EP1602024A1 - Secure computer data storage method and device - Google Patents

Abstract

The invention relates to a secure, tamper- and forgery-proof peripheral for the storage of authenticated, dated computer data with an evidential value. The invention is based on an optionally-rewritable, fully-integrated storage peripheral which is functionally protected by an electronic system which only authorises the addition of data in the unused free space and prohibits the deletion, alteration or overwriting of pre-recorded data. At the end of each file-write operation, the date, which is provided by a non-modifiable internal clock that is calibrated only at the time of production, is associated therewith. Each device comprises a unique integrated secret internal digital identifier which is associated with a unique public identifier and said identifiers are used to personalise the device and render same unreproducible. The aforementioned association can be verified using several third methods which never divulge the secret internal identifier. According to the invention, a physical and electronic peripheral protection system activates, incorporates and protects the assembly of above-mentioned internal elements and destroys the unique internal identifiers in the event of an attempted intrusion, compromise, corruption or internal analysis. For said purpose, the authenticity of the equipment assembly is invalidated and, consequently, so too are the stored data which nonetheless remain readable. The inventive device is particularly suitable for the private storage and archiving of computer data having an authentic, juridical and/or legal weight of evidence in respect of official authorities. Moreover, said device uses no heavy cryptography methods.

Description

 Method and device for secure computer data storage

The present invention relates to methods and devices for storing secure and tamper-proof computer data intended to guarantee the authenticity of the data recorded on a given date, as well as to methods and methods of production, verification, authentication, operation. associated.

Traditionally, such systems have been implemented on single-write storage devices, which rely on the ability of the medium used to be physically changed irreversibly. These media (fixed or mostly removable) are easy to reproduce in their entirety. For example, non-rewritable discs CDR DVDR very common type and are not, most of the time, identified individually and can therefore be handled on an ^external system, and completely copied after tampering on an identical blank media . Similarly, it is possible to corrupt the recording of a write-only media by forcing the rewriting on tracks already containing data, the reader-recorder offering little or no protection against sending native commands d access to the media. The unique identification of the media often comes up against the ease of reproduction or falsification, and the fact that the media is not self-powered. It therefore generally has only a very low level of security. Magnetic media are denser and more powerful, but are natively multi-write, so they are even easier to corrupt.

One of the ways to overcome the weaknesses of removable or rewritable media is to use methods cryptographic which effectively allow at the time of writing, to authenticate the data but which increase the vulnerability to unique errors and often make data sharing between users tedious. These algorithmic methods, often complex, however, do not directly protect the media and / or the reader, remain costly in computing resources and are subject to obsolescence over time. If they prove to be efficient and necessary in the real-time transactional field, they have many disadvantages since either you want to use a large amount of data, or the sharing of data between several users is necessary, or that the data must be kept for several years. Access to the media makes any cryptographic method obsolete in any case, the media can be duplicated and falsified a posteriori.

The problem of dating is added to the preceding ones, the systems which date the files often presenting a great vulnerability, and seldom preventing the voluntary or involuntary corruption of the dates of the computer files. The structure . commonly used file systems are known and the associated dating is very rarely protected or encrypted, therefore vulnerable.

The ultimate solution is to use third-party organizations, which store redundant data, possibly encrypted, and ensure their physical protection. These methods can be effective, but often require costly logistics depending on the amount of data to be transmitted. They use communication links subject to unavailability and / or poorly protected public networks. Other similar data authentication solutions go through digital signature methods, the third party organization only keeping the signatures of the files. Authenticity can be established by comparison of signatures. This does not prevent the problems linked to the possibility that the original data will be voluntarily or involuntarily corrupted in the long term or that the evolution of the computers makes it possible to falsify them. Also, the use of such methods requires the use of signature software - itself vulnerable.

Each method cited covers a particular field of application, and does not provide a simple solution to independently record internal, confidential, private, or strategic data of an individual, a professional, or a company, from when the quantity of data becomes significant, and when it must be possible to present or request it as authentic legal and legal proof a posteriori. The natural or legal person who owns, manages or is responsible for them and does not necessarily have the will or the technical and / or organizational means to transmit these data via - or to - an external third party, because of their strategic, confidential or private. The approved or institutional organizations which can request within the framework of control or search, the presentation of data, do not have the means to copy, authenticate or guarantee the data of all the users, and are forced to leave custody and control of the data to their transmitters.

The set of vulnerabilities makes it extremely difficult to give legal value to private computer records based on standard elements, and that in a contradictory manner it appears more and more necessary to guarantee the authenticity of the data whatever. they are, computer storage generally constituting the modern means of preserving information.

The methods and device according to the present invention seek to compensate for the ease of reproduction and falsification of standard audio, video and computer recordings, the control of which is entirely ensured by the people who store and control information, and who can therefore falsify or truncate it, l '' Date backwards, or be victims of internal or external attack on this data. The fact that this information is private can make it impossible, long or tedious, to cross-check certain unique information. More specifically, the invention consists of a method for storing computer data, characterized in that it consists in integrating entirely into a device:

- a data storage device,

a suitable input-output controller transforming said data storage device into a single-write or restricted storage device by deriving all commands from the input-output interface coming from a host system, towards the internal interface of the integrated data storage device, so as to: - exclusively authorize the addition of new files in the free space by dating them independently thanks to a secure internal real-time clock, and

- to prohibit the erasure, modification and rewriting of data previously written and therefore previously dated,

- to protect the whole of said device on the one hand against falsification or duplication by a unique internal secret identifier in correspondence with a unique public serial number, on the other hand by a perimeter protection enclosure intended to prevent intrusion in the equipment, the modification of any component and internal data, as well as the analysis of the unique secret identifier without irreversibly compromising the validity and authenticity, - to detect an attempt at intrusion, modification, analysis, said detections being able to be verified by several methods third to the said process, the latter being thus secure, and suitable for storing authenticated and dated computer data at value probative.

According to an advantageous characteristic, the method according to the invention also consists in electronically reproducing internally said unique public serial number, to allow electronic control thereof. According to an advantageous characteristic, the method according to the invention also consists in physically protecting all the internal components by leaving accessible only the external functional interfaces of input-output and supply, thanks to a perimeter sealed enclosure encompassing all the components, consisting of passive mechanical and electromagnetic protection including an active intrusion detection sensor and associated electronics, the latter ensuring immediate destruction of the secret internal unique identifier and the internal serial number if necessary, by instantaneous erasure , in the event of voluntary or involuntary deterioration changing the vulnerability or the autonomy of the device, or an internal investigation attempt detected by the sensor which can make it possible to compromise, control, modify, analyze data, interfaces or internal components.

According to an advantageous characteristic, the method according to the invention also consists in keeping active, the internal real-time clock, the secret identifier, the internal serial number if applicable, and the electronics for intrusion detection and destruction, outside of any external power source for the service life specified for the equipment, thanks to a self-contained internal power subsystem, possibly seconded externally to increase the service life. According to an advantageous characteristic, the method according to the invention also consists in making it possible to guarantee the unique and non-reproducible personalization of the equipment while publishing its existence, possibly before certification, by generating randomly by internal calculation of the equipment, the secret unique identifier, by internally calculating a one-way digital fingerprint of this identifier which cannot be traced back to it, by establishing the uniqueness of the fingerprint by interrogating a public database containing the fingerprints of all existing equipment, the identifier being recalculated in the event of a collision and the device being formally identified as soon as the fingerprint is unique, then adding the references of the to the 'public ^' database equipment: unique serial number and imprint.

