T 0632/10 (Re-signing electronic documents/SAP) 06-12-2013

Admission of late-filed requests

1. Compared with the previous version, the claims according to the pre­sent main and first to third auxiliary requests were amended to overcome the clarity objections raised with the summons to oral proceedings, and the board accepts the claims according to the fourth auxiliary request as a genuine attempt to overcome the board's inventive step objection. The board therefore exercises its dis­cretion under Rule 13 (1) RPBA to admit the new re­quests into the procedure.

The invention

2. The application relates to the question of how to pro­vide long-term authenticity proof of electronic do­cu­ments based on what is known as "electronic signatures" or, equi­valently, "digital signa­tures" (see e.g. the original application, p. 4, 2nd par.).

2.1 An electronic signature is typically generated based on a hash value calculated from the electronic do­cument and encrypted with a private key of the sig­ning party. Users of the document can validate the sig­nature by decrypting the signature with the pub­lic key of the signing party and comparing the value so-obtained with a hash value re-generated from the docu­ment. In case of a match the document is deemed to be au­thentic. Public keys and corresponding certificates may have a li­mited validity or may be revoked because the private key has become publicly known or safer encryption methods have become standard (see also p. 3, 2nd par.). Also the associated digital signatures may thus become in­valid.

2.2 The application explains that in Germany electronic sig­­­natures may be acknowledged as documents in the le­gal sense if they comply with the German signature law (p. 2, 2nd par.). The relevant law is the above-men­­tioned Ger­man Digital Sig­nature Law SigG and its Or­di­nance­ SigV, in view of the priority date of the pre­sent application both in their versions issued in 2001. The appli­cation further ex­plains that for certain kinds of documents a proof of au­­thenti­ci­ty over many years is re­quired, and that to this end it is prescribed to re­apply "secure methods and algo­rithms ... periodically by re-signing or time-stam­ping the electronic document and its digital sig­na­ture" (see p. 5, last par. - p. 6, 2nd par.).

2.3 The application states that in state of the art time stamping processes the document itself must be avai­lable for the time stamping process (p. 6, lines 25-28). This is said to be inefficient, require expen­sive ar­chi­ving technology, and be unsafe as it re­quires the hand­ling of the electronic document (p. 6, lines 28-31). The invention sets out to address this problem.

2.4 The claimed invention (claim 1 of the main request) spe­­ci­fies that the electronic document is stored "in a first data archive" and "hash information data" is stored "in a second data archive ... different from the first" one. In the cha­rac­terising portion, the first ar­chive is referred to as "standard archive", the second one as a "re-sign archive". It is further claimed that "a new digital signature [is pro­vided] to the hash in­for­mation data" and stored in "a data archive", and it is speci­fied that the "hash infor­ma­tion data com­pri­ses the hash value ... and the digital signature of the electronic document".

2.5 In claim 1 of the first auxiliary request it is further specified that the "hash information data" comprises "a reference to the corresponding electronic document for a later re­trie­val of the electronic document".

In claim 1 of the second auxilia­ry request it is yet further specified that the "re-signed hash information data comprises a time stamp from a trusted third party".

Claim 1 of the third auxiliary request contains the additional requirement that the electronic document "is a set of electronic documents which comprises a plurality of single electronic documents".

Claim 1 of the fourth auxiliary request further defines a so-called "document (B)" which comprises for each of the set of documents a refe­rence, the hash value and "a description of" the used hashing algorithms, and which is time-stamped as a whole.

The prior art

3. D1 is a short excerpt of a standard text book on cryp­to­gra­phy.

3.1 It explains that signing long docu­ments may be in­effi­cient and that, therefore, hash func­tions are used to map documents to a short hash value which is signed instead of the document. For mathematical reasons, the sig­na­ture of the hash can safely be "equated" with the sig­na­ture of the document (see p. 38, lower half).

