METHOD AND SYSTEM FOR THE SECURE EXECUTION OF CONTROL APPLICATIONS AND VERIFICATION DEVICE

DE502022002134D1Inactive Publication Date: 2024-12-05SIEMENS AG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
DE502022002134
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2021-05-26
Filing Date
2022-05-04
Publication Date
2024-12-05
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Industrial automation systems face challenges in securely executing control applications due to vulnerabilities in program code, leading to potential false alarms and costly disruptions, as existing exploit protection concepts are not applicable and often result in unnecessary process interruptions.

Method used

A method is introduced where a program flow control device monitors deviations in control application sequences and triggers events, reporting them to a separate checking device for analysis using updatable test rules, allowing the system to transition into a safe operating state without immediate shutdown, enabling reliable protection from attacks and reducing false alarms.

Benefits of technology

This approach ensures reliable protection of industrial automation devices from attacks by accurately assessing potential threats and minimizing disruptions, allowing controlled operation or safe shutdown, thus reducing costly interruptions and safety risks.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader
Need to check novelty before this filing date? Find Prior Art

Description

[0001] The present invention relates to a method and system for the secure execution of control applications, in particular within an industrial automation system, and to a verification device. The system and the verification device are intended for implementing the method according to the invention.

[0002] Industrial automation systems typically comprise a multitude of automation devices interconnected via an industrial communication network. They are used to control, regulate, or monitor systems, machines, or devices within the context of manufacturing or process automation or another automation domain. Due to the time-critical conditions in industrial automation systems, real-time communication protocols such as PROFINET, PROFIBUS, Real-Time Ethernet, or Time-Sensitive Networking (TSN) are predominantly used for communication between automation devices. In particular, control services or applications can be automated and distributed among currently available servers or virtual machines of an industrial automation system, depending on the load.

[0003] WO 2014 / 109645 A1 relates to a method for monitoring an industrial control system, in which data is collected from one or more sources external to the industrial control system. Furthermore, data is collected from one or more internal sources of the industrial control system. Furthermore, the data collected from the internal sources or from the external sources is aggregated. The collected data is further analyzed by correlating it with previously collected data to monitor a security status of the industrial control system.

[0004] WO 2020 / 182627 A1 describes a method for monitoring the integrity of an industrial cyber-physical system, in which measurement data acquired with various sensors of the cyber-physical system or control data intended for various actuators of the cyber-physical system are provided or retrieved. Furthermore, at least one measurement data correlation parameter between the measurement data acquired with the various sensors is determined, or at least one control data correlation parameter between the control data intended for the various actuators is determined. The at least one measurement data correlation parameter is compared with a measurement data correlation reference, or the at least one control data correlation parameter is compared with a control data correlation reference. Based on the comparison, the integrity of the cyber-physical system to be monitored is assessed.

[0005] WO 2020 / 212051 A1 discloses an industrial automation device comprising a monitoring unit for checking and monitoring the integrity state of the industrial automation device. Furthermore, at least one device component is provided that communicates non-interactively with the monitoring unit via a communication connection. The non-interactive communication comprises providing at least one device component parameter from the device component to the monitoring unit. The monitoring unit is configured to log and process the provided device component parameter of the device component of the industrial automation device for checking and monitoring the integrity state of the industrial automation device.Furthermore, the monitoring unit is designed to log or provide the integrity state of the industrial automation device as a result of the processed device component parameter of the device component of the industrial automation device. Furthermore, the monitoring unit is designed as a trusted device component in the industrial automation device, protected against manipulation by a tamper protection feature.

[0006] From the earlier international patent application WO 2021 / 164911 A1, a computer-assisted method for monitoring the integrity status of a computing device is known. In this method, an initialization value for an overall integrity value is first defined. The overall integrity value is then stored in a trusted storage area of ​​a storage unit. Subsequently, an integrity status is determined by an integrity detection system and an integrity status value is generated. The integrity status value is then stored in the overall integrity value in the trusted storage area as an integrity status.

