Method for securely transmitting data from a device

Abstract

The invention relates to a method for securely transmitting data from a device by means of one or more data characterizations, by means of which a respective data set can be generated from available data in each case and to each of which an encryption key is assigned, having the following steps which are carried out by means of a computing unit of the device: acquiring data; determining transmission data sets from the acquired data using the data characterizations; generating falsification data sets, a falsification data set being generated for each transmission data set; applying a falsification mapping, which is based on the falsification data set generated for the respective transmission data set, in order to map each transmission data set to a falsified transmission data set; encrypting the falsification data sets, each falsification data set being encrypted using the encryption key assigned to the data characterization used to determine the transmission data set for which the respective falsification data set was generated; and transmitting the falsified transmission data sets and the encrypted falsification data sets to one or more recipients.

Description

[0001] R.413671

[0002] - 1 -

[0003] Description

[0004] title

[0005] Methods for secure data transmission from a device

[0006] The present invention relates to a method for secure data transmission from a device, as well as a computing unit and a computer program for carrying it out.

[0007] Background of the invention

[0008] Vehicles and other devices can process image data and other data captured by sensors. In vehicles, this is particularly relevant in automated driving and driver assistance systems (AD / DA systems, AD: "assisted driving", DA: "driver assistance"). The resulting or captured data may include parts, such as specific image sections, that require protection, meaning they must not be disclosed to unauthorized parties. Examples include personal data, which may be subject to legal regulations, such as vehicle license plates or faces of individuals in camera recordings, or data used as the basis for the development of AD / DA systems. Relevant data can be transferred from vehicles, such as test vehicles with additional measurement technology, to various parties, e.g.,Developers of driving functions and / or AD / DA systems receive data, for example via mobile networks, to database infrastructures such as cloud or server.

[0009] Disclosure of the invention

[0010] According to the invention, a method for secure data transmission from a device, as well as a computing unit and a computer program, are described in R.413671.

[0011] - 2 - the implementation of which is proposed with the features of the independent claims. Advantageous embodiments are the subject of the dependent claims and the following description.

[0012] The invention employs the method of generating transmission data records from acquired data using a computing unit of the device, generating a falsification data record for each transmission data record, and applying a falsification image to each transmission data record, which depends on the falsification data record generated for the respective transmission data record, in order to determine a corresponding falsified transmission data record. Furthermore, the falsification data records or corresponding generation information are encrypted with encryption keys. The encryption keys are each assigned to the data characterization by which the transmission data record for which the respective falsification data record was generated was determined. The falsified transmission data records and the encrypted falsification data records are then...Generation information is transmitted to one or more recipients. This approach makes it possible to provide different recipients with different transmission data sets in unaltered form, depending on which decryption keys corresponding to the encryption keys they know.

[0013] In particular, the encryption keys for different data characterizations differ. The encryption of the falsified datasets or their generation information is carried out using a symmetric cryptographic encryption method (e.g., AES: Advanced Encryption Standard) or an asymmetric cryptographic encryption method (e.g., RSA: Rivest-Shamir-Adleman).

[0014] It is assumed that the data characterizations are not trivial in the sense that they always generate an empty set as the data set, regardless of the available data. R.413671

[0015] - 3 -

[0016] The falsified data sets are, for example, random strings or indistinguishable from random strings, where the term "character" is understood to mean a bit, a number or another character (byte, word, letter, ...).

[0017] In particular, the data manipulation is reversible if the source data set is known. The data manipulation can be chosen such that, when using a suitable source data set, the data of a transmitted data set is falsified in such a way that the data in the falsified transmitted data set is no longer recognizable; that is, without knowledge of the source data set, the original information of the data cannot be extracted.

[0018] According to one embodiment, a predetermined data structure for elements is given, which specifies an arrangement of the elements, wherein data elements of the captured data are arranged according to the data structure, wherein the data characterizations include at least one data characterization that defines a subset of the elements of the data structure or specifies a method by which a subset of the elements of the data structure can be defined using information contained in the data, and wherein the data elements of the captured data are selected as the transmission data set that correspond to the subset of the elements of the data structure.

[0019] In particular, the subset of elements of the data structure defined by a data characterization is a proper subset of the set of all elements of the data structure; that is, the subset is neither empty nor does it include all elements of the data structure. The union of all subsets defined by data characterizations may be equal to the set of all elements of the data structure, but need not be; that is, the union of all subsets defined by data characterizations may include only a subset of all elements of the data structure. In other words, the union of all subsets defined by data characterizations is a (proper or improper) subset of the set of all elements of the data structure, excluding the empty set. R.413671

[0020] - 4 -

[0021] An example of a data structure is the arrangement of pixels (where the data elements are, for example, color values ​​and / or grayscale values ​​and / or other properties of the pixels) in an image captured, for instance, by a device's camera. The pixels are arranged in a two-dimensional grid with M pixels in a horizontal (or first) direction and N pixels in a vertical (or second) direction. The subset of a data characterization is then a specific subset of the pixels. For example, a first data characterization might specify that every m-th pixel in the horizontal direction and every n-th pixel in the vertical direction is selected, where m and n are equal to or greater than 2. The resulting transmitted data set corresponds to a more or less blurred version of the original image, depending on m and n.Another, second data characterization, which can be used to determine a further transmission data set, can then specify as a subset that all pixels are selected except for every m-th pixel in the horizontal direction and every n-th pixel in the vertical direction. Both transmission data sets together then again yield the entire image, or rather, all pixels of the image.

[0022] The selection of specific pixels from an image described above is an example of the use of image masks (also called pixel masks), which each specify a subset of pixels to be selected from an image (as data) and which are disjoint (i.e., no pixel is included in two image masks). Image masks can be statically defined or fixed (for example, directly as data characterization in transmission orders), meaning they always affect the same sets of pixels for multiple images, or they can be dynamically generated or fixed, meaning they may affect different sets of pixels for multiple images.

