CN117478697A - Industrial Internet data sharing optimization method based on intelligent slicing decision block chain - Google Patents
Industrial Internet data sharing optimization method based on intelligent slicing decision block chain Download PDFInfo
- Publication number
- CN117478697A CN117478697A CN202311396787.3A CN202311396787A CN117478697A CN 117478697 A CN117478697 A CN 117478697A CN 202311396787 A CN202311396787 A CN 202311396787A CN 117478697 A CN117478697 A CN 117478697A
- Authority
- CN
- China
- Prior art keywords
- blockchain
- transaction
- data sharing
- nodes
- mec
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 86
- 238000005457 optimization Methods 0.000 title claims abstract description 32
- 230000008569 process Effects 0.000 claims abstract description 28
- 230000009471 action Effects 0.000 claims abstract description 22
- 238000004364 calculation method Methods 0.000 claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 17
- 230000002787 reinforcement Effects 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 24
- 239000012634 fragment Substances 0.000 claims description 21
- 238000012795 verification Methods 0.000 claims description 15
- 238000004891 communication Methods 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 11
- 230000008901 benefit Effects 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 9
- 230000035945 sensitivity Effects 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 230000007613 environmental effect Effects 0.000 claims description 4
- 238000013467 fragmentation Methods 0.000 claims description 4
- 238000006062 fragmentation reaction Methods 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 239000003795 chemical substances by application Substances 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 239000000470 constituent Substances 0.000 claims description 2
- 230000001276 controlling effect Effects 0.000 claims description 2
- 230000002596 correlated effect Effects 0.000 claims description 2
- 239000011159 matrix material Substances 0.000 claims description 2
- 230000007704 transition Effects 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 abstract description 3
- 238000004088 simulation Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 8
- 230000000875 corresponding effect Effects 0.000 description 5
- 239000010432 diamond Substances 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000000903 blocking effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010004142 Bartholinitis Diseases 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/092—Reinforcement learning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/104—Peer-to-peer [P2P] networks
- H04L67/1074—Peer-to-peer [P2P] networks for supporting data block transmission mechanisms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses an industrial Internet data sharing optimization method based on an intelligent slicing decision block chain. By adaptively slicing the blockchain system, the method can effectively break through the expandability bottleneck of the blockchain. Specifically, the method adds the block chain segmentation decision factors into the action space of a deep reinforcement learning algorithm, jointly considers the segmentation decision, the unloading decision, the block size and the block interval, introduces a cloud-edge cooperative calculation paradigm, models the optimization problem as a Markov decision process, and adopts the deep reinforcement learning method to solve the problem. Simulation results show that the total delay of industrial Internet data sharing and the throughput of blockchain transactions obtained by the method are superior to those of other methods.
Description
Technical Field
The invention relates to a data sharing optimization method based on an intelligent slicing decision block chain in an industrial Internet (Industrial Internet of Things, IIoT) system. By carrying out self-adaptive slicing decision on the blockchain system, the method can effectively break through the expandability bottleneck of the blockchain. Specifically, the method adds the slicing decision factors into the action space of the deep reinforcement learning algorithm, and performs joint optimization by taking the total data sharing delay and the blockchain transaction throughput as the rewarding function, thereby reducing the delay generated in the data sharing process to the maximum extent and improving the blockchain transaction throughput while guaranteeing the data authenticity, and belonging to the technical field of communication networks.
Background
In recent years, information and communication technology (Information and Communication Technology, ICT) has gradually penetrated the aspects of social life, which has promoted the formation of new forms of everything interconnection. The development of ICT has led to industry 4.0, while IIoT has become one of the most representative applications in industry 4.0. With the development of IIoT, network access of massive IIoT devices can generate huge data volume and huge-scale data sharing task.
IIoT builds its system framework on the basis of traditional industrial facilities, aiming at improving industrial production efficiency by realizing the comprehensive interconnection of industrial equipment. However, open networks and traditional centralized storage, represented by databases, may present serious security and privacy risks to industrial data. Furthermore, the wide distribution of IIoT devices and open data interactions also make it difficult to effectively protect the security and privacy of industrial data.
Blockchain technology is widely recognized as a secure, prospective data storage technology. Blockchains are a new model of distributed computing that uses a distributed architecture to validate and store data. The characteristics of decentralization, traceability and the like of the blockchain can effectively solve the problems of data security and data privacy in IIoT. However, blockchain consensus algorithms consume significant computational resources, such as the failsafe mechanism of bartholinitis (Practical Byzantine Fault Tolerance, PBFT). This tends to cause a huge computational overhead, resulting in computational inefficiency. In addition, although the blockchain enhances the security of data transmission and storage, the conventional blockchain system has a problem that the nodes are overburdened due to its high redundancy storage mechanism. Thus, there are serious scalability limitations for blockchain technology-based IIoT systems.