[0007] To prevent the exploitation of existing vulnerabilities in program code, compiler-based exploit protection concepts can be used, for example, which insert additional protection or verification routines into the program code during the compilation process. Such protection concepts can be implemented purely software-based or with hardware support. In particular, these protection concepts can detect threats such as code injection, buffer overflows, return-oriented programming (ROP), and jump-oriented programming (JOP) at runtime. Upon detection of a potential attack, for example, an error message is displayed in a command line or program execution is immediately stopped. This makes exploiting existing vulnerabilities more difficult.Typically, methods for detecting attacks and measures for responding to detected attacks are defined during the compilation process and are therefore no longer changeable at runtime.

[0008] Despite the application of protective measures, malicious manipulation of program code for control applications provided by industrial automation devices cannot always be prevented. Patches aimed at correcting vulnerabilities in the program code are often not available in a timely manner or cannot be applied promptly. In particular, ongoing technical processes or operational restrictions within a plant do not allow this at all times. Furthermore, eliminating vulnerabilities through code adaptations can be more or less complicated depending on the programming language used. Even though many protection concepts achieve a good balance between software performance and level of protection, false positives can occur, signaling an attack on program code even though no attack has occurred.In industrial automation systems, it is extremely problematic when automation processes are stopped due to such false alarms. This typically leads to costly, unnecessary interruptions, quality issues, or even security risks. Due to these problems, existing exploit protection concepts are often inapplicable to industrial automation devices.

[0009] The present invention is based on the object of creating a method for the secure execution of control applications which, in particular when using exploit protection concepts, enables increased reliability through correctly detected threats and avoidance of process interruptions caused by false alarms, as well as to specify a suitable implementation for carrying out the method.

[0010] This object is achieved according to the invention by a method having the features specified in claim 1, by a system having the features specified in claim 10, and by a checking device having the features specified in claim 11. Advantageous developments of the present invention are specified in the dependent claims.

[0011] According to the method according to the invention for the secure execution of control applications, at least one event is defined for at least one control application, which event is triggered in the event of potential manipulation of the program code assigned to the control application or of at least one peripheral device connected to a program sequence control device that processes the program code. The program sequence control device monitors a sequence of the control application for deviations from an expected sequence behavior and triggers the defined event in the event of a deviation. The event can, in particular, comprise a designation of a code location within the program code, a time of occurrence of the defined event, an event type, or an error type.

[0012] According to the invention, after the specified event is triggered, the program code is further processed by the program sequence control device, and the event is reported to a checking device separate from the program sequence control device. Upon reporting the event, the checking device analyzes the execution behavior of the control application and the control components that are dependent on it using updatable check rules. The check rules are updated independently of the execution of the control application. Preferably, the checking device analyzes manipulations of the program code assigned to the control application and the control components that are dependent on it in combination.In addition, the checking device can advantageously analyze the execution behavior of the control application and the control components that are dependent on it depending on an operating state of the program sequence control device or on at least one selected state.

[0013] If the control application detects behavior that is impermissible according to the test rules, the verification device transfers the control application and the control components dependent on it to a predefined safe operating state. This offers the advantage that software running on industrial automation devices, in particular, can be reliably protected from attacks without having to be immediately stopped. Rather, it can first ensure that an attack is not mistakenly recognized as such. This allows controlled measures to be initiated in suspected cases of potentially safety-critical error messages.This makes it possible to shut down a controlled technical system in a controlled manner, continue operating it in a limited operating mode (fall operational), or transfer it to a safe operating state (fall safe) in a controlled manner. By outsourcing the analysis of a potential attack to the testing facility, the testing rules can be reliably learned or defined, particularly during a test phase. No adaptation of the software itself is necessary. Overall, detected potential attacks can be correctly assessed with a significantly higher probability thanks to the analysis outsourced to the testing facility.

