blockchain-based computing power network control method of this application further considers using an appropriate parallel distributed architecture to improve the algorithm, so as to propose a more scalable and lightweight consensus algorithm. For the computing power network based on the blockchain, the same view unit task can be calculated in parallel through multiple master nodes. When one of the master nodes first completes a complete consensus process, the master control layer will notify other master nodes to stop the corresponding process. To reduce the additional consensus cost.

Based on the control method of the blockchain-based computing power network described above in this application, this application also proposes a control system 10 for a blockchain-based computing power network. As shown in FIG. 2 , the control system 10 of the blockchain-based computing power network includes a processor 11 and a memory 12 , and the processor 11 and the memory 12 are connected in communication. The memory 12 stores a computer program, and when the computer program is executed by the processor 11, the processor 11 realizes the control method of the computing power network based on blockchain in the foregoing embodiments of the present application.

The present application also proposes a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the method for controlling a computing power network based on blockchain in the foregoing embodiments of the present application is implemented.

The above descriptions are only preferred embodiments of the application, and are not intended to limit the application. Any modifications, equivalent replacements and improvements made within the spirit and principles of the application should be included in the protection of the application. within range.

Claims (9)

1. A method for controlling a blockchain-based computing power network, characterized in that the method includes: modeling the interaction between computing power providers in the computing power network as taking resource utilization as a parameter The revenue function of is an optimization problem with the optimization objective and the constraint conditions of operating cost and task waiting time for allocation. The optimization problem is solved to obtain the optimal match between the required tasks and the computing power resources. Among them, the The overall revenue function of each computing power provider in the computing power network is expressed as:

The constraints are: Subject to Tim e _i ≥tim e _i ,i∈(1,n),

Among them, Reward _j is the total income of Service _j , j ∈ {1,2,3,…} of the computing power provider, and QoE is the satisfaction measure index of the computing power demander.

Electricity charges involved in operating costs,

is the environmental processing fee involved in the operating cost, Cost _j is a constant, and refers to the cost budget of the computing power provider Service _j , j∈{1,2,3,…}, n is the number of tasks to be allocated by the computing power demander, Time _i is the longest waiting time that the demander of computing power can accept to be allocated, and time _i is the actual waiting time of the demander of computing power. 2. The method according to claim 1, characterized in that, the satisfaction index QoE of the computing power demand side is calculated by the following formula:

Among them, s represents the number of services provided by the computing power demand side. 3. The method according to claim 1, wherein the total revenue Reward _j of the computing power provider Service _j , j∈{1,2,3,…} is expressed as:

Among them, Rev _i is the computing power resource required by the computing power demander, TR _j is the sum of resources of the computing power provider Service _j , j∈{1,2,3,…}, Com _i ={1,0} means The completion of the task. 4. The method according to claim 1, characterized in that the method further comprises: based on the contribution value of each computing power provider participating in the consensus process in the consensus process, assigning corresponding weighted income to each computing power provider, Among them, the contribution value of a computing power provider Service _j ,j∈{1,2,3,…} to a task i is:

Among them, Rev _i is the computing power resource required by the computing power demander, and Resource _j is the available resources of the computing power provider Service _j , j∈{1,2,3,…} current cooperative members. 5. The method according to claim 4, characterized in that the method further comprises: according to a certain period T, the average contribution value of the tasks of the computing power provider Service _j , j∈{1,2,3,...} Perform calculations, and use the average contribution value of the task as the basis for dynamically updating the credit value of the computing power provider, where the average contribution value of the task is calculated as follows:

6. The method according to claim 1, further comprising: dynamically updating the real-time credit value of each computing power provider according to the participation of each computing power provider in the computing power network in the consensus process, based on the credit value High and low divide each computing power provider into the main control layer, coordination layer and peripheral layer and select a new master node from the main control layer. Among them, the main control layer is composed of trusted high-performance computing nodes, and the coordination layer is composed of the main control layer The nodes are dynamically selected and marked based on the credit value, and the peripheral layer is composed of nodes other than the control layer and the coordination layer. 7. The method according to claim 6, characterized in that, defining the credit value of node l (l≥1) in the k (k≥1) round consensus process is Reputation _l,k , then the credit value Reputation _l, The polynomial weighting formula for _k is: Reputation _l,k ＝αA _l,k +βB _l,k +γC _l,k +ηD _l,k +μE _l,k +Reputation _l,0 , Among them, A _l represents the computing power, B _l represents the memory capacity, C _l represents the bandwidth level, D _l represents the online stability, E _l represents the trust degree of interactive scoring, Reputation _{l, 0} represents the initial credit value, α, β, γ, η , μ represent the weights of these five dimensions respectively. 8. A control system based on a blockchain computing power network, characterized in that the system includes a processor and a memory, the memory stores a computer program, and when the computer program is executed by the processor, it realizes The control method of the computing power network based on blockchain described in any one of 1-7. 9. A computer-readable storage medium, characterized in that a computer program is stored, and when the computer program is executed by a processor, the blockchain-based computing power network according to any one of claims 1-7 is realized control method.

Images