Block chain consensus protocol privacy protection method based on equity proof

Abstract

The invention relates to the technical field of network space security, in particular to a block chain consensus protocol privacy protection method based on a right proof. Through a node capability construction mechanism, hardware acceleration data cache data, environment electromagnetic data and load data are acquired, a comprehensive influence value is calculated through an exponential attenuation correction model and the like, a real-time capability level is analyzed and output, and then a complete set of lightweight multi-level privacy policies and other multi-level privacy policies are matched, so that low-end nodes are prevented from being overloaded due to insufficient computing power, and the security of the low-end nodes is improved. High-end nodes are enabled to exert the computing power advantage, and the adaptability of privacy protection and node computing power is improved; according to the method, through an adaptive scheduling mechanism, attack protection monitoring, intermediate sensitive data and key data monitoring, comprehensive risk value calculation, risk level determination, low-risk maintenance generation, intermediate-risk degradation generation and other adjustment schemes, privacy execution is dynamically optimized, and the overall efficiency is balanced while privacy security is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of cyberspace security technology, specifically to a privacy protection method for a blockchain consensus protocol based on proof-of-stake. Background Technology

[0002] Proof-of-Stake (PoS) blockchain consensus protocols face challenges in the field of privacy protection, such as identity leakage, tagging attacks, and efficiency bottlenecks. Existing technologies achieve privacy protection through cryptographic means such as ring signatures, zero-knowledge proofs, and homomorphic encryption. For example, ring signatures are used to anonymize the identity of the verifier, zero-knowledge proofs are used to hide the stake holdings, and homomorphic encryption is used to protect the transaction amount.

[0003] To avoid the enormous computational resource consumption required for basic processing, existing proof-of-stake-based blockchain consensus protocols often deploy edge computing nodes to share the computational burden, making privacy protection at these nodes particularly important. However, edge computing nodes are deployed in a dispersed manner, and the devices vary significantly in terms of computing power, storage, and network bandwidth. Low-end edge computing nodes are prone to high loads that slow down global aggregation efficiency; moreover, low-end edge computing nodes cannot bear the computational overhead when implementing privacy protection, making it difficult to uniformly implement privacy mechanisms. Summary of the Invention

[0004] This invention provides a privacy protection method for blockchain consensus protocols based on proof-of-stake, which addresses the aforementioned technical problems.

[0005] The first aspect of this invention provides a privacy protection method for a blockchain consensus protocol based on proof-of-stake, comprising the following steps:

[0006] Step A1: The block control server collects information from each edge computing node through the information collection unit to obtain node data, and inputs the node data into the node capability building mechanism for analysis and output of the node's real-time capability level.

[0007] As a further improvement of the present invention, the node capability construction mechanism includes an information identification unit, an information analysis unit, and a capability analysis unit.

[0008] Furthermore, the information identification unit is used to identify the received node data, extract capability-related data from the node data, and send the capability-related data to the information analysis unit. The capability-related data includes hardware acceleration data, node cache data, environmental electromagnetic data, and load data.

[0009] Furthermore, the information analysis unit specifically comprises:

[0010] The received data is used to construct relevant data, and hardware acceleration data, node cache data, environmental electromagnetic data, and load data are obtained based on the data constructed from the data.

[0011] The real-time acceleration time for each edge computing node is obtained based on hardware acceleration data. A timer records the continuous runtime of the current hardware; when the continuous runtime exceeds a preset upper limit, the corresponding continuous runtime is marked as a risky continuous runtime. The real-time acceleration time and the risky continuous runtime are then substituted into the formula of the exponential decay correction model. The hardware acceleration impact value JS is calculated; where t s and T JS These represent the real-time acceleration time and the continuous risk duration, respectively; t0 represents the generation time of the hardware under the theoretically optimal state; k JS This is represented as the preset hardware attenuation coefficient.

[0012] Based on the node cache data, the final-level cache data and memory bus occupancy rate of each edge computing node are obtained; based on the final-level cache data, the current final-level cache and the corresponding preset standard final-level cache are obtained, and the final-level cache miss rate is calculated by comparing the current final-level cache with the standard final-level cache; the corresponding Lyapunov exponent matrix is ​​constructed from the final-level cache miss rate and memory bus occupancy rate using the Lyapunov exponent model, and the corresponding spectral norm is obtained based on the Lyapunov exponent matrix; the Lyapunov exponent matrix, spectral norm, and final-level cache miss rate are substituted into the chaotic attractor model. The cache impact value HC is calculated; where Λ and ||Λ||2 represent the Lyapunov exponent matrix and spectral norm, respectively; hc represents the last-level cache miss rate, and σ represents the preset standard deviation of the last-level cache miss rate.