[0023] In the latter case, i.e., a dynamic determination of image masks, the data characterization of a transmission order specifies, for example, a method by which image masks are generated. Information contained in images can be used for this purpose. For instance, an image recognition algorithm can be used to identify certain objects, such as license plates or faces, or generally objects requiring protection, in captured images (R.413671).

[0024] - 5 - and specify image areas in which detected objects are located. In a first data characterization, an image mask can then be generated for each image that includes all pixels except for the image areas in which detected objects are located. A further, second data characterization can also generate an image mask that includes all pixels located in the image areas containing the detected objects.

[0025] Of course, dynamic definition or selection of a subset of the elements or data elements of the data structure is also possible with data structures other than images (using a method specified in the data characterization that determines a subset). When dynamically defining image masks, or more generally, a subset of the elements or data elements of the data structure, it can be stipulated that this subset and / or information about the corresponding image areas or the subset itself is also transmitted to one or more recipients. Various possibilities are conceivable here. For example, in the case of images, the recognition data or corresponding image area data as determined by the image recognition algorithm can be transmitted, or more generally, the corresponding data with which the method selects the subset.It is also conceivable that the information about the image areas corresponding to the image masks, or more generally about the subset of the data structure, is contained in the falsification data record(s) or in the structure in which they are transmitted. For example, the falsification data records can be transmitted in a single, overall falsification data record, with elements of the overall falsification data record arranged according to the data structure. For each data element of a selected subset, a falsification element is determined and, after encryption, stored at the corresponding position in the overall falsification data record (according to the arrangement of the elements of the data structure). If several subsets are selected (e.g., through different data characterizations), this procedure is repeated for each selected subset (using the same overall falsification data record).Elements at positions in the overall falsification dataset that do not correspond to any subset can then be assigned a specific value (e.g., all with '1' or all with '0', especially in the case of an R.413671).

[0026] - 6 -

[0027] The XOR operation (as a falsification mapping) can be used. Alternatively, this can also be done in an initialization operation in which all elements of the overall falsification dataset are initialized with the specified value. This initialization is performed before the encrypted falsification data elements are stored, and the specified value stored during initialization is overwritten when the encrypted falsification data elements are stored. The resulting overall falsification dataset is an encrypted (since the falsification data elements are encrypted) dataset that can be transmitted.

[0028] Another example of data arranged in a data structure is numbers in a specific format, such as integers, floating-point numbers, or fixed-point numbers. The data structure is thus a predefined number format. Data characterizations can specify which digits (of the respective number format) are selected as a subset. For example, two data characterizations might be given: the first specifies the digits of the number format except for a certain number of significant digits that correspond to higher precision, and the second specifies the remaining significant digits. Knowing only the transmission data set generated by the first of these two data characterizations, the numbers are relatively imprecise. Knowing both transmission data sets generated by the data characterizations, the numbers have their original precision. The numbers can, for example,These could be measurement values ​​or position data already present in the device.

[0029] Data elements of data structures that are not selected by any of the data characterizations can optionally be transmitted in their unaltered form. That is, a data set or residual data set can be created from the data elements of data structures that are not selected by any of the data characterizations, which is then transmitted (in its original form) as an unaltered data set (or residual data set) to one or more recipients. Referring to the preceding examples, the first data characterizations could also be omitted, and only the second data characterization could be used to define a subset. R.413671

[0030] - 7 -

[0031] Subsets that would correspond to the initial data characterizations are then transmitted unaltered according to this optional procedure, so that the corresponding information (blurred image, less sensitive objects, inaccurate numbers) is available in unaltered form and is accessible even without a decryption key.

[0032] According to one embodiment, falsification elements are generated that are arranged according to the data structure. The falsification data set is generated by selecting those falsification elements that correspond to the defined subset of elements in the data structure. The falsification elements are thus determined in a one-to-one mapping with the data elements of the subsets of the captured data defined by the data characterization. In particular, the falsification elements are given in the same format as the data elements and / or have the same number of bits as the data elements.

[0033] According to one embodiment, the predetermined data structure is a two-dimensional grid (or array) with M elements in a first direction and N elements in a second direction, wherein the elements are in particular pixels or picture points; or wherein the predetermined data structure is a numerical format.

[0034] According to one embodiment, the predetermined data structure is a three-dimensional or multi-dimensional grid. Possible applications include, for example, a three-dimensional image from a stereo camera or lidar, where the elements can be pixels or image points in space and, if applicable, certain properties of these image points (color, density, reflectance properties, temperature, etc.).

[0035] More generally, the predetermined data structure can be a one- or multi-dimensional grid or array, with the data elements arranged according to the grid or array elements. For example, the data structure, as described above, is a one- or multi-dimensional recording (image) from a camera, with the data elements corresponding to picture points or pixels. The data elements can be any data values, for example, R.413671.

[0036] - 8 -

[0037] Color values ​​of the pixels and / or gray values ​​of the pixels and / or other properties assigned to the pixels (such as reflection properties, temperatures).

[0038] According to one embodiment, at least one of the data characterizations selects a subset of the captured data as the transmission data set and / or at least one of the data characterizations includes a characterization mapping, wherein a transmission data set is determined by applying the characterization mapping to the captured data. A characterization mapping can, for example, include a mathematical function that calculates data values ​​(also referred to as derived data) from the captured data, which are then contained in the transmission data set.

[0039] According to one configuration, the collected data includes first data and second data that are associated with the first data, and the data characterizations include a first and a second data characterization, whereby the first data is selected as a first transmission data record by the first data characterization, and the second data is selected as a second transmission data record by the second data characterization. The second data may, for example, include additional information about the first data.

