Patent Application
20150264080
The invention relates to a method and a testing system for checking integrity of property data relating to a device by means of a testing device inside a network.
The inclusion of distributed measuring or control devices in the field of system automation is becoming increasingly important. Intelligent field devices for implementing intelligent power supply systems as well as devices with a control function inside automation systems or meters for billing for services used, such as communication, power, gas or water, are required in this case, for example.
In this case, safety-relevant data, for example collected measurement data or switching commands, are often transmitted inside a network.
Manipulation of such devices is a serious problem for safe, reliable and economical operation of infrastructures, for example automation systems or intelligent energy supply networks. In this case, manipulation can be understood as meaning unauthorized modification of data on devices such as measuring or control devices. This manipulation includes, for example, changing the device properties or functionalities, for example replacing device software (firmware) or changing individual areas of the device software. In addition, configuration data relating to a device should be protected from unintentional and unauthorized modification.
The applicant is aware of a system and a method for detecting a state change in an automation system from the document having the official file reference 10 2011 077 289.8 of the German patent and trademark office. In this case, the integrity of the device software or of the configuration data is queried at regular intervals using a challenge-response protocol. If a fingerprint calculated by a device does not match the fingerprint expected by a monitoring device, the device has been changed in an unauthorized manner.
The patent specification DE 10 2007 034 525 B4 discloses a method which checks data stored in a predetermined memory area of devices by comparing a fingerprint locally calculated by a device with a fingerprint calculated by a testing entity.
For a large number of devices to be checked inside a network, a high computing power is expected of a testing entity, in particular in the case of a parallel or virtually simultaneous check. At the same time, the check is supposed to ensure that manipulation is discovered.
Against this background, the object of the present invention is to provide a method and a testing system which make it possible for the integrity of a respective device to be checked by a testing entity in a simplified and tamperproof manner in the case of a large number of devices.
This object is achieved by means of a method and a testing system according to the features stated in the independent claims. Advantageous embodiments and developments are stated in the subclaims.
The advantages mentioned below need not necessarily be achieved by the subject matters of the independent patent claims. Rather, they may also be advantages which are achieved only by individual embodiments or developments.
According to the invention, a method for checking integrity of property data relating to a device, the property data comprising a program data record and a configuration data record, by means of a testing device inside a network comprising the device, the testing device and at least one further device having further property data, the further property data comprising a further program data record and a further configuration data record, has the following steps:
A station which is coupled at least to the testing device by means of the network provides the property data and the further property data. The testing device evaluates the property data and the further property data with regard to at least identity of the program data record and the further program data record, and allocates an identification on the basis of an evaluation result.
The testing device determines a parameter which is suitable for influencing a hash value and a test hash value of a hash function, the parameter being configured on the basis of the identification, and transmits the parameter to the device.
The device calculates the hash value on the basis of the hash function and a character string, the character string comprising the parameter and the property data and a sequence inside the character string being able to be predefined, and transmits the hash value to the testing device.
The testing device calculates the test hash value on the basis of the hash function and the character string, the character string comprising the parameter and the property data and a sequence inside the character string being able to be predefined.
The testing device confirms the integrity of the property data if the hash value matches the test hash value.
In the present application, the integrity of property data relating to a device is understood as meaning a state which does not imply manipulation of the device or its property data. This means, for example, that property data have not been changed or replaced during a check in comparison with an earlier check of the property data or correspond to an expected, stipulated or stored state.
The program data record of the property data is typically understood as meaning the device software or firmware of a device or program data in the form of software. A configuration data record typically contains parameters such as network addresses, device identifiers or cryptographic key material and certificates as well as parameters which configure the functionality of the software; these are specific to a device.
The proposed method for checking a device inside a network makes it possible to take into account one or more further devices inside the network by virtue of the testing device evaluating both the property data relating to the device and the property data relating to the further devices and initiating the further checking method steps on the basis of the evaluation result. This makes it possible to check a device inside a network for manipulation, the prerequisites for efficiently checking a large number of devices being created by taking into account further devices inside the network. Manipulation is therefore detected promptly.
In this case, a device to be checked determines the hash value anew each time, for example, that is to say the hash value is calculated anew during each check using the parameter newly predefined for each check.
The testing device likewise calculates the test hash value for the device to be checked for each check. In this case, the testing device receives the property data such as, in particular, the program data record and the configuration data record from a station, which has the function of an engineering station inside an industrial automation system for example, in a state which is up-to-date at the time of the check. In this case, the station is generally part of the network and provides both the configuration data relating to the devices to be checked and the firmware and software versions installed on the devices.