According to an advantageous characteristic, the method according to the invention also consists in ensuring the authenticity of the device after manufacture, the test, the final assembly and the activation of the perimeter protection, until acquisition by the user. final, by making unique the operations of assigning the unique serial number, the synchronization of the real time clock with respect to a reliable external time reference, the internal self-generation of the unique secret identifier, the copying of the unique secret identifier in a secure environment from the unique secret identifier to a certifying body empowered to guarantee the secrecy of this identifier and therefore the authenticity of the equipment, the uniqueness of each operation being guaranteed by design in irreversibly inhibiting or destroying and as soon as possible in the initialization sequence, the hardware elements and / or the microcode allowing this initialization.

According to an advantageous characteristic, the method according to 1 the invention further consists in allowing without limitation on the occurrence, the preliminary verification of authenticity and integrity, by checking and cross-checking the unique serial numbers listed in the public database, by checking the validity and the evolution of ^f internal clock back to a reliable external time reference taking into account the specified maximum drift of the internal clock and by visual inspection and non-destructive integrity of the outer shell. According to an advantageous characteristic, the method according to the invention also consists in allowing authorized organizations to authenticate the equipment on demand by verification by third-party methods using the internal secret identifier without revealing it, by recalculating internally. dynamically the imprint of the identifier and by directly querying the public database, or by internally coupling the secret identifier to the internal date and to an external message provided before internal computation of the overall imprint, l 'all known elements being provided. for remote certification by the authorized certifying body, which alone can verify consistency by copying the unique secret identifier available to it, these functions involving the secret identifier, being timed and .limited in internal occurrence of the equipment so as to avoid the reverse determination of the identifier by analysis or successive tests.

According to an advantageous characteristic, the method according to the invention also consists in allowing the certifying body to authenticate itself under its control and as a last resort the equipment by injecting its copy of the secret identifier for internal comparison equipment which detects perfect similarity, and sends in response a single message which can be verified by said certifying body, this function involving 1 secret identifier being timed and limited in internal occurrence of the equipment so as to avoid the reverse determination of the identifier by analysis or tests According to an advantageous characteristic, the method according to the invention also consists in allowing, as a variant not necessarily implemented, to manage autonomously the expiration of data in the past, by authorizing -only in the event of complete filling of the device internal storage - the writing of new files by overwriting within the limit of sufficient and in chronological order, the data before a relative date, dynamically updated compared to the current date provided by the internal clock, derogating thus to the single writing rule without however compromising or modifying the useful data still valid, and dated after the expiration date.

According to an advantageous characteristic, the method according to the invention also consists in presenting any internal parameter and indicator specific and specific to said device in the form of virtual files with reserved names, viewed by the host system as conventional files and avoiding the setting in implementation of specific drivers or operating modes at the host system level for the management, administration and monitoring of said device.

According to an advantageous characteristic, the method according to the invention also consists in partitioning said data storage device with a different expiration time per partition. The present invention further relates to a device for storing computer data, characterized in that it fully integrates:

- a data storage device with an internal interface, - a suitable input-output controller,

- an input-output interface,

- a secure internal real-time clock,

- a unique internal secret identifier, - a unique public serial number,

- a perimeter protection enclosure, said adapted input-output controller being able to transform said data storage device into a single or restricted write storage device by deriving all commands from the input-output interface coming from from a host system, to the internal interface ^' of the integrated peripheral, so:

- to authorize exclusively the addition of new files in the free space by dating them independently thanks to the internal secure real-time clock, and

- to prohibit the erasure, modification and rewriting of data previously written and therefore previously dated, the whole of said device being protected on the one hand against falsification or duplication by the unique internal secret identifier in correspondence with the number single public serial, on the other hand by the perimeter protection enclosure preventing the intrusion into the equipment, the modification of any component and internal data, as well as the analysis of the unique secret identifier without compromising in a way irremediable validity and authenticity, attempts at intrusion, modification, analysis being able to be verified by several means third to the whole of said device, the latter being thus secure, tamper-proof and tamper-proof suitable for storing authenticated computer data and dated with probative value.

The device according to the present invention is fully protected against rewriting and able to give a probative and / or legal and / or legal authentic value to digital information. The data is irreversibly recorded and dated with sufficient precision completely independently. The device is self-protected and uniquely identified. Any attempt to modify or compromise the device results in its invalidation. The identical duplication of the device is made reasonably impossible.

The use of the device allows, without specific constraint and without using a third party during its operation, to any person anxious to be able to establish their good faith and the legality of their activity, to do so, leaving them free will on the data that 'she wants or must protect as well. The device does not in any case prevent the concealment, complete destruction of data, information or evidence, or a priori possible falsification in all cases, but establishes a posteriori that the data, when it is presented, were recorded on the date indicated, and could not be falsified or replaced later.

The originality of the device is that its functional principle is independent of any heavy cryptographic method, subject to obsolescence or expiration due to the unpredictable progress of cryptanalysis. Cryptography is not necessary, in the sense that the user physically controls and owns the device, and must decide the degree of confidentiality and data protection regardless of the physical storage device that he uses. It minimizes the intervention of third parties for authentication of data, avoids the transmission of strategic or critical information on remote third-party networks, and makes it possible to protect more data, or to migrate paper data or microfilmed to a computer medium that is easier to use and archive.

The advantage of the device described is the possibility of using any rewritable or non-rewritable random access storage device as physical storage element, without degrading performance, reliability or environmental resistance and giving it transparent behavior. neutral vis-à-vis the host system, thus leaving the freedom to the end user to use the external software methods of compression, cryptographic protection, and reliability of the storage made available to him.

The device which is the subject of the invention includes the peripheral equipment constituted (hereinafter referred to as 'the equipment'), but also the methods and processes which make it possible to manufacture and verify the equipment by third parties, and the description of certain particular modes of operation.

-The device completely integrates a standard commercial data storage device with non-removable media or made non-removable (by way of nonlimiting examples: a hard disk with magnetic technology, a flash memory disk, a CD / DVD burner, the ejection mechanism is deactivated loaded with writable media) whose interface is protected and isolated by a specific controller, by design limiting storage operations to the sole function of adding data. The equipment is a priori seen by a host computer system as peripheral computer storage equipment, except that it is functionally impossible to modify, erase or rewrite logical records previously written. It is only possible to read them. A date clock calibrated at the time of manufacture, non-modifiable and autonomous dates in real time and independently of the host system any recorded file. File dating is recorded by 1 control electronics on the internal device, either in addition to recording user data with a reference to the basic descriptor of the logical record or file with which it is associated as metadata, or in the descriptor itself file instead of the timing information normally provided by the host system, either in unused or unspecified spaces in this file descriptor. All of this dating information is accessible in read-only mode by the host system, which makes it possible to verify the dating certified by the device for each recorded record.

The equipment generally behaves like a device of the single writing type, which re-dates the information with respect to a relative reference time, sufficiently precise, and the drift of which can be measured, even authenticated. The integrated peripheral therefore only fills up, as if it integrated a physically non-rewritable medium. By way of a variant which is not necessarily implemented and under very restrictive conditions, the equipment can easily manage an expiration date in the past relating to its internal clock, therefore to its own dating, which eventually makes it possible to free the space occupied by obsolete data to register new records when the device is 'full'. This so-called 'expiration' functionality makes it possible to respond to the specific need for data storage with a short lifespan compared to the overall lifespan of the equipment, which is of interest in certain fields of applications.

An autonomous electronic identification subsystem contains the unique and secret authentication elements of the equipment, and prevents the duplication or production of false equipment from new parts or recovered from equipment voluntarily compromised. Internal unique identifiers are systematically destroyed during unauthorized physical access attempts. An active perimeter protection enclosure physically fully protects the assembly and detects any penetration or attempted intrusion or investigation aimed at compromising the integrity of the equipment, accessing internal interfaces allowing data to be modified and extract the identifiers. In the event of an intrusion attempt, the equipment destroys instantly and irreversibly. unique identifiers and possibly the clock. The equipment is invalidated and finds itself in an initial mode which continues to 'protect the data written on the storage device outside destruction of the equipment, by invalidating any write operation. The data remains legible in all circumstances.

