DE102019201728A1 - Method for securing data using at least two processing units and a decision unit in communication with the at least two processing units - Google Patents

Abstract

The invention relates to a method for securing data (D _E ) using at least two processing units (110, 120) and a decision unit (130) in communication with the at least two processing units (140), whereby each of the processing units (110, 120) received data (D _E ) are processed, an intermediate check value (P _AB ) is generated for the processed data (D _A ) and the generated intermediate check value (P _AB , P _AC ) is transmitted to the decision unit (130) , wherein a final test value (P _A ) is formed by means of the decision unit (130) from the intermediate test values (P _AB ) received from the processing units (110, 120), and the processed data (D _A ) with the final test value ( P _A ) are provided and output, as well as a computing system (100).

Description

The present invention relates to a method for securing data using at least two processing units and a decision unit which is in communication with the at least two processing units, as well as a computing system and a computer program for its implementation.

State of the art

Systems or computing systems can be required for a wide variety of applications that make calculations on the basis of input data (e.g. from sensors) and send or provide the results, for example, via a communication medium (e.g. Ethernet, CAN, etc.), i.e. in particular without direct control of an actuator. Such systems (often so-called domain or central computers) are increasingly being used, particularly in the automotive sector, but also in rail technology (both in the rail network and in trains) and in industrial applications.

One of the challenges in the design of such systems is to guarantee the required integrity (often classified with integrity levels or safety integrity levels according to the Safety Integrity Level or SIL) without changing the communication protocols, as e.g. So-called legacy systems (i.e. in particular existing systems that have to receive a certain type of data or instructions) can be used as receivers and / or actuators.

Disclosure of the invention

According to the invention, a method for securing data and a computing system and a computer program for carrying it out are proposed with the features of the independent claims. Advantageous configurations are the subject of the subclaims and the description below.

A method according to the invention serves to secure data using at least two processing units and a decision unit that is in communication with the at least two processing units. Both the computing units and the decision unit can be separate processors or microprocessors. A decision unit is to be understood here in particular as a unit by means of which different data or values can be compared with one another and, in particular, based on this comparison, a decision can be made as to whether these data or values are the same or whether, for example, a difference, there is an error. A comparator or voter is often used here.

It goes without saying that - for example depending on the desired requirement and / or security - it is also possible to use more than two computing units, for example three, four, five, six or even more.

For safety requirements from a certain integrity level (SIL, in the automotive sector also ASIL), critical calculations can be carried out multiple times in such systems or computing systems, i.e. by means of each of several or each of the at least two computing units, in order to be able to detect accidental hardware errors (sometimes also systematic errors) and to be able to avoid hazards.

In order to gain safety from the multiple execution, the processing units should be as independent as possible (avoidance of so-called common cause faults, CCF). In general, the entire system should not be prone to individual errors (errors on individual components). In the case of high safety integrity levels (SIL 4, see e.g. IEC61508-2: 2011, Appendix E), it should be noted that on a single processor or microprocessor (or a single processing unit) even complex errors must be viewed as single errors. In other words, redundancies on a (shared) chip must not be considered independent.

In general, such systems can work with a cross comparison, i.e. there is a mutual check of the output of the respective other computing units or with a comparator or voter external to the computing units, as mentioned above. The fact that a single secured message, i.e. However, data secured or provided with a test value to be output or sent has proven to be a challenge.

Depending on how the typically required functions or steps for calculating a check value (usually determined according to the so-called CRC or cyclical redundancy check), the comparison of check values and the sending or provision of the secured data to the outside world multiple computing units and, if necessary, the (external) decision-making unit, there are susceptibilities to individual errors or at least high internal data streams result.

In the proposed method, data received by means of each of the processing units are now processed for the processed data an intermediate test value is generated (by means of each of the computing units) and the intermediate test value generated is transmitted (by the respective computing unit) to the decision unit. The decision unit then forms a final test value from the intermediate test values received from the processing units, and the processed data is then provided with the final test value and output or provided (in particular via an interface or a communication medium).