The practice of checking devices in a system by means of a common test server which is optimized for checking a large number of devices in a parallel manner reduces the additional complexity caused by a plurality of individual testing systems for one device or a few devices.
According to one embodiment, the parameter at the start of the character string is transferred.
Therefore, it is not possible for the device to be checked to store an inner state of the hash method, which is calculated only on the basis of the program data record and the configuration data record, for the purpose of calculating the hash value, to then change the program data record and/or the configuration data record and to then again nevertheless calculate a correct hash value which matches the test hash value. This would be easily possible if the parameter at the end of the character string to be hashed were transferred.
Therefore, during a check, the device to be checked must calculate the hash value on the basis of the parameter transferred for the current check and on the basis of the program data record and configuration data record currently on the device.
According to one embodiment, the parameter is in the form of a random number whose length is adapted to the hash value.
In this case, the random number is typically a number having a length of 128 bits to 512 bits (16 to 128 bytes). In this case, the length should ideally correspond to the length of the hash value in the cryptographic hash method used. In this case, SHA-256, SHA-384 or SHA-512, in particular, is possible as the secure hash algorithm (SHA). If the determined hash value is shortened before being transmitted back because the amount of transmitted data is intended to be kept small and also because a shorter hash value already provides a sufficient security level, the random number can also be correspondingly short.
According to one embodiment, the parameter and/or the hash value is/are transmitted in a cryptographically secure manner.
If the parameter is transmitted in encrypted form, the risk of other devices, for which the information relating to the parameter, for example the random number, is not intended, acquiring knowledge of the random number is reduced. Only a device having a suitable key can therefore acquire knowledge of the random number.
Signed transmission of the hash value from the device to be checked to the testing device ensures the authenticity of the response from the device to be checked. The testing device can therefore use a public key to discern when the hash value is not transmitted by the device to be checked. This is useful, in particular, in order to detect or prevent denial-of-service attacks by further devices in the network.
As an alternative to the digital signature, the use of a keyed hash method is also conceivable, in which method a message authentication code (MAC) is calculated for the transmitted message both by the device to be checked and by the testing device using a secret key and the respectively calculated MAC must match.
Encrypted transmission of the hash value from the device to be checked to the testing device also prevents other devices in the network, which have received the same random number and should have the same program and configuration data record but no longer have the latter on account of manipulation and therefore can no longer themselves calculate the correct hash value, from simply eavesdropping the transmission of the hash value and then themselves transmitting this hash value to the testing device in order to deceive the testing device.
According to one development, an identical parameter is transmitted to the device and to the at least one further device.
During parallel checking of a large number of devices in particular, an identical parameter, for example a random number which is sent to all devices to be checked, simplifies the check since the computing complexity on the testing device is kept low.
According to one development, the testing device calculates an inner state on the basis of the hash function, the parameter and the program data record for the purpose of calculating the test hash value on the basis of the identification and stores the inner state.
The size of a program data record is greatly dependent on the device and may range from a few kB—for example in the case of an RFID tag—to several dozen MB—for example in the case of intelligent field devices or programmable logic controllers (PLCs)—for devices to be checked. In contrast, the size of the configuration data record is usually considerably smaller. It is usually in the range of at most a few percent of the program data record. The size of the parameter is negligibly small in comparison with that of the program data record.
Therefore, the program data record gives rise to the greatest computing complexity when calculating the test hash value.
A system generally contains relatively few different program data records since the devices in a series usually all have the same firmware or software or there are at most a few different versions inside a system. The number of different program data records used in a system is therefore highly limited even if a very large number of devices are installed in the system.
If, when evaluating the property data and the further property data, the testing device now discerns that the respective program data records are identical, the testing device, when sending an identical parameter to the devices with an identical program data record, can calculate an inner state which jointly applies to devices with a common program data record. This reduces the computing complexity for the test server since, after calculating the inner state, it must only respectively calculate the test hash value on the basis of the inner state and the configuration data record which is now individually calculated for each device.
According to another embodiment, the testing device transmits a unique parameter to the device and transmits a further unique parameter which differs therefrom to the at least one further device, the unique parameter and the further unique parameter being transmitted within a predefinable time window.
An advantage of this virtually simultaneous checking of a large number of devices to be checked with different parameters, in particular random numbers, is that forwarding attacks are made more difficult. A device which is installed in the system for a forwarding attack and to which the calculation of a hash value is diverted in order to thus conceal manipulation of the device to be checked now has an extremely high testing complexity: it now receives a different parameter for calculating the hash value from each manipulated device whose request to calculate the hash value is forwarded. For example, a forwarding attack can be detected by unexpectedly long response times of the checked devices.