[0040] According to one embodiment, the falsification mapping is a bitwise exclusive-OR operation of the respective falsification record with the respective transmission record. Bitwise exclusive-OR operations (XOR operation; i.e., as usual, a binary operation of two bits such that the result is "1" if the two bits are different, and "0" if the two bits are the same) can be performed quickly and efficiently; for example, XOR operations in processors are typically implemented as a machine instruction at the register level. Accordingly, the falsification records are generated, in particular, such that they have the same number of bits as the transmission records for which they are generated, and that a mapping of the bits to each other (between the bits of the transmission record and the R.413671

[0041] - 9 -

[0042] The bits of the corresponding falsification data set are given. The XOR operation is its own inverse.

[0043] According to one embodiment, the falsified data sets are generated as random strings and / or random bit sequences, and / or the falsified data sets are generated as pseudorandom strings (or pseudorandom number sequences) and / or pseudorandom bit sequences, whereby, in particular, starting values ​​of the pseudorandom strings or pseudorandom bit sequences are used as generation information. The falsified data sets generated in this way can be considered noise (e.g., image noise if the captured data consists of images recorded with a camera) that is superimposed on the data of the transmitted data sets, thus falsifying and rendering them unrecognizable. The starting values ​​can be chosen randomly and / or based on information stored in the device that is not precisely known outside the device, e.g., the time at which the generation of the pseudorandom number sequence begins.

[0044] According to one implementation, the data characterizations each generate a subset of the captured data as a dataset. Elements or parts of the captured data that are not included in any of the subsets generated by the data characterizations are transmitted to one or more recipients in unencrypted and / or unaltered form. Alternatively or additionally, a residual data characterization is provided, which generates a dataset from the captured data that is also transmitted to one or more recipients in unencrypted and / or unaltered form. Parts of the data that are not intended to be protected can thus be transmitted easily and without additional computational effort.

[0045] According to one embodiment, the method comprises transmitting, by at least one external computing unit, one or more data characterizations and one or more encryption keys to the device or to the device's computing unit. Alternatively, these may already be stored in the device's computing unit. R.413671

[0046] - 10 -

[0047] According to one embodiment, the method comprises, by a respective computing unit of a receiver that knows a corresponding decryption key or keys for part or all of the one or more encryption keys: receiving the falsified transmission data records and the encrypted falsification data records or the encrypted generation information; decrypting those encrypted falsification data records or encrypted generation information for which the receiver knows the decryption key, with the respective decryption keys, in order to form a set of available falsification data records that contains the one or more decrypted falsification data records.Contains one or more falsification records generated from decrypted generation information; and for each falsification record in the set of available falsification records, a calculation of the inverse of the falsification map using the respective falsification record and application of the calculated inverse of the falsification map to the corresponding falsified transmission record to obtain the unfalsified transmission record from which the falsified transmission record was determined. The corresponding falsified transmission record to which the inverse of the falsification map is applied is the falsified transmission record determined by means of the falsification map using the respective falsification record. Overall, depending on knowledge of the decryption keys, a recipient can determine certain transmission records in unfalsified form.When generation information is used, the recipients are aware of the procedure by which the falsification data records can be determined from this information.

[0048] A computing unit according to the invention or an arrangement of several computing units, e.g. a control unit or control unit group of a motor vehicle, is, in particular in terms of programming, equipped to carry out a method according to the invention.

[0049] The implementation of a method according to the invention in the form of a computer program or computer program product with program code for R.413671

[0050] - 11 -

[0051] Performing all process steps is advantageous because it incurs particularly low costs, especially if the executing control unit is already used for other tasks and is therefore already available. Finally, a machine-readable storage medium is provided with a computer program stored on it as described above. Suitable storage media or data carriers for providing the computer program include magnetic, optical, and electrical storage devices, such as hard drives, flash memory, EEPROMs, DVDs, etc. Downloading a program via computer networks (Internet, intranet, etc.) is also possible. Such a download can be wired or wireless (e.g., via a WLAN network, a 3G, 4G, 5G, or 6G connection, etc.).

[0052] Further advantages and embodiments of the invention will become apparent from the description and the accompanying drawing.

[0053] The invention is schematically illustrated in the drawing using exemplary embodiments and is described below with reference to the drawing.

[0054] Brief description of the drawings

[0055] Figure 1 shows a flowchart of the method for secure data acquisition from a device according to embodiments of the invention.

[0056] Figure 2 illustrates an example of a method for the secure transmission of data from a vehicle.

[0057] Figure 3 illustrates another example of a method for the secure transmission of data from a vehicle.

[0058] Figure 4 illustrates another example of a method for the secure transmission of data from a vehicle.

[0059] Embodiment(s) of the invention R.413671

[0060] - 12 -

[0061] Figure 1 shows a flowchart of the method for secure data transmission from a device (e.g. a vehicle) according to embodiments of the invention.

[0062] In optional step 100, one or more data characterizations, each associated with an encryption key, are transmitted from an external computing unit to a computing unit within the device. The data characterizations and the encryption keys are mapped to each other in a one-to-one relationship.

[0063] Each data characterization specifies how a dataset can be derived or generated from existing, especially previously collected, data (e.g., as a subset of the existing data and / or by means of a mapping from the existing data). The multiple data characterizations are disjoint in the sense that certain datasets derived from the same data with different data characterizations are contravalent or disjoint (i.e., they do not share a common element).

[0064] The encryption keys can be cryptographic keys used in either a symmetric or an asymmetric cryptographic method. In the latter case, the encryption keys are, for example, public keys of respective key pairs consisting of a public key (used to encrypt data) and a private key (used to decrypt the encrypted data). Using an asymmetric method is advantageous because the encryption key does not need to be kept secret.

