A Data Transmission Method for Billing at New Energy Vehicle Charging Stations Based on Mobile Edge Computing and Blockchain Technologies

Abstract

The present invention discloses a method for transmitting charging data of a new energy vehicle charging station based on mobile edge computing and blockchain technology. By constructing a blockchain model and a transmission model, the data throughput in the blockchain and the consensus process are calculated, and the delay and energy consumption generated by local calculation and unloading calculation, as well as the delay and energy consumption generated by the charging station sending output results to the vehicle are calculated. Thus, by training a deep neural network, the adjustment of the master node and replica node unloading decision, block size and block interval is guided to complete the optimal resource allocation in the scene. The present invention overcomes the problems of safe storage and transmission of charging billing data, excessive energy consumption of equipment due to processing computing tasks, excessive delay, and low data throughput in the blockchain. Simulation experiments show that the present invention has certain advantages in saving system delay, energy consumption and improving data throughput in the blockchain.

Description

Technical Field

[0001] The present invention relates to a method for data transmission of charging station charges for new energy vehicles based on mobile edge computing and blockchain technology. By adopting a deep reinforcement learning algorithm, a method is designed to optimize resource allocation strategies in the consensus process of the blockchain system for transaction data of each charging station, effectively reduce system delay and energy consumption and improve the data throughput of the blockchain system, while ensuring the safe storage and effective transmission of charging data. The method belongs to the relevant fields of resource allocation and system decision-making in the Internet of Things. Background Art

[0002] In recent years, with the increasing severity of environmental pollution and the depletion of fossil energy, green travel has become one of the ways of energy conservation and emission reduction advocated by the world. As a green means of transportation, electric vehicles (EVs) have obvious advantages over traditional fuel vehicles, such as less air pollution and high energy utilization, which provides strong support for promoting the green and intelligent development of the transportation industry. At present, a large number of EV charging infrastructures are being built and serve EVs to facilitate the timely charging and electricity trading of EVs.

[0003] In the charging billing process, it is crucial to ensure the accuracy and security of charging information. However, when EVs and charging stations use different smart meters to measure electricity demand, different results may be produced, resulting in invalid or incorrect billing. In addition, charging information is easily leaked or manipulated, and attackers can track users' billing data and obtain information such as location privacy. Therefore, the security and privacy of charging transaction information cannot be ignored.

[0004] In order to solve the privacy and security issues in charging transactions, blockchain technology is considered to be a promising technology. It can be widely used in power market transactions and other energy transaction interconnection scenarios. It has the characteristics of decentralization, trustlessness and tamper-proof, and promotes the interconnection of information. At present, although a lot of work has been done in blockchain applications, the problem of consuming a lot of computing resources in the blockchain consensus process is still a difficult problem that needs attention and solution.

[0005] In this regard, edge computing can be used to solve the problem of insufficient computing resources of blockchain nodes. By providing computing offload services for nodes, computing efficiency can be significantly improved and energy consumption can be reduced. Among them, mobile edge computing (MEC) is one of the most promising research paradigms and has been widely used in many outstanding research works, including resource allocation, offload strategies, and collaborative cooperation. In addition, since blockchain and MEC have the same decentralized characteristics and interdependent functions, the combination of the two becomes natural.

[0006] However, there are still some potential problems and challenges when integrating blockchain and MEC into the charging station billing data transmission system. For example, how to balance the system latency and energy consumption by considering different offloading decisions of the master node device and the replica node device. Another issue is how to improve the data throughput performance while maintaining the security of the blockchain system. In addition, since the introduction of MEC increases the total system latency and the coverage of charging stations is limited, how to select appropriate charging stations to transmit billing results to EVs is also a problem that needs to be considered. Therefore, these issues should be carefully considered when designing the system.

[0007] At the same time, in view of the high dynamics and high dimensionality of the environmental state in the charging station billing data transmission system, deep reinforcement learning (DRL) has gradually become a popular and efficient optimization method in recent years. It combines the perception ability of deep learning and the decision-making ability of reinforcement learning to train models and optimize decisions in a way close to the human brain. Since its proposal, DRL has been widely studied due to its versatility and effectiveness, and has been continuously applied in various fields. However, there is a lack of relevant research on the charging station system. How to use DRL to solve the performance optimization problems of the system's latency, energy consumption and data throughput still needs careful consideration.