3.2 D1 further discloses that hashing in this context also increases privacy by making it possible that the signa­ture is kept separate from the document. A central da­ta­base could just store the hash values while the do­cu­ment could be kept secret elsewhere. The central data­base is disclosed to perform the time-stamping and the authentication (see p. 39, first full par.). It is fur­ther disclosed that a time-stamp is effectively a digi­tal signature including date and time information (see p. 38, 5th par.).

4. The relevant regulations of the German Signature Act are §§ 2 and 6 SigG as well as § 17 SigV which, for ease of reading, are reproduced here:

§ 2 SigG - Begriffsbestimmungen

Im Sinne des Gesetztes sind

1. "elektronische Signaturen" Daten in elektronischer Form, die anderen elektronischen Daten beigefügt oder logisch mit ihnen verknüpft sind und die zur Authentifizierung dienen,

§ 6 SigG - Unterrichtspflicht

(1) Der Zertifizierungsdiensteanbieter hat den Antrag­steller nach § 5 Abs. 1 über die Maßnahmen zu unter­richten, die erforderlich sind, um zur Sicherheit von qualifizierten elektroni­schen Signa­turen und zu deren zu­ver­lässiger Prüfung beizutra­gen. Er hat den Antrag­steller darauf hinzuweisen, dass Daten mit einer quali­fi­zier­ten elektro­ni­schen Signatur bei Bedarf neu zu sig­nieren sind, bevor der Sicher­heits­wert der vorhan­denen Signatur durch Zeitablauf gerin­ger wird.

§ 17 SigV - Zeitraum und Verfahren zur langfristigen Datensicherung

Daten mit einer qualifizierten elektronischen Signatur sind nach § 6 Abs. 1 Satz 2 des Signaturgesetzes neu zu signieren, wenn diese für längere Zeit in signierter Form benötigt werden, als die für ihre Erzeugung und Prü­fung eingesetzten Algorithmen und zugehörigen Para­me­ter als geeignet beurteilt sind. In diesem Falle sind die Daten vor dem Zeitpunkt des Ablaufs der Eig­nung der Al­gorithmen oder der zugehörigen Parameter mit einer neu­­en qua­li­fi­zierten elek­tronischen Signatur zu versehen. Diese muss mit geeigneten neuen Algorithmen oder zuge­hörigen Parametern erfolgen, frühere Signa­tu­ren ein­schließen und einen qualifizierten Zeitstempel tragen.

Objective technical problem

5. D1 discloses all features of the preamble of claim 1 (all requests) but is not con­cerned with provi­ding long-term authenti­city proofs (see grounds of appeal, p. 13, 1st par.). As a con­se­quence, D1 also does not dis­close that or how a do­cu­ment should be re-signed as spe­cified in the characterizing portion of claim 1 (all requests).

5.1 Yet, the board disagrees with the appellant that D1 teaches away from considering long-term authenticity proofs because hash functions are disclosed to be very safe (see p. 38, last par.). Apart from the fact that hash functions have been cracked despite their high safety, a digital signature may also become useless for other reasons, e.g. because a private key has leaked.

5.2 Any de­veloper of digital sig­nature software for the Ger­­man market at the relevant priority date had to com­ply with the German Signature Law, and developers in­te­rested in providing software supporting long-term digi­tal signatures had to comply with the German Signa­ture Law, especially with § 17 SigV, quite inde­pen­dent of any technical considerations they might also have had. The board notes that the law applies in­dependent­ly of tech­nical considerations even though the law itself relates to a technical issue.

5.3 The board therefore considers that an objective tech­nical problem solved by the invention is to implement a di­gital signature system according to D1 suitable for long-term authenticity proof compliant with the re­quire­­ments of § 17 SigV. This also appears to conform with the background of the invention as presented in the application (loc. cit.).

Relevance of a German law for the assessment of inventive step of a European patent

6. During oral proceedings, the appellant argued that the German Signature Law might not be relevant for a Euro­pean patent application such as the present one for which states other than Germany may be designated, because, as the board understands the argument, compli­ance with German law is of no concern outside Germany.