[0014] According to the invention, the program code of the control application is created using a first compile flag. The first compile flag activates at least one code sequence in the program code for performing checks during indirect function jumps, whereby the checks are performed each time the control application is executed. During function calls, the checks check a function prototype of a called function. Function prototypes include, in particular, the number and type of call parameters and return values.

[0015] The predefined safe operating state preferably comprises stopping the control application, stopping the control components that are dependent on the control application, activating a fault operating mode of an industrial automation device implemented by the control application, or reporting an alarm to an operator of an industrial automation system that includes the automation device. Alternatively or additionally, a reconfiguration of an output interface, in particular an activation of a fail-safe mode, can be performed in the predefined safe operating state. Furthermore, the control application and the control components that are dependent on it can be stopped by transitioning to the predefined safe operating state within a specified period of time after the event is reported.

[0016] According to a further preferred embodiment of the present invention, the program code of the control application is created using a second compile flag. The second compile flag activates at least one code sequence in the program code, which causes the program flow control device to process the program code of the control application until further notice if the flow of the control application deviates from the expected flow behavior. Output and processing of error messages are still possible.

[0017] Advantageously, events triggered by the program sequence control device during operation of the control application in a secure environment are trained by the verification device as false positive alarms, which are to be ignored by the verification device when a corresponding event is reported after operation in the secure environment has ended. Thus, the test rules can be created or updated according to the trained false positive alarms. Furthermore, events reported to the verification device during a training phase can be selectively classified as critical or non-critical depending on user interaction. Accordingly, the test rules can be reliably created or updated according to a classification of the events as critical or non-critical.

[0018] The system according to the invention is intended for carrying out a method according to the preceding embodiments and comprises at least one program sequence control device and a checking device separate from the program sequence control device. The program sequence control device is designed to process program code assigned to control applications and to define at least one event for at least one control application that is triggered in the event of potential manipulation of the program code assigned to the control application or of at least one peripheral device connected to a program sequence control device processing the program code.

[0019] According to the invention, the program sequence control device is additionally designed to monitor a sequence of the control application for deviations from an expected sequence behavior, to trigger the specified event in the event of a deviation, to further process the program code after triggering the specified event, and to report the event to the checking device. In contrast, the checking device is designed to analyze the sequence behavior of the control application and the control components that are dependent on it using updatable test rules when the event is reported. In addition, the checking device is designed to transfer the control application and the control components that are dependent on it to a predefined safe operating state upon detection of sequence behavior of the control application that is inadmissible according to the test rules.

[0020] The checking device according to the invention is provided for a system according to the preceding embodiments and is designed to analyze the execution behavior of the control application and the control components dependent on it based on updatable test rules upon notification of a specified event triggered by a program flow control device in the event of potential manipulation of program code assigned to a control application. In addition, the checking device is designed to transfer the control application and the control components dependent on it to a predefined safe operating state upon detection of an execution behavior of the control application that is impermissible according to the test rules.

[0021] The present invention will be explained in more detail below using an exemplary embodiment with reference to the drawing. Figure 1 shows an industrial automation system with a plurality of program sequence control devices, each providing at least one control application, and with a checking device for analyzing potential manipulation attempts reported by the program sequence control devices. Figure 2 shows a schematic representation of a message traffic between a plurality of program sequence control devices and the checking device in the event of potential manipulation attempts.

[0022] The Figure 1The industrial automation system illustrated comprises a plurality of program sequence control devices 101-102, each implementing an automation device, as well as a separate checking device 300 for analyzing messages 201-202 from the program sequence control devices 101-102 regarding potential manipulation attempts. The program sequence control devices 101-102 and the checking device 300 are interconnected via an industrial communications network 600, which is preferably configured as a time-sensitive network, in particular in accordance with IEEE 802.3-2018, IEEE 802.1Q-2018, IEEE 802.1AB-2016, IEEE 802.1AS-2011, IEEE 802.1BA-2011, or IEEE 802.1CB-2017. A forwarding of time-critical data within the communication network 600 can be carried out, for example, by means of frame preemption according to IEEE 802.1Q-2018, time-aware shaper according to IEEE 802.1Q-2018, credit-based shaper according to IEEE 802.1Q-2018, burst limiting shaper, peristaltic shaper orPriority-based control.