Any equipment whose integrity can be confirmed by non-destructive analysis methods made available (essentially the verification of unique identification elements), ensures the validity and authenticity of the data read on the internal device.

In case of doubt or suspicion, the authorized certifying body, which alone has a copy of the secret identifier, may submit the copy of the secret identifier available to the equipment which itself performs the comparison. internal, or allow an authorized third party to indirectly verify this identifier by limiting the request intervals so as to avoid its determination by, as a non-limiting example, a brute force or screening method, or targeted analysis.

In the event of a critical internal failure, the voluntary dismantling of the equipment can be carried out so as to assess it or perform data recovery following the manufacturer's recommendations. For this purpose, a procedure must be able to allow the simple cutting of the equipment, which has the effect of instantly invalidating the authenticity, then extraction without deterioration of the internal storage device, but allows direct access to the latter, to the data, a priori not encrypted by the equipment, and possibly to the media.

Any damaged or corrupted equipment loses its authentic value: the data present is no more guaranteed than if it had been stored on any non-secure medium, on the other hand it remains accessible at least for reading as long as the equipment is functional.

Like any security method, the probability of non-detection of intrusion or the possibility of duplicating equipment is made negligible but not zero and reduced to a reasonable minimum depending on the objective cost targeted and the level of protection required.

The present invention includes the methods and the sequencing of operations subsequent to the material manufacturing of the equipment, which ensure irreversibly after sealing of the protective envelope of the equipment:

• safe and incorruptible dating,

• unique and secret identification, while making the equipment manufacturer, the authorized certifying body (for example one for each nation, consortium, community) and the end user independent.

The present invention includes several verification methods or methods associated with formal authentication of the equipment according to the invention. These methods or methods of verifying authenticity may or may not use external third parties, and / or may or may not use internal secret identification. Other characteristics and advantages will appear on reading the following of several examples of embodiments of methods and devices according to the invention, accompanied by the appended drawings, examples given by way of nonlimiting illustration.

According to a particular embodiment, the device is administered and configured using a method of virtual files, seen by the host system as standard files present on the integrated peripheral. This method presents any internal parameter and indicator specific to the device described in the form of files with reserved names accessible in a standard way by the host system, avoiding the implementation of specific drivers or operating modes at the host system level for the management, administration and monitoring of the device and of the variants and options described.

• Figure 1 represents an overall functional diagram of the equipment,

• Figure 2 illustrates the role of external third parties, • Figure 3 illustrates a suggested file structure with dating by adding metadata,

• Figure 4 illustrates a suggested file structure with replacement dating in the file descriptor, • Figure 5 illustrates the secure process of initialization and identification of each device individually by the manufacturer,

^• • Figure 6 illustrates the secure certification process for each piece of equipment, o Figure 7 illustrates the secure activation process by the end user, o Figure 8 illustrates the different authentication methods and methods related to carrying out the device described, • Figure 9 illustrates the different methods of expertise of the equipment, e »Figure 10 illustrates the suggestion of using virtual files to administer the equipment, o Figures 11 to 13 illustrate the filling of the media, o The Figures 14 to 16 illustrate the filling of the media with the establishment of the expiration mechanism, o Figure 17 shows a detailed diagram of suggested implementation. The indication [u] in some figures means that the action is functionally unique and irreversible. Certain indications are intentionally worded in English to conform to the use of the technique.

The device which is the subject of the invention integrates one or an assembly of standard peripherals for storing computer data 1. This peripheral is isolated from the host system by a controller 2 which limits the operation to adding data and ensures independent dating of information. stored by an internal clock 6. The device comprises a pair of unique digital electronic identifiers 7, 8, which prevent duplication of the equipment.

The assembly is protected in an active burglar-proof enclosure 10, which destroys the unique and secret elements of identification in the event of an attempted break-in. The device is then invalidated and loses its authenticity.

The integrated peripheral 1 is a read-write model, no prerequisites being necessary concerning the possibility or not of erasing or rewriting the data (by way of nonlimiting example a magnetic hard disk, a flash disk).

Generally, such peripherals address physical storage elements commonly called blocks, conventionally numbered in sequence, and higher level logical storage elements which can be manipulated by a system or host computer commonly called files, directories. All these entities are often accessible sequentially or randomly in read and write. A storage device does not normally have knowledge of the logical level, managed entirely by the host system via a file system (File System) subject to rigid specifications and easily analyzed.

In order to simplify the language, we will call blocks the physical entities addressable by the peripheral and files the logical entities manipulated by the host system, without detailing the concepts of registration, clusters, directories and associated attributes.

The functional interface 3 of the integrated storage peripheral 1 allows a priori any operation: conventionally, addressing, writing and reading of blocks, and a priori has no protection or limitation on these functions.

This interface is functionally isolated by an active intelligent controller 2 with an external interface 13 emulating a data storage device with a capacity substantially lower than the capacity of the real organ. The physical input / output interfaces 13 and 3 do not have to be necessarily identical, just as the peripheral recognized by the host system 20 does not have to be identical to the actually integrated peripheral 1, even if, in practice , the similarity on this specific point can simplify the overall operation.

When it is externally electrically powered 14, the equipment is in its nominal operating mode. The controller 2 translates in real time any request for addressing, reading and writing from the external interface 13 to the internal interface 3. The data proper is not processed, neither compressed nor encrypted, and is stored as is if the authorization to write them to the integrated device 1 is validated by the controller. 2 who basic knowledge of the level of logical storage, and therefore of the structure of files and directories.

In its nominal operating mode, the controller protects the data against rewriting, thus emulating an add-only storage system '. At the end of manufacturing, the equipment is empty of user data and pre-formatted, therefore suitable for receiving data. Overall, it therefore maintains a partition between the space ^ occupied '4 and the free space' 5. As a non-limiting example of implementation, it maintains a 'pointer' 15 which indicates from which data block , writing new blocks is possible, and globally inhibits any writing on previous blocks. In the actual implementation, partitioning can be more complex depending on the logical file system used, which can itself impose a specific partitioning between physical and logical information, as shown in Figure 17, which may require the simultaneous management of several pointers . Access to physical storage is therefore random in reading in the occupied space 4, and purely sequential in writing in the free space 5. An attempt to erase, modify or rewrite a file or previously written data thus results in an error reported to the host system 20, thus signaling the impossibility of carrying out the desired operation on the emulated device. Writing new data, new files or creating new directories is possible, the storage device therefore inexorably fills up to its maximum useful capacity, and rejects the writing of files exceeding the remaining size. It is then 'full', and must be completed or replaced by identical equipment since it is functionally impossible to delete the files entered. It can then be physically disconnected and archived outside any external power supply 14 of 1 device during the storage period thereof, an internal battery subsystem 19 allowing the maintenance of critical internal elements (clock, identifiers, pointers ...) purely for information, between ten years and twenty years following the life of internal elements subject to aging or perishable, such as batteries. Storage for longer periods may require either the use of external auxiliary batteries, taking over from the internal batteries, or the certified migration of these data with their creation date to new equipment.

The data entered is protected against any electronic attack (virus, malicious acts, falsification, etc.) since it cannot be rewritten or modified by design. The destruction of the data implies the destruction of the equipment.

Controller 2 is specifically adapted to the structure of a standard file system (by way of nonlimiting examples, FAT, NTFS, UDF, NFS, HFS, sequential ...), or to the structure of a proprietary files if the equipment wishes to ensure compatibility with several different standard file systems. The file system creates the link between logical storage and physical storage of data. The controller 2 detects by analysis of the writing orders sent, the end of a logical storage operation (for example, the end of writing of a file, or the creation of a directory entry). The sequential aspect of writing data, without the possibility of erasing, greatly simplifies this task. At the end of a successful logical write operation, the controller autonomously performs the following tasks: o Irreversible protection of the storage elements used for the file descriptor 24, and for the data themselves 25, by reserving the blocks used, for example fabrication and / or pointing 15 in a saved memory accessible to the controller, and redundant on a space reserved on the storage device itself. o Independent dating 26, thanks to the internal reference dating clock 6, either by writing additional blocks of masked metadata to the host computer 28, as shown in FIG. 3, or by rewriting this date instead place of the date written by the file system in the file descriptor 24 as defined by the file system used, as shown in FIG. 4.