6.1 The board remains unconvinced by this argument for two reasons.

6.2 Primarily, an objective assessment of inventive step prohibits any differentiation between skilled persons according to their nationality, residence, location or language (see also T 426/88, OJ EPO 1992, 427, reasons 6.4.; T 1688/08, unpublished, reasons 4). Thus, even if an invention happened to be obvious only for skilled persons of German nationality or residence, it would still lack an inventive step in the sense of Article 56 EPC 1973.

6.3 Moreover, the German Signature Law is available and accessible be­yond its region of validity. Digital sig­na­ture soft­ware for the German market must com­ply with the German Sig­na­ture Law. Compliance must be en­sured by any de­veloper of such software, in­de­pen­dent of its own natio­na­lity or residence. The fact that the Ger­­man Sig­na­ture Law is valid only within Germany thus has no bea­­ring on its status as prior art or its rele­vance for the assessment of inventive step outside Ger­many.

Inventive Step

Main request

7. D1 discloses an archival system which does not store the electronic document, but only its hash value in a central database (loc. cit.), i.e. in a "second data archive" as claimed. In this scenario, the hash value "re­pre­sents" the document which the copyright owner pre­­fers to keep secret in his or her own, separate lo­cal sto­rage, i.e. in a "first data archive" as claimed.

7.1 § 17 SigV prescribes that a digital signature system sui­table for long-term authenticity proof must be equipped to re-sign "the data" - i.e. the relevant elec­­tronic document - before the used algorithms or corres­­ponding parameters become useless, based on "sui­table new algorithms or corresponding parameters". § 17 SigV also prescribes that the new signature al­go­rithm include earlier signatures and a qualified time-stamp.

7.2 The skilled person modifying the system of D1 so as to comply with § 17 SigV would thus have to provide a way to re­new digital signatures. Naturally, the skilled person would enhance the "second data archive", responsible already for the primary signature, so as to become a "re-sign archive".

7.3 The skilled person would understand from § 17 SigV that re-signing could use the old algo­rithms and parameters as long and to the extent to which they are still safe and permissible. As long as possible, the skilled per­son would obviously consider using the same algo­rithms for re-signing that were used to produce the original digital signature, in particular the same hashing al­go­rithm.

7.4 The appellant argued that, according to D1, the two steps of generating a hash value and encrypting it were necessary parts of generating a digital signature (see nos. (1) and (2) on p. 38) and that D1 lacked any in­dication that either could be dispensed with. Also the requirement of § 17 SigV to re-sign "the data" had to be read as regenerating an electronic signature from the original document.

7.5 Therefore, so the argument, the available prior art taught the non-imaginative skilled person to refer to the original docum­ent whenever it had to be re-signed.

7.6 The board disagrees. As long as the same hashing alg­o­rithm is used, the skilled person would realize that it is not necessary to refer back to the original document because re-cal­cu­la­ting the hash value would only pro­duce the very same value which is already available. The skilled person would thus avoid this for obvious efficiency rea­sons and, in the system according to D1, for the addi­tional reason that the original document is not or not easily accessible anyway. In the board's view this is also not in contradiction with § 17 SigV due to the fact that § 2(1) SigG provides a rather broad definition of the term "electronic signature".

7.7 The skilled person would find it obvious to produce a new digital signature based on the existing, old hash value and, because § 17 SigV so provides, would have to include the ear­li­er digital signature.

7.8 The board concludes that claim 1 of the main re­quest is an obvious implemen­ta­tion of the system of D1 compliant with § 17 SigV, and there­fore does not in­volve an in­ventive step in the sense of Article 56 EPC 1973.

First auxiliary request

8. Even though according to D1 the electronic document is stored separately from the electronic signature it will, at some point, have to be retrieved. It is there­fore obvious that some suitable "reference" be provided that enables such retrieval. This might be a contact address for the copyright owner just as well as an in­dex into some storage location which might support au­to­matic retrieval.