[0013] Based on environmental electromagnetic data, IoT frequency band data and interference data corresponding to each edge computing node are obtained. The frequency band energy density is obtained by identifying the IoT frequency band data. A pre-set interference analysis time period is acquired, and the corresponding interference data is obtained. The number of interference events corresponding to the interference events is identified. When the number of interference events exceeds a preset interference event warning value, an interference penalty signal is generated. A signal monitoring program is executed. When an interference penalty signal is detected, the value corresponding to the number of interference events is recorded as the interference penalty value. The frequency band energy density and interference penalty value are then analyzed using the wavelet packet entropy integral model formula. The electromagnetic influence value DC was calculated; where WPD(f) represents the band energy density corresponding to the 2.4 / 5.2GHz band, f1 = 2.3GHz, f2 = 5.2GHz; CF represents the interference penalty value, ε CF df represents the interference penalty coefficient corresponding to the preset interference penalty value; df represents the infinitesimal increment with respect to frequency f.

[0014] Based on the load data, the current number of tasks to be processed and the historical average response latency for each edge computing node are obtained; the current number of tasks to be processed and the historical average response latency are then substituted into the formula of the nonlinear hyperbolic control function. The load impact value FZ is calculated; where R real R represents the number of tasks currently pending. max tanh represents the maximum task queue capacity corresponding to the node. -1 (x) is the inflection tangent function; T resp This is expressed as the historical average response time. The preset Sigmoid response threshold parameters are as follows:

[0015] The impact values ​​of hardware acceleration, cache, electromagnetic interference, and load are normalized and their values ​​are taken. A nonlinear coupling model is then used to... The overall impact value C is calculated; where, Represents the tensor product. X T This is represented as matrix transpose, specifically converting the vector corresponding to the cached impact value from a column vector to a row vector, so that it can be multiplied by matrix with the vector corresponding to the hardware acceleration impact value. Indicates perturbation addition. ⊙ represents modulation multiplication, DC⊙FZ = DC·FZ·e -γ|DC-FZ| γ is the preset modulation weight, with a value of 0.8; the comprehensive impact value is sent to the capability analysis unit.

[0016] Furthermore, the capability analysis unit specifically comprises:

[0017] The mean and standard deviation of the capability influence are calculated by taking the comprehensive influence value corresponding to each edge computing node, and are denoted as μ respectively. C and σ C ; Analyzing the mean and standard deviation of capability impact yields multiple level classification thresholds, and the comprehensive impact value corresponding to each edge computing node is matched with each level classification threshold; when When the real-time capability level is generated, it is a high computing power level; when At that time, the real-time capability level is medium computing power level; when The real-time capability level is generated based on the computing power level; when At that time, the generated real-time capability level is the low computing power level.

[0018] Furthermore, an analysis was conducted on the mean and standard deviation of the ability influence, specifically as follows:

[0019] The level classification threshold is marked as The system obtains pre-set threshold entropy increment and threshold entropy decrement coefficients, with the threshold entropy increment coefficient being greater than 1 and the threshold entropy decrement coefficient being (0,1). It then multiplies the standard deviation of capability influence with the threshold entropy increment coefficient, and sums the product with the mean of capability influence to obtain the level classification threshold. The mean influence of ability is marked as the threshold for classification. The threshold for classifying abilities is calculated by multiplying the standard deviation of the ability impact by the threshold entropy reduction factor, and then subtracting the product from the mean of the ability impact.

[0020] Step A2: Obtain the multi-level privacy policy corresponding to the preset privacy policy, and match the real-time capability level with the multi-level privacy policy to obtain the corresponding execution policy; specifically, the preset privacy policy is divided into multi-level privacy policies, including a full set of privacy policies, a lightweight privacy policy, and a basic privacy policy, and the real-time capability level corresponding to each edge computing node is obtained, and the real-time capability level is matched with the multi-level privacy policy to obtain the corresponding execution policy.

[0021] Step A3: Obtain the execution policy corresponding to each edge computing node, and perform privacy-preserving execution based on the execution policy. Monitor the privacy-preserving execution through an adaptive scheduling mechanism to obtain an adaptive adjustment scheme.

