Disclosure of Invention
In view of the above, the present invention aims to provide a power distribution network collaborative interaction method based on general sense calculation integration, which is used for solving the above problems existing in the prior art.
In order to solve the technical problems, the invention provides the following technical scheme:
A power distribution network collaborative interaction method based on general sense calculation integration comprises the following steps:
Acquiring an end-side data queue, wherein the end-side data queue prepares total data quantity transmitted to an edge server for all terminals;
Transmitting data through a channel connection terminal;
Generating a data calculation queue of the edge server, wherein the data calculation queue is the data quantity to be processed transmitted to the edge server by the terminal;
Acquiring various time delay information in the side end cooperative process, wherein the various time delay information at least comprises transmission related time delay information generated in the data transmission process of the total data volume of the side data queue and calculation related time delay information generated after the calculation resource allocation of the data volume to be processed of the data calculation queue;
the method is characterized in that the method takes the set time delay requirement as an optimization target, and continuously optimizes the edge data transmission process and the computing resource distribution process based on a channel selection principle and a computing resource distribution principle, so that the collaborative interaction of the distribution network based on general sense calculation integration is realized;
When the channel is subjected to terminal access countermeasure, the terminal access channels are sequentially selected according to the descending order of the signal to interference and noise ratio of the side end;
the computing resource allocation principle is to replace the block authentication computing allocated resources with the difference between the total computing resources and the data computing allocated resources.
Further, with the set time delay requirement as an optimization target, continuously optimizing an edge data transmission process and a computing resource allocation process based on a channel selection principle and a computing resource allocation principle, and constructing a first joint optimization problem model for solving, wherein the first joint optimization problem model is specifically as follows:
In the formula,For terminalsIs used for the channel aware time allocation ratio of (a),Is the firstTerminal with time slotsIs a channel selection indication variable of (a),Is the firstTerminal with time slotsThrough the channelThe transmit power of the transmitted data is set,For edge servers for handling incoming calls from terminalsIs a function of the computing resources allocated by the data of (a),For a set of time slots,For a set of terminals,Queuing delay for end-side queue transmission,Queue queuing delay weights are calculated for the edge side,Is the firstThe calculated queuing delay of the stripe data calculation queue,A latency weight is calculated for the blockchain authentication,Calculating a delay for blockchain authentication; a constraint is imposed on the time distribution ratio,Is a terminal set; in order to be a constraint on the availability of the channel,Is the firstTerminal with time slotsChannel availability indicator variable of (a);、 for the channel selection constraint, it means that each terminal can only select at most one channel for transmission in each time slot, and each channel is at most transmitted in one time slotThe individual terminals are multiplexed and,For the total number of channels,Is the firstThe number of channels in a single channel,Is a set of channels; for power allocation constraints, representing terminal transmission power discretized into slaveTo the point ofA kind of electronic deviceA grade of, whereinGrade of,AndRespectively minimum and maximum values of transmission power; For computing resource allocation constraints, representation for processingThe sum of the computing resources of the individual terminal data must be less than or equal to the computing resources available to the edge server;Representing terminal energy consumption constraints, i.e. terminalsAt the position ofThe perceived and transmitted energy consumption sum of the individual time slots must be less than or equal to the maximum energy consumption constraint;Representing channel selection reliability constraints for a terminalSelecting a channelSINR of (c) must be greater than or equal to the threshold;Represent the firstData volume of data queue at end side of each time slotAnd the data amount of the data calculation queueIs average rate stable.
Further, based on Lyapunov optimization, the first joint optimization problem model is converted into a second joint optimization problem model, which is specifically as follows:
In the formula,Weights representing data transmission delays, data computation delays and blockchain authentication computation delays,The amount of data collected for the end-side data queue,For the amount of data transmitted by the end-side data queue,For the amount of data handled by the edge server,To the firstTerminal when time slots are formedThe amount of energy consumption exceeding the expected amount of energy consumption is accumulated,In order to sense the energy consumption of the vehicle,For the transmission of energy.
Further, the second joint optimization problem model is converted into a first sub-optimization problem model for solving so as to realize the optimization of the edge data transmission process, wherein the first sub-optimization problem model is specifically as follows:
In the formula,Representing the optimization objective of the first sub-optimization problem model.
Further, the first sub-optimization problem model is modeled as a Markov decision process for solving, the Markov decision process comprising the following elements:
State space:, For terminalsFirst, theStatus of each time slot;
Action space:, For terminalsFirst, theActions taken by the time slots;
Rewarding:。
Further, a first joint optimization method based on the countermeasure DQN network is adopted to solve the Kelvin decision process, the first joint optimization method comprises a main network and a target network, and corresponding parameter vectors are respectivelyAnd;
The main network is used for utilizing the state of the current time slotAnd primary network parametersGenerating policiesAccording to the policy adoptAlgorithm determines actions, policiesParameterization intoI.e.To indicate when the system is in stateWhen carrying parametersAction under policy of (a)When the terminal accesses the countermeasure, adopting a channel selection principle to solve the countermeasure problem;
the target network is used for rewarding based on environment feedback and is combined with the terminalCurrent state ofAnd the next stateAnd calculating TD error and loss function, and regulating strategy by the main network by using a gradient descent method based on the loss function.