According to one embodiment, the testing device transmits a unique parameter to the device and transmits a further unique parameter to the at least one further device on the basis of the identification.
Therefore, on devices with the same program data record, it is possible to select a parameter from a small number of different parameters, with the result that an identical parameter cannot be assumed for devices with the same program data record. Although this increases the computing complexity for the testing device in comparison with an identical parameter for all devices in a system, it is therefore more difficult for manipulated devices to exchange information on the correct hash value. If, for example, a plurality of manipulated devices have together also stored all parts of an original program data record, they can create the correct hash value in cooperation. However, if they receive different parameters, this coordination is made more difficult and the computing complexity is increased since another hash value must now be calculated for each parameter received by one of the devices.
According to one development, the integrity of the device is confirmed by the testing device if the device calculates and transmits the hash value within an individually predefinable response time window.
In the case of a forwarding attack, it is now virtually impossible for a plurality of manipulated devices to respond by transmitting the hash value within the predefined response time window since a device to which the calculation of the hash value is diverted has to perform an excessively large amount of computing in order to respond in the predefined time. In particular, if the checking method of a large number of devices is started simultaneously by the testing device, a forwarding attack can therefore be detected.
According to one development, the testing device calculates the test hash value before or after transmitting the unique parameter.
If the test hash value is calculated separately for each device, there is a high computing complexity for the testing device, with the result that it is advantageous to already calculate the test hash values before transmitting the parameter, with the result that the testing device can quickly check the correctness of the hash value immediately upon receiving the latter from a device to be checked and can trigger an alarm if necessary. It is likewise possible for the testing device to calculate the test hash value after receiving the hash value as a response from the device to be checked, but an incorrect hash value can be detected only later and an alarm can be triggered, if necessary, only later. If there is a warning or indications of the manipulation of a device or a group of devices, this method can nevertheless be selected in order to ensure a high degree of reliability during checking.
The invention also comprises a testing system for checking integrity of property data relating to a device, the property data comprising a program data record and a configuration data record, by means of a testing device inside a network comprising the device, the testing device and at least one further device having further property data, the further property data comprising a further program data record and a further configuration data record, comprising:
According to one embodiment, the testing system also comprises at least one further unit for use in one of the method steps according to the above-mentioned embodiments of the method.
According to another embodiment, the testing device and the station are in the form of a common unit.
Therefore, there is no need to interchange data via the network such as, in particular, the often comprehensive property data.
The invention is explained in more detail below with exemplary embodiments using the figures, in which:
In the figures, identical or functionally identical elements have been provided with the same reference symbols unless indicated otherwise.
The fundamental method steps and the essential components of the method according to the invention are presented using
An automation system contains numerous devices such as measuring or control devices which need to be checked for manipulation in order to safely and reliably operate the system. In order to check a device 100, which is a robot controller of an industrial robot for example, it is proposed for a testing device 300, which is coupled to the device 100 via a network 400 of the automation system, to be provided with property data ED1 relating to the device 100 by a station 200 which is coupled at least to the testing device 300 via the network 400. The property data ED1 make it possible to uniquely identify the device 100, that is to say the robot controller. In this case, the station is in the form of an update server which continuously provides the testing device 300 with a current version of the property data ED1. The testing device 300 is also provided with further property data ED2 relating to a further device 102 inside the system by the station 200.
The property data ED1 and further property data ED2 made available to the testing device 300 are now evaluated. The property data ED1 and the further property data ED2 comprise a program data record PD1 and a configuration data record KD1 and a further program data record PD2 and a further configuration data record KD2 relating to the further device 102. The program data record PD1 contains an indication of the firmware on the robot controller. The configuration data record contains a device identifier and parameters which have been predefined for the execution of process steps of the industrial robot.
The respective property data which are made available to the testing device 300 by the station 200, the update server in this case, are now evaluated with regard to at least identity of the respective program data records. An identification Z is allocated to a program data record relating to a device to be checked, the program data record PD1 relating to the device 100 in the described example, on the basis of an evaluation result. The identification Z of program data records for which identity has been determined is carried out in this case by the testing device 300 logically grouping property data made available to it, for example storing property data made available to it with provided property data having an identical program data record in a common memory area. Robot controllers in a common series with identical version numbers can therefore be logically combined by the testing device 300.