[0022] As a further improvement to the present invention, the adaptive scheduling mechanism is specifically analyzed as follows:

[0023] The adaptive scheduling mechanism includes a privacy detection unit and a privacy assessment unit;

[0024] The privacy detection unit is used to collect data during the execution process of privacy protection to obtain privacy assessment data, and then send the privacy assessment data to the privacy assessment unit. The privacy assessment data includes attack protection data, intermediate sensitive data, and privacy key data collected through a pre-deployed data detection layer.

[0025] The privacy assessment unit is used to perform privacy assessment analysis on the received privacy assessment data to obtain the privacy assessment status, obtain the risk level corresponding to the privacy assessment status, and generate a corresponding adaptive adjustment plan based on the risk level.

[0026] Furthermore, the privacy assessment analysis includes the following specific content:

[0027] Based on the privacy assessment data, we obtain the attack protection data, intermediate sensitive data, and privacy key data corresponding to each privacy protection implementation;

[0028] Based on the attack protection data, the number of attacks and the number of successful protections are obtained. The ratio of the number of successful protections to the number of attacks is calculated to obtain the protection success rate. The protection success rate is divided into multiple protection rate sets based on a preset proportional interval. The minimum protection success rate corresponding to each protection rate set is obtained. The protection rate set sequence number is obtained by sorting the minimum protection success rates of each protection rate set in ascending order. A corresponding attack risk value is set for each protection rate set. The protection success rate corresponding to each privacy protection execution is matched with each protection rate set to obtain the corresponding attack risk value. The attack risk value decreases as the protection rate set sequence number increases.

[0029] Based on intermediate sensitive data, the deletion delay and residual data amount of sensitive data are obtained; the deletion delay is compared with a preset safe deletion delay threshold, and when the deletion delay is greater than the safe deletion delay threshold, the difference is calculated to obtain the deletion delay overflow value; the ratio of the residual data amount to the total amount of sensitive data is calculated to obtain the data residual degree, and when the data residual degree is greater than a preset residual degree threshold, the difference between the data residual degree and the residual degree threshold is calculated to obtain the residual overflow value; the deletion delay overflow value and the residual degree overflow value are normalized and a modified Gaussian distribution model formula is applied. The sensitivity risk value M is calculated; where t qc and cl represent the clearing delay overflow value and the residual overflow value, respectively; σ qc σ cl These represent the preset standard deviations of the clearing delay and the residual overflow, respectively.

[0030] Based on the privacy key data, the entropy fluctuation coefficient and key update duration corresponding to the privacy key are obtained. The fluctuation risk value is calculated by comparing the entropy fluctuation coefficient with the preset fluctuation coefficient threshold. Based on the database, the standard key update cycle corresponding to various keys is obtained. The key update duration corresponding to the current privacy protection execution is matched with the corresponding standard key update cycle. The part of the key update duration that exceeds the standard key update cycle is recorded as the update risk value. The fluctuation risk value and the update risk value are calculated and summed to obtain the key risk value.

[0031] The attack risk value, sensitivity risk value, and key risk value are monitored. The monitored attack risk value, sensitivity risk value, and key risk value are normalized, and their values ​​are then calculated using a dynamic weighted geometric mean model. The overall risk value F is calculated; where GJ and MY represent the attack risk value and key risk value, respectively; S rw This indicates the privacy sensitivity level corresponding to the current privacy protection implementation; S C This represents the node's privacy capability level.

[0032] The comprehensive risk value is used to obtain three comprehensive risk value intervals based on the preset risk classification threshold. The comprehensive risk value intervals are marked as low risk, medium risk and high risk as the comprehensive risk value increases. The interval level corresponding to the current privacy protection execution is obtained and marked as the risk level. Based on the risk level, the corresponding preset adaptive adjustment scheme is obtained.

[0033] Step A4: Obtain the adaptive adjustment scheme for each edge computing node and carry out emergency response based on the adaptive adjustment scheme.

[0034] The beneficial effects of the technical solution provided by this invention compared with the prior art are as follows:

[0035] 1. This invention collects hardware acceleration data cache data, environmental electromagnetic data and load data through a node capability construction mechanism. It calculates the comprehensive impact value through an exponential decay correction model, classifies real-time capability levels, and then matches multi-level privacy strategies such as full set and lightweight. This avoids low-end nodes from being overloaded due to insufficient computing power, while allowing high-end nodes to leverage their computing power advantages, thereby improving the adaptability of privacy protection and node computing power.