Further, the second joint optimization problem model is converted into a second sub-optimization problem model for solving, so as to realize optimization of the computing resource allocation process, wherein the second sub-optimization problem model is specifically as follows:
In the formula,For terminalsIs used for the data computation complexity of (a),Is a time slot long.
Further, the second sub-optimization problem model is converted into a convex optimization problem based on the result of the first sub-optimization problem model, and a computing resource allocation optimality condition is obtained based on a dual decomposition and a Carro-Coulomb-Take condition, specifically as follows:
By means ofRepresenting computing resources for blockchain certification computation, translating the second sub-optimization problem model into a convex optimization problem is as follows:
In the formula,Representing an optimization objective of the convex optimization problem;
Structure of the deviceThe lagrangian function of (2) is:
In the formula,AndRespectively representAndLagrangian multipliers of (2);
by dual decompositionThe method comprises the following steps of:
the requirement for the Lagrangian function to take the extremum is based on the Carlo-Coulomb-Tack conditionIs equal to zero, expressed as:
and solving the above equation to obtain an optimal computing resource allocation decision.
Further, the solving process for obtaining the optimal computing resource allocation decision comprises the following steps:
Initializing the number of iterationsIterative step sizeAndConvergence accuracy;
Judging iteration condition ifOr alternativelyThen updateOptimum value of (2)Is that
Updating Lagrangian multipliersAndIs that
Order theRepeating the above steps until meeting the convergence accuracyThe obtainedI.e. as a result of an optimal allocation of computing resources, i.e. a command。
Further, the delay information at least includes:
calculation queuing delay of data calculation queue:
blockchain authentication computation latency:
。
In summary, the invention provides a power distribution network collaborative interaction method based on general sense calculation integration, which comprises the steps of obtaining an end-side data queue, transmitting data through a channel, generating a data calculation queue of an edge server, obtaining various time delay information in an edge-side collaborative process, continuously optimizing an edge data transmission process and a computing resource allocation process based on a channel selection principle and a computing resource allocation principle so as to achieve the power distribution network collaborative interaction based on general sense calculation integration, wherein the channel selection principle is that when a channel is subjected to terminal access countermeasure, terminal access channels are sequentially selected according to descending order of the signal-to-interference-noise ratio of the edge, and the computing resource allocation principle is that the difference value of total computing resources and data computing allocation resources is used for replacing the resources distributed by block authentication computation. When a certain channel is accessed to the countermeasure, the terminals are arranged in descending order based on the SINR, and the terminals with larger SINR are preferentially permitted to access the channel, so that the countermeasure is solved, and the side cooperative optimization of channel selection and the efficient transmission of low-delay and low-power consumption data are realized. Meanwhile, the difference value of the total computing resources and the data computing allocation resources is utilized to replace the block authentication computing allocation resources, so that decoupling between the data computing and the block authentication computing is realized.
Detailed Description
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is apparent that the embodiments described below are only some embodiments of the present invention, not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, an embodiment of the invention provides a power distribution network collaborative interaction method based on general sense calculation integration, which includes the following steps:
s1, acquiring an end-side data queue, wherein the end-side data queue prepares total data quantity to be transmitted to an edge server for all terminals;
S2, transmitting data through a channel connection terminal;
s3, generating a data calculation queue of the edge server, wherein the data calculation queue is the data quantity to be processed transmitted to the edge server by the terminal;
s4, acquiring various time delay information in the side-end cooperative process, wherein the various time delay information at least comprises transmission related time delay information generated in the data transmission process of the total data volume of the side-end data queue and calculation related time delay information generated after the calculation resource allocation of the data volume to be processed of the data calculation queue;
s5, continuously optimizing an edge data transmission process and a computing resource distribution process based on a channel selection principle and a computing resource distribution principle by taking a set time delay requirement as an optimization target, so as to realize the collaborative interaction of the power distribution network based on general sense calculation integration;
When the channel is subjected to terminal access countermeasure, the terminal access channels are sequentially selected according to the descending order of the signal to interference and noise ratio of the side end;
the computing resource allocation principle is to replace the block authentication computing allocated resources with the difference between the total computing resources and the data computing allocated resources.