[0008] In summary, the present invention aims at the problems of computing task offloading of master node devices and replica node devices and data blockchain consensus in the new energy vehicle charging station scenario, and proposes a new energy vehicle charging station billing data transmission method based on mobile edge computing and blockchain technology. By jointly considering the status, transmission rate and average transaction size of wireless access points (APs) and servers in the scenario, the system latency and energy consumption are optimized at the same time, and the data throughput of the blockchain system is significantly improved, thereby effectively performing the secure storage and transmission of billing data. Summary of the invention

[0009] The main purpose of the present invention is to optimize the allocation of resources in the scene from the perspective of optimal optimization, considering the presence of multiple charging stations, multiple APs, multiple EVs, a single macrocell base station and a single MEC server in the scene, with the optimization goals of reducing system latency, energy consumption and increasing data throughput in the blockchain, model the scene, and apply the DRL algorithm to iteratively learn the model to obtain an optimal resource allocation strategy that is energy-saving, time-saving and high-throughput. This method solves the problem of how to determine the optimal resource allocation strategy when there are multiple charging stations, multiple APs, multiple EVs, a single macrocell base station and a single MEC server in the scene, and effectively reduces system latency and energy consumption by executing the optimal resource allocation strategy, and improves the data throughput of the blockchain system, thereby effectively performing secure storage and transmission of charging billing data.

[0010] The multi-charging station environment scenario model adapted by the present invention is shown in Figure 1 .

[0011] The system operation principle flow chart of the technical solution of the present invention is shown in Figure 2 .

[0012] The relationship between the total delay of the system of the present invention and the number of charging stations is shown in Figure 3 .

[0013] The relationship between the total energy consumption of the system of the present invention and the number of charging stations is shown in Figure 4 .

[0014] The relationship between the data throughput and average transaction size of the blockchain system of the present invention is shown in Figure 5 .

[0015] The multi-charging station environment scenario model of the present invention is as follows Figure 1 As shown in the figure, a new energy vehicle charging station billing data transmission method based on mobile edge computing and blockchain technology is shown. In a certain communication scenario, there are c charging stations, each of which is deployed with an AP, and the center of the coverage area is deployed with a macro cell base station and a MEC server. In addition, there is a blockchain system formed by APs. When the number of charging stations is known, the blockchain model and transmission model are set according to the actual environment. Then the state space, action space and reward function in the DRL are constructed, and the policy parameters, value function parameters, network layer parameters and other parameters in the training network are set. Iterative learning is combined with the scenario model to train the deep neural network parameters to estimate the state value function. Finally, the optimal resource allocation strategy learned by the intelligent agent through interaction with the environment is executed, thereby effectively reducing the delay and energy consumption generated by the storage and transmission of charging billing data, and improving the data throughput of the blockchain system. It is implemented in the following steps:

[0016] Step (1): The master AP collects the charging billing data collected by the APs of each charging station, packages the data into new blocks, and transmits it to the blockchain system. The consensus nodes in the system verify and reach consensus on the transaction data sent by the master AP. The specific steps are as follows:

[0017] Step (1.1), there are c charging stations with data consensus requirements in the communication scenario. In a period of time t∈{1,2···T}, T is the moment when any AP in the AP group runs out of energy. The charging piles in the charging station transmit charging billing data to the equipped AP. The AP group designates an AP as the master node to receive the billing data collected by other APs and process and package them to generate blocks, and transmit them to the blockchain system formed by the AP. The consensus nodes in the system use the practical Byzantine fault-tolerant consensus mechanism to verify and reach consensus on blocks and transactions. The CPU cycles required to generate or verify a signature and generate or verify a message verification code are α and β respectively. The total computing cycle C required for each consensus node to complete a consensus process is:

[0018]

[0019] Among them, f is the maximum number of error nodes that can be accommodated, f = (N-1) / 3, g is the proportion of correct transactions sent by the master node for verification, d is the total transaction size, δ is the average transaction size, N is the number of consensus nodes in the blockchain system, and the total computing cycle C required for the master node to complete a consensus process p for:

[0020]

[0021] The total computing cycle C required for the replica node to complete a consensus process r for:

[0022]

[0023] In step (1.2), when the master node and the replica node choose to process data locally, the delays generated are expressed as:

[0024]

[0025] At the same time, the energy consumption generated by local processing of the master node and the replica node is expressed as:

[0026] E p = k(F l ) 2 C p

[0027] E r = k(F l ) 2 C r

[0028] Among them, F l is the CPU computing power of the AP, and k is set to 10 based on the actual measured value. -27 , Correspondingly, when the master node and the replica node choose to offload the data processing task to the MEC server for processing, the resulting delay is expressed as:

[0029]

[0030] At the same time, the energy consumption generated by offloading computing tasks from the master node and the replica node is expressed as:

[0031]

[0032]

[0033] Where R is the transmission rate between AP and MEC server, λ is the computing power of MEC server, and p t is the AP transmission power, p m To calculate the power of the MEC server, the master node chooses local processing, and the replica node chooses offload processing. The delays generated by the master node and the replica node are expressed as:

[0034]

[0035]

[0036] Among them, c r c is the computing period when the number of replica nodes in the consensus process is greater than that of the master node. p The computing cycle of the master node is more than that of the replica node. Correspondingly, the master node chooses to offload processing and the replica node chooses local processing. The delays generated by the master node and the replica node are expressed as:

[0037]

[0038]

[0039] Step (2), each AP selects a processing method for the computing task, and the delay T of the blockchain consensus process for charging billing data is n It is expressed as:

[0040]

[0041] Among them, T i is the generation interval of the block, t b is the broadcast delay between nodes, and the energy consumption of the consensus process is E n It is expressed as:

[0042] En =E p +(N-1)E r

[0043] The transaction data throughput ψ of the blockchain system is expressed as:

[0044]

[0045] Among them, S is the size of the generated block;

[0046] Step (3), after the charging billing information is stored on the chain, APs send the output results to EVs traveling within its coverage area so that EVs can make payments and refer to the billing information. The transmission delay is T t It is expressed as:

[0047]

[0048] Energy consumption during transmission E t It is expressed as:

[0049]

[0050] Among them, d * is the output data size, R av is the transmission rate between AP and EVs. The delay T generated by the chain storage and transmission process of charging billing data in the scenario is expressed as:

[0051] T=T n +T t

[0052] The energy consumption E generated is expressed as:

[0053] E=E n +E t

[0054] Step (4): According to steps (1)-(3), combined with the scenario and optimization objectives, set the state space, action space and reward function in DRL. The specific steps are as follows:

[0055] Step (4.1), according to the APs energy V(t) = {v1(t),v2(t),...,v c (t)}, the transmission rate between APs and MEC server R(t) = {r1(t),r2(t),...,r c (t)}, the dynamic changes of the computing resources λ(t) of the MEC server, and the average transaction size δ(t), setting the state space:

[0056] S(t)={V(t),R(t),λ(t),δ(t)}

[0057] Step (4.2), set the action space according to the number of charging stations in the scene, the number of APs and MEC servers supporting the consensus process:

[0058] A(t)={a p (t),a r (t),S(t),T i (t)}

[0059] Among them, a p (t)∈(0,1) and a r (t)∈(0,1) are the offloading decision actions of the master node and the replica node respectively, S(t)∈{1,2,...,S} is the block size adjustment action, T i (t)∈{0.2,0.5,...,I} is the action to adjust the block interval;

[0060] Step (4.3), set the reward function according to the optimization goal:

[0061]

[0062] stC1:

[0063] C2:T n (t)≤ε×T i (t)

[0064] C3:d(t)≤S(t)

[0065] Among them, ω1, ω2 and ω3 are the weighted coefficients of delay, energy consumption and throughput respectively. is the penalty value, is the minimum value of APs energy, ε is the time limit coefficient, C1 represents the APs energy limit, C2 represents the block completion time limit, and C3 represents the task data size limit;

[0066] Step (5), according to the state space, action space and reward function constructed in step (4), set the strategy parameters, value function parameters and the number of network layers, and train the deep neural network to parameterize and estimate the state value function:

[0067] V w (s)≈V π (s)

[0068] Among them, V π (s) represents the value of state s under strategy π, V w (s) represents the state value function approximated by the parameter vector w parameterized function. The update of the policy function parameter θ is guided by the estimated value. Δθ can be expressed as:

[0069]

[0070] Among them, α is the learning efficiency of strategy update, d(s) is the state distribution, and π θ (s,a) is a random strategy. A(s,a)=Q w (s,a)-V w (s) is the advantage function, which significantly reduces the variance in gradient calculation, where Q w (s,a) is the action value function Q π The approximate action value function is obtained by approximating the function of (s, a). By continuously updating the policy function parameter θ, the random policy is gradually adjusted to the optimal policy;

[0071] Step (6), according to the deep neural network trained in step (5), the optimal strategy for the optional actions in each state is obtained, the action generated by the strategy is used as the optimal action in the state, and the optimal action in each state is continuously executed until the execution of the instruction is completed.

[0072] The advantage of the present invention is that in a communication scenario with multiple charging stations and multiple EVs, by considering the energy of each charging station APs, the transmission rate between APs and MEC servers, the computing resources of MEC servers, and the average transaction size status, the weighted cost of system delay and energy consumption in the working cycle is effectively reduced, and the data throughput of the blockchain system is improved. Through simulation experiments, the impact of the new energy vehicle charging station billing data transmission method based on mobile edge computing and blockchain technology on the weighted cost of system delay and energy consumption in the scenario and the data throughput in the blockchain is investigated. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 ,The communication scenario model includes the structural diagram of EVs, charging stations, APs, macro base stations, MEC servers and blockchain systems.

[0074] Figure 2 , design flow chart of the billing data transmission method for new energy vehicle charging stations based on mobile edge computing and blockchain technology.

[0075] Figure 3 , a relationship diagram between the total system delay and the number of charging stations, in which the lower triangle represents the method of the present invention, the square represents the fixed block interval framework, the circle represents the masterless node unloading decision framework, the upper triangle represents the replica node unloading decision framework, and the five-pointed star represents the existing framework.

[0076] Figure 4 , a relationship diagram between the total energy consumption of the system and the number of charging stations, in which the lower triangle represents the method of the present invention, the circle represents the master-less node unloading decision framework, and the square represents the replica-less node unloading decision framework.

[0077] Figure 5 , a graph showing the relationship between data throughput and average transaction size of a blockchain system, wherein the lower triangle represents the method of the present invention, the circle represents a fixed block size framework, the square represents a fixed block interval framework, and the five-pointed star represents an existing framework. DETAILED DESCRIPTION

[0078] The following is a further explanation of the technical solution of the new energy vehicle charging station billing data transmission method based on mobile edge computing and blockchain technology with reference to the accompanying drawings and examples.

[0079] The process flow chart of the present invention is as follows: Figure 2 As shown, the following steps are included:

[0080] Step 1: Initialize the charging station billing system, set the number of charging stations, the number of devices used to serve the blockchain consensus process, the APs power, and the base station transmission rate;

[0081] Step 2: Calculate the time delay t of the master node to process data locally based on the actual system situation p and energy consumption E p , the latency of the replica node processing data locally t r and energy consumption E r , the delay t of the master node unloading computing tasks p and energy consumption E p , the delay t of replica node offloading computing tasks p and energy consumption E p 、Consensus delay of blockchain system T n and consensus energy consumption E n , the data throughput of the blockchain system ψ and the delay T for APs to send output results to EVs t and energy consumption E t ;

[0082] Step 3: According to the optimization goal, set the state space S(t), action space A(t) and reward function r(t) of the DRL algorithm;

[0083] Step 4: Solve the joint optimization problem according to the DRL algorithm and set the number of deep neural network layers, strategy parameters, and value function parameters;

[0084] Step 5: Train the deep neural network to approximate the V value and use the V value to guide the update of the strategy parameters;

[0085] Step six: Select the optimal action based on the optimal strategy obtained in each state to obtain the maximum benefit.