The final test value formed by means of the decision unit is in particular transmitted to one of the computing units, and by means of this computing unit the processed data are then provided with the final test value and output.

In order to use the decision unit to form the final test value from the intermediate test values received from the processing units, the intermediate test values received by the processing units are compared, in which case, if the received intermediate test values are the same, one of the intermediate test values of the final test value is formed. For this purpose, the intermediate test value is generated in accordance with the same rule by means of each of the computing units. The arithmetic units can thus be designed in the same way, a comparison of the two intermediate test values which are the same (at least if they are functioning properly) takes place in the decision unit.

Alternatively, however, it is also preferred if the final test value is formed from the intermediate test values received from the processing units by means of the decision unit, in that a final comparison test value is formed from the intermediate test values received from the processing units and these are compared with one another and then if the final comparison test values formed are the same, one of them is used as the final test value. For this purpose, in particular, intermediate test values are generated in accordance with different regulations by means of the computing units. The final test value ultimately corresponds to the same final comparison test values (at least when functioning properly), which are each determined separately in the decision unit for the intermediate test values. With this variant, although several final test values or final comparison test values (from different intermediate test values in each case) have to be determined, additional robustness against errors with a common cause (i.e. which can occur in all processing units at the same time) is generated.

In this way, it is possible to generate data with integrity, which can be sent to external recipients, without incorrect but plausible output values (i.e. not recognizable as faulty by the recipient) being sent. The proposed method makes it possible to decouple the comparison of the redundant results of a safety-relevant calculation (or general processing) of received data from the generation of the (final) test value and the sending of the data in a way that reduces the safety-relevant influence of individual errors in each of the components avoids. In particular, this is done in a compatible manner for use in legacy systems (sending individual data packets protected by a check value, e.g. in rail systems) and without relevant performance and cost disadvantages.

A particularly relevant aspect of this proposed method is the division of the formation of the test value into two stages, so that none of the multiple (redundant) computing units (which process the data or calculate the output data) calculate the (final) test value and thus uncontrollably incorrect ones can make plausible expenses. In addition, the decision unit must use the original output data, i.e. do not know the data processed in a processing unit, which results in performance advantages and individual errors cannot cause any danger on the decision unit either. In addition, existing end-to-end secured transport protocols and existing legacy systems do not have to be adapted. In particular, there are no restrictions for a choice of the externally visible transport polynomial or, in general, of the secured data to be output.

However, the type of processing is not relevant. In particular, this can also only involve passing the received data through, i.e. an identical figure.

In the case of identity in particular, verification of data received with a check value is also possible. If the processing units form an intermediate test value from the received data (ie the processing is the identity) and transmit it to the decision unit together with the received test value, the final test value formed by the decision unit from the intermediate test values received from the processing units can be used can be compared with the received test value. If the final test value corresponds to the received test value, the received data has been received correctly.

Advantageously, by means of each of the computing units at predetermined times, in particular regularly, predetermined correct data and predetermined (deliberately) incorrect data are transmitted to the decision unit, with the aid of the Decision unit received test values a functionality of the decision unit is checked. In this way, not only individual errors, but in particular also dormant or latent errors can be detected or recognized.

The proposed method can be implemented in particular with the aid of a test value formed by means of division, in particular with a remainder (modulo), by a test value polynomial. A test value formed by division by an intermediate polynomial, which is formed by multiplying the test value polynomial by a further polynomial, then comes into consideration as an intermediate test value. For an exemplary calculation of such polynomials, reference is made at this point to the description of the figures.

The invention also relates to a computing system for securing data, with at least two computing units and a decision unit in communication with the at least two computing units, the computing system being set up to carry out all method steps of a method according to one of the preceding claims.

The computing system can be, for example, a control device, for example for a motor vehicle, which has the at least two computing units and the decision unit, which in turn can each be formed by processors or microprocessors. The decision unit can in particular be implemented on an FPGA, ASIC, 8-bit controller or the like. For a coupling, i.e. For the communication connection, simple interfaces with low bandwidth such as UART can be considered, as well as network or bus interfaces.