It should be noted that, edge computing is an information transmission processing technology of cloud computing, and is an extension of cloud computing at a user side. The edge calculation is to sink part or all of the data calculation and data storage to the network edge side close to the user, so that the data transmission delay is reduced, the service response time is shortened, the network broadband is lightened, and the calculation efficiency is improved. The distributed edge calculation meets the requirements of a large number of energy terminals in the energy Internet on response speed and Q0 S service quality. Blockchain technology implemented based on consensus algorithms is essentially a distributed billing technique, common consensus algorithms include workload certification PoW (ProofofWork) and the bayer fault tolerance PBFT (PracticalByzantinefaultTolerance) algorithm. In the energy internet communication architecture and application scenario, blockchain technology and edge computing have wide application. Integrating blockchain and edge computation into one system can achieve reliable access and control of the network, with storage and computation distributed across the edges, providing large-scale network servers, data storage, and near-end availability computation in a secure manner.
In a power distribution network collaborative interaction system based on a blockchain technology and an edge computing technology, the embodiment provides a power distribution network collaborative interaction method based on a channel selection principle and a computing resource allocation principle for the problem that access countermeasure of a channel generating terminal and computing resource allocation are unreasonable.
In the method, aiming at the problem of terminal access countermeasure in the channel occurrence process of the collaborative interaction of the power distribution network, a channel selection principle is adopted for solving the problem. When the channel generates terminal access countermeasure, terminal access channels are sequentially selected according to the descending order of the signal to interference plus noise ratio of the side ends, namely when the terminal access countermeasure of a certain channel is generated, namely the number of terminals for selecting the same channel to perform data transmission exceeds the maximum allowed access terminal number threshold of the channel, the edge server can collect the information of the signal to interference plus noise ratio (Signal to Interference plus Noise Ratio, SINR) of the side ends, and descending order is performed on the terminals based on the SINR, and the terminals with larger SINR are preferentially permitted to access the channel, so that the countermeasure is solved.
Aiming at the problems that the computing resource allocation in the cooperative interaction process of the power distribution network involves the coupling of data computing time delay and block authentication computing time delay, and the relation between dynamic balance time delay and safety is not considered, the method is solved by adopting a computing resource allocation principle. And the difference value of the total computing resources and the data computing allocation resources is utilized to replace the block authentication computing allocation resources, so that decoupling between the data computing and the block authentication computing is realized, and the whole interactive collaborative process is optimized by combining the time delay information in the collaborative interaction process, so that the balance and compromise between the data processing time delay and the block chain consensus safety are realized.
The edge server acquires an end-side data queue, and generates a data calculation queue after the end-side data queue starts transmitting data according to a channel selection principle. According to the data volume in the queue, such as time delay, the execution optimization condition of the collaborative interaction method provided by the embodiment can be reflected.
The embodiment provides a power distribution network collaborative interaction method based on general sense calculation integration, which is characterized in that when a certain channel is subjected to terminal access countermeasure, terminals are arranged in descending order based on SINR, and terminals with larger SINR are preferentially permitted to access the channel, so that countermeasure is solved, and side collaborative optimization of channel selection and low-delay and low-power consumption data efficient transmission are realized. Meanwhile, the difference value of the total computing resources and the data computing allocation resources is utilized to replace the block authentication computing allocation resources, so that decoupling between the data computing and the block authentication computing is realized.
In a preferred embodiment of the present invention, with the set delay requirement as an optimization target, the edge data transmission process and the computing resource allocation process are continuously optimized based on the channel selection principle and the computing resource allocation principle, and are constructed as a first joint optimization problem model for solving, where the first joint optimization problem model is specifically as follows:
(1)
In the formula,For terminalsIs used for the channel aware time allocation ratio of (a),Is the firstTerminal with time slotsIs a channel selection indication variable of (a),Is the firstTerminal with time slotsThrough the channelThe transmit power of the transmitted data is set,For edge servers for handling incoming calls from terminalsIs a function of the computing resources allocated by the data of (a),For a set of time slots,For a set of terminals,Queuing delay for end-side queue transmission,Queue queuing delay weights are calculated for the edge side,Is the firstThe calculated queuing delay of the stripe data calculation queue,A latency weight is calculated for the blockchain authentication,Calculating a delay for blockchain authentication; a constraint is imposed on the time distribution ratio,Is a terminal set; in order to be a constraint on the availability of the channel,Is the firstTerminal with time slotsChannel availability indicator variable of (a);、 for the channel selection constraint, it means that each terminal can only select at most one channel for transmission in each time slot, and each channel is at most transmitted in one time slotThe individual terminals are multiplexed and,For the total number of channels,Is the firstThe number of channels in a single channel,Is a set of channels; for power allocation constraints, representing terminal transmission power discretized into slaveTo the point ofA kind of electronic deviceA grade of, whereinGrade of,AndRespectively minimum and maximum values of transmission power; For computing resource allocation constraints, representation for processingThe sum of the computing resources of the individual terminal data must be less than or equal to the computing resources available to the edge server;Representing terminal energy consumption constraints, i.e. terminalsAt the position ofThe perceived and transmitted energy consumption sum of the individual time slots must be less than or equal to the maximum energy consumption constraint;Representing channel selection reliability constraints for a terminalSelecting a channelSINR of (c) must be greater than or equal to the threshold;Represent the firstData volume of data queue at end side of each time slotAnd the data amount of the data calculation queueIs average rate stable.