[0086] Figure 3 is the relationship diagram between the total system delay and the number of charging stations. Figure 3It can be seen that the system delay increases with the increase in the number of charging stations. When the number of charging stations is 7, the delay corresponding to the method of the present invention is as low as 1.17s, while the lowest delay of the other methods is 1.35s. It can be concluded that the total system delay is related to the number of charging stations. As the number of charging stations increases, the system computing tasks become heavier, which makes the task processing time longer, but the delay optimized based on the method of the present invention is always lower than that based on other methods.

[0087] Figure 4 is the relationship diagram between the total energy consumption of the system and the number of charging stations. Figure 4 It can be seen that when the number of charging stations increases, the system energy consumption also increases due to the increase in the amount of computing tasks generated. The system energy consumption corresponding to the method described in the present invention is always lower than that of other methods, and when the same system energy is consumed, more charging stations can be served under the method described in the present invention. For example, when the number of charging stations is 8, the system energy consumption corresponding to the method described in the present invention is only 8J, while the system energy consumption corresponding to the other methods is up to 41J.

[0088] Figure 5 This is the relationship between the data throughput and the average transaction size of the blockchain system. Figure 5 It can be seen that when the average transaction volume is 100B, the data throughput in the blockchain corresponding to the method of the present invention can reach 280TPS, while the data throughput of other methods can only reach 240TPS at most. It can be concluded that the data throughput of the blockchain system is related to the average transaction size. When the average transaction volume increases, the number of transactions contained in a block decreases, and the data throughput will decrease accordingly. However, the data throughput optimized based on the method of the present invention is always higher than that based on other methods.