The implementation of a method according to the invention in the form of a computer program or computer program product with program code for performing all method steps is advantageous, since this causes particularly low costs, especially if an executing control device is used for other tasks and is therefore available anyway. Suitable data carriers for providing the computer program are in particular magnetic, optical and electrical memories, such as e.g. Hard disks, flash memories, EEPROMs, DVDs, etc. It is also possible to download a program via computer networks (Internet, intranet, etc.).

Further advantages and configurations of the invention emerge from the description and the accompanying drawing.

The invention is shown schematically in the drawing using exemplary embodiments and is described below with reference to the drawing.

Figure list

• 1 shows schematically a computing system according to the invention in a preferred embodiment.
• 2 shows schematically a computing system according to the invention in a further preferred embodiment.

Embodiment (s) of the invention

In 1 is a schematic of a computing system according to the invention 100 shown in a preferred embodiment with which a method according to the invention can be carried out. In the following, the 1 both the computing system and the method are explained in more detail.

The computing system 100 , for example a control unit, has two processing units 110 and 120 , which are designed as microprocessors, for example, and a decision unit 130 , which is also designed as a microprocessor, for example.

In addition, the computing system 100 a communication medium or a communication link 140 on what, on the one hand, the arithmetic units 110 , 120 with the decision-making unit 130 communicate, ie exchange data, and on the other hand, data from an external transmitter 180 and receive data to an external recipient 190 , for example an actuator or another control device, can be provided. It goes without saying that corresponding interfaces are typically provided for this purpose.

As part of the proposed procedure, data D _E from the external transmitter 180 , which can be a sensor, for example, to the computing system 100 transmitted. Likewise, the data can also be processed by means of corresponding computing or processing steps in the computing system 100 have been self-generated. These dates D _E are then in the sense of input or input data within the computing system 100 to the computing unit 110 and in parallel to the processing unit 120 transmitted or handed over.

Both in the arithmetic unit 110 as well as in the arithmetic unit 120 will be the data D _E then processed or a desired calculation is carried out with it. The processed data is here with D _A and ultimately intended for output.

Furthermore, both in the processing unit 110 as well as in the arithmetic unit 120 from the processed data D _A or for these processed data D _A an intermediate test value each P _AB determined or calculated. This calculation or the rule on which it is based (here division by the remainder by the same test value polynomial) and thus the intermediate test value obtained in the case of proper function P _AB are for the arithmetic unit 110 and the arithmetic unit 120 equal.

From the computing unit 110 and from the computing unit 120 become the intermediate test values P _AB then to the decision-making unit 130 transmitted. This is where the received interim test values P _AB then in a comparison and / or decision module 131 compared and it is then if both intermediate test values P _AB are the same, from this or from one of them a final test value P _A determined or calculated.

From the decision-making unit 130 becomes the final test value P _A then to the computing unit 110 transmitted. There are by means of a transmission module 113 those previously in the arithmetic unit 110 processed data D _A with the final test value P _A provided and then as secured data D _{A, S} provided or to the external recipient 190 transmitted.

The following is the process of determining or calculating the final test value P _A about the intermediate test value P _AB will be explained using an example with polynomial divisions, as they are also based on conventional CRC methods.

• Analog gilt: D_E\(A*B)=P_AB und D_E = (A*B)*x_A + P_AB.
• Dann gilt aber auch: P_AB\A=(D_E-(A*B)*x_AB)\A.

The data D _E , ie the data to be protected or the data packet to be protected can be used as a polynomial with coefficients from the field {0,1; Addition modulo 2; Multiplication}. A can be a CRC or check value polynomial that is to be used to secure the data transport to the outside, while B can be another check value polynomial that differs from A. A * B is then a test value polynomial resulting from the multiplication of A and B. The respective test value is created as the remainder of the polynomial division of D _E by the test value polynomial. With * as the polynomial multiplication and \ the modulo operation (remainder of the polynomial division), the result is: $${\text{D}}_{\text{E}}\backslash \text{A}={\text{P}}_{\text{A}}{\text{and D}}_{\text{E}}={\text{A * x}}_{\text{A}}+{\text{P}}_{\text{A}}.$$

• Similarly: D _E \ (A * B) = P _AB and D _E = (A * B) * x _A + P _AB .
• But then also: P _AB \ A = (D _E - (A * B) * x _AB ) \ A.

