A verifiable runtime output authorization method for cloud-based security-critical systems

By using a multi-condition judgment method of the security gateway, the problem of incorrect authorization of outputs with consistent dual-channel outputs but erroneous computational semantics in cloud-based virtualization environments is solved. This enables the identification and prevention of erroneous outputs, improving the security of output authorization and the continuity of health status across cycles.

CN123069401APending Publication Date: 2026-11-10BEIJING JIAOTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611286073.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-24
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

In cloud-based virtualization environments, when dual-channel outputs are consistent but computational semantics are incorrect, the traditional two-out-of-two mechanism may result in incorrect authorization.

Method used

A verifiable runtime output authorization method is provided. The method obtains authentication input, system-level execution status and channel output parameters through a security gateway, determines the channel evidence predicate, the consistency result of the dual-channel functional side output digest and the consistency result of the candidate successor state commitment, and combines the lateness penalty of the closed link and the health status to comprehensively determine the final system mode to prevent erroneous output from being authorized.

Benefits of technology

While ensuring the consistency of dual-channel output, it identifies and prevents erroneous outputs from being authorized, improves the security of output authorization, and ensures cross-cycle continuity by latching the health status.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123069401A_ABST
    Figure CN123069401A_ABST
Patent Text Reader

Abstract

This application discloses a verifiable operational output authorization method for cloud-based security-critical systems, relating to the field of secure computer technology. Based on authentication input, system-level execution status, and channel output parameters, it determines the channel evidence predicate, the consistency result of the dual-channel functional-side output digest, and the consistency result of the dual-channel candidate successor state commitments. It then determines the grace period deadline and normalized late arrival penalty; based on the channel evidence predicate, closure time, normal release deadline, grace period deadline, historical health conditions, the consistency result of the dual-channel functional-side output digest, and the consistency result of the dual-channel candidate successor state commitments, it determines the final system mode and performs output disposal and status submission operations, generating an execution token; finally, it determines the health status and latches it. This application can solve the problem of incorrect authorization by the traditional two-out-of-two mechanism when dual-channel outputs are consistent but computational semantics are incorrect in cloud-based virtualization environments.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of secure computer technology, and in particular to a verifiable runtime output authorization method for cloud-based security-critical systems. Background Technology

[0002] As urban rail transit train operation control systems migrate to private clouds, multi-core processors, and virtualization software stacks, train control applications originally deployed on dedicated safety computers are gradually running on shared processor resources, unified scheduling strategies, virtual machine monitors, and cloud-based operation and maintenance environments. Cloudification and virtualization can improve resource utilization, system resilience, and operational efficiency, but at the same time, they change the fault independence conditions that traditional safety computers rely on.

[0003] Rail transit safety-critical control systems typically employ a redundant architecture consisting of two-out-of-two (BOT) units or multiple BOT units. Each BOT unit contains two parallel channels. These two channels execute control logic separately for the same input, and the outputs are compared to determine whether an output is allowed. Traditional mechanisms assume that random hardware failures or localized software anomalies in a single channel will usually cause inconsistencies in the outputs of the two channels, which can therefore be detected through comparison and voting.

[0004] In a multi-core virtualization environment, two channels may share processor microarchitecture resources, virtual machine monitors, underlying software components, scheduling policies, and configuration data. Shared component failures, common software defects, scheduling coupling, or microarchitectural interference can simultaneously affect both channels, causing them to produce erroneous outputs for the same input that are identical but do not conform to the established functional model and security control rules. These errors satisfy the traditional two-out-of-two consistency condition at the numerical level, but are erroneous at the computational semantic level, and thus may bypass a pass-through mechanism that relies solely on output comparison.

[0005] Existing technologies such as input / output storage management unit isolation, virtual machine isolation, and resource partitioning are mainly used to restrict out-of-bounds access, illegal storage access, and some resource interference. They are difficult to directly prove that the output of a certain control cycle is indeed calculated from a specified functional model, specified inputs, and specified preceding states. Traditional operation monitoring, health checks, and log auditing usually rely on resource status, thresholds, abnormal events, or post-event logs for judgment, which may miss transient calculation semantic errors or only complete the detection after the erroneous output has already occurred.

[0006] Verifiable computation technology allows the proving end to generate cryptographic proofs for the computation process, and the verifying end to verify whether the computation satisfies a pre-encoded verifiable relationship without re-executing the entire computation. Existing verifiable computation schemes mostly focus on single-channel computation correctness, communication trustworthiness, or large-scale computation task slicing, and have not yet formed a secure output authorization closed loop that simultaneously combines two-out-of-two output consistency, computation proof, output and proof binding, cross-cycle state submission, late proof handling, and tamper-proof runtime history.

[0007] Therefore, in cloud-based and virtualized environments, how to further verify the correctness of computational semantics, the continuity of states, and the validity of time while ensuring the consistency of dual-channel output, and prevent consistent but erroneous outputs from being authorized and released, is a technical problem that urgently needs to be solved in the field of output authorization for safety-critical systems. Summary of the Invention

[0008] The purpose of this application is to provide a verifiable runtime output authorization method for cloud-based security-critical systems, in order to solve the problem of incorrect authorization by the traditional two-out-of-two mechanism when dual-channel outputs are consistent but computational semantics are incorrect in a cloud-based virtualization environment.

[0009] To achieve the above objectives, this application provides the following solution.

[0010] This application provides a verifiable runtime output authorization method for cloud-based security-critical systems, wherein the cloud-based security-critical system includes two secure computing channels and a security gateway, and the method is executed in the security gateway, the method comprising: Designate any secure computing channel as the current channel; Obtain the authentication input, system-level execution status, and channel output parameters of the current channel in the current control cycle; the channel output parameters include functional-side service output, candidate successor states, public verification instances, and computational proofs; the public verification instances include cycle input context commitments, initial state commitments, proof-side output summaries, and candidate successor state commitments; Based on the functional side service output of the current channel in the current control cycle, determine the functional side output summary of the current channel in the current control cycle; Based on the authentication input, public verification instance, system-level execution status, computational proof, and functional side output summary of the current channel in the current control cycle, determine the channel evidence predicate of the current channel in the current control cycle; Based on the functional side output summary of the two secure computing channels in the current control cycle, the consistency result of the dual-channel functional side output summary in the current control cycle is determined. Based on the candidate successor state commitments of the two secure computing channels in the current control cycle, the consistency result of the dual-channel candidate successor state commitments in the current control cycle is determined. Acquire control cycle duration, lateness tolerance ratio, normal release deadline of the current control cycle, closure time of each closed link in the current control cycle, and historical health conditions; Based on the control cycle duration, the lateness tolerance ratio, and the normal release deadline of the current control cycle, determine the grace period deadline of the current control cycle. Based on the closing time of each closed link in the current control cycle and the normal release deadline and grace period deadline of the current control cycle, the normalized lateness penalty of each closed link in the current control cycle is determined. Based on the channel evidence predicates of the two secure computing channels in the current control cycle, the closure time and historical health conditions of each closed link in the current control cycle, the normal release deadline, the grace period deadline, the dual-channel functional side output digest consistency results, and the dual-channel candidate successor state commitment consistency results, the final system mode of the current control cycle is determined. Specifically, when initially determined to be a candidate for the grace period acceptance mode, state alignment processing is performed, and the final system mode is determined based on the state alignment results. The final system mode can be a normal acceptance mode, a grace period acceptance mode, or a rejection mode. Based on the final system mode of the current control cycle, execute the corresponding output handling and status submission operations. Specifically, when the final system mode of the current control cycle is the normal acceptance mode, authorize the release of the functional-side service outputs of the current control cycle that are consistent between the two secure computing channels, submit the candidate successor states of the current control cycle of the two secure computing channels as the system-level execution state of the next control cycle, and generate an execution token for the next control cycle. When the final system mode of the current control cycle is the grace period acceptance mode, output the fault safety result of the current control cycle, do not authorize the release of the functional-side service outputs of the current control cycle, submit the candidate successor states after the status alignment result of the current control cycle is passed as the system-level execution state of the next control cycle, and generate an execution token for the next control cycle. When the final system mode of the current control cycle is the rejection mode, refuse to authorize the release of the functional-side service outputs of the current control cycle, do not submit candidate successor states, and do not generate an execution token for the next control cycle. Based on the closing time, grace period, and historical health conditions of each closed link in the current control cycle, as well as the normalized lateness penalty of each closed link in each control cycle within the preset period segment, the health status of each closed link at the end of the current control cycle is determined and latched, serving as the historical health conditions of each closed link in the next control cycle.