[0065] A pair formed by a data characterization and its associated encryption key, or a set of several such pairs, can be considered a capture job. Generally, one or more such capture jobs can be transmitted to the device. R.413671

[0066] - 13 -

[0067] It is also possible that one or more data characterizations and their associated encryption keys are already present or stored in the device, i.e., in the device's processing unit. In this case, step 100 can be omitted.

[0068] The following steps 110 to 160 are performed by the device's computing unit.

[0069] In step 110, data is acquired; that is, data present in the device is collected, such as sensor data and / or camera data and / or data from a device included in the device (e.g., a motor) and / or position data (if, for example, a GPS sensor or similar is provided in the device). Since the device can generally contain a very large amount of data, for example, in various categories (sensor, camera, etc.), it may be specified, for example, as part of an acquisition order, which data, and of which category, should be acquired. However, this can also be part of the data characterization.

[0070] In step 120, transmission data records are determined. Each data characterization generates a transmission data record from the collected data. This means that each data characterization is applied to the collected data, if data characterization is understood in a general sense as a mapping (e.g., projection onto a subset and / or function).

[0071] In step 130, a falsification record is generated for each transmission record. After generation, each falsification record is initially known only within the device or its processing unit. For example, each falsification record is generated as a random or pseudorandom string (where a character can be a single bit, a number, or another type of character). In the case of a pseudorandom string, this can be determined, in particular, from a seed value, which is determined as a random value or as a seed value unknown outside the device or its processing unit (e.g., derived from a time and / or a device position at the time the seed value is determined). The seed value is an R.413671

[0072] - 14 -

[0073] Generation information from which the pseudorandom string can be generated.

[0074] In step 140, a corruption mapping, dependent on the corruption record generated for the respective transmission record in step 130, is applied to each of the transmitted data records to determine a corresponding corrupted transmission record. The corruption mapping is reversible if the corruption record is known; that is, if the corruption record is known, the original transmission record can be recovered from the corrupted transmission record. An example of a corruption mapping is a bitwise XOR operation between bits of the corruption record and bits of the transmitted record. The reversal consists of applying the bitwise XOR operation again.The falsification image should be cryptographic in the sense that, without knowledge of the falsification data set, the transmission data set can only be recovered from the falsified transmission data set by guessing the falsification data set or the transmission data set itself.

[0075] In step 150, the falsification data records are encrypted. For each falsification data record, the encryption key used is that which corresponds to the data characterization with which the transmission data record for which the respective falsification data record was generated. If the falsification data records can be generated from generation information (e.g., the aforementioned starting value of a pseudorandom string), the corresponding generation information can alternatively be encrypted instead of the falsification data records themselves. The encryption process results in either encrypted falsification data records or encrypted generation information.

[0076] In step 160, the falsified transmission data records and the encrypted falsification data records or the encrypted generation information are transmitted to one or more recipients. R.413671

[0077] - 15 -

[0078] In addition to the corrupted transmission datasets, certain datasets, e.g., parts of the captured data that are not selected by any of the data characterizations, can be transmitted in their original form, i.e., as uncorrupted datasets or uncorrupted transmission datasets. There may also be an additional data characterization, referred to as residual data characterization, which generates datasets from the captured data that are to be transmitted in their original form.

[0079] Steps 110 to 160 can be repeated for each new piece of data generated in the device (indicated by an arrow from step 160 to 110).

[0080] The one or more recipients can include various types of recipients. For example, a recipient might not possess a decryption key capable of decrypting data encrypted with one of the encryption keys. Such a recipient cannot recover the original transmission data from any of the corrupted transmission data and therefore only has access to the transmitted, uncorrupted data. In the case of a symmetric encryption method, a decryption key is the encryption key itself; in the case of an asymmetric encryption method, it is the private key that forms a key pair with the public key that serves as the encryption key.

[0081] The subsequent steps 170 and 180 refer to recipients who either know the decryption keys for all encryption keys (first recipient) or know the decryption keys for at least one but not all encryption keys (second recipient).

[0082] In step 170, the recipients decrypt the encrypted falsification data records or the encrypted generation information using the decryption keys known to them. The first recipient can therefore decrypt all encrypted falsification data records or generation information, and the second recipient can decrypt those encrypted falsification data records or generation information for which R.413671

[0083] - 16 - the second recipient knows the decryption keys, i.e., at least one. If generation information is used, after decryption of the generation information, the (unencrypted) falsification data records are generated using the generation information, assuming that the recipients know the method (e.g., random number generator) with which the falsification data records (e.g., as a pseudorandom character and / or bit sequence) were obtained from the generation information.

[0084] In step 180, each recipient takes each corrupted data record (to the extent known to that recipient) and constructs the inverse of the corrupting mapping, applying it to the corresponding corrupted transmitted data record to obtain the original, uncorrupted transmitted data record. The first recipient can thus determine all transmitted data records. The second recipient can only determine those transmitted data records for which they possess a corresponding decryption key (i.e., the decryption key that can decrypt data encrypted with the encryption key associated with the data characterization used to generate the transmitted data record).

[0085] Overall, different recipients can come into possession of different transmission data records, depending on which decryption keys they have.

[0086] Figure 2 illustrates an example of a method for secure data transmission from a vehicle 2 as an exemplary device.