Because (A * B) * x _AB is divisible by A without a remainder, we get: $${\text{P}}_{\text{FROM}}\backslash \text{A}={\text{D}}_{\text{E}}\backslash \text{A}={\text{P}}_{\text{A}}.$$

This means that a check value P _A , formed with the test value polynomial A on the original data D _E , is identical to a test value P _AB \ A, formed with the test value polynomial A on the intermediate test value P _AB , the one with the test value polynomial A * B on the original data D _E was calculated. An arithmetic unit and the decision unit (or two units in general) can thus generate the (final) test value P _A split into two steps so that the processing unit does not know the algorithm of the (final) test value and the decision unit never knows the data D _E receives, but only the intermediate test value P _AB , formed using the polynomial A * B.

In particular, a system consisting of two processing units and a decision unit is thus obtained that is highly robust against individual errors. The processing units can therefore not send any dangerous, faulty packets because they do not know the (final) test value. The decision unit cannot send dangerous faulty packets because it does not receive the data and does not have to have any logic for sending data to the outside. Therefore, with such a computing system or such a method, high security integrity levels can be achieved. The decision unit only receives a small data stream.

In 2 is a schematic of a computing system according to the invention 100 ' shown in a further preferred embodiment. The computing system 100 ' largely corresponds to the computing system 100 according to 1 so that only the differences will be discussed below. Otherwise, refer to the description 1 referenced.

In contrast to the computing system 100 according to 1 are in the arithmetic units 110 and 120 of the computing system 100 ' the intermediate test values are not determined according to the same regulation, but according to different regulations (here division with remainder by different test value polynomials A * B and A * C). This is done in the arithmetic unit 110 an intermediate test value P _AB determined in the arithmetic unit 120 however, an intermediate test value P _AC . The determination of the intermediate test value P _AB can as above too 1 explained, for the determination of the intermediate test value P _AC Correspondingly, a polynomial C can be used instead of the polynomial B.

The arithmetic units 110 and 120 transmit the respective intermediate test value P _AB or. P _AC then to the decision-making unit 130 . There will then - in contrast to the one in relation to 1 explained procedure - initially from each of the intermediate test values P _AB and P _AC each a final comparison test value P _A or a final test value in each case P _A determined according to P _AB \ A or P _AC \ A.

Then the two final comparison test values are determined P _A in a comparison and / or decision module 131 compared and it is when both final comparison test values P _A are the same, this or one of these final comparison test values P _A used as the final test value and sent to the computing unit 110 transmitted.

Although these are compared to that in relation to 1 The computing system or method described requires two test value calculations in the decision unit, but a higher level of security is achieved, as already mentioned above.

The described embodiments of the invention can, for example, for the case that the processing is the identity (ie D _A = D _E ), also for the verification of data obtained with the test value P _E D _E can be used.

In this case, the data received would be used D _E in each of the arithmetic units 110 , 120 analogous to the procedure with internally generated, i.e. processed, data with the polynomial A * B ( 1 ) or A * B and A * C ( 2 ) Test values formed. Each of the computing units now sends the intermediate value P _AB or. P _AC and the received check value P _{E of} the received data to the decision unit 130 .

This either compares the received test values P _AB and with those of the arithmetic units 110 and 120 forwarded test values P _E ( 1 ) or trains P _AB and P _AC each P _A and compares this with that of the computing units 110 and 120 forwarded test values P _{E of} the received data ( 2 ). To ensure that both processing units have received correct, identical data, all test values must ultimately be identical.