In the process of uploading terminal data to an edge server, a plurality of terminals select the same channel, so that channel conflict is easy to generate, serious interference influence is caused, and further data transmission quality and system performance are influenced. The existing method cannot effectively mine and solve the state information of the countermeasure access from the feedback of the edge server, so that poor learning performance, low convergence speed, unreasonable resource allocation and high data transmission queuing delay are caused. Therefore, the embodiment constructs the joint optimization problem of the general computation multidimensional resource allocation. The optimization objective is to minimize the weighted sum of the end-side transmission queue queuing delay, the side-side computation queue queuing delay and the block authentication computation delay by jointly optimizing the time allocation ratio, the channel selection, the power allocation and the computation resource allocation.
Specifically, the problem of joint optimization of general sense calculation multidimensional resource allocation is built by considering low time delay and safety requirements of power distribution network business on data acquisition, transmission and processing. The optimization objective is to minimize the weighted sum of the end-side transmission queue queuing delay, the side-side computation queue queuing delay and the block authentication computation delay by jointly optimizing the time allocation ratio, the channel selection, the power allocation and the computation resource allocation. The joint optimization problem is constructed as shown in formula 1.
The general sense of a single time slot is needed to calculate multidimensional resource optimization decisionAnd the long-term terminal energy consumption constraint of the time slots ensures that the decoupling between the time slots is carried out. In response to this problem, in a preferred embodiment of the present invention, virtual queues in Lyapunov optimization are introduced, constraining long-term terminal energy consumptionAnd converting into queue stability constraint to realize decoupling between time slots. Definition and definition of the inventionThe corresponding virtual energy consumption red word queue isThe queue backlog evolution formula is
(2)
In the formula,Representative to the thTerminal when time slots are formedThe amount of energy consumption exceeding the expected amount is accumulated.The larger means the terminalThe more energy consumption beyond expectations, the more difficult it is to meet long-term terminal energy consumption constraintsOtherwise, it means the terminalThe less energy consumption is unexpected.
Definition of the definitionRepresenting the connection vector of the data queue and the virtual queue. The Lyapunov function is defined as
(3)
Lyapunov drift is defined asThe condition in two consecutive time slots expects to change, and by minimizing Lyapunov drift and adding a punishment upper bound, the lower queue backlog value can be effectively ensured. Thus, optimization problemIs converted into
(4)
In the formula,Weights representing data transmission delays, data computation delays and blockchain authentication computation delays,The amount of data collected for the end-side data queue,For the amount of data transmitted by the end-side data queue,For the amount of data handled by the edge server,In order to sense the energy consumption of the vehicle,For the transmission of energy.
In a specific implementation manner of the embodiment, a collaborative interaction model of a power distribution network based on general sense calculation integration is provided, and joint optimization problems P1 and P2 are solved for the model, wherein the model provides a feasible calculation mode of each parameter in the joint optimization problem, so that how the constructed joint optimization problem is specifically implemented by the collaborative interaction model of the power distribution network can be understood by matching with other embodiments.
The model comprises two parts, namely a channel collaborative interaction perception model and a data transmission model and an edge collaborative interaction calculation model, and the two parts are specifically described below.
(1) Channel collaborative interactive perception and data transmission model
Assume a commonThe 5G cognitive Internet of things terminal is assembled into. The perception layer comprisesA plurality of channels, which are assembled into. The time slot set isEach time slot is as long as. In the first placeEach terminal perceives the availability of the channel based on the cognitive radio technology and performs data transmission. Definition terminalChannel aware time allocation ratio of (2) isWhereinFor the purpose of channel sensing,For data transmission. Assume that the energy consumption required to perceive a channel isThe required time isTerminal thenIs the perceived energy consumption of (2)
(5)
Wherein, theThe representation is rounded down and up,Indicating the number of channels perceived by the terminal.
Definition of the first embodimentTerminal with time slotsChannel selection indicator variable of (2) is,Time representation represents the firstTerminal with time slotsSelecting an access channel. Definition of the first embodimentTerminal with time slotsChannel availability indication variable of (2) is. Channel availability is derived based on channel co-operative perception, wherein,Representing channelsFor terminalsAvailable, otherwise,Indicating unavailability, only the available channels can be selected for data transmission, and therefore,。
TerminalThe collected data is stored in a local cache and can be modeled as an end-side data queue. The input of the queue is the collected data volumeThe queue output is the amount of data transferred. Thus, the end-side data queue backlog evolves to
(6)
In the formula,Indicating that the maximum value is taken.