[0087] The vehicle 2 has a camera 4 that captures images of the vehicle's surroundings and a processing unit 6, e.g., a control unit of the vehicle, which performs steps 110 to 160 with reference to Figure 1. The camera 4 captures, for example, two-dimensional images, but could also be a stereo camera that captures three-dimensional images. R.413671

[0088] - 17 -

[0089] A client 20 or a party transmits, e.g., via an external computing unit over a mobile network connection or similar, a first data characterization 22a and one of its associated first encryption keys 24a, as well as a second data characterization 22b and one of its associated second encryption keys 24b, to the vehicle 2 or its computing unit 6, which stores them. This corresponds to step 100 of Figure 1. The data characterizations 22a, 22b and the associated encryption keys 24a, 24b can together be considered a data acquisition order 26.

[0090] Data characterizations 22a and 22b are presented as image masks, where the black dots or areas illustrate the pixels that are to be selected from the images by the respective data characterization. For the first data characterization, 22a, this involves a relatively small number of pixels, while for the second data characterization, 22b, it involves a relatively large number of pixels—namely, those not selected by the first data characterization, 22a. The image masks are, for example, complementary to each other (i.e., they share no pixels and together represent all the pixels), which, for graphical purposes, is illustrated by different pixel densities.

[0091] The encryption keys 24a, 24b are, for example, public keys of key pairs that can be used in asymmetric cryptographic methods.

[0092] As an example, an image 10 captured by the camera is recorded as data by the processing unit 10. Using the data characterizations, a first transmission data record 28a (with the first data characterization 22a) and a second transmission data record 28b (with the first data characterization 22a) are determined or generated from the recorded data, i.e., the image 10. Since the data characterizations are given as image masks, this can be done, as symbolically illustrated, by an AND operation; that is, specific pixels are selected from the image by the respective image mask, e.g., those indicated as black dots in the image mask. If the first data characterization selects relatively few pixels, as illustrated, the first R.413671

[0093] - 18 -

[0094] Transmission data set 28a can be viewed as a blurred version of image 10.

[0095] Furthermore, a first falsification data set 30a and a second falsification data set 30b are generated, wherein the individual elements of the falsification data sets are arranged according to the corresponding data characterization or image mask, so that each element of the transmission data sets corresponds to an element of the corresponding falsification data set, i.e., each element of the first transmission data set 28a corresponds to an element of the first falsification data set 30a and each element of the second transmission data set 28b corresponds to an element of the second falsification data set 30b.

[0096] In the depicted case, where image masks are used for data characterization, a total falsification dataset 32 ​​is first created or generated. From this, as symbolically represented as an AND operation, the first falsification dataset 30a is generated using the first data characterizations 22a, and the second falsification dataset 30b is generated using the second data characterizations 22b, whereby the respective elements are selected from the total falsification dataset 32. The falsification datasets 30a, 30b, and the total falsification dataset 32 ​​are generated specifically as random strings or pseudorandom strings or pseudorandom sequences of random numbers.

[0097] A distortion mapping 44, dependent on the corresponding distortion data set 30a, 30b, is applied to each of the transmission data sets 28a, 28b. The distortion mapping 44 is, in particular, a bitwise XOR operation. Accordingly, the elements of the distortion data sets 30a, 30b should have the same number of bits as the corresponding elements (pixels) of the transmission data sets 28a, 28b. In the example of Figure 2, where image masks are used as data characterizations, the XOR operation (as shown) can also be performed using the overall distortion data set 32, assuming that the elements of the transmission data sets 28a, 28b and the overall R.413671

[0098] - 19 -

[0099] The falsification data set 32 ​​is arranged according to the pixel structure of the image and processed together. By applying the falsification mapping 44, a first falsified transmission data set 34a is obtained from the first transmission data set 28a and the first falsification data set 30a, and a second falsified transmission data set 34b is obtained from the second transmission data set 28b and the second falsification data set 30b. The falsification data sets 30a and 30b can be considered noise, so that the original content of the (unfalsified) transmission data sets 28a and 28b is no longer recognizable in the falsified transmission data sets 34a and 34b. The falsified transmission data sets 34a and 34b can be combined into a single falsified transmission data set 34 (as a union).

[0100] The falsification data records 30a and 30b are encrypted with their respective encryption keys 24a and 24b. That is, the first falsification data record 30a is encrypted with the first encryption key 24a to obtain the first encrypted falsification data record 36a, and the second falsification data record 30b is encrypted with the second encryption key 24b to obtain the second encrypted falsification data record 36b. The encrypted falsification data records 36a and 36b can be combined into a single encrypted falsification data record 36 (as a union).

[0101] The falsified total transmission record 34 and the encrypted total falsification record 36, i.e., the falsified transmission records 34a, 34b and the encrypted falsification records 36a, 36b, are transmitted to recipients. Of course, the falsified transmission records 34a, 34b and the encrypted falsification records 36a, 36b can also be transmitted separately, i.e., without first creating the falsified total transmission record 34 and the encrypted total falsification record 36. In this case, the total transmission record 34 and the encrypted total falsification record 36 can, for example, be transmitted to a database or database server 38 (which may also be R.413671).

[0102] - 20 - can be implemented as a cloud server) are transmitted, from which the recipients can retrieve them.

[0103] The recipients are, for example, a first recipient 46 and a second recipient 47, who is, for instance, the client 20 and who possesses a first decryption key 40a and a second decryption key 40b. The first recipient 46 possesses only the first decryption key 40a. The first decryption key 40a forms a key pair with the first encryption key 24a, and the second decryption key 40b forms a key pair with the first encryption key 24b. A key pair is defined as a pair consisting of a public key (for encryption) and a private key (for decryption) that can be used in asymmetric cryptographic encryption methods. It is also conceivable to use a symmetric cryptographic encryption method, in which case the decryption key is the same as the encryption key.