[0023] Automation devices can be designed, in particular, as programmable logic controllers, operator control and monitoring stations, I / O controllers, or I / O modules. Operator control and monitoring stations are used, for example, to visualize process data or measurement and control variables that are processed or acquired by programmable logic controllers, input / output units, or sensors. In particular, operator control and monitoring stations can be used to display values ​​of a control loop and to change control parameters. Operator control and monitoring stations comprise at least a graphical user interface, an input device, a processor unit, and a communications module.

[0024] In the present exemplary embodiment, the program sequence control devices 101-102 each implement a programmable logic controller. Programmable logic controllers typically each comprise a processor 111, 121 for processing program code for control applications 113, 123, a working memory 112, 122 for loading the program code, a communication module, and at least one input / output unit. Input / output units can also be configured as decentralized peripheral modules located remotely from a programmable logic controller. A programmable logic controller can be connected to a switch or router or additionally to a fieldbus via the communication module.The input / output unit is used to exchange control and measurement variables between the programmable logic controller 101, 102 and a machine or device 501, 502 controlled by the programmable logic controller 101, 102. The control applications are particularly intended for determining suitable control variables from recorded measurement variables.

[0025] Separate I / O modules 510 can also be provided for exchanging control and measurement variables with connected machines or devices 501, 502. I / O modules can be controlled, in particular, by means of an I / O controller per automation cell. Alternatively, I / O modules can also be controlled by an associated programmable logic controller.

[0026] In the present exemplary embodiment, for the secure execution of the control applications 113, 123 provided by the program sequence control devices 101, 102, at least one event is defined for each control application 113, 123 and stored in a configuration database 110, 120. This event is triggered in the event of potential manipulation of the program code assigned to the respective control application 113, 123 or of a peripheral device 510 connected to a program sequence control device. In the present exemplary embodiment, such a peripheral device is the I / O module 510 connected to the program sequence control device 101. However, the application of the present invention is by no means limited to this type of peripheral device.

[0027] The program sequence control devices 101, 102 each monitor a sequence of their respective control application 113, 123 for deviations from an expected sequence behavior and trigger the respective specified event in the event of a deviation. The respective specified event can, in particular, include a designation of a code location within the program code, a time of occurrence of the specified event, an event type, an error type, or an error frequency. Depending on this information and, in particular, based on experience from previously reported events, the checking device 300 can decide whether an event is critical and whether further measures to contain the risk are required, for example, stopping the respective control application 113, 123.

[0028] After triggering the respective specified event, the program code is further processed by the program sequence control devices 101, 102 and the event is reported to the checking device 300. In this case, a message 201, 202 about the triggering of the respective specified event is transmitted to the checking device 300 (see also Figure 2 ).

[0029] Upon notification of the respective specified event, the checking device 300 analyzes the execution behavior of the respective control application 113, 123 as well as control components that are dependent on it, for example, the I / O module 510, using updateable test rules that are stored in a rule database 310 of the checking device 300. The test rules can be updated independently of the execution of the respective control application 113, 123. For this purpose, the checking device 300 has an update module 302 (see Figure 2 ).

[0030] If the respective control application detects an impermissible operating behavior according to the test rules, the checking device 300 transfers the control application 113, 123 and the control components dependent on it to a predefined safe operating state. For this purpose, the checking device 300 transmits a corresponding message 401, 402 with a control sequence to the respective control application 113, 123 or program sequence control device 101, 102. Advantageously, the checking device 300 analyzes, in combination, any manipulation of the program code assigned to the respective control application 113, 123 and the control components dependent on it.In addition, the checking device 300 analyses the execution behaviour of the respective control application 113, 123 as well as the control components dependent thereon, preferably depending on an operating state of the respective program sequence control device 101, 102 or on at least one selected state.