Based on Little's Law, the transmission queuing delay of the end-side queue is
(7)
In the formula,To get toThe average data arrival rate up to each time slot is calculated by the following formula。
First, theTime slot, terminalThrough the channelIs the transmission data rate of
(8)
In the formula,Indicating terminalSelecting a channelThe magnitude of the signal-to-interference-and-noise ratio of (c),Indicating terminalThrough the channelThe transmitted data is subject to co-channel interference,Representing the power of the gaussian white noise,Represent the firstTerminal with time slotsSelected channelIs used for the transmission of the bandwidth of (a),Represent the firstTerminal with time slotsThrough the channelThe transmit power of the transmitted data is set,Represent the firstTerminal with time slotsSelected channelIs provided.
TerminalThe data quantity transmitted to the edge server through the 5G channel is the minimum value of queue backlog and theoretical transmission capacity, and the expression is
(9)
TerminalIn the first placeThe transmission energy consumption of the time slot is
(10)
(2) Edge collaborative interactive computing model
At each time slot, the edge server optimizes the computing resource allocation for processing data from the 5G cognitive internet of things terminal and blockchain authentication computation required to complete the consensus. Definition of the definitionFor edge servers for handling incoming calls from terminalsIs allocated to the data of the computer system. Definition of the definitionFor the edge server total available computing resources, the computing resources allocated for performing the block authentication computation are。
On edge servers there areA data queue corresponding toAnd a terminal. Wherein, the terminalData computation queuing at edge servers
(11)
In the formula,For terminalsIn the first placeThe amount of data transmitted to the edge server in a single slot,Is the firstFrom terminals handled by a single slot edge serverIs expressed as
(12)
In the formula,For terminalsIs used for calculating the complexity of the data.
On the edge server, the firstThe calculated queuing delay of the stripe data calculation queue is as follows
(13)
In the formula,To get toThe average data arrival rate at the edge side of each time slot is calculated by the following formula。
Consider the alliance blockchain consensus based on the utility Bayesian fault tolerance (PRACTICAL BYZANTINE FAULT TOLERANCE, PBFT) algorithm, in the consensus process, the alliance blockchain consensus comprises a master node and slave nodes, wherein the master node is responsible for proposing new transactions and blocks, broadcasting the transactions and the blocks to other slave nodes, playing the role of interactively coordinating the operation of other nodes, and the slave nodes are responsible for receiving the transactions and the blocks broadcasted by the master node and verifying the legality of the transactions and the blocks. PBFT includes five steps altogether, request, prepare, commit and reply.
Defining the computing resources required to complete the consistency authentication computation asThe blockchain authentication computation delay is
(14)
In the formula,Is a computing resource for blockchain authentication.
The invention provides a power distribution network collaborative interaction method based on communication and calculation integration of countermeasure learning. The method is divided into two stages. The first stage is a method proposed for channel generation terminal access countermeasure, and the second stage is a method proposed for computing resource allocation involving coupling of data computation delay and block authentication computation delay.
In a preferred embodiment of the present invention, for the first stage, a combined optimization method (i.e., a first combined optimization method) of time allocation, channel allocation and energy control based on a Deep Q-Network (DQN) is provided, and when terminal access countermeasure occurs, countermeasure sensing is performed based on side SINR information, so that a terminal access channel with a larger SINR is preferentially permitted, thereby solving the terminal countermeasure problem. The terminal countermeasure perception information is fed back to the countermeasure DQN network in a rewarding mode, the countermeasure DQN main network realizes high-efficiency learning of the countermeasure perception information under the assistance of the target network, so that the countermeasure DQN learning rate is effectively improved, and the terminal access countermeasure conflict is reduced. The method comprises the following specific scheme:
(1) Time allocation, channel allocation and energy control combined optimization method based on countermeasure DQN
Problems after transformationFurther decomposed intoA sub-problem is that,The sub-problem is a sub-problem of joint optimization of time allocation, channel allocation and power control.
The sub-problem is expressed as
(15)
In a further embodiment of the invention, the following will be providedThe sub-problem is modeled as a Markov decision process (Markov Decision Process, MDP), as described in detail below.
State space: time slotMiddle terminalComprises an optimization of the state space of (a)Information required, including terminalEnd-side queue backlog of (1)Calculating queue backlog at a terminal of an edge serverVirtual energy consumption red word queueWeight ofAndComposition is prepared. Thus, the state space can be represented as。
Motion space-motion space is perceived by channel time distribution ratioChannel selection indicator variableAnd controlling powerComposition, expressed as。
Rewards due toTo minimize the problem, the reward for MDP is therefore defined asNegative values of the optimization objective, i.e。
In a further embodiment of the invention, the proposed first stage method solves the constructed MDP problem with an anti-DQN network. The countermeasure DQN network comprises a main network and a target network, wherein the parameter vectors are respectivelyAnd. Wherein the primary network utilizes the state of the current time slotAnd primary network parametersGenerating policiesAccording to the policy adoptThe algorithm determines the action. StrategyParameterization intoI.e.To indicate when the system is in stateWhen carrying parametersAction under policy of (a)Is a probability of (2). When terminal access countermeasure occurs, the edge server performs countermeasure perception based on the collected edge SINR information, and preferentially grants terminals with larger SINR access to the channel, so that the terminal countermeasure problem is solved. Rewards based on environment feedback by the target network and combined with the terminalCurrent state ofAnd the next stateAnd calculating TD error and loss function, and regulating strategy by the main network by using a gradient descent method based on the loss function.