The written information is dated to the second independently of the internal clock 6 of the device, by the equipment itself, independently of the host file system which applies its own dating, often insecure and unreliable. This internal dating is irreversible and cannot be forged, the internal clock being set once only at the time of manufacture and cannot materially be set to the time thereafter, and the recordings containing this information being protected against rewriting . The accuracy of the dating relies entirely on the precision and the drift, of this internal clock which can be correlated externally without however possibility of correcting it. The dating is done in Universal Time.

In the case where the device requires dating by adding metadata 28, these are made invisible in normal use of the files and are themselves protected against rewriting. The metadata are at least composed of a pointer 27 allowing access to the descriptor of the file or of the directory and of a structure comprising the internal date and time 26 at the end of writing. All of the metadata of all of the written files is sent in a 'system' file 29 accessible in read only. A tool or else the analysis of this file makes it possible to correlate the metadata and the user files, and therefore to verify the internal date that the device had assigned to each file at registration, which alone is authentic by design.

In the case where the device imposes the dating of the files by replacement 26 of the date of creation in the descriptor 24 of the file, in accordance with the requirements of the file system used, the behavior of the device is perfectly standard. The date imposed by the equipment differs from the system date, and can, for practical reasons, be corrected dynamically internally at the time of reading to take into account the difference between local time and universal time (time zone). Most of the file systems currently used manage several dates (for example the modification date which is useless in our case), it is therefore possible to also keep the date of the external system for information only.

The equipment itself is made tamper-proof thanks to a unique identification method implemented internally of the equipment. It is reasonably impossible to duplicate the equipment without full knowledge of the unique identification which is never disclosed by the device, apart from an initialization phase where only an authorized certifying body 22 receives the full copy of this document. identifier, part of which remains secret. The identification elements are destroyed as soon as an attempt to intrude, break in, or compromise the equipment is detected. The identification is only used for the validation and certification of the equipment and has no functional role for the end user.

The identification system is made up of the following elements: • Unique public identification element 8, 18: number of series of complete equipment, marked visibly and indelibly on equipment 18, and reproduced electronically internally 8 for reasons of ease of operation and direct authentication. o Unique identification element 7 secret generated and imperatively stored internally by the equipment and verifiable only by third-party methods which never broadcast this identifier. o Optionally elements of manufacturer identification not shown: for example, electronic serial number or marked internal components, date code, lot numbers, revision, often traced by the manufacturer as part of its quality system , which depend on the implementation and which may allow additional verifications or expertise, possibly after dismantling the equipment.

The correspondence between the public identifier and the secret identifier is known only by a third authorized certifying body 22 which is involved at the end of the manufacturing process. The secret identifier is distributed once and only once.17 outside the equipment to the attention of this single body 22, which stores all the information relating to all the equipment it manages and certifies. This organization is the only third party that is involved in the certification of each piece of equipment, only initially at the end of the manufacturing of the piece of equipment and possibly during use in the event of a thorough check of the authenticity of the piece of equipment if any doubt persists this authenticity, or if the issue or the legislator justifies such a recourse.

The unique electronic identifiers of the equipment 7, 8 (excluding manufacturer identification numbers not described as linked to the implementation and to the components used) are stored in a volatile memory circuit backed up by battery 19 and having an internal and / or external rapid erasure function 9. The absence or inconsistency of one of these identifiers indicates the compromise of the equipment. Only a one-way imprint 23 of the unique secret identifier associated with the equipment serial number 18 is publicly disseminated 21 outside of the authorized certifying body or bodies 22 which have sole access to the correspondence between the serial number 18 and a copy 17 of the unique secret identifier.

The unique secret identifier is a very long binary message (for example 4096 bits or more) certain values which can be reserved and cannot be used. The internal clock 6 is set to the time in universal time only once during the manufacture of the product in relation to a certified reference clock 16, and cannot be reset by design later without corruption of the other parameters d integrity of the system.

The system thus described is only valid if it is physically self-protected against intrusion and internal analysis. To this end, the assembly described is entirely confined in a protected enclosure 10 by a perimeter sensor 12 sensitive to intrusion, powered by a battery 19 having a long lifespan (for example, lithium battery). Any attempt at penetration and internal microprobe analysis is, on the one hand made extremely difficult by external mechanical and electromagnetic protection 11, and on the other hand triggers an alert in the event of penetration.

-beyond a limit envelope 12 which includes any functional component of the equipment. The alert results in the instant destruction 9 (erasure / reset) of the internal identifiers 7, 8, making the equipment non-duplicable thanks to the secret part 7 inaccessible which cannot be analyzed without entering the equipment, therefore without trigger an alert.

The equipment is fully protected by an active physical enclosure which provides protection against intrusion by any known method, standardized in certain countries for the protection of critical security parameters and / or cryptographic modules (for example in the States - United FIPS 140-2 standards, or internationally ISO 15408 standard / AVL Common Criteria). The use of such methods can be made very inexpensive for mass production, even for the highest levels of safety. Such sensors exist commercially (for example D ³ sensors of the Gore brand, FIPS 140 level 4 certified) and their general principle is described for information purposes only, sometimes referring to FIG. 17. Conventionally, protection perimeter 10 is composed of an external mechanical protection 11 or sealed external box (for example in machined, molded, folded metal, assembled in a non-removable manner by welding, crimping) leaving space only for the functional communication and supply interfaces , and a sensitive sensor 12 detecting intrusion (drilling, machining, electrical or laser mechanics, selective chemical attack) by modification of an electrical parameter (resistance 121 and / or capacity 122 and / or contact 123). Often this sensor is based on a network of conductors and insulators sensitive ^'to cuts, breaks and short circuits, covering the entire vessel to be protected uniformly and provided with' baffle 'to protect external functional interfaces. The detection and alert electronics 118 is very low consumption so as to be able to provide a large autonomy to the protection supplied autonomously by battery 124. Usually, all of the electronics and the internal components are embedded in a resin 130 intended to ensure thermal conduction in the absence of convection internal, and prohibiting direct access to internal components or signals. The purpose of the protection system is to prevent the insertion of a micro probe allowing the analysis or direct control of internal electrical interfaces making it possible to modify the protected data, to analyze the elements kept secret with a view to their duplication .

In order to eliminate any thermal attack, a temperature sensor 120 analyzes excessive high or low temperatures which can inhibit or limit the effectiveness of the sensor.

The failure of the batteries 124 immediately puts the system in an alert state.

The protection also includes the external electrical interfaces in overvoltage / overcurrent / frequency or signals out of specification 129. To facilitate protection, the functional interfaces are reduced to a minimum by the use of fast serial functional buses (for example reflecting the current state of the art Fire ire / i-Link / IEEE1394, USB, Serial ATA, Ethernet), a continuous external power supply, and static system state interfaces.

In the particular case of using a magnetic hard disk which must operate at atmospheric pressure, the functional depressurization cover of the disk is carried outwards thanks to an integrated tube allowing the free passage of air without increasing the vulnerability. of external protection 132.

The detection of intrusion into the enclosure results in the destruction of the integrity parameters 7, 8 of the equipment, by instantaneous erasure 119, 9 of memories 112 saved by internal battery 113, 19, this that the device is powered. or not externally. This destruction of the system identification parameters does not prevent read access to all of the data. Controller 2 continues protect data, which loses its authentic value as soon as the equipment is reported as compromised.