Fortunately, the slicing technique can overcome the scalability problem faced by blockchains. As a method for expanding the capacity on the chain, the slicing technology can improve the performance of a block chain system under the condition of not influencing the decentralization of the block chain. However, existing research often ignores the dynamics of IIoT scenes when the sliced blockchain is applied to IIoT scenes. Furthermore, existing research lacks active fragmentation strategies, which can lead to low system throughput, excessive delay, and increased security risks. Meanwhile, the blocking chain nodes are distributed into a plurality of independent fragments by the fragment technology, so that the size of each fragment is greatly reduced, and the probability of tampering and attack of data by malicious nodes is greatly increased, thereby causing security threat. In order to effectively solve the problems, the method designs a block chain dynamic fragmentation decision strategy based on the principle of minimum transaction cost. By dynamically adjusting the tiles, the security of the system can be maintained while improving the performance of the system.
Admittedly, the blocking technique of the blockchain can increase the performance of the IIoT system and simultaneously generate huge computing overhead and consume a great amount of computing resources. Furthermore, in the IIoT application scenario, a large number of device nodes may result in inefficiency of the blockchain consensus process. Therefore, the IIoT system based on the sliced blockchain inevitably faces the problems of large computing overhead and limited computing resources.
Mobile edge computing (Mobile Edge Computing, MEC) is an effective method to overcome the problems of low computational power and excessive computational overhead of the sliced blockchain and large-scale IIoT. The MEC server is arranged at an edge layer closer to the IIoT equipment, so that the calculation efficiency of the IIoT system is improved. Meanwhile, by combining MEC with traditional centralized cloud computing, a cloud-edge cooperative solution can be obtained, so that the computing efficiency is further improved. Furthermore, it is natural to combine both blockchain and MEC technologies because of their decentralization characteristics.
In summary, the present invention provides an IIoT data sharing optimization method based on an intelligent slicing decision block chain, which aims at the above challenges and problems. By adding the slicing decision factors into the action space of the deep reinforcement learning algorithm and taking the total delay of data sharing and the transaction throughput of the blockchain as the rewarding function to carry out joint optimization, a cloud edge cooperative computing paradigm is introduced, and the purposes of reducing the delay generated in the data sharing process and improving the transaction throughput of the blockchain to the greatest extent while guaranteeing the authenticity of the data are achieved.
Disclosure of Invention
The invention mainly aims to execute slicing and intelligent slicing decision on a blockchain system under the condition that mass industrial equipment, multi-block chain nodes, multi-block chain slicing, multi-MEC servers and a single cloud computing server exist in a scene in an IIoT system with large-scale characteristics and dynamic characteristics, model the scene by taking the total delay of an IIoT data sharing process and blockchain transaction throughput as optimization targets and apply a depth reinforcement learning algorithm of an asynchronous dominant actor criticizing algorithm (Asynchronous Advantage Actor-Critic, A3C) to perform iterative learning on the model so as to obtain an optimal industrial data sharing strategy with high throughput and low delay. The method solves the problems that under the condition that mass industrial equipment, multi-block chain nodes, multi-block chain fragments, multi-MEC servers and a single cloud computing server exist in the scene, the total delay of an IIoT system is effectively reduced and the transaction throughput of a block chain is improved by executing an optimal industrial data sharing strategy, so that industrial data sharing in the IIoT is safely and rapidly carried out.
The IIoT scene model based on the intelligent slicing decision block chain, which is adapted by the invention, is shown in fig. 1.
The flow chart of the system operation principle in the technical scheme of the invention is shown in fig. 2.
The total delay of the system of the invention is plotted against the number of blockchain slices in FIG. 3.
The blockchain transaction throughput versus blockchain fragment number of the present invention is shown in fig. 4.
The blockchain transaction throughput versus average transaction size of the present invention is shown in fig. 5.