[0011] In one embodiment, the cloud-based safety-critical system is a rail transit train operation control system.

[0012] In one embodiment, the channel output parameters of the current channel in the current control cycle are determined within the current channel; the process of determining the channel output parameters of the current channel in the current control cycle includes: Obtain the authentication input and system-level execution status of the current channel in the current control cycle; Based on the authentication input of the current channel in the current control cycle, determine the functional side service output of the current channel in the current control cycle; The proof-side service output and candidate successor status of the current channel in the current control cycle are determined based on the authentication input and system-level execution status of the current channel in the current control cycle. Based on the authentication input of the current channel in the current control cycle, determine the cycle input context commitment of the current channel in the current control cycle; Based on the system-level execution state of the current channel in the current control cycle, determine the initial state commitment of the current channel in the current control cycle; Based on the proof-side service output of the current channel in the current control period, determine the proof-side output summary of the current channel in the current control period; Based on the authentication input, candidate successor states, and proof-side output digest of the current channel in the current control cycle, determine the candidate successor state commitment of the current channel in the current control cycle; The current channel's periodic input context commitment, initial state commitment, proof-side output summary, and candidate successor state commitment are written into the public part of the complete proof witness of the current control period, thus constituting a public verification instance of the current channel in the current control period; Determine the non-public portion of the complete proof witness for the current channel in the current control period; the non-public portion is used to prove that the business calculation and state transition satisfy the business constraint relationship. Based on the complete proof witness of the current channel in the current control period and the aforementioned business constraint relationship, generate the computational proof of the current channel in the current control period.

[0013] In one implementation, based on the authentication input, public verification instance, system-level execution state, computational proof, and functional output digest of the current channel in the current control cycle, the channel evidence predicate for the current channel in the current control cycle is determined, including: Based on the authentication inputs and public verification instances of the current channel in the current control cycle, determine the cycle input context consistency result of the current channel in the current control cycle; Based on the public verification instance of the current channel in the current control cycle and the system-level execution state of the current control cycle, determine the proof starting anchor consistency result of the current channel in the current control cycle; Based on the publicly available verification examples and computational proofs of the current channel in the current control period, determine the proof verification results of the current channel in the current control period; Based on the functional side output summary and the proof side output summary of the current channel in the current control cycle, determine the consistency result of the output summary on both sides of the current channel in the current control cycle; Based on the consistency results of the current channel's input context, the consistency results of the proof start anchor, the proof verification results, and the consistency results of the output summaries on both sides, determine the channel evidence predicate for the current channel in the current control period.

[0014] In one embodiment, the normalized lateness penalty for each closed link in the current control cycle is determined based on the closing time of each closed link in the current control cycle, the normal release deadline time of the current control cycle, and the grace period deadline time, including: Using the normalized lateness penalty calculation formula, the normalized lateness penalty for each closed link in the current control cycle is calculated based on the closing time of each closed link in the current control cycle, as well as the normal release deadline and grace period deadline of the current control cycle. The normalized lateness penalty calculation formula is as follows: ; in, This represents the normalized lateness penalty for the closed link r in the t-th control cycle; This represents the closing time of the closed link r in the t-th control cycle; This represents the normal release deadline for the t-th control cycle; Indicates the maximum grace period; This represents the grace period for the t-th control cycle; This indicates that the closure is not complete.

[0015] In one implementation, the final system mode for the current control cycle is determined based on the channel evidence predicates of the two secure computing channels in the current control cycle, the closure time and historical health conditions of each closed link in the current control cycle, the normal release deadline, the grace period deadline, the dual-channel functional side output digest consistency results, and the dual-channel candidate successor state commitment consistency results, including: When both security computing channels have valid channel evidence predicates in the current control period, the closure time of the system-level closed link in the current control period is less than or equal to the normal release deadline of the current control period, the historical health conditions of each closed link in the current control period allow normal operation, and the consistency results of the dual-channel functional side output digest and the dual-channel candidate successor state commitment are consistent in the current control period, the final system mode of the current control period is determined to be the normal acceptance mode. When both security computation channels have valid channel evidence predicates in the current control cycle, the closure time of the system-level closed link is greater than the normal release deadline of the current control cycle but less than or equal to the grace period deadline of the current control cycle, the historical health conditions of each closed link in the current control cycle allow normal operation, and the consistency results of the dual-channel functional side output digest and the dual-channel candidate successor state commitments in the current control cycle are consistent, state alignment processing is performed. If the state alignment result of the current control cycle is successful, the final system mode of the current control cycle is determined to be the grace period acceptance mode; if the state alignment result of the current control cycle is unsuccessful, the final system mode of the current control cycle is determined to be the rejection mode. If any secure computation channel has an invalid channel evidence predicate in the current control cycle, or the closure time of the system-level closed link is greater than the grace period deadline of the current control cycle, or any closed link is not allowed to operate normally under the historical health conditions of the current control cycle, or the consistency result of the dual-channel functional side output summary of the current control cycle is inconsistent, or the consistency result of the dual-channel candidate successor state commitment of the current control cycle is inconsistent, the final system mode of the current control cycle is determined to be the rejection mode.

[0016] In one embodiment, the state alignment process includes: Output the fault-safe result of the current control cycle and obtain the actual system-level state after executing the fault-safe result of the current control cycle; Perform a preset security recovery or state alignment process on the actual system state to obtain the aligned state of the current control cycle, and calculate the commitment value of the aligned state. The state alignment result of the current control cycle is obtained by comparing the commitment value with the candidate successor state commitments of the two secure computing channels.

[0017] In one embodiment, based on the closing time, grace period, and historical health conditions of each closed link in the current control cycle, as well as the normalized lateness penalty of each closed link in each control cycle within a preset period segment, the health status of each closed link at the end of the current control cycle is determined and latched, serving as the historical health condition of each closed link in the next control cycle, including: Based on the closing time of each closed link in the current control cycle and the grace period cutoff time, determine whether each closed link has a fault-level time anomaly, and obtain the fault-level time anomaly judgment result of each closed link in the current control cycle. Based on the normalized lateness penalty of each closed link in each control cycle within the preset period segment, the window lateness intensity of each closed link in the current control cycle is determined. Based on the normalized lateness penalty of each closed link in the current control period and the long-term lateness debt of the previous control period, the long-term lateness debt of each closed link in the current control period is determined; when the current control period is the initial control period, the long-term lateness debt of the previous control period is the initial long-term lateness debt. Based on the fault level time anomaly judgment results, window lateness intensity, long-term lateness debt, and historical health conditions of each closed link in the current control cycle, determine the health status of each closed link at the end of the current control cycle and latch it.