The equipment is designed to be autonomous in normal use. It therefore does not require any intervention by third-party authenticators in use, and in fact allows it to be used entirely privately, without prejudging the possibility or the will of the end user to authorize direct access or indirect to an outside network. The following section describes the methods and methods used in the initialization phase, which makes it possible to switch from neutral non-personalized equipment 30 leaving manufacturing, to unique equipment certified 55 then operational at the end user 60.

The equipment at the end of assembly is completely integrated and protected, the perimeter protection 10, 9 is activated. It is in a neutral initialization state where. the internal numbers are blank and an initialization microcode is preloaded in random access memory 107 of the controller. An internal state logic machine 117 can assist the hardware protection of non-modifiable resources. The operational operational program is itself loaded in read-only memory or programmable in a unique manner 106, and cannot be modified. The contents of all internal memories can be analyzed and checked at any time. At the end of initialization, the initialization program is completely erased, the RAM can no longer be used to execute any program and is only used for temporary storage of internal data. As a suggestion, the initialization program is erased step by step at each stage of initialization, which ensures the uniqueness of the operations.

On power-up, the functional program detects the initialization state and executes the initialization sequence or resumes it at the current stage. The initialization sequence takes place in three phases: • the personalization phase, as shown in Figure 5, o the certification phase, as shown in Figure 6, o the activation phase, as shown in Figure 7 Until these three phases have been completed, the equipment is inoperative as a storage device, and no addressing, reading or writing operations on the internal device are authorized.

The following description illustrates a sequencing suggestion.

The personalization phase includes the following operations using neutral equipment 30:

- 31: Marking of the unique external serial number 18 in the structure of the equipment case (external protective envelope) indelibly (for example by laser marking or mechanical engraving). - 32: Entry of the serial number in electronic form via the external interface.

- 33: Copy of the serial number 8 in the identification memory.

- 34: Entry of the reference date and time, compared to a calibrated and certified external time reference 16 (atomic clock and / or time server). The date and time are programmed in Universal Time.

- 35: Synchronization of the internal dating clock within a fraction of a second. By design, time setting can only be done once.

- 36: Random internal generation of the equipment of two passwords: these optional passwords ensure minimal protection of the equipment, and are for single use: they allow the protection of the equipment during storage or transit phases. Passwords are transmitted only once through the interface to the manufacturer, 37 and 38, which transmits them independently to the authorized certifying body and to the end user.

- 39: Self-generation of the secret identifier, internally of the equipment: a random generator calculates a very long identifier; so as to avoid any analysis, the encoding is on a number of bits such that the interrogation of the equipment for analysis of the secret identifier, is unthinkable by any screening or brute force methods, reducing the probability of find this identifier by chance at an improbable value, and this, even knowing the correspondence between the third party elements provided (signatures) and the unique identifier. This identifier is stored in a volatile random-access memory saved by battery, instantly destroyed in the event of a break-in. - 40: Calculation of a one-way digital fingerprint, by a hashing algorithm (for example SHAl), reproducing a one-way abstract 23 of a few tens of bytes (20 in the case of SHAl, algorithm of hash normalized FIPS 180-2) from a minimum message of a few hundred bytes (512 in the case of SHAl).

- 41: The equipment delivers the digital imprint calculated with the internal serial number for verification of uniqueness. To this end, a central and public database 21 contains all the serial numbers of all the equipment produced and the fingerprints used. The interrogation makes it possible to know if the identification is unique by an indirect method (fingerprint) which does not reveal any direct information on the internal secret identifier 7. Only manufacturers can enter this information in the database, which is available for consultation.

- 42: If the impression is already used, there is a doubt about the uniqueness, a message is sent to the equipment so that it submits a new identification.

- 43: Otherwise, the database 21 is updated, and the initialization process can continue.

- 44: All the manufacturer initialization functions are then physically inhibited, for example by deletion (replacement by a null instruction

(NOP)) of all the code located in RAM which made it possible to carry out this part of the initialization, and by possibly inhibiting this memory space for the subsequent execution of programs.

- 45: The equipment is therefore identified and can be sent and / or made available to the certifying body 22, which alone is authorized to know the correspondence between the serial number and the unique identifier.

As an example, a secret message of 4096 (512 bytes) bits seems to give an almost total invulnerability to the system. Due to one interrogation per day, limited in this case by the equipment itself, it takes approximately 12 years to recover 4096 different fingerprints, which, in spite of everything, would not allow to go back to the source, the usual fingerprints being 'mathematically' one way and each having a huge number of digital backgrounds

(> 10 ¹⁰⁰ ), and a tiny probability of collision (<10 ^{~ 50} ), quantities which in themselves constitute 'physical infinites'.

Then begins the certification phase:

The authorized certifying body 22 independently receives the identified equipment 45 and the password question mark 37.

- 46: Entry of the certifier password 37.

- 47: Unique transmission of the unique secret identifier

7 on the functional interface. - 48: Keeping of a copy of the unique secret identifier 17 by the authorized certifying body, associated with the serial number 8.

- 49: Calculation of the print of the copy 17 according to the method used in the equipment 40. - 50: Query and update of the certification status by request in the public identification database 21.

- 51: Confirmation of the correspondence between the two fingerprints of the internal identifier and its copy. - 54: In the event of a correspondence failure, or the impossibility of carrying out the operation, the equipment is corrupt and must be rejected and dismantled 53, either because the equipment has already issued the identifier uncontrollably unique, either because it has broken down. The certifying body then publishes the rejection of the numbered equipment in the public database 21.

- 55: If successful, the authorized body declares the certification of the equipment, and publishes it in the public database 21.

- 52: The internal certifier password is deleted, as well as the associated initialization programs and functions.

Only authorized certifying bodies can modify the certification status of equipment in the public database.

The authorized certifying body guarantees the secrecy of correspondence between public and secret identifiers, and can establish in case of doubt and as a last resort the authenticity of the equipment. The authorized certifying body and the manufacturer can be completely independent. There may be several manufacturers and several authorized certification bodies approved by different bodies giving authentic value to the equipment. The authorized certifying bodies may optionally share copies of the unique secret identifiers with each other.

The equipment is then certified. It can be marketed and delivered to an end user, who can use it after activation.

The end user who acquires certified equipment 55, independently receives the initial password 38 issued by the manufacturer, and must activate it: - 56: At the first power-up, he must connect the equipment and must enter the password provided.

- 57: The initialization functions are then all disabled: only the functional program remains resident and can be executed from this moment. - 58: The equipment is activated functionally.

- 59: This state is indicated visually, even without external power supply (for example, flashing at low duty cycle of a battery-type LED indicator). - 60: Once the equipment is activated, the end user can use the equipment as a partitioned storage computer device, initially blank. He can verify the authenticity and validity of his equipment himself by the methods explained.

This methodology therefore guarantees that no equipment can be diverted, subtracted or created without control and certification.

The following section describes the methods for to establish the validity and authenticity of the equipment, made available to the end user and authorized bodies authorized to carry out control of the equipment and of the authenticated data recorded. Since the equipment and its unique identification are physically protected by the protective enclosure 10, the probability that a user can present equipment that is voluntarily corrupted and that has an authentic appearance is negligible, however even this extreme case is provided. The failure of one of these methods indicates that the equipment is invalid, the success of a method being able to be deepened by another method, in the event of persistent doubt on the authenticity of the equipment. • Elementary methods not limited in occurrence: - 61: Verification by comparison of the internal serial number 8 and the external serial number 18.

- 62: Verification of conformity of the equipment identified by its serial number 18 in the public database 21. - 63: Verification of the evolution of the internal clock 6.

- 64: Checking the consistency of the internal clock 6 with respect to an external reference 16, taking into account the specified drift of the internal clock.