[0104] On the receiver side, i.e., by a processing unit of the respective receiver, the encrypted falsification data records 36a, 36b are extracted from the encrypted total falsification data record 36 using the data characterizations 22a, 22b, i.e., the image masks, employing an AND operation, analogous to the formation of the transmission data records 28a, 28b from image 10 and the formation of the falsification data records 30a, 30b from the total falsification data record 32. Thus, the first encrypted falsification data record 36a is obtained from the encrypted total falsification data record 36 using the first data characterization 22a, and the second encrypted falsification data record 36b is obtained from the encrypted total falsification data record 36 using the second data characterization 22b.Similarly, on the recipient side, the first corrupted transmission data record 34a and the second corrupted transmission data record 34b are extracted from the corrupted overall transmission data record 34 using the data characterizations 22a and 22b. If the corrupted transmission data records 34a and 34b and the encrypted corrupted data records 36a and 36b are transmitted separately, the extraction steps can be omitted. R.413671.

[0105] - 21 -

[0106] Provided that the recipient knows the respective decryption key, the first encrypted falsification data record 36a is decrypted using the first decryption key 40a and the second encrypted falsification data record 36b is decrypted using the second decryption key 40b in order to obtain the first and / or the second falsification data record.

[0107] If available at the respective recipient, the first falsification record 30a is used in an inverse 45 of the falsification map, which is applied to the first falsified transmission record 34a to obtain the first transmission record 28a, and the second falsification record 30b is used in the inverse 45 of the falsification map, which is applied to the second falsified transmission record 34b to obtain the second transmission record 28b. The inverse 45 of the falsification map is also a bitwise XOR operation if the falsification map is a bitwise XOR operation.

[0108] The first receiver 46, who only knows the first decryption key 40a, can only recover the first transmission data record 28a, i.e., a blurred version of the image. The second receiver 47, who knows both decryption keys 40a and 40b, can recover both transmission data records 28a and 28b and thus form the complete image 10 as a union.

[0109] The procedure illustrated in Figure 2 can be applied more generally to data whose data elements are arranged in a predetermined data structure, whereby operations are always assumed to preserve the arrangement of the elements in processed data sets, etc., according to the data structure. In Figure 2, the images, which are, for example, pixels arranged in a two-dimensional grid (or array) with M pixels in a horizontal direction and N pixels in a vertical direction, represent such a data structure. R.413671

[0110] - 22 -

[0111] Figure 3 illustrates another example of a method for secure data transmission from a vehicle 2 as an exemplary device, which, as in Figure 1, includes a camera 4 that captures images 10 of the vehicle's surroundings and a processing unit 6, e.g., a control unit of the vehicle, which performs steps 110 to 160 with reference to Figure 1. The procedure shown largely corresponds to that of Figure 1, so that the same or similar reference symbols are used and reference is also made to the description of Figure 1 for the description of identical steps or elements.

[0112] As shown in Figure 1, a client 20 transmits a first data characterization 52a and an associated first encryption key 24a, as well as a second data characterization 52b and an associated second encryption key 24b, to the computing unit 6 of the vehicle 2. The descriptions given in relation to Figure 1 apply to the encryption keys 24a, 24b and the corresponding decryption keys 40a, 40b.

[0113] The first data characterization 52a, for example, selects from the images 10 or image sequences captured by camera 4 those as transmission data set 28a in which predetermined objects or situations occur, whereby their occurrence can be recognized by a suitable image recognition algorithm. Predetermined objects or situations can be, for example, people, traffic lights, certain traffic signs (e.g., stop signs), etc. The second data characterization 52b, for example, determines additional information for the images 10 or image sequences selected by the first data characterization 52a as the second transmission data set 28b. Additional information can be, for example, a vehicle position or vehicle-internal data, e.g., from an autonomous driving algorithm. Based on such additional information, the behavior of certain functionalities of the vehicle, e.g., the autonomous driving algorithm, can be determined when the predetermined objects or situations occur.Situations are recognized and further developed.

[0114] For the first transmission data record 28a, a first falsification data record 30a is determined, and for the second transmission data record 28b, an R.413671 is determined.

[0115] - 23 - second falsification data set 30b determined, e.g. as random or pseudorandom strings.

[0116] As described in connection with Figure 1, the falsified data records 30a and 30b are used in the falsification transformation 44 (for example, again an XOR operation) and encrypted with the corresponding encryption key. A first falsified transmission data record 34a is determined from the first transmission data record 28a and the first falsified data record 30a using the falsification transformation 44, and a second falsified transmission data record 34b is determined from the second transmission data record 28b and the second falsification data record 30b using the falsification transformation 44.Furthermore, the first falsification data record 30a is encrypted with the first encryption key 52a to determine a first encrypted falsification data record 36a, and the second falsification data record 30b is encrypted with the second encryption key 52b to determine a second encrypted falsification data record 36b.

[0117] The two falsified transmission data records 34a, 34b and the two encrypted falsification data records 36a, 36b are transmitted to recipients 46, 47, whereby, as shown, the falsified transmission data records 34a, 34b and the encrypted falsification data records 36a, 36b can first be transmitted to a database server 38 and from this server to the recipients.

[0118] As in Figure 1, a first recipient 46 knows only the first decryption key 40a (which can be used to decrypt data encrypted using the first encryption key 52a) but not the second decryption key 40b (which can be used to decrypt data encrypted using the second encryption key 52b). The second recipient 47, who is, for example, the client 20, knows both decryption keys 40a and 40b. R.413671

[0119] - 24 -

[0120] Depending on their knowledge of the decryption keys, recipients 46 and 47 decrypt the first encrypted falsification record 36a with the first decryption key 40a to obtain the first falsification record 30a, and / or the second encrypted falsification record 36b with the second decryption key 40b to obtain the second falsification record 30b. The first falsification record 30a is used in the inverse operation 45 of the falsification mapping to determine the first transmission record 28a from the first falsified transmission record 34a. The second falsification record 30b is used in the inverse operation 45 of the falsification mapping to determine the second transmission record 28b from the second falsified transmission record 34b.