The provided time allocation, channel allocation and energy control combined optimization method based on the countermeasure DQN comprises five stages of network initialization, action selection, countermeasure perception, action execution and network parameter updating, wherein the execution steps are specifically described as follows
1) Initializing primary network parametersInitializing target network parameters for random values。
2) In the first placeTime slot, terminalPolicies based on antagonizing DQN networksBy usingPolicy selection actions。Policy toIn order to be able to explore the degree,Select actions for availability, with ongoing training, exploratory degreesContinuously decaying to a very small value, and finally, the model is with high probabilityAnd executing the current optimal strategy.
3) Definition setRepresenting a selection channelFor the terminal set of (2)Terminal in (a)Calculation ofBy usingA kind of electronic deviceWhen (when)When allowingSelecting a channelAnd orderOn the contrary, whenWhen the terminal is not allowedSelecting a channelAnd orderFrom the collectionIs removed from. Judging whether terminal access countermeasure occurs in any channel, if soTerminal access countermeasure, i.e. occursThen pair the collectionBased on the terminal in (a)Arranging in descending order before selectingPersonal terminal access channelAnd the rest terminals are assembled fromIs removed.
4) Terminals, e.g.Decision making based on motionTime distribution ratio in (2)Power controlChannel selection strategyChannel sensing and data transmission are carried out, and rewards are observedAnd further updates the queue according to equations (2) and (11),。
Calculating a TD error based on the current state and the next state, expressed as
(16)
In the formula,Representing the degree to which the target network considers future rewards value as a discount factor; Is thatQ-functions of the primary and target networks through different network parametersThe Q value of the network is calculated, and the Q value of the target network is introduced to improve the learning stability; Is a penalty function, expressed as
(17)
In the formula,For channelsThe number of times selected.
The TD error represents the prediction error of the environment in the current state, so as to guide the intelligent agent to learn and adjust the strategy, and the loss function is calculated based on the TD error and expressed as
(18)
Main network update based on loss function
(19)
Wherein, theRepresenting the update step size.
6)Up toThe iteration ends. Wherein each target networkUpdating by time slots。
In a preferred embodiment of the present invention, for the second stage, a computing resource allocation method based on a compromise of time delay and security (i.e., a second joint optimization method) is provided to replace the resources allocated by the block authentication computation with the difference between the total computing resources and the data computing allocation resources, so as to implement decoupling between the data computing and the block authentication computing, and the channel allocation and power control joint optimization result is utilized to convert the original problem into a convex optimization problem. And then, obtaining a computing resource allocation optimality condition based on a dual decomposition and a Carlo-Coulomb-Tak (Karush-Kuhn-Tucker, KKT) condition, and carrying out iterative solution on the computing resource allocation optimality condition to realize balance and compromise between data processing delay and block chain consensus safety. The method comprises the following specific scheme:
(2) Computing resource allocation method based on time delay and security compromise
Problems after transformationFurther decomposed intoA sub-problem is that,The sub-problem is a side computing resource allocation optimization sub-problem.
Based onOptimal time allocation, channel allocation and power control result solving for sub-problemsSub-problems, expressed as
(20)
In a further embodiment of the present invention, data computation and blockchain authentication computation are decoupled, utilizingRepresenting computing resources for blockchain authentication computation, further to,,,Substitution into, toConversion to convex optimization problem
(21)
Structure of the deviceIs a Lagrangian function of
(22)
In the formula,AndRespectively representAndLagrangian multiplier of (c).
By dual decompositionConversion to
(23)
According to the KKT condition, the necessary condition that the Lagrangian function takes the extremum isIs equal to zero, expressed as
(24)
An optimal computing resource allocation decision can be obtained by solving the equation (24).
In a further embodiment of the present invention, a method for computing resource allocation based on a compromise between latency and security is proposed to solve the formula (24), which specifically includes the following steps:
1) Initializing the number of iterationsIterative step sizeAndConvergence accuracy。
2) Judging iteration condition ifOr alternativelyThen updateOptimum value of (2)Is that
(25)
3) Updating Lagrangian multipliersAndIs that
(26)
(27)
4) Order theRepeating the steps 2) -3) until meeting convergence accuracyThe obtainedI.e. as a result of an optimal allocation of computing resources, i.e. a command。
Compared with the prior art, the method has the following advantages:
1. The invention provides a time planning, channel allocation and energy control combined optimization method based on a countermeasure DQN. When a certain channel generates terminal access countermeasure, namely the number of terminals selecting the same channel for data transmission exceeds the threshold value of the maximum allowed access terminal number of the channel, the edge server can acquire edge SINR information, and descending order the terminals based on SINR, and preferentially permit the terminals with larger SINR to access the channel, thereby solving the countermeasure and realizing the edge collaborative optimization of channel selection and the efficient transmission of low-delay and low-power consumption data.