Since the same test value (e.g. P _A ) applies, can in particular by the embodiment based on 2 it must be ensured that the security concept for sending the data is not replaced by a CRC _A implementation required for receiving (ie an implementation of the final check value P _A ) on the computing units 110 and 120 being undermined.

To reduce the latency, the received data can be checked in the decision unit 130 also parallel to their processing in the computing units 110 and 120 respectively. If it then turns out that one of the processing units or even both have received incorrect input data and have therefore already processed it, the decision unit can 130 respond by using the generated test values P _A from then on is no longer available or these test values are deliberately and permanently falsified (since it can be assumed that the incorrectly processed input data will permanently affect the internal state of the processing unit 110 and or 120 affect).

Claims (13)

A method for securing data (D _E ) using at least two processing units (110, 120) and a decision unit (130) in communication with the at least two processing units (140), wherein data received by means of each of the processing units (110, 120) (D _E ) are processed, an intermediate check value (P _AB , P _AC ) is generated for the processed data (D _A ) and the generated intermediate check value (P _AB , P _AC ) is transmitted to the decision unit (130), wherein a final test value (P _A ) is formed by means of the decision unit (130) from the intermediate test values (P _AB , P _AC ) received from the computing units (110, 120), and the processed data (D _A ) with the final Test value (P _A ) can be provided and output. Procedure according to Claim 1 , the final test value (P _A ) formed by the decision unit (130) being transmitted to one of the computing units (110), and the processed data (D _A ) being provided with the final test value (P _A ) by means of this computing unit (110) and output. Procedure according to Claim 1 or 2 , the final test value (P _A ) being formed by means of the decision unit (130) from the intermediate test values (P _AB ) received by the processing units (110, 120) by comparing the intermediate test values (P _AB ) received by the processing units and then, if the received intermediate test values (P _AB ) are the same, the final test value (P _A ) is formed from one of the intermediate test values (P _AB ). Procedure according to Claim 3 , the intermediate test value (P _AB ) being generated by means of each of the computing units (110, 120) in accordance with the same rule. Procedure according to Claim 1 or 2 , the final test value (P _A ) being formed by means of the decision unit (130) from the intermediate test values (P _AB , P _AC ) received from the processing units (110, 120) by using the intermediate test values ( P _AB , P _AC ) each form a final comparison test value (P _A ) and this are compared with one another, and then, if the final comparison test values (P _A ) formed are the same, one of them is used as the final test values (P _A ). Procedure according to Claim 5 , an intermediate test value (P _AB , P _AC ) being generated in accordance with different regulations by means of the computing units (110, 120). Method according to one of the preceding claims, wherein by means of each of the computing units (110, 120) at predetermined points in time, respectively predetermined correct data and predetermined incorrect data are transmitted to the decision unit (130), and with the aid of the data based thereon, from the decision unit (130) a functionality of the decision unit (130) is checked. Method according to one of the preceding claims, wherein the processing of the received data (D _E ) by means of each of the computing units (110, 120) is an identity mapping. Method according to one of the preceding claims, wherein a check value received with the received data (D _E ) is transmitted to the decision unit (130), and wherein the intermediate value received by the decision unit (130) from the intermediate data received by the processing units (110, 120) Test values (P _AB , P _AC ) formed final test value (P _A ) is compared with the received test value. Method according to one of the preceding claims, wherein the intermediate test value (P _AB , P _AC ) is generated by means of division, in particular with the remainder, of the processed data (D _A ) by a test value polynomial and / or wherein the final test value (P _A ) is generated by means of division, in particular with the remainder, of an intermediate test value (P _AB , P _AC ) by a test value polynomial. Computing system (100, 100 ') for securing data, with at least two computing units (110, 120) and a decision unit (130) in communication with the at least two computing units (140), the computing system (100, 100') being set up for this purpose is to carry out all process steps of a process according to one of the preceding claims. Computer program that causes a computing system (100, 100 ') to carry out all process steps of a process according to one of the Claims 1 to 10 when it is executed on the computing system (100, 100 '). Machine-readable storage medium with a computer program stored thereon Claim 12 .

Images