[0121] Accordingly, the first recipient 46 receives only the first transmission data record 28a, i.e., the images or image sequences containing the predetermined objects or situations, but not the second transmission data record 28b, i.e., the additional information. The second recipient 47 receives both transmission data records 28a and 28b.

[0122] Figure 4 illustrates another example of a method for secure data transmission from a vehicle. This exemplary device, like the one in Figure 1, includes a camera 4 that captures images of the vehicle's surroundings and a processing unit 6, e.g., a control unit of the vehicle, which performs steps 110 to 160 as described in Figure 1. The procedure shown largely corresponds to that of Figure 1, so the same or similar reference symbols are used, and reference is also made to the description of Figure 1 for identical steps or elements.

[0123] A first data characterization 62a and an associated first encryption key 24a are transmitted from a first client 20 to the computing unit 6 of vehicle 2, and a second data characterization 62b and an associated second encryption key 24b are transmitted from a second client 21 to the computing unit 6 of vehicle 2. For the encryption keys 24a, 24b, and R.413671

[0124] - 25 - the statements made in relation to Figure 1 apply to the corresponding decryption keys 40a, 40b.

[0125] Data characterizations 62a and 62b select, for example, from the images 10a and 10b or image sequences captured by camera 4, those images or image sequences as transmission data sets 28a and 28b in which predetermined objects appear, the appearance of which can be recognized by a suitable image recognition algorithm. In the case of the first data characterization 62a, these are, for example, persons, so that the first transmission data set 28a comprises images or image sequences that show (at least) one person. In the case of the second data characterization 62b, these are, for example, animals, so that the second transmission data set 28b comprises images or image sequences that show (at least) one animal. Overlaps are possible here, i.e., an image containing both a person and an animal can belong to both transmission data sets 28a and 28b.

[0126] For the first transmission data record 28a, a first falsification data record 30a is determined, and for the second transmission data record 28b, a second falsification data record 30b is determined. Furthermore, from the first transmission data record 28a and the first falsification data record 30a, a first falsified transmission data record 34a is determined using the falsification mapping 44, and from the second transmission data record 28b and the second falsification data record 30b, a second falsified transmission data record 34b is determined using the falsification mapping 44.

[0127] The distortion data sets 30a, 30b are generated in such a way that only the areas containing detected objects are distorted. If the distortion mapping 44 is an XOR operation, approximately 10 data bits of the distortion data sets, which here have a data structure with M x N pixels similar to Figure 1, that lie outside the respective area containing the detected object, are set to 0.

[0128] Alternatively, this configuration can be implemented such that the first and second data characterizations 62a, 6b determine image masks that dynamically R.413671

[0129] - 26 - are, i.e., are changed in position and / or size and / or shape so that the respective detected object lies within the area selected by the respective image mask. The position and / or size and / or shape can be determined by means of the image recognition algorithm used to detect the objects.

[0130] The transmission data records 28a and 28b then contain the elements (pixels) selected by the image masks (i.e., in which the respective detected object is located). Additionally, the transmission data records 28a and 28b can include information specifying the position, size, and / or shape of the area, or the image mask itself.

[0131] The falsification data records 30a and 30b are also determined using the dynamic image masks, meaning they only include falsification elements that correspond to the pixels of the image masks. Accordingly, the falsification operation then only affects the areas of the image masks. The falsified transmission data records 34a and 34b also only include the areas of the image masks.

[0132] In this implementation, image areas that fall neither under the first nor under the second data characterization 62a, 62b can be transmitted unaltered, e.g. as a residual data transmission data set, in order to obtain these image areas on the receiver side.

[0133] The falsification data sets 30a, 30b are in any case encrypted using the encryption keys 24a, 24b, so that a first encrypted falsification data set 36a is obtained as the version of the first falsification data set 30a encrypted with the first encryption key 24a and a second encrypted falsification data set 36b is obtained as the version of the second falsification data set 30b encrypted with the second encryption key 24b.

[0134] The falsified transmission data records 34a, 34b can be combined into a falsified overall transmission data record 34, whereby all R.413671

[0135] - 27 -

[0136] Elements (pixels) in image areas that fall under no or only one data characterization are only included once in the falsified total transmission data set 34.

[0137] The corrupted total transmission record 34 and the two encrypted corrupting records 30a, 30b, or (in the alternative implementation) the two corrupted transmission records 34a, 34b, the data residue transmission record and the two encrypted corrupting records 30a, 30b, are transmitted to recipients, e.g. via a database server 38.

[0138] Shown are a first recipient, who is the first client 20 and who knows the first decryption key 40a but does not know the second decryption key, and a second recipient, who is the second client 21 and who knows the second decryption key 40b but does not know the first decryption key.

[0139] Depending on their knowledge of the decryption keys, the recipients can each decrypt one of the encrypted falsification data sets 30a, 30b and use in reverse 45 the falsification image, which is applied to the falsified total transmission data set 34 or the falsified transmission data sets 34a, 34b to obtain images 11a, 11b.

[0140] The first recipient, i.e., the first client 20, can reconstruct images 11a in which objects falling under the first data characterization 62a are clearly identifiable and objects falling under the second data characterization 62b are not identifiable. The second recipient, i.e., the second client 21, can reconstruct images 11b in which objects falling under the second data characterization 62b are clearly identifiable and objects falling under the first data characterization 62a are not identifiable.