[0036] 2. This invention uses an adaptive scheduling mechanism to monitor attack protection, intermediate sensitive data and key data, calculate a comprehensive risk value to determine the risk level, generate adjustment schemes such as low-risk maintenance and medium-risk downgrade, dynamically optimize privacy execution, and balance overall efficiency while ensuring privacy security. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. The following drawings are not deliberately drawn to scale according to the actual size, but are intended to show the main idea of ​​this application.

[0038] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 In this embodiment of the invention, one example of a privacy protection method for a blockchain consensus protocol based on proof-of-stake includes:

[0041] A1. Node Capability Building Analysis: The block control server collects information from each edge computing node through the information collection unit to obtain node data, and inputs the node data into the node capability building mechanism for analysis and output of the real-time capability level of the node.

[0042] The node capability building mechanism, and its specific analysis content are as follows:

[0043] The node capability building mechanism includes an information identification unit, an information analysis unit, and a capability analysis unit.

[0044] The information identification unit is used to identify the received node data, extract the capability building-related data from the node data, and send the capability building-related data to the information analysis unit. The capability building-related data includes hardware acceleration data, node cache data, environmental electromagnetic data, and load data.

[0045] The information analysis unit is used to analyze relevant data on the receiving capacity to obtain the load impact value, specifically as follows:

[0046] Based on hardware acceleration data, the real-time acceleration time of each edge computing node is obtained. The real-time acceleration time is the acceleration efficiency of zero-knowledge proof by the running unit of HSM or TEE, that is, the real-time acceleration time corresponds to the actual time taken for the current HSM to complete the generation of one zk-STARK (zero-knowledge scalable transparent knowledge proof). The continuous runtime corresponding to the current hardware is recorded by a timer. When the continuous runtime exceeds the preset continuous upper limit, the corresponding continuous runtime is marked as a risky continuous runtime. The real-time acceleration time and the risky continuous runtime are substituted into the formula of the exponential decay correction model. The hardware acceleration impact value JS is calculated; where t s and T JS These represent the real-time acceleration time and the continuous risk duration, respectively; t0 represents the generation time of the hardware under the theoretically optimal state; k JS This represents the preset hardware attenuation coefficient, with a value ranging from 0 to 1, which is preset by the corresponding hardware chip model.

[0047] Based on the node cache data, the last-level cache data and memory bus occupancy rate of each edge computing node are obtained; based on the last-level cache data, the current last-level cache and the corresponding preset standard last-level cache are obtained, and the ratio of the current last-level cache to the standard last-level cache is calculated to obtain the last-level cache miss rate; using the Lyapunov exponent model, the last-level cache miss rate and memory bus occupancy rate are used to construct the corresponding Lyapunov exponent matrix, and based on the Lyapunov exponent matrix, the corresponding spectral norm is obtained. For example, the data sampling for the last-level cache miss rate is as follows:

[0048] The values ​​are [0.22, 0.25, 0.31, 0.28, 0.19, 0.33, 0.26, 0.24, 0.29, 0.35], with a corresponding mean of 0.272; the memory bus occupancy data is sampled as follows:

[0049] The range is [0.58, 0.62, 0.71, 0.65, 0.53, 0.68, 0.61, 0.59, 0.66, 0.73], with a corresponding mean of 0.636. Substituting the sampled data into the Lyapunov model yields the Lyapunov exponent matrix, i.e. λ1 and λ2 are the corresponding exponential divergence rates in the exponential matrix; the spectral norm is ||Λ||2=max|λ k | 1≤k≤2 =max(|-1.652|,|-0.752|)=1.652, which represents the corresponding maximum divergence rate; Substituting the Lyapunov exponent matrix, spectral norm, and final-level buffer miss rate into the chaotic attractor model. The cache impact value HC is calculated; where Λ and ||Λ||2 represent the Lyapunov exponent matrix and spectral norm, respectively; hc represents the last-level cache miss rate, and σ represents the preset standard deviation of the last-level cache miss rate.