The IIoT scene model with large-scale characteristics and dynamic characteristics, which is adapted by the invention, is shown in figure 1. In the IIoT data sharing optimization method based on intelligent slicing decision block chain, under the communication scene of a large industrial park, N MEC nodes are commonly arranged in each park department to cover massive IIoT equipment, and each department is simultaneously provided with 1 controller for processing the management and control tasks of an edge layer. In addition, there are 1 blockchain system formed by MEC servers and 1 cloud server equipped at headquarters. The blockchain sharding algorithm will divide the N MEC nodes (verifiers in the blockchain) of each department into k blockchain shards on average, and k is ≡2, according to the principle of minimum transaction cost. Thus, the whole blockchain system has k blockchain networks to process the data sharing task of the IIoT device in parallel. And setting a segmentation model, a blockchain consensus model and a calculation model according to the actual environment condition after knowing the number of MEC nodes. And then constructing a state space, an action space and a reward function in the deep reinforcement learning method, setting parameters such as strategy parameters, value function parameters, network layer numbers and the like in a training network, carrying out iterative learning in combination with a scene model, and training a deep neural network for asynchronously updating global parameters. And finally, executing an optimal data sharing strategy learned by the agent through interaction with the environment, thereby effectively reducing the total delay of the system and improving the transaction throughput of the blockchain. The method is realized by the following steps:
the IIoT equipment requests data sharing to generate a calculation task, and uploads the task to an edge layer for processing, and the edge layer intelligently decides whether to perform block chain slicing or not according to the delay sensitivity of the data sharing task; the method comprises the following specific steps:
the IIoT device is responsible for uploading the data sharing task to the MEC server at the edge layer. There are N MEC nodes in the edge layer, n= { η 1 ,η 2 ,.....,η n And } represents a set of MEC nodes, where η represents a MEC node, subscript 1,2, & n is a specific number of MEC nodes. If the sensitivity of the current data sharing task to delay is low, the controller selects not to perform block chain fragmentation when processing the data sharing task so as to ensure the safety of the system to the maximum extent; if the sensitivity of the data sharing task to delay is high, N MEC nodes are evenly distributed into k blockchain slices according to the principle of minimum transaction cost, and k is more than or equal to 2. Set s= { S 1 ,s 2 ,......,s k And (c) represents a set of blockchain slices, wherein s represents a blockchain slice, subscripts 1,2, & gt, k is a specific number of blockchain slices; the blockchain slicing method based on the principle of minimum transaction cost is as follows:
first, the number of MEC nodes of the whole blockchain system is guaranteed to be N:
where i represents the order of MEC nodes and n is the specific number of MEC nodes. j represents the order of the blockchain shards, and k is the specific number of blockchain shards.Representing MEC node eta i Assigning blockchain slices s j Otherwise->
Second, to ensure that all nodes are evenly distributed into the various blockchain slices, the number of nodes in each blockchain slice is M:
finally, the total cost of the data sharing task between every two nodes is calculated as follows:
wherein omega ij Representing MEC node eta i Distribution to blockchain slices s j And:
where L represents the order of MEC nodes other than i, L represents the number of MEC nodes located in the same slice s j Every two MEC nodes eta within i And eta l The set of constituent, -L represents the set of every two MEC nodes located in different slices, and:
wherein N is il Representing every two MEC nodes eta of the whole edge layer i And eta l A set of components, and:
wherein c (i, l) represents node η i And eta l Trade costs between, and:
wherein f i,l And c i,l Node η from different blockchain slices i And eta l Frequency of communication between them and communication overhead. I il Is the cost of communication between two nodes located in the same blockchain shard.
Step (2), the controller is responsible for making a slicing decision on the blockchain system when processing the data sharing task, and the intelligent slicing decision divides the consensus mode (transaction type) of the blockchain into the following three cases:
step (2.1), uploading data tasks to the same MEC node eta by both IIoT equipment with data sharing requirement i Or in the same blockchain slice s j Two MEC nodes eta in (1) i And eta l The transaction type of the transaction is considered an intra-slice transaction. In this case, the transaction is divided by the blockchain shard s j Verifying the M nodes in the network.
Step (2.2), the IIoT equipment with data sharing requirement uploads the data task to the fragments s located in different block chains j Sum s k Is defined by two MEC nodes eta i And eta l The transaction type is a cross-fragment transaction. At this point, the transaction is fragmented s by the blockchain j Sum s k Is verified by the 2M nodes.
In addition, when processing the data sharing task with insensitive delay, the controller may choose not to slice the blockchain system to maximize the security of the data. At this point, the transaction is validated by all N MEC nodes at the edge layer.
Step (2.4), the consensus node of the blockchain system verifies and consensus transaction data generated after the IIoT equipment requests the data sharing task, and when the transaction type is intra-chip transaction, the consensus node number N is the same as the transaction type C M, N when the transaction type is cross-fragment transaction C =2m, transaction type N for non-fragmented transactions C =n. The common-knowledge node adopts a practical Bayesian and busy-family fault-tolerant mechanism to verify and common-knowledge the blocks and the transactions, and when the number of the fault nodes is not more than 1/3 of the number of the common-knowledge nodes, the PBFT can ensure the correctness of the common-knowledge process. CPU period required for generating or verifying a signature or a message verification code is alpha and beta respectively, then one common identification process is completedThe total calculation period required is:
where a (t) is the average transaction size, b (t) is the transaction batch size, ρ is the probability of verifying the transaction correctly at the request stage.