On the basis of this identification Z, the testing device 300 now determines a parameter P which, on the one hand, is transmitted to the device 100 for the purpose of calculating a hash value H1 by means of the device 100 and, on the other hand, is used by the testing device 300 to calculate a test hash value H′1. The calculation of the hash value H1 by the device 100 and the calculation of the test hash value H′1 by the testing device 300 are each carried out on the basis of a hash function and a character string. In this case, the character string comprises the parameter P and the property data ED1, the character string respectively being hashed such that the parameter P is transferred first and the program data record PD1 and the configuration data record KD1 are then transferred.
In this case, the device 100 can calculate a correct hash value, which matches the test hash value H′1 calculated by the testing device 300, only when it carries out the calculation using the property data ED1 currently on the device 100 following a request to calculate the hash value H1—this is effected by transmitting the parameter P to the device 100. An inner state of the hash method on the basis of the property data ED1 therefore cannot be meaningfully precalculated and stored. If a parameter P newly determined for each testing method, for example a random number, is concomitantly included in the calculation and is transferred at the beginning of the character string, originally unmanipulated property data ED1 cannot be used to calculate an inner state of the hash method which would still provide a correct hash value even after the property data ED1 have been manipulated, as would be possible in a method in which the parameter P is at the end of the character string.
Inclusion of the further devices inside the system when determining the parameter P reduces the computing complexity for the testing device, as explained in more detail below using
An exemplary embodiment is explained using three devices to be checked. In a similar manner to the situation described above, the station 200 provides property data, that is to say the property data ED1, the further property data ED2 and third property data ED3 in this case. The property data ED1 comprise the program data record PD1 and the configuration data record KD1, the further property data ED2 likewise comprise the program data record PD1, which is identical to the program data record PD1 from the device 100, and the further configuration data record KD2, and the third property data ED3 comprise a third program data record PD3 and a third configuration data record KD3. The testing device 300 now identifies the property data ED1 and the further property data ED2 since the evaluation of the respective property data has revealed the identity of the respective program data records.
An identical parameter PC is now transmitted to the device 100 and to the further device 102 on account of the identical program data records. If no identity of the third program data record PD3 with one of the other program data records relating to the devices to be checked has been determined for the third device 103, an arbitrary parameter P is transmitted to the third device 103. In a similar manner to the method described above, the device 100, the further device 102 and the third device 103 calculate the hash value H1, a further hash value H2 and a third hash value H3. The calculated hash values are respectively transmitted from the respective devices to the testing device 300.
The testing device 300 can now calculate a respective test hash value according to a simplified method. For the program data record PD1 transmitted to it, the testing device 300 now first of all calculates an inner state IZ which depends on the hash function, the respective parameter PC and the program data record PD1. This inner state IZ is stored and the test hash value H′1 is then calculated, the test hash value H′1 being calculated on the basis of the hash function, the inner state IZ and the configuration data record KD1.
The inner state IZ is likewise used to calculate the test hash value H′2 of the further device 102, with the result that the further configuration data record KD2 only needs to be hashed.
This method can be expanded to any desired numbers of devices, the inner state IZ always being calculated and stored for respectively transmitted program data. This combination of the calculation of the inner state is effected independently of the number of available devices. The more devices can be combined using this method, the greater the reduction in the computing complexity.
In the case of large numbers of devices, this method therefore constitutes a great simplification for the testing device 300, which shall be illustrated using the following numerical example.
A size of program data relating to a device, for example firmware or software, is greatly dependent on the device and may range from a few kB (for example for RFID tags) to several dozen MB (for example in the case of intelligent field devices or programmable logic controllers). For the numerical example, an average size of program data GPD of GPD=32 MB shall be assumed. It shall be assumed that a program data number M of different program data records, in particular different series, of M=10 exists inside a system.
An average size of configuration data GKD is typically smaller than that of the program data and is GKD=0.1 MB in the example. A number of different devices, for example a number of 1000 devices inside a system, means a configuration data number N of N=1000.
For a checking method without use of an inner state IZ for devices to be checked with an identical program data record, the data volume G1 to be hashed for the testing device 300 is composed of:
N·GPD+N·GKD=G1
This corresponds to the case in which a hash value must be calculated for each device in a system and the data volume G1 to be hashed, as a sum, is composed of the product of the configuration data number N and the average size of program data GPD and the product of the configuration data number N and the average size of configuration data GKD.