[0031] The predefined safe operating state can in particular comprise stopping the control application 113, 123, stopping the control components that are dependent on the control application 113, 123, for example the I / O module 510, activating an error operating mode of an industrial automation device implemented by means of the control application, or a message 410 of an alarm to an operator 500 of the industrial automation system (see Figure 2). Alternatively or additionally, the respective control application 113, 123 and the control components dependent on it can be stopped by transitioning to the predefined safe operating state within a specified period of time after the event has been reported.

[0032] According to a preferred embodiment, the program code for the control applications 113, 123 is each created using a first compile flag. Such a first compile flag, for example, "-fsanitize=cfi" in the Clang compiler, activates a code sequence in the program code for performing checks during indirect function jumps, whereby the checks are performed each time the respective control application 113, 123 is executed. In particular, the checks check a function prototype of a called function during function calls. Additionally, the program code of the respective control application 113, 123 can each be created using a second compile flag.This second compile flag, for example "-fsanitize-recover=all" in the Clang compiler, activates a code sequence in the program code, which causes the program flow control device to process the program code of the respective control application 113, 123 until further notice if the flow of the control application deviates from the expected flow behavior.

[0033] Accordingly Figure 2In addition to the update module 302, the checking device 300 comprises a training or learning module 301. By means of this training or learning module 301, events triggered by the respective program sequence control device 101, 102 during operation of the respective control application 113, 123 in a secure environment are learned by the checking device as false positive alarms, which are to be ignored by the checking device when a corresponding event is reported after operation in the secure environment has ended. The test rules are then created or updated according to the learned false positive alarms. Furthermore, in a learning phase, events reported to the checking device 300 can be selectively classified as critical or non-critical according to user interaction. The test rules are then created or updated according to a classification of the events as critical or non-critical.updated.

[0034] The checking device 300 can take different forms of implementation, namely as an integrated, separate software component on a program flow control device, as a component on an external device connected via an I / O interface, or as an implementation on an external server, for example an edge or cloud computing server.

[0035] In principle, further design variants are possible for updating the test rules for the testing device 300: The learning phase can be supported by a developer who analyzes errors in more detail and defines or updates the checking rules accordingly. The checking device can use methods based on artificial intelligence or machine learning to update the checking rules. By analyzing frequently occurring, identical error events in one or more binary files, and in particular based on analyses of subsequent events, the checking device learns and adapts its checking rules accordingly. This enables learning towards a correct analysis. The environment in which the respective control application is deployed can advantageously be taken into account in the analyses of the checking device and in any updated checking rules. Analyses can advantageously be carried out depending on the location of an error occurrence.For example, no response is made if an error occurs in a less critical code segment. For code segments with critical components, such as functions for a network protocol stack, an error can be assumed to be critical, and appropriate action can be initiated. A whitelist for the validator can also define which errors are non-critical and can be ignored. Such whitelists are preferably also updated. Training the validator's check rules for false positives can also be simulated and optimized, for example, in a digital twin before going live.

Claims

1. Method for the secure execution of control applications, in which - for at least one control application (113, 123), at least one event is defined which is triggered in the event of potential manipulation of program code assigned to the control application and / or of at least one peripheral device (510) which is connected to a program flow control device (101, 102) processing the program code, - the program flow control device monitors a flow of the control application for deviations from an expected flow behaviour and triggers the defined event in the event of a deviation, - following triggering of the defined event, the program code continues to be processed by the program flow control device and the event is signalled to an inspection device (300) separate from the program flow control device, - upon the signalling (201, 202) of the event, the inspection device analyses the flow behaviour of the control application and control components in a dependency relationship thereto using updatable inspection rules, the inspection rules being updated independently of the flow of the control application, - the inspection device transfers the control application and the control components in a dependency relationship thereto to a predefined safe operating state (401, 402) on detection of a flow behaviour of the control application contravening the inspection rules, - the program code of the control application is created by means of a first compile flag which activates at least one code sequence in the program code to carry out inspections for indirect function jumps, - the inspections, which are each used to inspect a function prototype of a called function during function calls, are carried out during each run of the control application.