- 65: Visual integrity check. • Third-party methods internally involving the unique secret identifier 7, made on demand and internally timed so as to avoid any analysis of the identifier (for example, limited to a request per 24 hours):

- 67: Internal recalculation of the digital fingerprint 66 of the internal unique secret identifier 7 and verification vis-à-vis the public database 21.

- 71: Internal calculation of the digital fingerprint of an external message 68 (for example a certificate) supplied by an approved body authorized to control the equipment, of a length at least equal to that of the unique identifier, and mixed in a deterministic but non-reversible manner 70 with the internal unique secret identifier 7, and possibly the internal date 6 before calculation of the imprint of the overall result 66. The calculation elements 68, 6, the serial number 8 and the result are supplied to the authorized certifying body 22 which keeps the copy of the secret identifier 17, which is therefore the only one capable of establishing the consistency of all, and which confirms the authenticity to the applicant after verification. The fact that the message 68 uniquely and self-identifying identifies the requesting body allows the certifying body to limit the requests to only the institutions authorized to carry out checks. The equipment itself accepts a priori any message or request without prejudice to the nature or content of the message.

• Direct method for checking the unique secret identifier:

- 72: verification of the internal unique identifier directly by the authorized certifying body. It is similar to the previous method, except that the message supplied is exactly the unique identifier 17. The exact correspondence is detected internally by the equipment 69 which provides for response in a particular case, a calculated fingerprint 66 solely on the basis of the unique identifier 7 and the internal date 6.

• Non-destructive appraisal methods allowing further analysis using the following methods:

- 73: Emptying internal memories (except secret identifier): ROM (Read Only Memory) 106, RAM (Random Access Memory) 107 or other rewritable memory (Flash)

- 74: Emptying any manufacturer identification number internal components accessible in electronic form. For example: models, revisions, unique serial numbers of the components used, binary code of the programmable components. Destructive appraisal methods requiring controlled dismantling of the equipment, after, if possible, controlled copy of the data on another medium.

- 75: The equipment is arranged to allow simple mechanical cutting so as to allow direct access to the native electrical interfaces 76 of the internal storage device 1, and to the other elements 77 for expertise. This destructive operation instantly invalidates the authenticity and any other functionality. As an implementation suggestion, all these methods can use virtual files managed directly by the controller, and appearing as files available on the device, whereas they are not in fact directly stored on the integrated device 1. These files can be accessed in read mode, and in some cases in write mode so as to allow the end user to change the global parameters of the equipment under his control, or to allow authenticity verification to third parties by entering settings. This method can also be used for the exchange of system data, identifiers or passwords during the initialization phases (identification, certification, activation).

As a non-limiting implementation suggestion, virtual files can contain:

- 78: the internal date 6, possibly associated with important dates such as the date of the first time setting, and the date of certification,

- 79: the internal electronic serial number 8, - 80: authentication elements,

- 81: system parameters, writing allowing to assign certain optional operating modes programmable by the user (for example time zone, expiration, control and access limitation ...),

- 82: the state of the system,

- 83: the log file of exceptional events,

- 84: the elements monitored by the equipment: internal voltages, temperatures, in the case of a hard disk

SMART (Self-Monitoring, Analysis and Reporting Technology),

- 85: the contents of the ROM memory,

- 86: the content of the RAM memory, - 87: the 'manufacturer' identification elements.

It is recommended that all these files be easily readable by a human operator and printable.

(use of alphanumeric coding, for example base 64, with limitation of the number of characters per line, or standard HTML or XML).

The following example illustrates the operation of virtual files: the authentication file 80 is encoded so as to be easily editable in a reserved system file named by way of example <AuthChek. sys>. Writing is possible, if the anti-analysis time delay conditions are met 88. Writing to this file triggers the internal calculation of one of the certification elements, depending on the nature of the file entered: »empty file 89, © file containing a formatted input message 68, o file containing exactly the unique secret identifier 17.

In reading, the system returns at the end of the calculation, a file of fixed length and structure including correspondence, according to the methods explained above, 'a formatted message ready to be sent for verification, and including all the elements necessary for verification: a fixed length imprint, the serial number, the message used, the date used.

As an option or as a variant, in the case where the application requires a remanence of the data much less than the subjective or objective lifespan of the equipment, an expiration option can make it possible to free up space on the media by automatically destroying, if necessary, obsolete files in the past compared to the internal clock. Operability limits can be set either at construction by design, or by the authorized certification body, or at initial start-up by the end user, and may include limitations on the space freed up, such as for example , necessarily protect a fraction of the total capacity. The expiration is typically expressed in number of days relative to the present, as dated by the internal clock, and can be calibrated on calendar events (start or day of the week, start or day of the month, start or day of the year) . Internal dating in universal time means that at least one day of margin is systematically added to take into account the local time difference and the drift of the internal clock.

The different modes of operation linked to expiration are explained in FIGS. 11 to 16. The storage space is represented in the form of a rectangle. The storage space is initially empty 90.

The addition of new 91 files gradually fills the storage space. The data previously entered is protected against modification 4, the new file takes its place in the free space 5 and is dated at the end writing on the current internal date 26, thus increasing the protected occupied space 4.

The files therefore fill the storage space sequentially over time 92. In the mode of operation without expiration, the storage space is completely filled: any attempt to write a file which exceeds the remaining free space in capacity is rejected and reports an error 93.

The operating mode with expiration is explained below. The expiration condition determines a pointer 94 below which the data can be deleted only if necessary. When the device is full, a file larger than the available free space can be added, completely deleting one or more expired files. The file could not be written due to lack of space can then be stored by overwriting the minimum necessary files have expired, even be split into two parts not ^'contiguous 95, 96. files are removed completely, thus leaving possibly an excess of free space 5, the size of the new written file not necessarily being exactly identical to the size of the old overwritten files.

Other files can be added following the same principle 97. The system protects old files beyond the expiration condition. If the writing of the new file requires the destruction of a file that does not meet the expiration conditions, the device is considered to be 'full', the operation is canceled 98 and reports an error. In the case where the expiration mechanism is implemented, the storage space has a circular structure 99 unlike the conventional linear structure 92 without expiration.

Device reports virtual free space corresponding to the space potentially freed by overwriting all the files that have 'expired'. The media can advantageously be partitioned with a variable expiration time. Once programmed, the expiration time cannot be changed. Partitioning advantageously makes it possible to manage different expiration times on different logical units. Partitioning can be fixed: number of partitions fixed at activation by the user , then size of fixed partitions. Partitioning can be dynamic:

- number of variable but limited partitions (no active data or partition can be deleted in all partitions). According to a particular embodiment linked to the current state of the art, available interfaces, and the existence of current basic components, FIG. 17 shows a more detailed block diagram of an implementation suggestion based on 'nonlimiting example, on a hard disk 100. Likewise, the functional breakdown described does not necessarily reflect the structure of the components used, each function being able to be in future implementations, distributed or integrated differently. For example, it is perfectly conceivable to integrate all or part of these functions into the internal electronics integrated into the hard disk, which already includes in current implementations a controller capable of covering part of the functions requested by modification of the microcode.

The controller as defined is based on a bridge interface component 101, usually integrating a processor capable of executing a program interpreting the commands, and usually two standard interface controllers for data storage 102, 103. In order to ensure optimal performance, these special controllers have specific internal hardware resources optimized so as not to degrade the overall performance in terms of data transfer. As a non-limiting example given the constant evolution of standards in this area, there are intelligent converters for external interfaces 105 of the FireWire / USB / Ethernet / SerialATA type to the interfaces of storage device ATA, IDE, SerialATA 102 Depending on the standard used, a specific active physical interface can be integrated or entrusted to an external component 104 depending on the transmission technology used.

This controller must be able to address a non-modifiable internal or external ROM program memory (ROM or OTP One Time Programmable) 106, and a RAM data memory (107) saved by battery 108 without external power supply thanks to a switching device 109, or possibly based on a Flash memory, or on a combination of the two technologies.