Step (3), the node verification process of the blockchain is provided with calculation support by the MEC server and the cloud server, and specifically comprises the following steps:
step (3.1), the delay of the MEC server providing computational support for the node verification process is:
where λ (t) is the CPU period required for the computing task, F m The calculation frequency for the MEC server.
At this time, the total delay of the system may be expressed by the following equation:
wherein i (t) is the block interval, t b Delays are broadcast for nodes.
Step (3.2), when the cloud server provides computing support for the node verification process, the delay of the verification process is:
where D (t) is the data task size, μ (t) represents the data transfer rate of the edge layer to cloud layer calculation offload wireless link, F c Is the computation frequency of the cloud server.
At this time, the total delay of the system may be expressed by the following equation:
in step (4), transaction throughput is an important indicator for evaluating the performance of the slicing decision block chain system, and is directly affected by the following two parameters. The first is the block size, i.e., the capacity of a block, which determines the transactions that a block contains. The other is the block interval, which represents the release rate of the block. Because the blockchain generates new blocks in a multi-threaded manner, blockchain transaction throughput is positively correlated with the number of tiles k. The blockchain transaction throughput is calculated by:
wherein S is B Representing the block size.
Step (5), according to the steps (1) - (4), setting a state space, an action space and a reward function in the A3C deep reinforcement learning method by combining scenes and optimization targets, wherein the specific steps are as follows:
in step (5.1), in one discrete time slot, the agent first senses the environmental state. Subsequently, learning experience is obtained based on the environmental state, and then the policy is updated. Thus, the state space is set to:
S(t)＝{D(t),b(t),μ(t)}
wherein the data task size D (t) ranges from 4MB to 8MB, and the transaction batch size b (t) ranges from 1MB to 2MB. Due to the diversity and dynamic nature of the computation offload radio link conditions, the data transfer rate is modeled as a finite state Markov decision process, the data transfer rate assigned to each computation offload task comes from the set {125,100,75,50} Mbps, with the probability transition matrix expressed as:
step (5.2), in order to obtain an optimal strategy for solving the above-mentioned scenario problem, dynamic adjustment of several parts in the system is required. The action space is thus set as follows:
A(t)＝{S B ,i(t),o(t),f(t)}
wherein S is B = {1,2,..b } and I (t) = {0.2,0.4,..i }. In addition, o (t) = {0,1}, o (t) = 0 indicates that the calculation support in the blockchain node verification process is provided by the MEC server, and conversely, if o (t) = 1 indicates that the calculation task is offloaded to the cloud layer for processing. f (t) = {0,1}, when f (t) = 1, the controller chooses to slice the blockchain system, otherwise, the controller does not slice.
And (5.3) optimizing the throughput and the total time delay in the IIoT system supported by the slicing decision block chain. According to the optimization objective, establishing an optimization problem as follows:
thus, the reward function is given by:
wherein w is 1 +w 2 ＝1，Is a punishment parameter.
Step (6), setting strategy parameters, value function parameters and network layer numbers according to the state space, action space and rewarding function constructed in the step (5), training the deep neural network, and estimating the advantage function:
A(S(t),A(t)；θ,θ v )＝R t -V(S(t)；θ v )
and is also provided with
Wherein Rt is the accumulated return of discount, V (S (t); θ) v ) Is an estimate of the cost function. k can be changed along with the change of the state and is represented by the local maximum iteration number t max Is the upper limit. Gamma epsilon (0, 1)]R is the discount factor t+i For instant rewards.
The local accumulated gradient update of the guiding policy function parameter θ is estimated from the resulting dominance, expressed as:
▽ θ L π (θ)＝▽ θ logπ(A(t)|S(t)；θ)A(S(t),A(t)；θ,θ v )+▽ θ βH(π(S(t)；θ))
wherein pi (A (t) |S (t); theta) is a random strategy, beta is a super parameter for controlling the intensity of the entropy regularization term, and H (pi (S (t); theta)) is the entropy of the strategy pi.
According to the RMSProp algorithm, the global parameter theta is asynchronously updated, and the random strategy can be gradually adjusted to the optimal strategy. The RMSProp asynchronous update algorithm is expressed as:
wherein eta is the learning rate, g is the gradient estimation value under RMSProp, and epsilon is a small positive number.
And (7) obtaining an optimal strategy of the optional action in each state according to the depth neural network trained in the step (6), and continuously executing the optimal action of each state by taking the action generated by the strategy as the optimal action in the state until the execution instruction is ended.