[0018] In one embodiment, the method further includes: Receive the raw input for the next control cycle; If the original input for the next control cycle arrives before the execution token for the next control cycle is generated, the original input for the next control cycle is stored in the replay buffer. After the execution token for the next control cycle is generated, based on the execution token for the next control cycle and the cycle number of the next control cycle, the original input in the replay buffer is rebound with the system-level execution state of the next control cycle to obtain the authentication input for the next control cycle. If the original input for the next control cycle arrives after the execution token for the next control cycle is generated, then based on the execution token and the cycle number of the next control cycle, the original input for the next control cycle is bound to the system-level execution state of the next control cycle to obtain the authentication input for the next control cycle.

[0019] In one embodiment, the method further includes: After the current control cycle ends, a cycle operation audit record for the current control cycle is generated. The cycle operation audit record includes: the cycle number of the current control cycle, the cycle input context commitment of the current control cycle, the final system mode of the current control cycle, the reason for the rejection mode of the current control cycle, the system-level state anchor in the execution token of the next control cycle, the functional output summary of the current control cycle, the proof output summary of the current control cycle, the proof verification results of each secure computing channel in the current control cycle, the health status of each closed link at the end of the current control cycle, whether replay buffering occurred in the current control cycle, whether state alignment processing was triggered in the current control cycle, and the state alignment result of the current control cycle.

[0020] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application discloses a verifiable operational output authorization method for cloud-based security-critical systems. The security gateway determines the channel evidence predicate, the consistency results of the dual-channel functional-side output digests, and the consistency results of the dual-channel candidate successor states by acquiring authentication inputs, system-level execution status, functional-side business outputs, candidate successor states, public verification examples, and computational proofs. Furthermore, by acquiring the closure time and historical health conditions of each closed link, determining the grace period deadline, and normalized late arrival penalties, the final system mode is comprehensively determined as a normal acceptance mode, a grace period acceptance mode, or a rejection mode. This expands output authorization from relying solely on output consistency to a comprehensive multi-condition judgment, enabling the identification and blocking of outputs that are consistent across both channels but have computational semantic errors. The system prevents erroneous outputs from being authorized. When the final system mode is the grace period acceptance mode, the security gateway outputs a fault-safe result, does not authorize the release of functional side service outputs, and only submits the current control cycle's state alignment result as a candidate successor state for the next control cycle's system-level execution state. When the final system mode is the rejection mode, the security gateway does not submit candidate successor states or generate execution tokens. Through these operations, the system rejects erroneous outputs and maintains the system's security state. By determining and latching the health status of each closed link as historical health conditions for the next control cycle, the output authorization determination for the next control cycle can be based on the health status latched in the previous cycle, ensuring the continuity of health status across cycles. This scheme, through the above multi-condition joint determination, prevents erroneous outputs from being authorized while ensuring the consistency of dual-channel outputs, thus improving the security of output authorization. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A schematic diagram of a verifiable operational output authorization method for a cloud-based security-critical system, provided as an embodiment of this application; Figure 2 This is a schematic diagram of the final system mode determination; Figure 3 A schematic diagram of the closed loop for state anchors, execution tokens, and replay buffers; Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The purpose of this application is to provide a verifiable operational output authorization method for cloud-based safety-critical systems, aiming to address the problem of incorrect authorization by the traditional two-out-of-two mechanism when dual-channel outputs are consistent but computational semantics are incorrect in a cloud-based virtualization environment.

[0025] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] In one exemplary embodiment, such as Figure 1 As shown, a verifiable operational output authorization method for cloud-based security-critical systems is provided. The cloud-based security-critical system includes two secure computing channels and a security gateway. The method is executed in the security gateway and includes the following steps.

[0027] Step 01: Select any secure computing channel as the current channel.

[0028] As an optional implementation method, cloud-based safety-critical systems are used in rail transit train operation control systems.

[0029] Specifically, the cloud-based security-critical system consists of a single 2-out-of-2 secure computing unit and a security gateway (SG). The 2-out-of-2 secure computing unit includes two parallel secure computing channels (Channel A and Channel B), with the security gateway connected to both channels. The security gateway is responsible for the unified maintenance and submission of functional-side business outputs, the final authorization of functional-side business outputs, and the cross-control cycle maintenance, update, and latching of the health status of the proof link. Both Channel A and Channel B include a functional executor, a prover, and an evidence output interface. The input processing components of Channel A and Channel B respectively obtain the authentication inputs of the same control cycle and the latest system-level execution status submitted by the security gateway. The functional executors of each channel process the authentication inputs along the original functional execution path, and the provers of each channel independently perform proof-side business calculations and state transition calculations based on the authentication inputs and the system-level execution status.

[0030] Step 02: Obtain the authentication input, system-level execution status, and channel output parameters of the current channel in the current control cycle.

[0031] The channel output parameters include functional-side service output, candidate successor states, public verification instances, and computational proofs; public verification instances include periodic input context commitments, initial state commitments, proof-side output summaries, and candidate successor state commitments.

[0032] As an optional implementation, the channel output parameters of the current channel in the current control cycle in step 02 are determined in the current channel; the process of determining the channel output parameters of the current channel in the current control cycle includes the following steps.

[0033] Step 0201: Obtain the authentication input and system-level execution status of the current channel in the current control cycle.

[0034] Step 0202: Based on the authentication input of the current channel in the current control cycle, determine the functional side service output of the current channel in the current control cycle.

[0035] Specifically, step 0202 is implemented in the function executor of the current channel. Function executor acquisition. Directly execute the business calculation function The original rail transit train operation control business program of the rail transit train operation control system is represented, and the functional side business output is obtained. : ; in, This indicates that channel X is in the t-th control cycle, according to pass The calculated functional side service output, X=A represents channel A, and X=B represents channel B; This represents the authentication input of channel X in the t-th control cycle (i.e., the train control authentication input used by channel X in the t-th control cycle that has completed input authentication and cycle context binding). The authentication input includes the running data and its timing identification information.

[0036] Step 0203: Determine the proof-side service output and candidate successor status of the current channel in the current control cycle based on the authentication input and system-level execution status of the current channel in the current control cycle.

[0037] Specifically, steps 0203-0210 are all implemented in the current channel's prover. The formula for determining the proof-side service output and candidate successor state is: ; in, This represents the proof-side service output of channel X in the t-th control cycle; This represents the candidate successor state of channel X in the t-th control cycle; This refers to the computational functions for proof-side rail transit train operation control deployed and running on cloud computing devices or virtualized computing environments; This represents the system-level execution state of the t-th control cycle (i.e., the system-level execution state that is verified and submitted by the security gateway after the (t-1)-th control cycle and used as the initial running state at the beginning of the t-th control cycle).

[0038] Step 0204: Based on the authentication input of the current channel in the current control cycle, determine the cycle input context commitment of the current channel in the current control cycle.

[0039] Step 0205: Based on the system-level execution state of the current channel in the current control cycle, determine the initial state commitment of the current channel in the current control cycle.

[0040] Step 0206: Based on the proof-side service output of the current channel in the current control period, determine the proof-side output summary of the current channel in the current control period.

[0041] Step 0207: Based on the authentication input, candidate successor states, and proof-side output digest of the current channel in the current control period, determine the candidate successor state commitment of the current channel in the current control period.

[0042] Step 0208: Write the current channel's periodic input context commitment, initial state commitment, proof-side output summary, and candidate successor state commitment into the public part of the complete proof witness of the current control period, thus forming a public verification instance of the current channel in the current control period.