[0050] Based on environmental electromagnetic data, IoT frequency band data and interference data corresponding to each edge computing node are obtained. The frequency band energy density is obtained by identifying the IoT frequency band data, specifically the energy densities of the commonly used industrial IoT frequency bands 2.4GHz and 5.2GHz. A pre-set interference analysis time period is acquired, and the corresponding interference data is obtained. The number of interference events corresponding to the interference events is identified. When the number of interference events exceeds a preset interference event warning value, an interference penalty signal is generated. A signal monitoring program is executed; when an interference penalty signal is detected, the value corresponding to the number of interference events is recorded as the interference penalty value. The frequency band energy density and interference penalty value are then analyzed using a wavelet packet entropy integral model formula. The electromagnetic influence value DC was calculated; where WPD(f) represents the band energy density corresponding to the 2.4 / 5.2GHz band, f1 = 2.3GHz, f2 = 5.2GHz; CF represents the interference penalty value, ε CF The interference penalty coefficient is the preset interference penalty value, which takes a value of 0-1 and increases with the number of interference events; df represents the infinitesimal increment of frequency f, which is used to incorporate the frequency domain continuity of WPD(f) into the calculation and accurately quantify the entropy value of electromagnetic energy distribution.

[0051] Based on the load data, the current number of tasks to be processed and the historical average response latency for each edge computing node are obtained; the current number of tasks to be processed and the historical average response latency are then substituted into the formula of the nonlinear hyperbolic control function. The load impact value FZ is calculated; where R real R represents the number of tasks currently pending. max tanh represents the maximum task queue capacity corresponding to the node. -1 (x) is the inflection tangent function; T resp This is expressed as the historical average response time. The preset Sigmoid response threshold parameters are as follows: It combines inverse hyperbolic tangent quantization of resource margin and uses Sigmoid activation to dynamically penalize high-latency nodes.

[0052] The impact values ​​of hardware acceleration, cache, electromagnetic interference, and load are normalized and their values ​​are taken. A nonlinear coupling model is then used to... The overall impact value C is calculated; where, Represents the tensor product. X T This is represented as matrix transpose, specifically converting the vector corresponding to the cached impact value from a column vector to a row vector, so that it can be multiplied by matrix with the vector corresponding to the hardware acceleration impact value. Indicates perturbation addition. ⊙ represents modulation multiplication, DC⊙FZ = DC·FZ·e -γ|DC-FZ| γ is the preset modulation weight, with a value of 0.8; the comprehensive impact value is sent to the capability analysis unit.

[0053] The capability analysis unit receives the comprehensive impact value and performs computing power level analysis to obtain the real-time capability level of the node, specifically as follows:

[0054] The mean and standard deviation of the capability influence are calculated by taking the comprehensive influence value corresponding to each edge computing node, and are denoted as μ respectively. C and σ C ; Analyzing the mean and standard deviation of capability impact yields multiple level classification thresholds, and the comprehensive impact value corresponding to each edge computing node is matched with each level classification threshold; when When the real-time capability level is generated, it is a high computing power level; when At that time, the real-time capability level is medium computing power level; when The real-time capability level is generated based on the computing power level; when At that time, the generated real-time capability level is the low computing power level.

[0055] An analysis was conducted on the mean and standard deviation of the influence of ability, specifically by marking the level division threshold as... The system obtains pre-set threshold entropy increment and threshold entropy decrement coefficients, with the threshold entropy increment coefficient being greater than 1 and the threshold entropy decrement coefficient being (0,1). It then multiplies the standard deviation of capability influence with the threshold entropy increment coefficient, and sums the product with the mean of capability influence to obtain the level classification threshold. The mean influence of ability is marked as the threshold for classification. The threshold for classifying abilities is calculated by multiplying the standard deviation of the ability impact by the threshold entropy reduction factor, and then subtracting the product from the mean of the ability impact.

[0056] A2. Privacy Policy Matching and Execution: The preset privacy policy is divided into multi-level privacy policies, including full privacy policy, lightweight privacy policy and basic privacy policy. The real-time capability level of each edge computing node is obtained, and the real-time capability level is matched with the multi-level privacy policy to obtain the corresponding execution policy.

[0057] A3. Privacy Task Scheduling Analysis: Obtain the execution strategy corresponding to each edge computing node, and perform privacy-preserving execution based on the execution strategy. Monitor the privacy-preserving execution through an adaptive scheduling mechanism to obtain an adaptive adjustment scheme.