2. The invention provides a computing resource allocation method based on time delay and security compromise. And replacing the block authentication calculation allocated resources by utilizing the difference value of the total calculation resources and the data calculation allocated resources, so that decoupling between the data calculation and the block authentication calculation is realized, and the original problem is converted into a convex optimization problem. And then, obtaining an optimal condition of computing resource allocation based on the dual decomposition and the KKT condition, and providing a computing resource allocation method based on time delay and security compromise to carry out iterative solution so as to realize balance compromise between data processing time delay and block chain consensus security.
The above is a detailed description of an embodiment of the power distribution network collaborative interaction method based on general sense calculation integration of the invention, the following describes an embodiment of the power distribution network collaborative interaction system based on general sense calculation in detail.
Referring to fig. 2, the present embodiment provides a collaborative interaction system of a power distribution network based on general sense calculation integration, which can be divided into two parts, namely a sensing calculation integration edge computing device and a 5G cognitive internet of things terminal device.
1) And the sensing and calculating integrated edge calculating device. The sensing and calculation integrated edge computing device comprises a time distribution, channel distribution and energy control combined optimization module based on the countermeasure DQN, a computing resource distribution module based on time delay and safety compromise, a communication module, a data storage module and a power supply module.
And the time distribution, channel distribution and energy control combined optimization module based on the countermeasure DQN performs five steps of network initialization, action selection, countermeasure perception, action execution and network parameter updating based on the side end SINR information acquired by the SINR information acquisition module, thereby realizing side end cooperative time distribution, channel distribution and energy control combined optimization and solving the problem of terminal access countermeasure.
And the computing resource allocation module based on the time delay and the security compromise utilizes an optimal strategy generated by the time allocation, channel allocation and energy control combined optimization module based on the anti-DQN to decouple the data calculation and the blockchain authentication calculation, converts the problem of computing resource allocation optimization of the edge server into a convex optimization problem, and obtains an optimal result of computing resource allocation based on the dual decomposition and KKT conditions.
And the communication module is responsible for communicating with the 5G cognitive Internet of things terminal, receiving the electrical equipment data acquired by the terminal, issuing a resource allocation strategy, and acquiring edge SINR information when terminal access conflict occurs.
The data storage module mainly comprises a Flash memory unit (Flash) and a random access memory unit (RAM). The RAM unit is used as a memory module capable of carrying the electronic positioning label device, provides a basis for high-speed operation of the sensing and calculating integrated edge computing device, and the flash memory unit is used as a long-term storage module of equipment data, and provides a sufficient storage space for the received electrical equipment data and the like.
The power module is responsible for providing stable voltage level for each module in the sensing and calculating integrated edge computing device and guaranteeing normal operation of a system circuit.
2) 5G cognition thing allies oneself with terminal device. The 5G cognitive Internet of things terminal device comprises a data acquisition module, a channel sensing module, a communication module, a data storage module and a power module.
And the data acquisition module is responsible for acquiring the data of the electrical equipment connected with the data acquisition module.
And the channel sensing module senses the availability of the channel based on the cognitive radio technology.
And the communication module is responsible for communicating with the sensing and calculating integrated edge computing device, uploading the acquired electrical equipment data and receiving the optimal resource allocation strategy issued by the sensing and calculating integrated edge computing device.
The data storage module mainly comprises a Flash memory unit (Flash) and a random access memory unit (RAM). The RAM unit is used as a memory module capable of carrying the electronic positioning tag device, provides a basis for high-speed operation of the 5G cognitive Internet of things terminal device, and the flash memory unit is used as a long-term storage module of equipment data, and provides a sufficient storage space for the received electrical equipment data and the like.
And the power supply module is responsible for providing stable voltage level for each module in the 5G cognitive Internet of things terminal device and guaranteeing normal operation of a system circuit.
The invention provides a power distribution network collaborative interaction system based on general sense calculation integration, which comprises a sense calculation integration edge computing device and a 5G cognition Internet of things terminal device. The sensing and calculating integrated edge computing device comprises a time distribution, channel distribution and energy control combined optimization module based on the countermeasure DQN, a computing resource distribution module based on time delay and safety compromise, a communication module, a data storage module and a power supply module, wherein the 5G cognitive Internet of things terminal device comprises a data acquisition module, a channel sensing module, a communication module, a data storage module and a power supply module. The power distribution network collaborative interaction system based on the general sense calculation can realize intelligent sensing by optimizing the time distribution ratio of channel sensing and data transmission based on the cognitive radio technology in the sensing aspect, and can realize low-time delay and low-power consumption data transmission by jointly optimizing channel selection and power distribution and realizing access countermeasure through side collaborative interaction in the transmission aspect, and finally, can realize low-time delay data processing and block chain consensus safety by optimizing side computing resource distribution in the computing aspect, thereby guaranteeing the balance and the balance between time delay and safety and providing a basis for efficient safe collaborative interaction of the power distribution network.