The invention has the advantages that under the IIoT communication scene with large-scale characteristics and dynamic characteristics, the total delay of the system is reduced to the maximum extent and the transaction throughput of the block chain is improved to the maximum extent by jointly considering the slicing decision, the unloading decision, the block size and the block interval. In addition, the method examines the influence of the intelligent slicing decision block chain-based industrial Internet data sharing optimization method on the total data sharing delay and the block chain transaction throughput in a scene through simulation experiments.
Drawings
Fig. 1 is a schematic diagram of a communication scenario model including IIoT devices, MEC servers, controllers, cloud computing servers, and a blockchain system.
FIG. 2 is a flow chart of an industrial Internet data sharing optimization method design based on an intelligent slicing decision block chain.
FIG. 3 is a graph of total system delay versus number of blockchain slices, with diamonds representing the method of the invention, squares representing fixed block size frames, circles representing fixed block spacing frames, and crosses representing existing fixed blockchain slice frames.
FIG. 4 is a graph of blockchain transaction throughput versus number of tiles, with diamonds representing the method of the present invention, crosses representing fixed block size frames, circles representing fixed block spacing frames, and squares representing existing fixed tile frames.
FIG. 5 is a graph of blockchain transaction throughput versus average transaction size, wherein diamonds represent the method of the present invention, and "X" shapes represent DQN-based deep reinforcement learning method frameworks, crosses represent existing fixed slice frameworks, and circles represent no blockchain slice frameworks.
Detailed Description
The technical scheme of the industrial Internet data sharing optimization method based on the intelligent slicing decision block chain is further described below with reference to the accompanying drawings and examples.
The flow chart of the method of the invention is shown in fig. 2, and comprises the following steps:
step one, setting the number of MEC server nodes and the number of blockchain fragments, and dynamically and intelligently making a fragment decision according to the actual delay sensitivity of a data sharing task;
judging the relationship between the MEC nodes connected with the IIoT equipment with the data sharing requirement, dividing the transaction types, and calculating the delay generated in the node consensus process according to different conditions of the transaction types;
step three, according to the data transmission rate of the wireless communication link from the edge layer to the cloud layer and the transaction batch size, executing calculation unloading decision and calculating delay generated in the node verification process;
calculating the transaction throughput of the blockchain according to the number of the blockchain fragments, the average transaction size, the block size and the block interval;
step five, setting a state space, an action space and a reward function of a deep reinforcement learning algorithm of A3C according to an optimization target;
step six, solving the joint optimization problem according to an A3C algorithm, setting the layer number, strategy parameters and value function parameters of the deep neural network, training the approximate advantage function of the deep neural network, guiding the local accumulated gradient update of the strategy parameters through the advantage estimation, and asynchronously updating the global parameters according to an RMSProp algorithm;
and step seven, selecting an optimal action according to the optimal strategy obtained in each state, and obtaining the maximum benefit.
FIG. 3 is a graph of total system delay versus the number of blockchain slices. As can be seen from fig. 3, the total system delay decreases with the number of slices. The total delay of the system corresponding to the method is always lower than that of other methods. For example, when the number of blockchain slices is 16, the total system delay corresponding to the method of the present invention is only 1.09s, and the total system delay corresponding to the other methods can reach 1.50s at most.
FIG. 4 is a graph of blockchain transaction throughput versus number of slices. As can be seen from fig. 4, the blockchain transaction throughput increases with the number of slices. When the number of fragments is 16, the transaction throughput corresponding to the method of the invention is up to 5925TPS, while the transaction throughput of the other methods can only up to 4798TPS. According to the method, the block chain transaction throughput is related to the number of the fragments, and with the increase of the number of the fragments, the calculation tasks generated in the transaction consensus verification process can be processed in parallel in a multithreading manner, so that the transaction throughput is greatly improved, and the transaction throughput optimized based on the method is always higher than that based on other methods.
FIG. 5 is a graph of blockchain transaction throughput versus average transaction size. As can be seen from fig. 5, when the average transaction size is 100Bytes, the blockchain transaction throughput corresponding to the method of the present invention can reach 2579TPS, and the highest transaction throughput reached by the other methods is only 2455TPS. It can be derived that the blockchain transaction throughput is related to the average transaction size, and when the average transaction size increases, the number of transactions contained in one block becomes smaller, and the transaction throughput is reduced accordingly, but the transaction throughput optimized based on the method of the invention is always higher than that based on other methods.