[0058] The adaptive scheduling mechanism, specifically analyzed below:

[0059] The adaptive scheduling mechanism includes a privacy detection unit and a privacy assessment unit;

[0060] The privacy detection unit is used to collect data during the execution process of privacy protection to obtain privacy assessment data, and then send the privacy assessment data to the privacy assessment unit. The privacy assessment data includes attack protection data, intermediate sensitive data, and privacy key data collected through a pre-deployed data detection layer.

[0061] The privacy assessment unit is used to perform privacy assessment analysis on the received privacy assessment data to obtain the privacy assessment status, obtain the risk level corresponding to the privacy assessment status, and generate a corresponding adaptive adjustment plan based on the risk level.

[0062] The privacy assessment analysis specifically includes the following:

[0063] Based on the privacy assessment data, we obtain the attack protection data, intermediate sensitive data, and privacy key data corresponding to each privacy protection implementation;

[0064] Based on the attack protection data, the number of attacks and the number of successful protections are obtained. The ratio of the number of successful protections to the number of attacks is calculated to obtain the protection success rate. The protection success rate is divided into multiple protection rate sets based on a preset proportional interval. The minimum protection success rate corresponding to each protection rate set is obtained. The protection rate set sequence number is obtained by sorting the minimum protection success rates of each protection rate set in ascending order. A corresponding attack risk value is set for each protection rate set. The protection success rate corresponding to each privacy protection execution is matched with each protection rate set to obtain the corresponding attack risk value. The attack risk value decreases as the protection rate set sequence number increases.

[0065] Based on intermediate sensitive data, the deletion delay and residual data amount of sensitive data are obtained; the deletion delay is compared with a preset safe deletion delay threshold, and when the deletion delay is greater than the safe deletion delay threshold, the difference is calculated to obtain the deletion delay overflow value; the ratio of the residual data amount to the total amount of sensitive data is calculated to obtain the data residual degree, and when the data residual degree is greater than a preset residual degree threshold, the difference between the data residual degree and the residual degree threshold is calculated to obtain the residual overflow value; the deletion delay overflow value and the residual degree overflow value are normalized and a modified Gaussian distribution model formula is applied. The sensitivity risk value M is calculated; where t qc and cl represent the clearing delay overflow value and the residual overflow value, respectively; σ qc σ cl These represent the preset standard deviations of the clearing delay and the residual overflow, respectively.

[0066] Based on the privacy key data, the entropy fluctuation coefficient and key update duration corresponding to the privacy key are obtained. The fluctuation risk value is calculated by comparing the entropy fluctuation coefficient with the preset fluctuation coefficient threshold. Based on the database, the standard key update cycle corresponding to various keys is obtained. The key update duration corresponding to the current privacy protection execution is matched with the corresponding standard key update cycle. The part of the key update duration that exceeds the standard key update cycle is recorded as the update risk value. The fluctuation risk value and the update risk value are calculated and summed to obtain the key risk value.

[0067] The attack risk value, sensitivity risk value, and key risk value are monitored. The monitored attack risk value, sensitivity risk value, and key risk value are normalized, and their values ​​are then calculated using a dynamic weighted geometric mean model. The overall risk value F is calculated; where GJ and MY represent the attack risk value and key risk value, respectively; S rw This represents the privacy sensitivity level corresponding to the current privacy protection implementation, with a quantification value of 1-3, where 3 represents high sensitivity; S CThis is represented by the node's privacy capability level, with a quantification value of 1-3. The computing power level corresponding to the edge computing node is used as the quantification basis, i.e., S represents the level of low computing power. C =1.

[0068] The comprehensive risk value is used to obtain three comprehensive risk value intervals based on the preset risk classification threshold. The comprehensive risk value intervals are marked as low risk, medium risk and high risk as the comprehensive risk value increases. The interval level corresponding to the current privacy protection execution is obtained and marked as the risk level. Based on the risk level, the corresponding preset adaptive adjustment scheme is obtained.

[0069] A4. Adaptive Adjustment: Obtain the adaptive adjustment scheme for each edge computing node and perform emergency response based on the adaptive adjustment scheme; the adaptive adjustment scheme includes the current strategy for low-risk maintenance, the privacy strategy for medium-risk degradation, and the migration task and memory erasure for high-risk tasks.