Claims

1. A method for transmitting charging data at a new energy vehicle charging station based on mobile edge computing and blockchain technology, characterized in that: The method comprises the following steps: Step 1: Initializing the charging station billing system, setting the number of charging stations, the number of devices used to serve the blockchain consensus process, the power of access points APs, and the transmission rate of base stations; Step 2: Calculate the time delay t of the master node to process data locally based on the actual system situation p1 and energy consumption E p1 , the latency of the replica node processing data locally t r1 and energy consumption E r1 , the delay t of the master node unloading computing tasks p2 and energy consumption E p2 , the delay t of replica node offloading computing tasks r2 and energy consumption E r2 、Consensus delay of blockchain system T n and consensus energy consumption E n , the data throughput of the blockchain system ψ and the delay T for APs to send output results to electric vehicles EVs t and energy consumption E t ; Step 3: According to the optimization goal, set the state space State(t), action space A(t) and reward function r(t) of the DRL algorithm; Step 4: Solve the joint optimization problem according to the DRL algorithm and set the number of deep neural network layers, strategy parameters, and value function parameters; Step 5: Train the deep neural network to approximate the V value and use the V value to guide the update of the strategy parameters; Step 6: Select the optimal action based on the optimal strategy obtained in each state to obtain the maximum benefit; In step 1, the master AP collects the charging billing data collected by the APs of each charging station, packages the data into new blocks, and transmits it to the blockchain system. The consensus nodes in the system verify and reach consensus on the transaction data sent by the master AP. The specific steps are as follows: Step (1.1), there are c charging stations with data consensus requirements in the communication scenario. In a period of time t∈{1,2···T1}, T1 is the moment when any AP in the AP group runs out of energy. The charging piles in the charging station transmit charging billing data to the equipped AP. The AP group designates an AP as the master node to receive the billing data collected by other APs and process and package them to generate blocks, and transmit them to the blockchain system formed by the AP. The consensus nodes in the system use the practical Byzantine fault-tolerant consensus mechanism to verify and reach consensus on blocks and transactions. The CPU cycles required to generate or verify a signature and generate or verify a message verification code are α and β respectively. The total computing cycle C required for each consensus node to complete a consensus process is: Among them, f is the maximum number of error nodes that can be accommodated, f = (N-1) / 3, g is the proportion of correct transactions sent by the master node for verification, d is the total transaction size, δ is the average transaction size, N is the number of consensus nodes in the blockchain system, and the total computing cycle C required for the master node to complete a consensus process p for: The total computing cycle C required for the replica node to complete a consensus process r for: In step (1.2), when the master node and the replica node choose to process data locally, the delays generated are expressed as: At the same time, the energy consumption generated by local processing of the master node and the replica node is expressed as: E p1 =k(F l ) 2 C p E r1 =k(F l ) 2 C r Among them, F l is the CPU computing power of the AP, and k is set to 10 based on the actual measured value. -27 , Correspondingly, when the master node and the replica node choose to offload the data processing task to the MEC server for processing, the resulting delay is expressed as: At the same time, the energy consumption generated by offloading computing tasks from the master node and the replica node is expressed as: Where R is the transmission rate between AP and MEC server, λ is the computing power of MEC server, and p t is the AP transmission power, p m To calculate the power of the MEC server, the master node chooses local processing, and the replica node chooses offload processing. The delays generated by the master node and the replica node are expressed as: Among them, c r c is the computing period when the replica nodes in the consensus process are more than the master nodes, p The computing cycle of the master node is more than that of the replica node. Correspondingly, the master node chooses to offload processing and the replica node chooses local processing. The delays generated by the master node and the replica node are expressed as: In step 2, each AP selects a processing method for the computing task, and the delay T of the blockchain consensus process for charging billing data is n It is expressed as: Among them, T i is the generation interval of the block, t b is the broadcast delay between nodes, and the energy consumption of the consensus process is E n It is expressed as: AND n =And p +(N-1)E r The transaction data throughput ψ of the blockchain system is expressed as: Among them, S is the size of the generated block; After the charging billing information is stored on the chain, APs send the output results to EVs traveling within their coverage area so that EVs can make payments and refer to the billing information. The transmission delay is T t It is expressed as: Energy consumption during transmission E t It is expressed as: Among them, d * is the output data size, R av is the transmission rate between AP and EVs. The total delay T generated by the chain storage and transmission process of charging billing data in the scenario is expressed as: T=T n +T t The total energy consumption E generated is expressed as: E=E n +E t According to steps 1 to 3, combined with the scenario and optimization objectives, set the state space, action space and reward function in DRL. The specific steps are as follows: Step (4.1), according to the APs energy V(t) = {v1(t),v2(t),...,v c (t)}, the transmission rate between APs and MEC server R(t) = {r1(t),r2(t),...,r c (t)}, the dynamic changes of the computing resources λ(t) of the MEC server, and the average transaction size δ(t), setting the state space: State(t)={V(t),R(t),λ(t),δ(t)} Step (4.2), set the action space according to the number of charging stations in the scene, the number of APs and MEC servers supporting the consensus process: A(t)={a p (t),a r (t),S(t),T i (t)} Among them, a p (t)∈(0,1) and a r (t)∈(0,1) are the offloading decision actions of the master node and the replica node respectively, S(t)∈{1,2,...,S} is the block size adjustment action, T i (t)∈{0.2,0.5,...,I} is the action to adjust the block interval; Step (4.3), set the reward function according to the optimization goal: s.t.C1: C2:T(t)≤ε×T i (t) C3:d(t)≤S(t) Among them, ω1, ω2 and ω3 are the weighted coefficients of delay, energy consumption and throughput respectively. is the penalty value, is the minimum value of APs energy, ε is the time limit coefficient, C1 represents the APs energy limit, C2 represents the block completion time limit, and C3 represents the task data size limit; In step 5, according to the state space, action space and reward function constructed in step 4, the policy parameters, value function parameters and number of network layers are set to train the deep neural network for parameterized estimation of the state value function: V w (s)≈V π (s) Among them, V π (s) represents the value of state s under strategy π, V w (s) represents the state value function approximated by the parameter vector w parameterized function, and the update of the policy function parameter θ is guided by the estimated value, Δθ is expressed as: Among them, γ is the learning efficiency of strategy update, d(s) is the state distribution, and π θ (s,a) is a random strategy; A(s,a)=Q w (s,a)-V w (s) is the advantage function, which significantly reduces the variance in gradient calculation, where Q w (s,a) is the action value function Q π The approximate action value function is obtained by function approximation of (s, a); by continuously updating the policy function parameter θ, the random policy is gradually adjusted to the optimal policy; In step six, the optimal strategy for optional actions in each state is obtained based on the deep neural network trained in step five, the action generated by the strategy is used as the optimal action, and the optimal action in each state is continuously executed until the execution of the instruction is completed.

Images