Claims

R.413671 - 28 - Claims 1. Method for secure data transmission from a device (2) using one or more data characterizations (22a, 22b, 52a, 52b, 62a, 62b) and one or more encryption keys (24a, 24b), wherein each encryption key is assigned to a data characterization, wherein each data characterization can generate a data set from available data, wherein the data sets generated from the same available data with different data characterizations are disjoint; the method comprising, by means of a computing unit (6) of the device (2): acquisition (110) of data; Determine (120) transmission data records (28a, 28b), wherein each of the transmission data records is generated from the captured data by one of the data characterizations (22a, 22b, 52a, 52b, 62a, 62b); Generating (130) falsification records (30a, 30b), wherein for each transmission record (28a, 28b) a falsification record is generated; Apply (140) to each of the transmission data records (28a, 28b) a falsification mapping (44) which is dependent on the falsification data record (30a, 30b) generated for the respective transmission data record and maps the respective transmission data record (28a, 28b) to a falsified transmission data record (34a, 34b); Encrypting (150) the falsification data records (30a, 30b) or of the generation information with which they were generated, wherein each falsification data record or its generation information is encrypted with the encryption key (24a, 24b) which is associated with the data characterization (22a, 22b, 52a, 52b, 62a, 62b) by which the transmission data record for which the respective falsification data record was generated was determined; and R.413671 - 29 - Transmit (160) the falsified transmission data records (34a, 34b) and the encrypted falsification data records (36a, 36b) or the encrypted generation information to one or more recipients (46, 47).

2. The method of claim 1, wherein a predetermined data structure for elements is given, specifying an arrangement of the elements; wherein data elements of the acquired data are arranged according to the data structure; wherein the data characterizations (22a, 22b, 52a, 52b, 62a, 62b) include at least one data characterization that defines a subset of the elements of the data structure or specifies a method by which a subset of the elements of the data structure can be defined using information contained in the data; wherein the data elements of the acquired data are selected as a transmission data set (28a, 28b) corresponding to the subset of the elements of the data structure.

3. Method according to claim 2, wherein falsification elements are generated which are arranged according to the data structure; wherein the falsification data set (30a, 30b) is generated by selecting those falsification elements which correspond to the specified subset of the elements of the data structure.

4. A method according to claim 2 or 3, wherein the predetermined data structure is a two-dimensional grid with M elements in a first direction and N elements in a second direction, wherein the elements are in particular pixels or picture points; or wherein the predetermined data structure is a numerical format.

5. A method according to any one of the preceding claims, wherein at least one of the data characterizations (22a, 22b, 52a, 52b, 62a, 62b) selects a subset of the recorded data as a transmission data set (28a, 28b); and / or wherein at least one of the data characterizations includes a characterization mapping, wherein a R.413671 - 30 - The transmission data set is determined by applying the characterization mapping to the captured data.

6. A method according to any of the preceding claims, wherein the acquired data includes first data and second data associated with the first data, and the data characterizations include a first and a second data characterization (52a, 52b); wherein the first data is selected by the first data characterization (52a) as a first transmission data record (28a) and the second data is selected by the second data characterization (52b) as a second transmission data record (28b).

7. Method according to one of the preceding claims, wherein the falsification image (44) is reversible upon knowledge of the falsification data set (30a, 30b).

8. Method according to one of the preceding claims, wherein the falsification image (44) is a bitwise exclusive-OR operation of the respective falsification data set (30a, 30b) with the respective transmission data set (28a, 28b).

9. Method according to one of the preceding claims, wherein the falsification data sets (30a, 30b) are generated as random strings and / or random bit sequences; and / or wherein the falsification data sets (30a, 30b) are generated as pseudorandom strings and / or pseudorandom bit sequences, wherein in particular starting values ​​of the pseudorandom strings or pseudorandom bit sequences are used as generation information.

10. Method according to any of the preceding claims, wherein the data characterizations (22a, 22b, 52a, 52b, 62a, 62b) each generate a subset of the captured data as a data set, wherein elements or parts of the captured data that are not contained in any of the subsets generated by the data characterizations are stored in unencrypted form. R.413671 - 31 - and / or in unaltered form to the one or more recipients (46, 47); and / or wherein a residual data characterization is given by which a data set is generated from the collected data which is transmitted in unencrypted and / or unaltered form to the one or more recipients (46, 47).

11. Method according to any of the foregoing claims, further comprising: Transmit (100) by at least one external computing unit, one or more data characterizations (22a, 22b, 52a, 52b, 62a, 62b) and one or more encryption keys (24a, 24b) to the device (2) or to the computing unit (6) of the device.

12. Method according to one of the preceding claims, further comprising a respective computing unit of a receiver (46, 47) which knows a corresponding decryption key or decryption key (40a, 40b) for part or for all of the one or more encryption keys (24a, 24b): Receiving the falsified transmission data records (34a, 34b) and the encrypted falsification data records (36a, 36b) or the encrypted generation information; Decrypting (170) those encrypted forgery records (34a, 34b) or encrypted generation information for which the recipient (46, 47) knows the decryption key, with the respective decryption keys (40a, 40b) to form a set of available forgery records containing one or more decrypted forgery records or containing one or more forgery records generated from decrypted generation information; for each forgery record (30a, 30b) in the set of available forgery records: Forming (180) the inverse (45) of the falsification image with the respective falsification data set (30a, 30b) and applying the formed inverse (45) of the falsification image to the corresponding R.413671 - 32 - falsified transmission record (34a, 34b) in order to obtain the transmission record (28a, 28b) from which the falsified transmission record was determined.

13. Computing unit or arrangement of several computing units configured to perform all process steps of a method according to any one of claims 1 to 12.

14. Computer program that causes a computing unit (2) or an arrangement of several computing units to perform all the procedural steps of a The method according to one of claims 1 to 10 can be carried out when it is performed on the computing unit or the multiple computing units.

15. Machine-readable storage medium with a computer program stored thereon according to claim 14.

Images