Claims (8)
1. The industrial Internet data sharing optimization method based on the intelligent slicing decision block chain is characterized by comprising the following steps of: the method comprises the following steps:
step one, setting the number of MEC server nodes and the number of blockchain fragments, and dynamically and intelligently making a fragment decision according to the actual delay sensitivity of a data sharing task;
judging the relationship between the MEC nodes connected with the IIoT equipment with the data sharing requirement, dividing the transaction types, and calculating the delay generated in the node consensus process according to different conditions of the transaction types;
step three, according to the data transmission rate of the wireless communication link from the edge layer to the cloud layer and the transaction batch size, executing calculation unloading decision and calculating delay generated in the node verification process;
calculating transaction throughput of the blockchain according to the blockchain fragment number, the average transaction size, the block size and the block interval;
step five, setting a state space, an action space and a reward function of a deep reinforcement learning algorithm of A3C according to an optimization target;
step six, solving the joint optimization problem according to an A3C algorithm, setting the layer number, strategy parameters and value function parameters of the deep neural network, training the approximate advantage function of the deep neural network, guiding the local accumulated gradient update of the strategy parameters through the advantage estimation, and asynchronously updating the global parameters according to an RMSProp algorithm;
and step seven, selecting an optimal action according to the optimal strategy obtained in each state, and obtaining the maximum benefit.
2. The intelligent slicing decision block-chain-based industrial internet data sharing optimization method of claim 1, wherein: in the first step, IIoT equipment requests data sharing to generate a calculation task, and the task is uploaded to an edge layer for processing, and the edge layer intelligently decides whether to perform block chain slicing or not according to the delay sensitivity of the data sharing task; the method comprises the following specific steps:
the IIoT equipment is responsible for uploading the data sharing task to the MEC server of the edge layer; there are N MEC nodes in the edge layer, n= { η 1 ,η 2 ,.....,η n -represents a set of MEC nodes, wherein η represents a MEC node, subscript 1,2,..n is the specific number of MEC nodes; if the sensitivity of the data sharing task to delay is low, the controller selects not to carry out block chain fragmentation when processing the data sharing task so as to ensure the safety of the system to the maximum extent; if the sensitivity of the data sharing task to delay is high, N MEC nodes are evenly distributed into k blockchain fragments according to the principle of minimum transaction cost, wherein k is more than or equal to 2; set s= { S 1 ,s 2 ,......,s k And (c) represents a set of blockchain slices, wherein s represents a blockchain slice, subscripts 1,2, & gt, k is a specific number of blockchain slices; the blockchain slicing method based on the principle of minimum transaction cost is as follows:
first, the number of MEC nodes of the whole blockchain system is guaranteed to be N:
wherein i represents the order of MEC nodes, and n is the specific number of MEC nodes; j represents the order of the blockchain slices, and k is the specific number of the blockchain slices;representing MEC node eta i Assigning blockchain slices s j Otherwise->
Second, to ensure that all nodes are evenly distributed into the various blockchain slices, the number of nodes in each blockchain slice is M:
finally, the total cost of the data sharing task between every two nodes is calculated as follows:
wherein omega ij Representing MEC node eta i Distribution to blockchain slices s j And:
where L represents the order of MEC nodes other than i, L represents the number of MEC nodes located in the same slice s j Every two MEC nodes eta within i And eta l The set of constituent, -L represents the set of every two MEC nodes located in different slices, and:
wherein N is il Representing every two MEC nodes eta of the whole edge layer i And eta l A set of components, and:
wherein c (i, l) represents node η i And eta l Trade costs between, and:
wherein f i,l And c i,l Node η from different blockchain slices i And eta l Communication frequency and communication overhead between them; i il Is the cost of communication between two nodes located in the same blockchain shard.
3. The intelligent slicing decision block-chain-based industrial internet data sharing optimization method of claim 2, wherein: in the second step, the controller is responsible for making a slicing decision on the blockchain system when processing the data sharing task, and the intelligent slicing decision divides the consensus mode of the blockchain into the following three cases:
both IIoT devices with data sharing requirement upload data tasks to the same MEC node eta i Or in the same blockchain slice s j Two MEC nodes eta in (1) i And eta l The transaction type of the transaction is considered as an intra-slice transaction; the transaction is fragmented s by a blockchain j Verifying M nodes in the network;
both IIoT devices with data sharing requirement upload data tasks to the tiles s located in different blockchains j Sum s k Is defined by two MEC nodes eta i And eta l When the transaction type is cross-fragment transaction; at this point, the transaction is fragmented s by the blockchain j Sum s k Verifying the 2M nodes in the network;
when processing a data sharing task with insensitive delay, the controller selects not to fragment the blockchain system so as to furthest improve the safety of the data; the transaction is verified by all N MEC nodes of the edge layer;
the consensus node of the blockchain system verifies and consensus transaction data generated after the IIoT equipment requests a data sharing task, and when the transaction type is intra-chip transaction, the consensus node number N is obtained C M, N when the transaction type is cross-fragment transaction C =2m, transaction type N for non-fragmented transactions C =n; the common-knowledge nodes adopt a practical Bayesian fault-tolerant common-knowledge mechanism PBFT to verify and realize common knowledge on the blocks and the transactions, and when the number of fault nodes is not more than 1/3 of the total number of the common-knowledge nodes, the PBFT ensures the correctness of the common-knowledge process; assuming that the CPU cycles required to generate or verify a signature or a message authentication code are α and β, respectively, the total computation cycle required to complete a consensus process is:
where a (t) is the average transaction size, b (t) is the transaction batch size, ρ is the probability of verifying the transaction correctly at the request stage.