[0070] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A privacy protection method for a blockchain consensus protocol based on proof-of-stake, characterized in that, Includes the following steps: Step A1: The block control server collects information from each edge computing node through the information collection unit to obtain node data, and inputs the node data into the node capability building mechanism for analysis and output of the real-time capability level of the node. Step A2: Obtain the multi-level privacy policy corresponding to the preset privacy policy, and match the real-time capability level with the multi-level privacy policy to obtain the corresponding execution policy; Step A3: Obtain the execution strategy corresponding to each edge computing node, and perform privacy-preserving execution based on the execution strategy. Monitor the privacy-preserving execution through an adaptive scheduling mechanism to obtain an adaptive adjustment scheme. Step A4: Obtain the adaptive adjustment scheme for each edge computing node and carry out emergency response based on the adaptive adjustment scheme.

2. The privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 1, characterized in that, The specific analysis of the node capability construction mechanism is as follows: The node capability building mechanism includes an information identification unit, an information analysis unit, and a capability analysis unit; The information identification unit is used to identify the received node data, extract the capability building-related data from the node data, and send the capability building-related data to the information analysis unit. The capability building-related data includes hardware acceleration data, node cache data, environmental electromagnetic data, and load data. The information analysis unit is used to analyze the relevant data on the receiving capability to obtain the load impact value; The capability analysis unit is used to receive the comprehensive impact value and perform computing power level analysis to obtain the real-time capability level of the node.

3. The privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 2, characterized in that, The information analysis unit is used to analyze the relevant data on the received capability to obtain the load impact value, specifically as follows: The real-time acceleration time of each edge computing node is obtained based on the hardware acceleration data; the continuous runtime corresponding to the current hardware is recorded by a timer, and when the continuous runtime exceeds the preset continuous upper limit, the corresponding continuous runtime is marked as a risky continuous runtime. The impact value of hardware acceleration is calculated by substituting the real-time acceleration time and the continuous duration of risk into the formula of the exponential decay correction model. The final-level cache data and memory bus occupancy rate of each edge computing node are obtained based on the node cache data. The current last-level cache and the corresponding preset standard last-level cache are obtained based on the last-level cache data. The last-level cache miss rate is calculated by comparing the ratio of the current last-level cache to the standard last-level cache. The Lyapunov exponent model is used to construct the corresponding Lyapunov exponent matrix by using the last-level cache miss rate and memory bus occupancy rate, and the corresponding spectral norm is obtained based on the Lyapunov exponent matrix. The cache impact value is calculated by substituting the Lyapunov exponent matrix, spectral norm, and final-level cache miss rate into the chaotic attractor model. Based on environmental electromagnetic data, the corresponding IoT frequency band data and interference data of each edge computing node are obtained, and the frequency band energy density is obtained by identifying the IoT frequency band data. Obtain a pre-set interference analysis time period, acquire the interference data corresponding to the interference analysis time period, identify the number of interference events corresponding to the interference events in the interference data, and generate an interference penalty signal when the number of interference events exceeds the preset interference event warning value; execute the signal monitoring program, and when the interference penalty signal is detected, record the value corresponding to the number of interference events as the interference penalty value. The electromagnetic influence value is calculated using the wavelet packet entropy integral model formula for the frequency band energy density and interference penalty value. Based on the load data, the current number of tasks to be processed and the historical average response latency for each edge computing node are obtained; The load impact value is calculated by substituting the current number of tasks to be processed and the historical average response delay into the formula of the nonlinear hyperbolic control function. The impact values ​​of hardware acceleration, cache, electromagnetic interference, and load are normalized and their values ​​are taken. The comprehensive impact value is calculated using a nonlinear coupling model and then sent to the capability analysis unit.

4. The privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 3, characterized in that, The capability analysis unit is used to receive the comprehensive impact value and perform computing power level analysis to obtain the real-time capability level of the node, specifically as follows: The mean and standard deviation of the capability influence are calculated by taking the comprehensive influence value corresponding to each edge computing node, and are denoted as μ respectively. C and σ C Analysis of the mean and standard deviation of the influence of ability yields multiple threshold values ​​for classifying levels. Match the comprehensive impact value corresponding to each edge computing node with the threshold values ​​for each level classification; when When the real-time capability level is generated, it is a high computing power level; when At that time, the real-time capability level is medium computing power level; when The real-time capability level is generated based on the computing power level; when At that time, the generated real-time capability level is the low computing power level.