The functional and operational microcode, outside the initialization phase, is only in ROM and cannot be modified or updated. The initialization microcode is preloaded in the data memory 107 during manufacture and destroyed before activation, in a non-reversible manner. No program or microcode can be loaded by standard interfaces.

The functional data memory is saved so as to keep the temporary or non-critical intermediate parameters of the state of the system, and / or the history of past operations (log or 'log') or in progress in the event of a break in external power supply. The controller 101 does not need to be powered independently.

To the controller 101 can be associated an independent controller 110 allowing the monitoring of the internal state of the system with or without any external power source, also powered by battery 108. Monitoring monitors the condition of the equipment, the voltages of the internal batteries, and possibly other significant physical parameters (temperature). It is capable of controlling the interfacing of an external indicator 111 (indicator / LED) visually giving the state of the system even without power supply, and capable of preventing equipment failures, or its invalidation. The state of validity of the equipment indicated in this way must be correlated by other methods of investigation under power.

The controller 101 indirectly addresses a volatile memory / real-time clock component 112, comprising a volatile memory erasable by an external signal 119. These components natively manage a battery-backed power supply 113 linked to their permanent functionality. The real time clock is associated with a low consumption temperature compensated oscillator 114, which is also saved by battery 115. It is the stability of this oscillator which determines the maximum drift of the clock. For example, there are compensated oscillators compatible with a battery power supply which provide a drift, guaranteed less than 1 minute per year, which is reasonable and not very critical since the drift can be measured at any time.

The built-in erasable memory contains the unique identifiers: serial numbers 8 and unique secret identifier 7, as well as possibly other parameters deemed critical (pointers, etc.). The real-time clock assembly 112, 114 is secured by a sequential logic 116, which can be integrated into a low-consumption programmable logic component supplied by battery 113. This logic manages the initialization modes, the operational mode, and the destruction of unique parameters, and integrates a state machine 117 ensuring the unique and irreversible sequencing of operations.

On initialization, the sequencer 116 assists the method described and provides sequential write protection for the parameters to be initialized only once.

(clock, identifiers) in component 112. The initialization sequence can be carried out only once, starting from an unstable state artificially forced during manufacture, the transition to the next state being ensured by the controller, up to the operational state, then possibly the invalidation state. The interruption of the saved power supply or the intrusion detection resets the state machine to an initial state corresponding to the equipment invalidation state.

Functionally, the sequencer 116 provides timed hardware protection for reading the secret unique identifier, by authorizing its access in a timed manner limited by the real-time clock to, for example, once per 24 hours.

In the event of intrusion detection, or power supply failure, it ensures the inhibition of the real-time clock by blocking the oscillator 114 towards it, and reassures the invalidation of the identification parameters which may not be erased. . Access to the serial number and the read clock from the controller remains possible, read access to the erased unique identifier is permanently inhibited, for invalidation control, as well as any write access. The intrusion alert system 118 is linked to the intrusion sensor used, and permanently controls the electrical parameters of the sensor (resistance 121, capacity 122 short circuit or open circuit 123), as well as the temperature 120. This part is fully autonomous and self-powered by battery 124. An alert condition, or the absence of power by battery 124 erases 119 the volatile memory containing the identifiers 112 and resets the logic element 116 which goes into an equipment disabled state.

The system is supplied for the functional part by an external power supply 125, 14 possibly seconded by switching converters or voltage regulators 126 which generate the internal supply voltages of the equipment. The permanent power supply saves the internal batteries thanks to an electronic switching system 127 which primarily uses the external power supply 125, an optional external auxiliary battery 128, and finally as a last resort the internal batteries 108, 113, 115, 124 which extends the service life of the equipment, which is essentially limited by the service life of these internal batteries.

The number of internal batteries depends on the consumption and the functional independence of each sub-assembly. The distribution is given for information only.

The battery of the intrusion detector is possibly doubled for redundancy, but remains independent of any other internal power source.

Components 100 to 106 are only supplied by the main external source.

External interfaces such as power supplies, input / output buses, control signals are protected against overvoltages, overcurrents and filtered 129 so as to prevent the system from being damaged by a defective source, or intentionally by applying levels extremes that can put the equipment in an unknown state of vulnerability.

The RAM 106 used by the controller must at no time include a substantial partial copy of The internal secret identifier 7.

All of the internal elements are embedded in an electrically insulating and thermally conductive resin 130. A second non-secure casing 131 protects the whole of the equipment against shocks so as to avoid any accidental false alarm, reports wear interfaces (mechanical interfaces, connectors), the interchangeable auxiliary battery, visual indicators (indicators) and possibly sound indicators (buzzer), and improves the external aesthetic and / or ergonomic aspect. This jacket must allow visual access to the external serial number, and should be easily removable so as to perform the integrity check of one equipment ^'by itself. The double envelope can be adapted to the use of the peripheral •: • removable peripheral or. fixed in a standard 3 "l / 2 or 5" l / 4 rack, external device, device integrated in a laptop, individual or server.

The depressurization cap of the hard disk is carried by a conduit 132 outside the enclosure 10 without degrading the vulnerability (tubing and capillarity).

The following devices and methods can be implemented with advantage and are listed in a non-exhaustive manner: An auxiliary serial interface 133 can allow limited independent access to the controller in the event of failure of the main input / output functional interface 105, for the analysis of the authenticity and validity of the equipment, or can be used for advanced access control or secure copy functions by interfacing, for example a smart card.

It is necessary to provide a certified method of data migration with conservation of legal validity: authenticity and dating. As an example, this can be implemented using an external cryptographic method of digital signature and certification. A “smart card” type interface can also be used in this case with a cryptographic card authenticating the transfer of data from the source equipment to the target equipment using the internal date and the secret identifiers of both.

According to a variant not necessarily implemented, in the typical case where the controller executes an on-board program or contains programmable circuits, the object code

(possibly source) can be published and in fact can be checked.

According to a variant not necessarily implemented, it seems useful to include an autonomous system for protecting access to the user's data by password or any other identification method (smart card), so as to inhibit the request:

• data write access

• data read access (returning empty content, for example a character string ^Λ null ')

• visibility of directories and file names

• all access to files

According to ^" a variant not necessarily implemented, it is wise to include an internal system of limitation on the filling quota, which can avoid deliberate or accidental filling of the equipment (expressed for example in megabytes by number of days ) or make it possible to anticipate the replacement of the equipment for extension of the storage capacity. According to a variant not necessarily implemented, it is possible, depending on the file system used, to provide a possibility of changing the names of the files and directories, to facilitate the use of data over time. This also provides the possibility of modify the directory tree structure, and any movement of logical data insofar as this does not affect the content, the date, or possibly the extension or version of the files. File renaming can be supplemented, only the basic header cannot be changed: renaming of the long name, without touching the short name (8.3 characters in many operating systems). Moving files can be implemented: change of directories, without touching the data. A read protection mode can advantageously be implemented by partition, ensuring data confidentiality protection, the switching of this function can be controlled by password or smart card, and by partition. The device described is particularly suitable for the private storage and archiving of computer data likely to have the value of authentic, legal and / or legal proof before authorized bodies.

Claims

1. Process for the storage of computer data, ca.za.ct z3.se <if it consists of integrating entirely into a device:

- a data storage device (1),

- a suitable input-output controller (2) transforming said. data storage device (1) into a write-only or restricted storage device by deriving all commands from the input-output interface (13) coming from a host system (20), towards the internal interface (3) of the integrated data storage device (1), so:

- to authorize exclusively the addition of new files in the free space (5) by dating them independently thanks to a secure internal real-time clock (6), and

- to prohibit the erasure, modification and rewriting of data previously written and therefore previously dated (4), - to protect the whole of said device on the one hand against falsification or duplication by a unique internal secret identifier (7 ) in correspondence with a unique public serial number, on the other hand by a perimeter protection enclosure (10) intended to prevent intrusion into the equipment, modification of any component and internal data, as well as analysis the unique secret identifier without irreversibly compromising validity and authenticity,

- To detect an attempt of said intrusion, modification, analysis, said detections being able to be verified by several methods third to the said process, the latter being thus secure, and suitable for storing authenticated and dated computer data with probative value.