[0043] Specifically, the formula for determining public verification instances is as follows: ; in, This represents the public verification instance of channel X in the t-th control cycle; This represents the periodic input context commitment of channel X in the t-th control cycle; This represents the initial state commitment of channel X in the t-th control cycle; This represents the candidate successor state commitment for channel X in the t-th control cycle; This represents the proof-side output summary for channel X in the t-th control cycle.

[0044] Step 0209: Determine the non-public portion of the complete proof witness for the current channel in the current control period; the non-public portion is used to prove that the business calculation and state transition satisfy the business constraint relationship.

[0045] Step 0210: Generate the computational proof of the current channel in the current control period based on the complete proof witness and business constraint relationship of the current channel in the current control period.

[0046] Specifically, the calculation proof of channel X in the t-th control cycle Used to prove that: the proof-side output summary and candidate successor state commitment written into the complete proof witness public part are authentication inputs based on the same control cycle (i.e., the t-th control cycle). System-level execution status Proof-side business output and candidate successor states It is calculated according to the business constraints corresponding to the business calculation function. The prover of channel X will calculate the proof. and public verification examples Together with the verification material for the t-th control cycle, the verifier corresponding to channel X in the security gateway.

[0047] Step 03: Based on the functional side service output of the current channel in the current control cycle, determine the functional side output summary of the current channel in the current control cycle.

[0048] Specifically, the verifier corresponding to channel X in the security gateway obtains the functional side service output generated by the functional executor. The results from the functional side service output are summarized to obtain the channel. The functional side outputs a summary during the t-th control cycle. The validator inputs the cycle input context consistency results. Prove the consistency of the initial anchor results Proof and verification results Functional side output summary and from publicly verified instances The output digest of the proof obtained in the middle and candidate successor state commitment As a channel The channel verification result in the t-th control cycle is submitted to the security gateway.

[0049] Step 04: Based on the authentication input, public verification instance, system-level execution status, computational proof, and functional output summary of the current channel in the current control cycle, determine the channel evidence predicate of the current channel in the current control cycle.

[0050] Specifically, step 04 is implemented in the authenticator corresponding to the current channel in the security gateway.

[0051] As an optional implementation, step 04 includes the following steps.

[0052] Step 041: Based on the authentication input and public verification instance of the current channel in the current control cycle, determine the periodic input context consistency result of the current channel in the current control cycle.

[0053] Specifically, the authenticator corresponding to channel X in the security gateway determines the authentication input based on the authentication input. Recalculate the periodic input context commitment for channel X in the t-th control cycle, and compare the recalculated periodic input context commitment with the public verification instance. Periodic input context commitment By comparison, the periodic input context consistency result of channel X in the t-th control cycle is obtained. When the two are consistent, ;otherwise, .

[0054] Step 042: Based on the public verification instance of the current channel in the current control period and the system-level execution status of the current control period, determine the proof starting anchor consistency result of the current channel in the current control period.

[0055] Specifically, the validator corresponding to channel X in the security gateway obtains authentication data from the public authentication instance. Obtaining initial state commitment and commit to the initial state By comparing the system-level execution state used at the start of the t-th control cycle, the consistency result of the initial anchor of channel X in the t-th control cycle is obtained. Prove the consistency of the initial anchor results. The following conditions must be met: .

[0056] Step 043: Based on the public verification examples and computational proofs of the current channel in the current control period, determine the proof verification result of the current channel in the current control period.

[0057] Specifically, the validator corresponding to channel X in the security gateway publicly verifies instances. As public input, using the verification key corresponding to the business constraints, the computation proof is performed. Perform cryptographic verification to obtain the proof verification result of channel X in the t-th control cycle. : .in, This indicates that the verifier performs computational proof verification processing using the verification key. When the proof is computed... Able to prove public verification instances When the publicly available data in the database satisfies the corresponding business constraints and state transition relationships, ;otherwise, .

[0058] Step 044: Based on the functional side output summary and the proof side output summary of the current channel in the current control cycle, determine the consistency result of the output summaries on both sides of the current channel in the current control cycle.

[0059] Specifically, the security gateway obtains the functional output summary from the channel verification result corresponding to channel X. and proof side output summary The two are compared to obtain the consistency result of the output summary on both sides of channel X in the t-th control cycle. It satisfies the following logical conditions: .

[0060] Step 045: Based on the consistency results of the current channel's periodic input context, the consistency results of the proof starting anchor, the proof verification results, and the consistency results of the output summaries on both sides, determine the channel evidence predicate for the current channel in the current control period.

[0061] Specifically, the security gateway periodically inputs context consistency results. Prove the consistency of the initial anchor results Proof and verification results Consistency results with output summaries from both sides Determine the channel evidence predicate for channel X in the t-th control period. : ; in, Represents the logical AND operator.

[0062] When the periodic input context consistency result Prove the consistency of the initial anchor results Proof and verification results Consistency results with output summaries from both sides When both are 1, the channel evidence predicate The security gateway determines the channel evidence for channel X in the t-th control cycle to be valid. When any constituent item is 0, the channel evidence predicate... The security gateway will determine that the channel evidence for channel X in the t-th control cycle is invalid.

[0063] Step 05: Based on the functional side output summary of the two secure computing channels in the current control cycle, determine the consistency result of the dual-channel functional side output summary in the current control cycle.

[0064] Specifically, the security gateway outputs a summary to the functional side of channel A during the t-th control cycle. The functional side output summary of channel B in the t-th control cycle The comparison yields the consistency result of the dual-channel functional side output summary for the t-th control cycle. It satisfies: .

[0065] when When, the dual-channel functional side output summary consistency result of the t-th control cycle is consistent; when At that time, the consistency result of the dual-channel functional side output summary in the t-th control cycle is inconsistent.

[0066] Step 06: Based on the candidate successor state commitments of the two secure computing channels in the current control cycle, determine the consistency result of the dual-channel candidate successor state commitments in the current control cycle.

[0067] Specifically, the security gateway commits to the candidate successor state of channel A in the t-th control cycle. The candidate successor state commitment of channel B in the t-th control period By comparing the results, the consistency of the dual-channel candidate successor state commitments in the t-th control cycle is obtained. It satisfies: .

[0068] when When, the dual-channel candidate successor state commitment consistency result of the t-th control cycle is consistent; when When the dual-channel candidate successor state commitment consistency result of the t-th control cycle is inconsistent.

[0069] Step 07: Obtain the control cycle duration, lateness tolerance ratio, normal release deadline of the current control cycle, closure time of each closed link in the current control cycle, and historical health conditions.

[0070] Specifically, the closed link includes the closed link for proving channel A, the closed link for proving channel B, and the system-level output authorization closed link of the security gateway, i.e. Channel A indicates a closed link. Channel B indicates a closed link. This indicates a closed-loop system-level output authorization for the security gateway.

[0071] Historical health conditions are expressed as follows: ; in, This indicates the historical health condition of the closed link r in the t-th control cycle; HEALTHY indicates that the corresponding closed link is allowed to continue participating in the normal output authorization determination (allowing normal operation); DEGRADED indicates that the corresponding closed link enters a degraded state due to long-term late debt or the window lateness intensity reaching a preset threshold; FAULT indicates that the corresponding closed link enters a fault state due to deterministic error, timeout rejection, or fault latch.

[0072] After the system completes trusted initialization and startup self-test, the security gateway establishes initial historical health conditions. , and At the start of the t-th control cycle, the security gateway reads the health status of each closed link latched at the end of the (t-1)-th control cycle from the cross-cycle health status record maintained by the security gateway itself. , and This is used as the historical health condition for the t-th control cycle. Normal authorization is allowed when the historical health condition of each closed link is HEALTHY; otherwise, normal authorization is not allowed.