The following describes advantages of the present invention over the prior art with a simulation analysis example.
The simulation experiment of the method and the system provided by the invention uses MATLAB software and is based on Intel Core i7-6900K CPU. The area of the test field is 100m multiplied by 100m, and the test field comprises 8 channels, 25 5G cognitive Internet of things terminal devices and 1 sensing calculation integrated edge calculation device. The 5G cognition Internet of things terminal device and the sensing calculation integrated edge calculation device are randomly distributed in the simulation area. In this section, the proposed method is compared with two different comparison methods. The comparative method 1 employs a simulated annealing algorithm (simulate ANNEAL ARITHMETIC, SAA), which is a random optimization method based on a Monte Carlo iterative strategy. The method can realize random global optimal solution searching in a feasible domain, get rid of a local optimal solution in a probability jump mode and finally converge to the global optimal solution. However, the comparative method 1 cannot perform channel challenge sensing, and adopts a random access policy when terminal access challenge occurs. The comparison method 2 adopts a traditional DQN algorithm, performs resource allocation based on queue backlog sensing, and ensures that the queue delay is minimum. But it lacks channel countering perception, resulting in excessive blockchain authentication computation delay and compromised security.
As can be seen from fig. 4, the weighted delay of the proposed method is reduced by 21.23% and 14.39%, respectively, compared to the two comparison methods. The method is based on countermeasure perception, so that the learning capability of DQN is enhanced, terminal access conflict is avoided, resource utilization efficiency is improved, and data transmission delay is reduced. Meanwhile, the proposed algorithm realizes the compromise allocation of data computing resources and blockchain authentication computing resources.
As can be seen from fig. 5, compared with the two comparison methods, the energy consumption of the 5G cognitive internet of things terminal device of the proposed method is reduced by 5.48% and 3.28%, respectively. It can be seen from the figure that the mean energy consumption of the proposed method fluctuates relatively little and remains within the energy consumption constraints at all times. The root cause of the reduction of the energy consumption is that the method optimizes the channel perception time distribution ratio of the channel distribution of the 5G cognitive Internet of things terminal device. The optimization realizes comprehensive channel perception, reduces terminal access countermeasure, and ensures reliable and stable data transmission.
Based on the same inventive concept, the embodiment of the application also provides a power distribution network collaborative interaction system based on the general sense calculation integration, which is used for realizing the power distribution network collaborative interaction method based on the general sense calculation integration. The implementation scheme of the system for solving the problems is similar to the implementation scheme recorded in the method, so the specific limitation in the embodiment of the power distribution network collaborative interaction system based on the general sense calculation integration can be referred to the limitation of the power distribution network collaborative interaction method based on the general sense calculation integration, and the description is omitted here.
The embodiment of the invention provides a power distribution network collaborative interaction system based on general sense calculation integration, which comprises the following steps:
an edge server and a terminal;
The edge server and the terminal perform power distribution network cooperative interaction, and the cooperative interaction steps based on the edge server are as follows:
Acquiring an end-side data queue, wherein the end-side data queue prepares total data quantity transmitted to an edge server for all terminals;
Transmitting data through a channel connection terminal;
Generating a data calculation queue of the edge server, wherein the data calculation queue is the data quantity to be processed transmitted to the edge server by the terminal;
Acquiring various time delay information in the side end cooperative process, wherein the various time delay information at least comprises transmission related time delay information generated in the data transmission process of the total data volume of the side data queue and calculation related time delay information generated after the calculation resource allocation of the data volume to be processed of the data calculation queue;
the method is characterized in that the method takes the set time delay requirement as an optimization target, and continuously optimizes the edge data transmission process and the computing resource distribution process based on a channel selection principle and a computing resource distribution principle, so that the collaborative interaction of the distribution network based on general sense calculation integration is realized;
When the channel is subjected to terminal access countermeasure, the terminal access channels are sequentially selected according to the descending order of the signal to interference and noise ratio of the side end;
the computing resource allocation principle is to replace the block authentication computing allocated resources with the difference between the total computing resources and the data computing allocated resources.
The foregoing embodiments are merely for illustrating the technical solution of the present invention, but not for limiting the same, and although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that modifications may be made to the technical solution described in the foregoing embodiments or equivalents may be substituted for parts of the technical features thereof, and that such modifications or substitutions do not depart from the spirit and scope of the technical solution of the embodiments of the present invention in essence.