2. Method according to Claim 1, in which the event comprises a designation of a code location within the program code, a time of occurrence of the defined event, an event type, and / or an error type.

3. Method according to either of Claims 1 and 2, in which the predefined safe operating state comprises stopping the control application, stopping the control components in a dependency relationship to the control application, activating a fault operating mode of an industrial automation device implemented by means of the control application and / or signalling an alarm to an operator of an industrial automation system comprising the automation device.

4. Method according to any one of Claims 1 to 3, in which the control application and the control components in a dependency relationship thereto are stopped by a transfer into the predefined safe operating state within a specified time period after the event has been signalled.

5. Method according to any one of Claims 1 to 4, in which the inspection device analyses manipulations of the program code assigned to the control application and of the control components in a dependency relationship thereto in combination.

6. Method according to any one of Claims 1 to 5, in which the inspection device analyses the flow behaviour of the control application and the control components in a dependency relationship thereto depending on an operating state of the program flow control device and / or on at least one selected state.

7. Method according to any one of Claims 1 to 6, in which the program code of the control application is created by means of a second compile flag and in which the second compile flag activates at least one code sequence in the program code, by means of which the program flow control device is caused to process the program code of the control application until further notice in the event of a deviation in the flow of the control application from the expected flow behaviour.

8. Method according to any one of Claims 1 to 7, in which events triggered by the program flow control device during operation of the control application in a secure environment are learned by the inspection device as false positives, which should be ignored by the inspection device when a corresponding event is signalled after completion of operation in the secure environment, and in which the inspection rules are created and / or updated according to the learned false positives.

9. Method according to any one of Claims 1 to 8, in which events signalled to the inspection device in a learning phase are classified selectively as critical or non-critical according to user interaction and in which the inspection rules are created and / or updated according to a classification of the events as critical or non-critical.

10. System for carrying out a method according to any one of Claims 1 to 9, wherein - the system comprises at least one program flow control device and an inspection device separate from the program flow control device, - the program flow control device is designed to process program code respectively assigned to control applications and to define at least one event for at least one control application, said event being triggered in the event of potential manipulation of program code assigned to the control application and / or of at least one peripheral device which is connected to a program flow control device processing the program code, wherein the program code of the control application is created by means of a first compile flag which activates at least one code sequence in the program code to carry out inspections for indirect function jumps, - the program flow control device is additionally designed to monitor a flow of the control application for deviations from an expected flow behaviour, to trigger the defined event in the event of a deviation, to continue to process the program code following triggering of the defined event and to signal the event to the inspection device, - the inspection device is designed, when the event is signalled, to analyse the flow behaviour of the control application and control components in a dependency relationship thereto using updatable inspection rules, the inspection rules being updated independently of the flow of the control application, - the inspection device is additionally designed to transfer the control application and the control components in a dependency relationship thereto into a predefined safe operating state on detection of a flow behaviour of the control application contravening the inspection rules and to carry out the inspections, which are each used to inspect a function prototype of a called function during function calls, during each run of the control application.

11. Inspection device for a system according to Claim 10, wherein - the inspection device is designed, on the signalling of a defined event which is triggered by a program flow control device in the event of potential manipulation of program code assigned to a control application, to analyse a flow behaviour of the control application and control components in a dependency relationship thereto using updatable inspection rules, the inspection rules being updated independently of the flow of the control application, and the program code of the control application being created by means of a first compile flag which activates at least one code sequence in the program code to carry out inspections for indirect function jumps, - the inspection device is additionally designed to transfer the control application and the control components in a dependency relationship thereto into a predefined safe operating state on detection of a flow behaviour of the control application contravening the inspection rules and to carry out the inspections, which are each used to inspect a function prototype of a called function during function calls, during each run of the control application.