[0073] Step 08: Based on the control cycle duration, lateness tolerance ratio, and the normal release deadline of the current control cycle, determine the grace period deadline of the current control cycle.

[0074] Specifically, the formula for calculating the grace period deadline includes: ; ; in, This indicates the maximum grace period during which late evidence can be used for state alignment processing; This indicates the tolerance rate for lateness; This indicates the duration of the control cycle of the rail transit train operation control system; This represents the grace period for the t-th control cycle; This represents the normal release deadline for the t-th control cycle.

[0075] Step 09: Determine the normalized late penalty for each closed link in the current control cycle based on the closing time of each closed link in the current control cycle, as well as the normal release deadline and grace period deadline of the current control cycle.

[0076] As an optional implementation, step 09 includes the following steps.

[0077] Using the normalized lateness penalty calculation formula, the normalized lateness penalty for each closed link in the current control cycle is calculated based on the closing time of each closed link in the current control cycle, as well as the normal release deadline and grace period deadline of the current control cycle. The normalized lateness penalty calculation formula is as follows: ; in, This represents the normalized lateness penalty for the closed link r in the t-th control cycle; This represents the closing time of the closed link r in the t-th control cycle; This represents the normal release deadline for the t-th control cycle; Indicates the maximum grace period; This represents the grace period for the t-th control cycle; This indicates that the closure is not complete.

[0078] Specifically, A value of 1 indicates a fault-level time anomaly where the closed link r experiences a closure failure or exceeds the grace period in the t-th control cycle.

[0079] Step 10: Based on the channel evidence predicates of the two secure computing channels in the current control cycle, the closing time and historical health conditions of each closed link in the current control cycle, the normal release deadline, grace period deadline, dual-channel functional side output digest consistency results, and dual-channel candidate successor state commitment consistency results, determine the final system mode of the current control cycle.

[0080] When a candidate for the grace period acceptance mode is initially determined, a state alignment process is performed, and the final system mode is determined based on the state alignment result. The final system mode is the normal acceptance mode, the grace period acceptance mode, or the rejection mode.

[0081] As an optional implementation method, such as Figure 2 As shown, step 10 includes the following steps.

[0082] Step 101: When both security computing channels have valid channel evidence predicates in the current control period, the closure time of the system-level closed link in the current control period is less than or equal to the normal release deadline of the current control period, the historical health conditions of each closed link in the current control period allow normal operation, and the consistency results of the dual-channel functional side output digest and the dual-channel candidate successor state commitments in the current control period are consistent, the final system mode of the current control period is determined to be the normal acceptance mode NORMAL_ACCEPT.

[0083] Step 102: When both security computation channels have valid channel evidence predicates in the current control cycle, the closure time of the system-level closed link is greater than the normal release deadline of the current control cycle and less than or equal to the grace period deadline of the current control cycle, the historical health conditions of each closed link in the current control cycle allow normal operation, and the consistency results of the dual-channel functional side output digest and the dual-channel candidate successor state commitment are consistent in the current control cycle, perform state alignment processing; if the state alignment result of the current control cycle is successful, the final system mode of the current control cycle is determined to be the grace period state acceptance mode GRACE_STATE_ACCEPT; if the state alignment result of the current control cycle is unsuccessful, the final system mode of the current control cycle is determined to be the rejection mode REJECT.

[0084] As an optional implementation, step 102, the state alignment process, includes the following steps.

[0085] Step 1021: Output the fault safety result of the current control cycle and obtain the actual system-level state after executing the fault safety result of the current control cycle.

[0086] Step 1022: Perform a preset security recovery or state alignment process on the actual system state to obtain the aligned state of the current control cycle, and calculate the commitment value of the aligned state.

[0087] Step 1023: Compare the commitment value with the candidate successor state commitments of the two security calculation channels to obtain the state alignment result of the current control cycle.

[0088] Specifically, the formula for calculating the state alignment result is as follows: ; in, This represents the state alignment result in the t-th control cycle; This indicates the preset safety recovery or state alignment process used in the t-th control cycle; This represents the commitment computation function that transforms the running state into a state commitment.

[0089] Step 103: When any secure computation channel's channel evidence predicate is invalid in the current control cycle, or the closure time of the system-level closed link is greater than the grace period deadline of the current control cycle, or any closed link is not allowed to operate normally under the historical health conditions of the current control cycle, or the consistency result of the dual-channel functional side output summary of the current control cycle is inconsistent, or the consistency result of the dual-channel candidate successor state commitment of the current control cycle is inconsistent, the final system mode of the current control cycle is determined to be the rejection mode REJECT.

[0090] Step 11: Based on the final system mode of the current control cycle, execute the corresponding output processing and status submission operations.

[0091] Specifically, when the final system mode of the current control cycle is the normal acceptance mode, the functional-side service outputs of the current control cycle that are consistent across the two secure computing channels are authorized for release. The candidate successor states of the current control cycle for the two secure computing channels are submitted as the system-level execution state for the next control cycle, and an execution token for the next control cycle is generated. When the final system mode of the current control cycle is the grace period acceptance mode, the fault-safe result of the current control cycle is output. The functional-side service outputs of the current control cycle are not authorized for release. The candidate successor states that have passed the state alignment result of the current control cycle are submitted as the system-level execution state for the next control cycle, and an execution token for the next control cycle is generated. When the final system mode of the current control cycle is the rejection mode, the functional-side service outputs of the current control cycle are not authorized for release. No candidate successor states are submitted, and no execution token for the next control cycle is generated.

[0092] Step 12: Based on the closing time, grace period, and historical health conditions of each closed link in the current control cycle, as well as the normalized lateness penalty of each closed link in each control cycle within the preset period segment, determine and latch the health status of each closed link at the end of the current control cycle, so as to serve as the historical health conditions of each closed link in the next control cycle.

[0093] As an optional implementation, step 12 includes the following steps.

[0094] Step 121: Based on the closing time and grace period of each closed link in the current control cycle, determine whether each closed link has a fault-level time anomaly, and obtain the fault-level time anomaly judgment result of each closed link in the current control cycle.

[0095] Step 122: Determine the window delay intensity of each closed link in the current control cycle based on the normalized delay penalty of each closed link in each control cycle within the preset period segment.

[0096] Specifically, the formula for calculating window lateness intensity is: ; in, This represents the window delay intensity of closed link r in the most recent W control cycles up to the t-th control cycle, i.e., the window delay intensity of closed link r in the t-th control cycle. Indicates the number of control cycles contained in the sliding window (preset cycle segment); Indicates the number of the control cycle involved in the summation within the sliding window; This represents the normalized lateness penalty for the closed link r in the i-th control cycle.

[0097] Step 123: Determine the long-term late debt of each closed link in the current control period based on the normalized late penalty of each closed link in the current control period and the long-term late debt of the previous control period; when the current control period is the initial control period, the long-term late debt of the previous control period is the initial long-term late debt.

[0098] Specifically, the formula for calculating long-term late payment debt is as follows: ; in, This represents the long-term late debt accumulated by the closed link r after the end of the t-th control period, i.e., the long-term late debt of the closed link r in the t-th control period. This represents the long-term late debt accumulated in the closed link r after the (t-1)th control cycle ends; This indicates the credit recovery rate used when the closed link r completes the closure on time. All closed links use the same credit recovery rate.

[0099] When any closed link meets any of the following conditions, the security gateway will set the corresponding link to a degraded state and latch it: or .

[0100] in, This indicates a long-overdue debt ceiling that allows for normal operations; This indicates the maximum allowed window delay intensity for normal operation.

[0101] Step 124: Based on the fault level time anomaly judgment results, window lateness intensity, long-term lateness debt and historical health conditions of each closed link in the current control cycle, determine the health status of each closed link at the end of the current control cycle and latch it.