2. Method according to claim 1, characterized in that qμ 'it consists in electronically reproducing internally (8) said unique public serial number, to allow electronic control thereof.

3. Method according to claim 1 or 2, c & srsictésr ± & é e C <qμ ^r it further consists in physically protecting all internal components by leaving accessible only the external functional interfaces of input-output (13) and power supply ( 14), thanks to a perimeter sealed enclosure (10) encompassing all the components, consisting of a passive mechanical and electromagnetic protection (11) including an active intrusion detection sensor (12) and the associated electronics (9), the latter ensuring the immediate destruction of the secret internal unique identifier (7) and of the internal serial number (8) if necessary, by instant deletion, in the event of voluntary or involuntary deterioration changing the vulnerability or the autonomy of the device, or of internal investigation attempt detected by the sensor (12) which can make it possible to compromise, control, modify, analyze data, interfaces or internal components.

4. Method according to claim 3, characterized in that it also consists in keeping active, the internal real-time clock (6), the secret identifier (7), the internal serial number (8) if necessary ^. if applicable, and the intrusion and destruction detection electronics (9), outside of any external power source for the life specified for the equipment, thanks to a self-contained internal power subsystem ( 15), possibly seconded externally to increase its lifespan.

5. Method according to any one of claims 1 to 4, characterized <SΏ. this qμ '11 also consists in making it possible to guarantee the unique and non-reproducible personalization of the equipment while publishing its existence, possibly before certification, by randomly generating, by internal calculation of the equipment, the secret unique identifier (7), by internally calculating a one-way digital fingerprint (23) of this identifier which does not allow it to be traced back to it , by establishing the uniqueness of the fingerprint by interrogating a public database (21) containing the fingerprints of all the existing equipment, the identifier being recalculated if there is a collision and the device being formally identified as soon as that the fingerprint is unique, then adding to the public database (21), the references of the equipment: unique serial number (8, 18) and fingerprint (23).

6. Method according to claim 4 or 5, characterized in that qμ 'it further consists in ensuring the authenticity of the device after manufacture, the test, the final assembly and the activation of the perimeter protection (9), up to to the acquisition by the end user, by making unique the operations of assigning the unique serial number (8, 18), the synchronization of the real time clock (6) with respect to a reliable external time reference (16), internal self-generation of the unique secret identifier (7), the external copy of the unique secret identifier (17) in a secure environment from the unique secret identifier to a certifying body (22) authorized to guarantee the secrecy of this identifier and in fact the authenticity of the equipment, the uniqueness of each operation being guaranteed by design by irreversibly inhibiting or destroying and as soon as possible in the initialization sequence, the material elements and / or the microc ode allowing this initialization.

7. Method according to claim 2, and one of claims 5 or 6, characterized in that it further consists in allowing without limitation on the occurrence, the preliminary verification of authenticity and integrity, by checking and cross checking of serial numbers unique (8, 18) listed in the public database (21), by checking the validity and the evolution of the internal clock (6) compared to a reliable external time reference (16) by taking into account counts the maximum specified drift of the internal clock, and by visual or non-destructive inspection of the integrity of the external envelope (12).

8. Method according to claim 5 or 6, characterized in that it comprises, in addition, allowing authorized organizations to authenticate the equipment on request by verification by third-party methods using the internal secret identifier (7) without revealing it, by dynamically recalculating internally the imprint of the identifier (23) and by directly querying the public database (21), or by internally coupling the secret identifier to the internal date (6) and to an external message provided before internal calculation of the footprint of the assembly, all of the known elements being supplied for remote certification by the authorized certifying body (22) which alone can verify the consistency by means of the copy of the unique secret identifier (17) at its disposal, these functions implying the secret identifier, being timed and limited in internal occurrence of the equipment so as to avoid the reverse determination of the identifier by a nalysis or successive tests.

9. Method according to claim 6, characterized in that it further consists in allowing the certifying body (22) to authenticate itself under its control and as a last resort the equipment by injecting its copy of the secret identifier (17) for internal comparison of the equipment which detects perfect similarity, and sends in response a unique message which can be verified by said certifying body (22), this function involving the secret identifier being timed and limited by internal occurrence of equipment so as to avoid reverse determination of the identifier by analysis or successive tests.

10. Method according to any one of claims 1 to 8, characterized in that it also consists in allowing, as a variant not necessarily implemented, to manage autonomously the expiration of data in the past, by authorizing , only if the internal storage device (1) is completely filled, writing new files overwriting, within the limit of sufficient and in chronological order, the data prior to a relative date (94), update dynamically with respect to the current date provided by the internal clock (6), thereby derogating from the single writing rule without however compromising or modifying the useful data still having validity, and dated after the expiration date.

11. Method according to any one of the preceding claims, characterized in that it also consists in presenting any internal parameter and indicator specific to the said device in the form of virtual files with reserved names, seen by the host system as conventional files and avoiding the implementation of specific drivers or operating modes at the host system level for the management, administration and monitoring of said device.

12. Method according to any one of the preceding claims, characterized in that it also consists in partitioning said data storage device with a different expiration time per partition.

13. Device for storing computer data, characterized in that it fully integrates:

- a data storage device (1) comprising an internal interface (3),

- a suitable input-output controller (2),

- an input-output interface (13), - a secure internal real-time clock (6),

- a unique internal secret identifier (7),

- a unique public serial number,

- a perimeter protection enclosure (10), said adapted input-output controller (2) being able to transform said data storage device (1) into a single-write or restricted storage device by deriving all commands from the input-output interface

(13) coming from a host system (20), towards the internal interface (3) of the integrated peripheral (1), so:

- to authorize exclusively the addition of new files in the free space (5) by dating them independently thanks to the secure internal real-time clock (6), and

- to prohibit the erasure, modification and rewriting of data previously written and therefore previously dated (4), the whole of said device being protected on the one hand against falsification or duplication by the unique internal secret identifier (7 ) in correspondence with the unique public serial number, on the other hand by the perimeter protection enclosure (10) preventing intrusion into the equipment, modification of any component and internal data, as well as the analysis of the unique secret identifier without irreversibly compromising the validity and authenticity, the intrusion, modification, analysis attempts being able to be verified by several means third of the whole of said device, the latter being thus secure, inviolable and tamper-proof suitable for storing authenticated and dated computer data with probative value.

14. Device according to claim 13, characterized in that it comprises a serial number reproduced electronically internally (8) reproducing said unique public serial number.

15. Device according to claim 13 or 14, characterized in that it comprises means of physical protection of all the internal components while leaving accessible only the external functional interfaces of inputs-outputs (13) and supply (14) , thanks to a perimeter sealed enclosure (10) encompassing all the components, consisting of a passive mechanical and electromagnetic protection (11) including an active intrusion detection sensor (12) and the associated electronics (9), the latter ensuring the immediate destruction of the secret internal unique identifier (7) and of the internal serial number (8) if necessary by instantaneous erasure, in the event of voluntary or involuntary deterioration changing the vulnerability or the autonomy of the device, or of attempt of internal investigation detected by the sensor (12) which can make it possible to compromise, control, modify, analyze data, interfaces or internal components.

16. Device according to claim 14, characterized in that it comprises active holding means, the internal real time clock (6), the secret identifier (7), the internal serial number (8) the if applicable, and intrusion and destruction detection electronics (9), outside of any external power source for the life specified for the equipment, thanks to an internal power subsystem autonomous (15), possibly seconded externally to increase its lifespan.