4. The intelligent shard decision blockchain-based industrial internet data sharing optimization method of claim 3, wherein the method comprises the following steps of: in the third step, the node verification process of the blockchain is provided with calculation support by the MEC server and the cloud server, and specifically comprises the following steps:
when the MEC server provides computational support for the node verification process, the delay of the verification process is:
where λ (t) is the CPU period required for the computing task, F m Calculating frequency for the MEC server;
the total delay is represented by:
wherein i (t) is the block interval, t b Broadcasting a delay for the node;
when the cloud server provides computing support for the node verification process, the delay of the verification process is as follows:
where D (t) is the data task size, μ (t) represents the data transfer rate of the edge layer to cloud layer calculation offload wireless link, F c Calculating frequency for a cloud server;
the total delay is represented by:
5. the intelligent slicing decision blockchain-based industrial internet data sharing optimization method of claim 4, wherein: in the fourth step, the transaction throughput is an important index for evaluating the performance of the slicing decision block chain system and is directly influenced by the following two parameters; the first is the block size, i.e., the capacity of a block, which determines the transactions contained in a block; the other is the block interval, which represents the release rate of the block; since the sliced blockchain generates new blocks in a multithreading manner, the transaction throughput of the blockchain is positively correlated with the number k of slices; the blockchain transaction throughput is calculated by:
wherein S is B Representing the block size.
6. The intelligent slicing decision blockchain-based industrial internet data sharing optimization method of claim 5, wherein: in the fifth step, according to the first to fourth steps, setting a state space, an action space and a reward function in the A3C by combining the scene and the optimization target, wherein the specific steps are as follows:
in a discrete time slot, firstly, sensing the environmental state by an intelligent agent; subsequently, learning experience is obtained according to the environmental state, and then the strategy is updated; the state space is set to:
S(t)＝{D(t),b(t),μ(t)}
wherein the data task size D (t) ranges from 4MB to 8MB, and the transaction batch size b (t) ranges from 1MB to 2MB; due to the diversity and dynamic nature of the computation offload radio link conditions, the data transfer rate is modeled as a finite state Markov decision process, the data transfer rate assigned to each computation offload task comes from the set {125,100,75,50} Mbps, with the probability transition matrix expressed as:
in order to obtain an optimal strategy for solving the above-mentioned scene problem, dynamic adjustment is required; the action space is set as follows:
A(t)＝{S B ,i(t),o(t),f(t)}
wherein S is B = {1,2,..b } and I (t) = {0.2,0.4,..i }; in addition, o (t) = {0,1}, o (t) = 0 represents that the calculation support in the blockchain node verification process is provided by the MEC server, and conversely, if o (t) = 1 represents that the calculation task is unloaded to the cloud layer for processing; f (t) = {0,1}, when f (t) = 1, indicating that the controller chooses to slice the blockchain system, otherwise, not slice;
in the IIoT system supported by the slicing decision block chain of the design, the optimization target is throughput and total time delay; according to the optimization objective, establishing an optimization problem as follows:
s.t.C1:D(t)≤S B
C2:T(t)≤β·i(t)
the bonus function is given by:
wherein w is 1 +w 2 ＝1，Is a punishment parameter.
7. The intelligent slicing decision blockchain-based industrial internet data sharing optimization method of claim 6, wherein: and step six, setting strategy parameters, value function parameters and network layer numbers according to the state space, the action space and the rewarding function constructed in the step five, and training the deep neural network to estimate the advantage function.