[0102] Specifically, when or If the fault-level time anomaly judgment result of closed link r in the t-th control cycle is abnormal, the security gateway will record the health status of closed link r after the end of the t-th control cycle. Set it directly to FAULT and latch it, instead of redetermining the health status based on the long-term late debt or window lateness intensity in the t-th control period.

[0103] If there are no deterministic errors (proof verification failure, input authentication error, proof initial state anchor inconsistency, inconsistent output digests on both sides, or conflicting results between two channels are deterministic errors), the window lateness intensity and long-term lateness debt have not reached the corresponding thresholds, and there are no unreleased health state latches, then... Set HEALTHY; when the window lateness intensity or long-term lateness debt reaches the corresponding threshold, Set DEGRADED; when a proof verification fails, the cycle input context consistency check fails, the state anchor is inconsistent, the output digest is inconsistent, there is a dual-channel conflict, or the closure is not completed after the grace period deadline, the corresponding event will be set. Set FAULT.

[0104] The latched DEGRADED or FAULT state will only be released after the preset recovery conditions are met and the corresponding recovery process is completed; it will not be automatically released due to the successful closure of a subsequent single control cycle. Write the cross-cycle health status record maintained by the security gateway to form the historical health conditions of the (t+1)th control cycle, without retroactively changing the system mode determination result of the already completed tth control cycle.

[0105] As an optional implementation method, such as Figure 3 As shown, after step 12, the method further includes the following steps.

[0106] Step S11: Receive the raw input for the next control cycle.

[0107] Step S12: If the original input of the next control cycle arrives before the execution token of the next control cycle is generated, the original input of the next control cycle is stored in the replay buffer; after the execution token of the next control cycle is generated, based on the execution token of the next control cycle and the cycle number of the next control cycle, the original input in the replay buffer is rebound with the system-level execution state of the next control cycle to obtain the authentication input of the next control cycle.

[0108] Step S13: If the original input of the next control cycle arrives after the execution token of the next control cycle is generated, then based on the execution token of the next control cycle and the cycle number of the next control cycle, the original input of the next control cycle is bound to the system-level execution state of the next control cycle to obtain the authentication input of the next control cycle.

[0109] Specifically, the security gateway determines the effective system state after the t-th control cycle ends and calculates the corresponding state anchor for use in the (t+1)-th control cycle. The initial state commitments of channel A and channel B in the (t+1)-th normal control cycle should be consistent with the latest system-level state anchor submitted by the security gateway. ; in, This represents the initial state commitment of channel A in the (t+1)th control cycle; This represents the initial state commitment of channel B in the (t+1)th control cycle; This represents a system-level state anchor point submitted by the security gateway after the end of the t-th control cycle and used in the (t+1)-th control cycle.

[0110] As an optional implementation, after step 12, the method further includes the following steps.

[0111] Step S21: After the current control cycle ends, generate the cycle operation audit record for the current control cycle. The cycle operation audit record includes: the cycle number of the current control cycle, the cycle input context commitment of the current control cycle, the final system mode of the current control cycle, the reason for the rejection mode of the current control cycle, the system-level state anchor in the execution token of the next control cycle, the functional output summary of the current control cycle, the proof output summary of the current control cycle, the proof verification results of each secure computing channel in the current control cycle, the health status of each closed link at the end of the current control cycle, whether replay buffering occurred in the current control cycle, whether the state alignment process of the current control cycle was triggered, and the state alignment result of the current control cycle.

[0112] Furthermore, this application also describes specific embodiments.

[0113] (1) In this embodiment, taking the t-th control cycle as an example, channel A and channel B respectively execute the train control security calculation program deployed and running on the cloud, and respectively generate the business output, public verification instance, state commitment, output summary and verifiable calculation proof for this control cycle. The security gateway saves the system state anchor points submitted in the previous control cycle. And receive evidence data sent by channel A and channel B during the t-th control period.

[0114] Under normal circumstances, the control cycle sequence number sent by both channel A and channel B is t, proving that the starting point is equal to t. If the authentication input matches the publicly verified instance, the SNARK verification passes, the function execution output digest matches the verifiable relationship output digest, and the service outputs and subsequent states of both channels are consistent, then the security gateway determines that the control cycle meets the normal output authorization conditions and allows the release of the function-side service output.

[0115] In the first type of fault, the service output of channel A is tampered with on a single channel, causing the output of channel A to be altered. With channel B output Inconsistency. After receiving the dual-channel output, the security gateway performs a consistency check on the dual-channel output and finds... Therefore, the function side service output is rejected, and the fault is converted into a rejection output event.

[0116] In the second type of fault, the same deterministic perturbation is injected into both channel A and channel B simultaneously, causing the two channels to produce consistent but erroneous service outputs. At this point, a traditional two-out-of-two output comparison might determine that the two channel outputs are consistent. The security gateway continues to verify the verifiable computation proofs for both channels and checks whether the function execution output digest matches the verifiable relation output digest. Due to erroneous output... If the pre-coded train control security calculation relationship, proof validity, and output digest binding conditions cannot be met simultaneously, the security gateway determines that the proof is invalid or the output digest binding has failed, and therefore rejects the functional side business output.

[0117] In the third type of failure, the service output remains unchanged, but the verifiable computational proof sent by channel A is changed. Replace with invalid proof. The security gateway calls the proof verification module to verify... The verification was performed, and the verification result was unsuccessful. Therefore, the evidence for channel A was deemed invalid, and the business output of the control cycle function was rejected.

[0118] In the fourth type of fault, the function execution output summary is enabled. Output summary of verifiable relations Inconsistency. The security gateway compares the output digest of the function execution with the output digest obtained from the verifiable relation calculation. Finding an inconsistency, it indicates that the service output is not correctly bound to the verifiable relation, and therefore rejects the service output from the function side.

[0119] In the fifth type of fault, the proof starting state of channel A or channel B is rolled back to the historical state anchor point. Where k≥2. The security gateway compares the proof starting point with the system state anchor point submitted in the previous control cycle. If they are inconsistent, the state anchor point continuity check is deemed to have failed. Therefore, the functional side service output is rejected to prevent historical states from being reused as the execution starting point of the current control cycle.

[0120] In the sixth type of fault, the control cycle sequence number is set to either a repeat sequence number t-1 or a jump sequence number t+2. The security gateway compares the received control cycle sequence number with its maintained expected control cycle sequence number t. If the control cycle sequence number is inconsistent, it determines that the input context binding has failed and therefore rejects the functional side service output to prevent replay data or out-of-order data from entering the normal output authorization process.

[0121] As can be seen from the above embodiments, single-channel output tampering can be detected by traditional two-out-of-two output comparison; however, two-channel inconsistency errors, state rollback, and control cycle context replay may bypass traditional two-out-of-two output comparison. This application converts the above-mentioned faults into output rejection, degraded operation, or failsafe events by means of proof validity checks, output digest binding, state anchor continuity checks, and input context binding.

[0122] (2) In this embodiment, the control cycle is set to The normal output deadline is The grace period ends at [date]. Set a threshold for long-term late debt. The threshold value is 3, representing the window lateness intensity threshold. The credit recovery rate is 2. The value is 1, and the sliding window length W is 5. The security gateway records the proof closure time of channel A, channel B, and the system-level closure process, and updates the late debt and window lateness intensity according to the lateness situation.

[0123] In the first scenario, no additional delay is injected. Evidence processing for both Channel A and Channel B occurs within the normal output deadline. Previously completed, with consistent dual-channel outputs, valid proof, continuous state anchors, and correct output summary binding, the security gateway determines that this control cycle meets the normal output authorization conditions. The system enters the NORMAL_ACCEPT state and releases the functional side service output.

[0124] In the second scenario, a short-term scheduling delay is injected to ensure that evidence processing occurs within the normal output deadline. Completed later, but still within the grace period deadline. Previously completed. The security gateway determines that the control cycle has missed the normal output deadline, and therefore does not trace the authorized function side business output for this control cycle; if the state anchor, proof, and output summary are all checked correctly, the late evidence is used for state alignment and audit logging, and the system enters the GRACE_STATE_ACCEPT state.

[0125] In the third scenario, the scheduling delay is further increased, causing the evidence processing time to exceed the grace period deadline. The security gateway determines that the evidence closure for the control cycle has timed out, rejects the service output from the function side of the control cycle, and records the event in the late health assessment process.

[0126] In the fourth scenario, multiple consecutive control cycles are slightly delayed, causing evidence processing in each control cycle to be slightly later than the normal output deadline. However, it did not exceed the grace period deadline. The security gateway accumulates long-term lateness debt for each late arrival. When the long-term lateness debt reaches a threshold... When the value is 3, the security gateway determines that the closed link is in an unhealthy state, the system enters the DEGRADED state, and performs degradation control.

[0127] In the fifth scenario, multiple sudden late arrivals are injected within a sliding window W=5. The security gateway calculates the late arrival intensity within this window, and when the late arrival intensity reaches a threshold... When the value is 2, the security gateway determines that there is a risk of concentrated late arrivals within a short period of time, the system enters the DEGRADED state, and performs degradation control.

[0128] In the sixth scenario, the input for the next control cycle arrives at the security gateway prematurely, before the current control cycle has received an execution token issued by the security gateway. The security gateway does not allow the channel to directly initiate the next control cycle computation based on its locally submitted state. Instead, it first places the prematurely arrived input into a replay buffer. After the current control cycle completes its state submission and generates an execution token, the security gateway rebinds the buffered input to the new execution state, and then allows the next control cycle to enter the normal evidence generation and output authorization process.

[0129] As can be seen from the above embodiments, this application not only rejects outputs for proof verification failures, inconsistent output digests, discontinuous state anchors, and control cycle context errors, but also imposes time boundary constraints and health status assessments on proof delays, proof missing, continuous minor delays, and sudden delays, thereby preventing cloud-deployed programs from continuing to release functional side business outputs when the proof link is unstable or the state is unauthorized.

[0130] Advantages of this application: (1) It can detect and block dual-channel consistency error output caused by defects in virtualized shared environment or common source software, and make up for the defect of traditional two-out-of-two mechanism that only checks output consistency and does not check the correctness of computational semantics; (2) The normal control cycle can only start from the state that the security gateway has authenticated and submitted, which can prevent unsubmitted states, state rollbacks, state jumps and cross-cycle historical breaks. (3) Strictly distinguish between the three modes: normal acceptance mode, grace state acceptance mode and rejection mode. This can absorb short-term scheduling jitter and avoid late proofs from retroactively authorizing outputs that have missed the deadline. (4) Transform persistent and sudden lateness into quantifiable health indicators, identify performance degradation before the closed loop fails completely, and ensure the stability of the handling through a latching mechanism; (5) Output proof binding chain can discover post-event modifications and selective omissions in the operation history, improving the credibility of security demonstration, incident tracing and operation audit; (6) The proof generation can be done in a bypass mode without changing the original main business execution path of the dual channels; the proof verification and security gateway side closure overhead is low, which is suitable for periodic security-critical control tasks.

[0131] In one exemplary embodiment, a computer device is provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a verifiable operational output authorization method for a cloud-based security-critical system.

[0132] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements a verifiable operational output authorization method for cloud-based security-critical systems.

[0133] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements a verifiable runtime output authorization method for cloud-based security-critical systems.

[0134] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a verifiable operational output authorization method for cloud-based safety-critical systems.

[0135] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0136] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0137] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0139] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0140] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A verifiable runtime output authorization method for cloud-based security-critical systems, characterized in that, The cloud-based security-critical system includes two secure computing channels and a security gateway. The method is executed in the security gateway and includes: Designate any secure computing channel as the current channel; Obtain the authentication input, system-level execution status, and channel output parameters of the current channel in the current control cycle; the channel output parameters include functional-side service output, candidate successor states, public verification instances, and computational proofs; the public verification instances include cycle input context commitments, initial state commitments, proof-side output summaries, and candidate successor state commitments; Based on the functional side service output of the current channel in the current control cycle, determine the functional side output summary of the current channel in the current control cycle; Based on the authentication input, public verification instance, system-level execution status, computational proof, and functional side output summary of the current channel in the current control cycle, determine the channel evidence predicate of the current channel in the current control cycle; Based on the functional side output summary of the two secure computing channels in the current control cycle, determine the consistency result of the dual-channel functional side output summary in the current control cycle; Based on the candidate successor state commitments of the two secure computing channels in the current control cycle, the consistency result of the dual-channel candidate successor state commitments in the current control cycle is determined. Acquire control cycle duration, lateness tolerance ratio, normal release deadline of the current control cycle, closure time of each closed link in the current control cycle, and historical health conditions; Based on the control cycle duration, the lateness tolerance ratio, and the normal release deadline of the current control cycle, determine the grace period deadline of the current control cycle. Based on the closing time of each closed link in the current control cycle and the normal release deadline and grace period deadline of the current control cycle, the normalized lateness penalty of each closed link in the current control cycle is determined. Based on the channel evidence predicates of the two secure computing channels in the current control cycle, the closure time and historical health conditions of each closed link in the current control cycle, the normal release deadline, the grace period deadline, the dual-channel functional side output digest consistency results, and the dual-channel candidate successor state commitment consistency results, the final system mode of the current control cycle is determined. Specifically, when initially determined to be a candidate for the grace period acceptance mode, state alignment processing is performed, and the final system mode is determined based on the state alignment results. The final system mode can be a normal acceptance mode, a grace period acceptance mode, or a rejection mode. Based on the final system mode of the current control cycle, execute the corresponding output handling and status submission operations. Specifically, when the final system mode of the current control cycle is the normal acceptance mode, authorize the release of the functional-side service outputs of the current control cycle that are consistent between the two secure computing channels, submit the candidate successor states of the current control cycle of the two secure computing channels as the system-level execution state of the next control cycle, and generate an execution token for the next control cycle. When the final system mode of the current control cycle is the grace period acceptance mode, output the fault safety result of the current control cycle, do not authorize the release of the functional-side service outputs of the current control cycle, submit the candidate successor states after the status alignment result of the current control cycle is passed as the system-level execution state of the next control cycle, and generate an execution token for the next control cycle. When the final system mode of the current control cycle is the rejection mode, refuse to authorize the release of the functional-side service outputs of the current control cycle, do not submit candidate successor states, and do not generate an execution token for the next control cycle. Based on the closing time, grace period, and historical health conditions of each closed link in the current control cycle, as well as the normalized lateness penalty of each closed link in each control cycle within the preset period segment, the health status of each closed link at the end of the current control cycle is determined and latched, serving as the historical health conditions of each closed link in the next control cycle.

2. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 1, characterized in that, The cloud-based safety-critical system is the rail transit train operation control system.

3. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 1, characterized in that, The channel output parameters for the current channel in the current control cycle are determined within the current channel; The process of determining the channel output parameters for the current channel in the current control cycle includes: Obtain the authentication input and system-level execution status of the current channel in the current control cycle; Based on the authentication input of the current channel in the current control cycle, determine the functional side service output of the current channel in the current control cycle; The proof-side service output and candidate successor status of the current channel in the current control cycle are determined based on the authentication input and system-level execution status of the current channel in the current control cycle. Based on the authentication input of the current channel in the current control cycle, determine the cycle input context commitment of the current channel in the current control cycle; Based on the system-level execution state of the current channel in the current control cycle, determine the initial state commitment of the current channel in the current control cycle; Based on the proof-side service output of the current channel in the current control period, determine the proof-side output summary of the current channel in the current control period; Based on the authentication input, candidate successor states, and proof-side output digest of the current channel in the current control cycle, determine the candidate successor state commitment of the current channel in the current control cycle; The current channel's periodic input context commitment, initial state commitment, proof-side output summary, and candidate successor state commitment are written into the public part of the complete proof witness of the current control period, thus constituting a public verification instance of the current channel in the current control period; Determine the non-public portion of the complete proof witness for the current channel in the current control period; the non-public portion is used to prove that the business calculation and state transition satisfy the business constraint relationship. Based on the complete proof witness of the current channel in the current control period and the aforementioned business constraint relationship, generate the computational proof of the current channel in the current control period.

4. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 3, characterized in that, Based on the authentication inputs, public verification instances, system-level execution status, computational proofs, and functional-side output summaries of the current channel in the current control cycle, determine the channel evidence predicates for the current channel in the current control cycle, including: Based on the authentication inputs and public verification instances of the current channel in the current control cycle, determine the cycle input context consistency result of the current channel in the current control cycle; Based on the public verification instance of the current channel in the current control cycle and the system-level execution state of the current control cycle, determine the proof starting anchor consistency result of the current channel in the current control cycle; Based on the publicly available verification examples and computational proofs of the current channel in the current control period, determine the proof verification results of the current channel in the current control period; Based on the functional side output summary and the proof side output summary of the current channel in the current control cycle, determine the consistency result of the output summary on both sides of the current channel in the current control cycle; Based on the consistency results of the current channel's input context, the consistency results of the proof start anchor, the proof verification results, and the consistency results of the output summaries on both sides, determine the channel evidence predicate for the current channel in the current control period.

5. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 1, characterized in that, Based on the closing time of each closed link in the current control cycle, as well as the normal release deadline and grace period deadline of the current control cycle, the normalized lateness penalty for each closed link in the current control cycle is determined, including: Using the normalized lateness penalty calculation formula, the normalized lateness penalty for each closed link in the current control cycle is calculated based on the closing time of each closed link in the current control cycle, as well as the normal release deadline and grace period deadline of the current control cycle. The normalized lateness penalty calculation formula is as follows: ; in, This represents the normalized lateness penalty for the closed link r in the t-th control cycle; This represents the closing time of the closed link r in the t-th control cycle; This represents the normal release deadline for the t-th control cycle; Indicates the maximum grace period; This represents the grace period for the t-th control cycle; This indicates that the closure is not complete.

6. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 4, characterized in that, Based on the channel evidence predicates of the two secure computing channels in the current control cycle, the closure time and historical health conditions of each closed link in the current control cycle, the normal release deadline, grace period deadline, dual-channel functional side output digest consistency results, and dual-channel candidate successor state commitment consistency results, the final system mode of the current control cycle is determined, including: When both security computing channels have valid channel evidence predicates in the current control period, the closure time of the system-level closed link in the current control period is less than or equal to the normal release deadline of the current control period, the historical health conditions of each closed link in the current control period allow normal operation, and the consistency results of the dual-channel functional side output digest and the dual-channel candidate successor state commitment are consistent in the current control period, the final system mode of the current control period is determined to be the normal acceptance mode. When both security computation channels have valid channel evidence predicates in the current control cycle, the closure time of the system-level closed link is greater than the normal release deadline of the current control cycle but less than or equal to the grace period deadline of the current control cycle, the historical health conditions of each closed link in the current control cycle allow normal operation, and the consistency results of the dual-channel functional side output digest and the dual-channel candidate successor state commitments in the current control cycle are consistent, state alignment processing is performed. If the state alignment result of the current control cycle is successful, the final system mode of the current control cycle is determined to be the grace period acceptance mode; if the state alignment result of the current control cycle is unsuccessful, the final system mode of the current control cycle is determined to be the rejection mode. If any secure computation channel has an invalid channel evidence predicate in the current control cycle, or the closure time of the system-level closed link is greater than the grace period deadline of the current control cycle, or any closed link is not allowed to operate normally under the historical health conditions of the current control cycle, or the consistency result of the dual-channel functional side output summary of the current control cycle is inconsistent, or the consistency result of the dual-channel candidate successor state commitment of the current control cycle is inconsistent, the final system mode of the current control cycle is determined to be the rejection mode.

7. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 6, characterized in that, State alignment processing includes: Output the fault-safe result of the current control cycle and obtain the actual system-level state after executing the fault-safe result of the current control cycle; Perform a preset security recovery or state alignment process on the actual system state to obtain the aligned state of the current control cycle, and calculate the commitment value of the aligned state. The state alignment result of the current control cycle is obtained by comparing the commitment value with the candidate successor state commitments of the two secure computing channels.

8. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 1, characterized in that, Based on the closing time, grace period, and historical health conditions of each closed link in the current control cycle, as well as the normalized lateness penalty of each closed link in each control cycle within the preset cycle segment, the health status of each closed link at the end of the current control cycle is determined and latched, serving as the historical health conditions for each closed link in the next control cycle, including: Based on the closing time of each closed link in the current control cycle and the grace period cutoff time, determine whether each closed link has a fault-level time anomaly, and obtain the fault-level time anomaly judgment result of each closed link in the current control cycle. Based on the normalized lateness penalty of each closed link in each control cycle within the preset period segment, the window lateness intensity of each closed link in the current control cycle is determined. Based on the normalized lateness penalty of each closed link in the current control period and the long-term lateness debt of the previous control period, the long-term lateness debt of each closed link in the current control period is determined; when the current control period is the initial control period, the long-term lateness debt of the previous control period is the initial long-term lateness debt. Based on the fault level time anomaly judgment results, window lateness intensity, long-term lateness debt, and historical health conditions of each closed link in the current control cycle, determine the health status of each closed link at the end of the current control cycle and latch it.

9. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 6, characterized in that, The method further includes: Receive the raw input for the next control cycle; If the original input for the next control cycle arrives before the execution token for the next control cycle is generated, the original input for the next control cycle is stored in the replay buffer. After the execution token for the next control cycle is generated, based on the execution token for the next control cycle and the cycle number of the next control cycle, the original input in the replay buffer is rebound with the system-level execution state of the next control cycle to obtain the authentication input for the next control cycle. If the original input for the next control cycle arrives after the execution token for the next control cycle is generated, then based on the execution token and the cycle number of the next control cycle, the original input for the next control cycle is bound to the system-level execution state of the next control cycle to obtain the authentication input for the next control cycle.

10. The verifiable operational output authorization method for cloud-based security-critical systems according to claim 9, characterized in that, The method further includes: After the current control cycle ends, a cycle operation audit record for the current control cycle is generated. The cycle operation audit record includes: the cycle number of the current control cycle, the cycle input context commitment of the current control cycle, the final system mode of the current control cycle, the reason for the rejection mode of the current control cycle, the system-level state anchor in the execution token of the next control cycle, the functional output summary of the current control cycle, the proof output summary of the current control cycle, the proof verification results of each secure computing channel in the current control cycle, the health status of each closed link at the end of the current control cycle, whether replay buffering occurred in the current control cycle, whether state alignment processing was triggered in the current control cycle, and the state alignment result of the current control cycle.