5. A privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 4, characterized in that, The analysis of the mean and standard deviation of the influence of ability specifically involves: marking the level division threshold as... The system obtains pre-set threshold entropy increment and threshold entropy decrement coefficients, with the threshold entropy increment coefficient being greater than 1 and the threshold entropy decrement coefficient being (0,1). It then multiplies the standard deviation of capability influence with the threshold entropy increment coefficient, and sums the product with the mean of capability influence to obtain the level classification threshold. The mean influence of ability is marked as the threshold for classification. The threshold for classifying abilities is calculated by multiplying the standard deviation of the ability impact by the threshold entropy reduction factor, and then subtracting the product from the mean of the ability impact.

6. The privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 1, characterized in that, Specifically, A2 involves dividing the preset privacy policy into multi-level privacy policies, including a full set of privacy policies, a lightweight privacy policy, and a basic privacy policy. It also involves obtaining the real-time capability level of each edge computing node and matching the real-time capability level with the multi-level privacy policies to obtain the corresponding execution policy.

7. A privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 1, characterized in that, The specific analysis of the adaptive scheduling mechanism is as follows: The adaptive scheduling mechanism includes a privacy detection unit and a privacy assessment unit; The privacy detection unit is used to collect data during the execution process of privacy protection to obtain privacy assessment data, and then send the privacy assessment data to the privacy assessment unit. The privacy assessment data includes attack protection data, intermediate sensitive data, and privacy key data collected through a pre-deployed data detection layer. The privacy assessment unit is used to perform privacy assessment analysis on the received privacy assessment data to obtain the privacy assessment status, obtain the risk level corresponding to the privacy assessment status, and generate a corresponding adaptive adjustment plan based on the risk level.

8. A privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 7, characterized in that, The privacy assessment analysis specifically includes the following content: Based on the privacy assessment data, we obtain the attack protection data, intermediate sensitive data, and privacy key data corresponding to each privacy protection implementation; Based on the attack protection data, the number of attacks and the number of successful protections are obtained. The ratio of the number of successful protections to the number of attacks is calculated to obtain the protection success rate. The protection success rate is divided into multiple protection rate sets based on a preset proportional interval. The lowest protection success rate corresponding to each protection rate set is obtained. The protection rate sets are sorted in ascending order based on the lowest protection success rates of each protection rate set to obtain the protection rate set number. A corresponding attack risk value is set for each protection rate set. The protection success rate of each privacy protection execution is matched with each protection rate set to obtain the corresponding attack risk value. The attack risk value decreases as the protection rate set index increases; Based on intermediate sensitive data, the deletion delay and residual amount of sensitive data are obtained; the deletion delay is compared with a preset safe deletion delay threshold, and when the deletion delay is greater than the safe deletion delay threshold, the difference is calculated to obtain the deletion delay overflow value; the residual amount of data is calculated as the ratio of the total amount of sensitive data to obtain the data residual degree, and when the data residual degree is greater than a preset residual degree threshold, the difference between the data residual degree and the residual degree threshold is calculated to obtain the residual overflow value; the deletion delay overflow value and the residual degree overflow value are normalized and the sensitive risk value is calculated using the modified Gaussian distribution model formula; Based on the privacy key data, the entropy fluctuation coefficient and key update time corresponding to the privacy key are obtained. The fluctuation risk value is obtained by calculating the ratio of the entropy fluctuation coefficient to the preset fluctuation coefficient threshold. Based on the database, the standard key update cycle corresponding to various keys is obtained. The key update duration corresponding to the current privacy protection execution is matched with the corresponding standard key update cycle. The part of the key update duration that exceeds the standard key update cycle is recorded as the update risk value. The key risk value is obtained by summing the volatility risk value and the update risk value.

9. A privacy protection method for a blockchain consensus protocol based on proof-of-stake as described in claim 8, characterized in that, The attack risk value, sensitivity risk value, and key risk value are monitored. The monitored attack risk value, sensitivity risk value, and key risk value are normalized and their values ​​are taken. The comprehensive risk value is calculated using the dynamic weighted geometric mean model formula. The comprehensive risk value is used to obtain three comprehensive risk value intervals based on the preset risk classification threshold. The comprehensive risk value intervals are marked as low risk, medium risk and high risk as the comprehensive risk value increases. The interval level corresponding to the current privacy protection execution is obtained and marked as the risk level. Based on the risk level, the corresponding preset adaptive adjustment scheme is obtained.

Images