8.1 Neither D1 nor § 17 SigV discloses or prescribes that such reference be included in the new digital signa­ture. The description is silent about the reason for doing this, but the board considers that the inclusion of any information in the digital signature protects that information against tampering.

8.2 In the board's judgement it would be evident for the skilled person that information relevant to retrieve a protected electronic document must also be pro­tected against tampering: An archiving system such as that of D1 would not achieve its purpose if it were to authen­ticate a document via its hash value but then point an interested reader to the wrong document. The board there­fore deems it obvious that all security-re­levant information that happens to be stored in the "se­cond da­ta archive" be included in the digital sig­na­ture, the reference included. Also § 17 SigV contains a pertinent hint in requiring that the new signature should "include" the old one.

8.3 The appellant submits that the "reference value does not only serve for retrieving the document [but also] for increasing security by enhancing the amount of struc­­tured data re-signed". The board first notes that if the provision of a feature is obvious as a means to achieve one effect, it does not become less obvious if it also has another effect. Beyond that, the board is not convinced that the inclusion of additional infor­ma­tion into the data being signed can be said to increase security: The primary effect of signing additional in­for­mation is that of providing authenticity proof for the additional in­for­ma­tion. Moreover it appears ques­tio­­nable whether a di­gital signature indeed becomes sa­fer, or in what res­pect, when applied to additional information: By the same logic it would appear that a di­gital signa­ture would be the safer the longer the signed electronic document. If a hashing al­gorithm were broken by, say, a collision attack, digi­tal signa­tures relying on this hashing algorithm would be compromised independent of how much "addi­­­tional in­for­mation" the signed data contained.

8.4 Therefore, the board comes to the conclusion that also claim 1 of the first auxiliary request lacks an inventive step, Article 56 EPC 1973.

Second auxiliary request

9. During oral proceedings, the appellant confirmed that a time stamp according to the claim should be construed to subsume a normal electronic signature including sui­table date and time information. This is, in fact, the definition for time stamping given in D1 (loc. cit.). The board further considers that the central database according to D1 must be considered as a "trusted third party". Therefore, also claim 1 of the second auxiliary request lacks an inventive step, Article 56 EPC 1973.

Third auxiliary request

10. The appellant points out that D1 illustrates electronic signatures only by reference to small "documents", name­ly contracts or checks and thus neither discloses nor suggests that a signed document could "comprise a plurality of single electronic documents".

10.1 The board, however, considers it obvious for the skilled person that the principles of electronic signa­tures apply independent of the size and form of the document, and well-known that digital signatures have been applied to all sorts of documents (email, books, music, video, etc.).

10.2 In the board's judgment it is also obvious that "docu­ments" to be protected as a whole may con­sist of se­ve­ral individual files, i.e. documents in the "technical sense": For instance, the individual chapters of an elec­tronic book may be stored in separate files, as may be a contract and its potential annexes.

10.3 The board considers it obvious to apply a common elec­tro­nic signature to all components of a document to be protected. Claim 1 of the third auxiliary request thus also lacks an inventive step, Article 56 EPC 1973.

Fourth auxiliary request

11. The claim makes reference to "a description of one or more algorithms used to calculate the hash values" but neither the claim nor the application as a whole define what kind of "description" is meant. The board con­strues this term broadly as any information relevant to identify the pertinent algorithms.

11.1 In order to validate the hash value encrypted in an electronic signature, it must be re-ge­nerated from the signed document. Therefore, a digi­tal signature must identify, one way or another, the hashing algorithm re­lied on. Moreover, this information, too, must not be tampered with.

11.2 The board therefore considers for the above reasons (point 8.2) that it is obvious to include "a de­scrip­tion of" the relevant algorithms in the hash in­for­ma­tion being signed.

11.3 The use of time-stamping and the joint signing of se­ve­ral documents was separately found obvious above. The board considers that this also applies to their combination.

12. Thus, also claim 1 of the fourth auxiliary request lacks an inventive step, Article 56 EPC 1973.

Summary

13. There being no allowable request, the appeal must be dismissed.