If the testing device 300 first of all calculates an inner state IZ only for each different program data record and then calculates the hash value using the configuration data, the following further data volume G2 to be hashed results:
M·GPD+N·GKD=G2
For the variant without use of the inner state IZ, this means, in the numerical example, a data volume G1 to be hashed of:
1000·32 MB+1000·0.1 MB=32100 MB
In contrast, the further data volume G2 to be hashed, when using the inner state IZ, is only:
10·32 MB+1000·0.1 MB=420 MB
In this case, the data volume to be hashed is respectively correlated with a computing complexity which, in the case of calculating an inner state IZ, is likewise lower in a manner corresponding to the smaller data volume to be hashed, with the result that devices are checked more quickly.
For an increased security requirement, in a further exemplary embodiment,
The computing complexity for the testing device according to this method is increased since an inner state must now be calculated for each possible random number for each program data record occurring in the system. A third data volume G3 to be hashed results:
L·M·GPD+N·GKD=G3
In the numerical example selected above, a third data volume G3 to be hashed of:
10·10·32 MB+1000·0.1 MB=3300 MB
therefore results when selecting the random number from L=10 different random numbers.
This is a method with increased security since an agreement between manipulated devices, which exchange information in order to determine a correct hash value, is now made more difficult. For example, a plurality of manipulated devices may together reconstruct an original program data record. Although the program data record has been respectively changed for each device, it is nevertheless possible to calculate a correct hash value in cooperation. If, however, different random numbers are transmitted to the devices, this coordination is made more difficult. Each device must now already calculate an individually correct hash value on account of the random number which cannot be assumed to be the same, in which case the inner state IZ must also already be individually correctly determined, in particular on the basis of the random number and the program data record.
In the case of a large amount of available computing capacity of a testing entity, it is also conceivable to transmit a different parameter P to all devices to be checked inside a system. This method can be selected by the testing device 300, in particular, if there is sufficient time to calculate the test hash values or a general check with very high security requirements is pending. The parameters P specific to each device to be checked are then transmitted within a predefinable time window.
Integrity of the property data relating to a device to be checked is confirmed only if a response in the form of the hash value determined and transmitted by the device to be checked is given within a predefined response time window.
In this case, the testing device can take account of the fact that some devices to be checked which have an increased utilization of their processor—for example on account of extensive functions or a lower available computing capacity—require an accordingly longer response time and can either adapt the response time window in a device-specific manner or can accordingly start a checking request in the form of a transmitted parameter earlier for these devices. For the testing device 300, this method means an enormously high computing complexity, with the result that it is advantageous to calculate test hash values to be calculated in advance, that is to say before transmitting the parameters.
Forwarding attacks inside a system are made more difficult by this method since a virtually simultaneous check of all devices inside the system is started and each device must respond within a predefined response time window.
The presented exemplary embodiments can be combined with one another in an advantageous manner: for this purpose, the testing device, as a standard setting, can select the method in which a test hash value calculation is combined for devices with identical program data records, an identical parameter PC is transmitted to these devices and an inner state IZ is calculated, with the result that the computing complexity for the testing device is kept low even in the case of a large number of devices to be checked in a parallel manner. If manipulation of a group of devices with identical program data records is suspected, it is possible to deliberately resort to the method in which different parameters PS1, PS2 are transmitted to devices having identical program data records. This may be advantageous even when a new group of devices is installed inside the system.
Different security levels for different security requirements can be complied with inside a system by varying the number of parameters used, in particular the random numbers used. Unintentional software changes, for example caused by memory errors, or discrepancies between the assumed software or configuration stored in the engineering station and the actual software or configuration in the device which are produced by in-situ configuration by a service engineer and a lack of comparison with the engineering station are also detected. The combination of the different methods presented allows an efficient check in systems having a large number of installed devices.
The units of the device 100, further device 102, third device 103, testing device 300 and station 200 and the further unit can be implemented and executed using software, hardware or a combination of software and hardware.
The steps implemented by the units can be stored as program code on a storage medium, in particular a hard disk, a CD-ROM or a memory module, the individual instructions of the program code being read and processed by at least one computing unit comprising a processor. The processor is connected to the storage medium via a bus for the purpose of interchanging data.
Furthermore, an input/output unit can be connected via the bus, the input/output unit being able to be used to receive and/or transmit data, for example property data, a parameter, a hash value or a test hash value.
The described embodiments and developments can be freely combined with one another.
| Number | Date | Country | Kind |
|---|---|---|---|
| 102012217743.4 | Sep 2012 | DE | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2013/066648 | 8/8/2013 | WO | 00 |