A(S(t),A(t)；θ,θ v )＝R t -V(S(t)；θ v )
And is also provided with
Wherein Rt is the accumulated return of discount, V (S (t); θ) v ) An estimate of a cost function; k varies with the state and varies with the local maximum number of iterations t max Is an upper limit; gamma epsilon (0, 1)]R is the discount factor t+i Is an instant rewards;
the local accumulated gradient update of the guiding policy function parameter θ is estimated from the resulting dominance, expressed as:
wherein pi (A (t) |S (t); theta) is a random strategy, beta is a super parameter for controlling the intensity of an entropy regularization term, and H (pi (S (t); theta)) is the entropy of the strategy pi;
according to the RMSProp algorithm, asynchronously updating the global parameter theta, and gradually adjusting the random strategy to the optimal strategy; the RMSProp asynchronous update algorithm is expressed as:
wherein eta is the learning rate, g is the gradient estimation value under RMSProp, and epsilon is a small positive number.
8. The intelligent shard decision blockchain-based industrial internet data sharing optimization method as in claim 7, wherein: and step seven, obtaining an optimal strategy of the optional action under each state according to the deep neural network trained in the step six, and continuously executing the optimal action of each state by taking the action generated by the strategy as the optimal action under the state until the execution instruction is ended.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311396787.3A CN117478697A (en) | 2023-10-26 | 2023-10-26 | Industrial Internet data sharing optimization method based on intelligent slicing decision block chain |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311396787.3A CN117478697A (en) | 2023-10-26 | 2023-10-26 | Industrial Internet data sharing optimization method based on intelligent slicing decision block chain |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117478697A true CN117478697A (en) | 2024-01-30 |
Family
ID=89635670
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311396787.3A Pending CN117478697A (en) | 2023-10-26 | 2023-10-26 | Industrial Internet data sharing optimization method based on intelligent slicing decision block chain |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117478697A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117880331A (en) * | 2024-03-13 | 2024-04-12 | 北京邮电大学 | Block chain data sharing method and device for ocean satellite Internet of things |
-
2023
- 2023-10-26 CN CN202311396787.3A patent/CN117478697A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117880331A (en) * | 2024-03-13 | 2024-04-12 | 北京邮电大学 | Block chain data sharing method and device for ocean satellite Internet of things |
| CN117880331B (en) * | 2024-03-13 | 2024-05-24 | 北京邮电大学 | Block chain data sharing method and device for ocean satellite Internet of things |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111445111B (en) | Electric power Internet of things task allocation method based on edge cooperation | |
| CN108920280B (en) | Mobile edge computing task unloading method under single-user scene | |
| CN111507601B (en) | Resource optimization allocation decision method based on deep reinforcement learning and block chain consensus | |
| CN112380008B (en) | Multi-user fine-grained task unloading scheduling method for mobile edge computing application | |
| CN112598150B (en) | Method for improving fire detection effect based on federal learning in intelligent power plant | |
| CN112995289B (en) | Internet of vehicles multi-target computing task unloading scheduling method based on non-dominated sorting genetic strategy | |
| CN111901862A (en) | User clustering and power distribution method, device and medium based on deep Q network | |
| CN117478697A (en) | Industrial Internet data sharing optimization method based on intelligent slicing decision block chain | |
| CN114189892A (en) | Cloud-edge collaborative Internet of things system resource allocation method based on block chain and collective reinforcement learning | |
| CN112118312B (en) | Network burst load evacuation method facing edge server | |
| CN108650131B (en) | Processing system for multi-controller deployment in SDN network | |
| CN112214301B (en) | Smart city-oriented dynamic calculation migration method and device based on user preference | |
| CN112511614B (en) | Intelligent networking vehicle task unloading method under vehicle-mounted edge environment based on reinforcement learning | |
| CN115935442A (en) | Block chain performance optimization method based on multi-agent deep reinforcement learning | |
| CN112287990A (en) | Model optimization method of edge cloud collaborative support vector machine based on online learning | |
| CN113573363A (en) | MEC calculation unloading and resource allocation method based on deep reinforcement learning | |
| CN116405569A (en) | Task unloading matching method and system based on vehicle and edge computing server | |
| Zhao et al. | Communication-efficient federated learning for digital twin systems of industrial Internet of Things | |
| CN117113274A (en) | Heterogeneous network data-free fusion method and system based on federal distillation | |
| CN111930435A (en) | Task unloading decision method based on PD-BPSO technology | |
| CN113157344B (en) | DRL-based energy consumption perception task unloading method in mobile edge computing environment | |
| CN112764932B (en) | Deep reinforcement learning-based calculation-intensive workload high-energy-efficiency distribution method | |
| CN114006817B (en) | VGDT construction method and device oriented to SDN and readable storage medium | |
| CN116702583B (en) | Method and device for optimizing performance of block chain under Internet of things based on deep reinforcement learning | |
| He et al. | Client selection and resource allocation for federated learning in digital-twin-enabled industrial Internet of Things |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |