Network Working Group T. Gondrom Request for Comments: 4998 Open Text Corporation Category: Standards Track R. Brandner InterComponentWare AG U. Pordesch Fraunhofer Gesellschaft August 2007
Evidence Record Syntax (ERS)
Status of This Memo
This document specifies an Internet standards track protocol for the Internet community, and requests discussion and suggestions for improvements. Please refer to the current edition of the "Internet Official Protocol Standards" (STD 1) for the standardization state and status of this protocol. Distribution of this memo is unlimited.
Copyright Notice
Copyright (C) The IETF Trust (2007).
Abstract
In many scenarios, users must be able prove the existence and integrity of data, including digitally signed data, in a common and reproducible way over a long and possibly undetermined period of time. This document specifies the syntax and processing of an Evidence Record, a structure designed to support long-term non- repudiation of existence of data.
In many application areas of electronic data exchange, a non- repudiable proof of the existence of digital data must be possible. In some cases, this proof must survive the passage of long periods of time. An important example is digitally signed data. Digital signatures can be used to demonstrate data integrity and to perform source authentication. In some cases, digitally signed data must be archived for 30 years or more. However, the reliability of digital signatures over long periods is not absolute. During the archival period, hash algorithms and public key algorithms can become weak or certificates can become invalid. These events complicate the reliance on digitally signed data after many years by increasing the likelihood that forgeries can be created. To avoid losing the desired security properties derived from digital signatures, it is necessary to prove that the digitally signed data already existed before such a critical event. This can be accomplished using a timestamp. However, some timestamps rely upon mechanisms that will be subject to the same problems. To counter this problem, timestamps are renewed by simply obtaining a new timestamp that covers the original data and its timestamps prior to the compromise of mechanisms used to generate the timestamps. This document provides a syntax to support the periodic renewal of timestamps.
It is necessary to standardize the data formats and processing procedures for such timestamps in order to be able to verify and communicate preservation evidence. A first approach was made by IETF within [RFC3126], where an optional Archive Timestamp Attribute was specified for integration in signatures according to the Cryptographic Messages Syntax (CMS) [RFC3852].
Evidence Record Syntax (ERS) broadens and generalizes this approach for data of any format and takes long-term archive service requirements [RFC4810] into account -- in particular, the handling of large sets of data objects. ERS specifies a syntax for an EvidenceRecord, which contains a set of Archive Timestamps and some additional data. This Evidence Record can be stored separately from the archived data, as a file, or integrated into the archived data, i.e., as an attribute. ERS also specifies processes for generation and verification of Evidence Records. Appendix A describes the integration and use of an EvidenceRecord in context of signed and enveloped messages according to the Cryptographic Message Syntax (CMS). ERS does not specify a protocol for interacting with a long- term archive system. The Long-term Archive Protocol specification being developed by the IETF LTANS WG addresses this interface.
ERS is designed to meet the requirements for data structures set forth in [RFC4810].
The basis of the ERS are Archive Timestamps, which can cover a single data object (as an RFC3161 compliant timestamp does) or can cover a group of data objects. Groups of data objects are addressed using hash trees, first described by Merkle [MER1980], combined with a timestamp. The leaves of the hash tree are hash values of the data objects in a group. A timestamp is requested only for the root hash of the hash tree. The deletion of a data object in the tree does not influence the provability of others. For any particular data object, the hash tree can be reduced to a few sets of hash values, which are sufficient to prove the existence of a single data object. Similarly, the hash tree can be reduced to prove existence of a data group, provided all members of the data group have the same parent node in the hash tree. Archive Timestamps are comprised of an optional reduced hash tree and a timestamp.
An EvidenceRecord may contain many Archive Timestamps. For the generation of the initial Archive Timestamp, the data objects to be timestamped have to be determined. Depending on the context, this could be a file or a data object group consisting of multiple files, such as a document and its associated digital signature.
Before the cryptographic algorithms used within the Archive Timestamp become weak or timestamp certificates become invalid, Archive Timestamps have to be renewed by generating a new Archive Timestamp. (Note: Information about the weakening of the security properties of public key and hash algorithms, as well as the risk of compromise of private keys of Time Stamping Units, has to be closely watched by the Long-Term Archive provider or the owner of the data objects himself. This information should be gathered by "out-of-band" means and is out of scope of this document.) ERS distinguishes two ways for renewal of an Archive Timestamp: Timestamp Renewal and Hash-Tree Renewal.
Depending on the conditions, the respective type of renewal is required: The timestamp renewal is necessary if the private key of a Timestamping Unit has been compromised, or if an asymmetric algorithm or a hash algorithm used for the generation of the timestamps is no longer secure for the given key size. If the hash algorithm used to build the hash trees in the Archive Timestamp loses its security properties, the Hash-Tree Renewal is required.
In the case of Timestamp Renewal, the timestamp of an Archive Timestamp has to be hashed and timestamped by a new Archive Timestamp. This mode of renewal can only be used when it is not
Gondrom, et al. Standards Track [Page 4]
RFC 4998 ERS August 2007
necessary to access the archived data objects covered by the timestamp. For example, this simple form of renewal is sufficient if the public key algorithm of the timestamp is going to lose its security or the timestamp authority certificate is about to expire. This is very efficient, in particular, if Archive Timestamping is done by an archiving system or service, which implements a central management of Archive Timestamps.
Timestamp renewal is not sufficient if the hash algorithm used to build the hash tree of an Archive Timestamp becomes insecure. In the case of Hash-Tree Renewal, all evidence data must be accessed and timestamped. This includes not only the timestamps but also the complete Archive Timestamps and the archived data objects covered by the timestamps, which must be hashed and timestamped again by a new Archive Timestamp.
Archived data object: A data unit that is archived and has to be preserved for a long time by the Long-term Archive Service.
Archived data object group: A set of two or more of data objects, which for some reason belong together. For example, a document file and a signature file could be an archived data object group, which represent signed data.
Archive Timestamp: A timestamp and typically lists of hash values, which allow the verification of the existence of several data objects at a certain time. (In its most simple variant, when it covers only one object, it may only consist of the timestamp.)
Archive Timestamp Chain: Part of an Archive Timestamp Sequence, it is a time-ordered sequence of Archive Timestamps, where each Archive Timestamp preserves non-repudiation of the previous Archive Timestamp, even after the previous Archive Timestamp becomes invalid. Overall non-repudiation is maintained until the new Archive Timestamp itself becomes invalid. The process of generating such an Archive Timestamp Chain is called Timestamp Renewal.
Archive Timestamp Sequence: Part of the Evidence Record, it is a sequence of Archive Timestamp Chains, where each Archive Timestamp Chain preserves non-repudiation of the previous Archive Timestamp Chains, even after the hash algorithm used within the previous Archive Timestamp's hash tree became weak. Non-repudiation is preserved until the last Archive Timestamp of the last chain becomes invalid. The process of generating such an Archive Timestamp Sequence is called Hash-Tree Renewal.
Gondrom, et al. Standards Track [Page 5]
RFC 4998 ERS August 2007
Evidence: Information that may be used to resolve a dispute about various aspects of authenticity of archived data objects.
Evidence record: Collection of evidence compiled for one or more given archived data objects over time. An evidence record includes all Archive Timestamps (within structures of Archive Timestamp Chains and Archive Timestamp Sequences) and additional verification data, like certificates, revocation information, trust anchors, policy details, role information, etc.
Long-term Archive (LTA) Service: A service responsible for preserving data for long periods of time, including generation and collection of evidence, storage of archived data objects and evidence, etc.
Reduced hash tree: The process of reducing a Merkle hash tree [MER1980] to a list of lists of hash values. This is the basis of storing the evidence for a single data object.
Timestamp: A cryptographically secure confirmation generated by a Time Stamping Authority (TSA). [RFC3161] specifies a structure for timestamps and a protocol for communicating with a TSA. Besides this, other data structures and protocols may also be appropriate, e.g., such as defined in [ISO-18014-1.2002], [ISO-18014-2.2002], [ISO-18014-3.2004], and [ANSI.X9-95.2005].
An Archive Timestamp relates to a data object, if the hash value of this data object is part of the first hash value list of the Archive Timestamp. An Archive Timestamp relates to a data object group, if it relates to every data object of the group and no other data objects. An Archive Timestamp Chain relates to a data object / data object group, if its first Archive Timestamp relates to this data object/data object group. An Archive Timestamp Sequence relates to a data object / data object group, if its first Archive Timestamp Chain relates to this data object/data object group.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].
As many open ASN.1 compilers still support the 1988 syntax, this standard offers to support two versions of ASN.1 1997-ASN.1 and 1988- ASN.1. (For specification of ASN.1 refer to [CCITT.X208.1988], [CCITT.X209.1988], [CCITT.X680.2002] and [CCITT.X690.2002].) This specification defines the two ASN.1 modules, one for 1988 conform ASN.1 and another in 1997-ASN.1 syntax. Depending on the syntax version of your compiler implementation, you can use the imports for the 1988 conformant ASN.1 syntax or the imports for the 1997-ASN.1 syntax. The appendix of this document lists the two complete alternative ASN.1 modules. If there is a conflict between both modules, the 1988-ASN.1 module precedes.
2.1.1. ASN.1 Module Definition for 1988 ASN.1 Syntax
The specification exports all definitions and imports various definitions. Depending on the ASN.1 syntax version of your implementation, you can use the imports for the 1988 conform ASN.1 syntax below or the imports for the 1997-ASN.1 syntax in Section 2.2.2.
Gondrom, et al. Standards Track [Page 7]
RFC 4998 ERS August 2007
2.2.1. Imports and Exports Conform with 1988 ASN.1
-- EXPORTS ALL --
IMPORTS
-- Imports from RFC 3852 Cryptographic Message Syntax ContentInfo, Attribute FROM CryptographicMessageSyntax2004 -- FROM [RFC3852] { iso(1) member-body(2) us(840) rsadsi(113549) pkcs(1) pkcs-9(9) smime(16) modules(0) cms-2004(24) }
An Evidence Record is a unit of data, which can be used to prove the existence of an archived data object or an archived data object group at a certain time. The Evidence Record contains Archive Timestamps, generated during a long archival period and possibly useful data for validation. It is possible to store this Evidence Record separately from the archived data objects or to integrate it into the data itself. For data types, signed data and enveloped data of the CMS integration are specified in Appendix A.
CryptoInfos ::= SEQUENCE SIZE (1..MAX) OF Attribute
The fields have the following meanings:
The 'version' field indicates the syntax version, for compatibility with future revisions of this specification and to distinguish it from earlier non-conformant or proprietary versions of the ERS. The value 1 indicates this specification. Lower values indicate an earlier version of the ERS has been used. An implementation conforming to this specification SHOULD reject a version value below 1.
Gondrom, et al. Standards Track [Page 9]
RFC 4998 ERS August 2007
digestAlgorithms is a sequence of all the hash algorithms used to hash the data object over the archival period. It is the union of all digestAlgorithm values from the ArchiveTimestamps contained in the EvidenceRecord. The ordering of the values is not relevant.
cryptoInfos allows the storage of data useful in the validation of the archiveTimeStampSequence. This could include possible Trust Anchors, certificates, revocation information, or the current definition of the suitability of cryptographic algorithms, past and present (e.g., RSA 768-bit valid until 1998, RSA 1024-bit valid until 2008, SHA1 valid until 2010). These items may be added based on the policy used. Since this data is not protected within any timestamp, the data should be verifiable through other mechanisms. Such verification is out of scope of this document.
encryptionInfo contains the necessary information to support encrypted content to be handled. For discussion of syntax, please refer to Section 6.1.
ArchiveTimeStampSequence is a sequence of ArchiveTimeStampChain, described in Section 5.
If the archive data objects were encrypted before generating Archive Timestamps but a non-repudiation proof is needed for unencrypted data objects, the optional encryptionInfos field contains data necessary to unambiguously re-encrypt data objects. If omitted, it means that data objects are not encrypted or that a non-repudiation proof for the unencrypted data is not required. For further details, see Section 6.
The generation of an EvidenceRecord can be described as follows:
1. Select a data object or group of data objects to archive.
2. Create the initial Archive Timestamp (see Section 4, "Archive Timestamp").
3. Refresh the Archive Timestamp when necessary, by Timestamp Renewal or Hash-Tree Renewal (see Section 5).
The process of generation depends on whether the Archive Timestamps are generated, stored, and managed by a centralized instance. In the case of central management, it is possible to collect many data objects, build hash trees, store them, and reduce them later. In case of local generation, it might be easier to generate a simple Archive Timestamp without building hash trees. This can be
Gondrom, et al. Standards Track [Page 10]
RFC 4998 ERS August 2007
accomplished by omitting the reducedHashtree field from the ArchiveTimestamp. In this case, the ArchiveTimestamp covers a single data object. Using this approach, it is possible to "convert" existing timestamps into ArchiveTimestamps for renewal.
An Archive Timestamp is a timestamp and a set of lists of hash values. The lists of hash values are generated by reduction of an ordered Merkle hash tree [MER1980]. The leaves of this hash tree are the hash values of the data objects to be timestamped. Every inner node of the tree contains one hash value, which is generated by hashing the concatenation of the children nodes. The root hash value, which represents unambiguously all data objects, is timestamped.
The fields of type ArchiveTimeStamp have the following meanings:
Gondrom, et al. Standards Track [Page 11]
RFC 4998 ERS August 2007
digestAlgorithm identifies the digest algorithm and any associated parameters used within the reduced hash tree. If the optional field digestAlgorithm is not present, the digest algorithm of the timestamp MUST be used. Which means, if timestamps according to [RFC3161] are used in this case, the content of this field is identical to hashAlgorithm of messageImprint field of TSTInfo.
attributes contains information an LTA might want to provide to document individual renewal steps and the creation of the individual ArchiveTimeStamps, e.g., applied policies. As the structure of the ArchiveTimeStamp may be protected by hash and timestamps, the ordering is relevant, which is why a SET is used instead of a SEQUENCE.
reducedHashtree contains lists of hash values, organized in PartialHashtrees for easier understanding. They can be derived by reducing a hash tree to the nodes necessary to verify a single data object. Hash values are represented as octet strings. If the optional field reducedHashtree is not present, the ArchiveTimestamp simply contains an ordinary timestamp.
timeStamp should contain the timestamp as defined in Section 1.3. (e.g., as defined with TimeStampToken in [RFC3161]). Other types of timestamp MAY be used, if they contain time data, timestamped data, and a cryptographically secure confirmation from the TSA of these data.
The lists of hash values of an Archive Timestamp can be generated by building and reducing a Merkle hash tree [MER1980].
Such a hash tree can be built as follows:
1. Collect data objects to be timestamped.
2. Choose a secure hash algorithm H and generate hash values for the data objects. These values will be the leaves of the hash tree.
3. For each data group containing more than one document, its respective document hashes are binary sorted in ascending order, concatenated, and hashed. The hash values are the complete output from the hash algorithm, i.e., leading zeros are not removed, with the most significant bit first.
4. If there is more than one hash value, place them in groups and sort each group in binary ascending order. Concatenate these values and generate new hash values, which are inner nodes of
Gondrom, et al. Standards Track [Page 12]
RFC 4998 ERS August 2007
this tree. (If additional hash values are needed, e.g., so that all nodes have the same number of children, any data may be hashed using H and used.) Repeat this step until there is only one hash value, which is the root node of the hash tree.
5. Obtain a timestamp for this root hash value. The hash algorithm in the timestamp request MUST be the same as the hash algorithm of the hash tree, or the digestAlgorithm field of the ArchiveTimeStamp MUST be present and specify the hash algorithm of the hash tree.
An example of a constructed hash tree for 3 data groups, where data groups 1 and 3 only contain one document, and data group 2 contains 3 documents:
h1 = H(d1) where d1 is the only data object in data group 1 h3 = H(d3) where d3 is the only data object in data group 3 h12 = H( binary sorted and concatenated (h1, h2abc)) h123 = H( binary sorted and concatenated (h12, h3)) h2a = H(first data object of data object group 2) h2b = H(second data object of data object group 2) h2c = H(third data object of data object group 2) h2abc = H( binary sorted and concatenated (h2a, h2b, h2c))
Gondrom, et al. Standards Track [Page 13]
RFC 4998 ERS August 2007
The hash tree can be reduced to lists of hash values, necessary to have a proof of existence for a single data object:
1. Generate hash value h of the data object, using hash algorithm H of the hash tree.
2. Select all hash values, which have the same father node as h. Generate the first list of hash values by arranging these hashes, in binary ascending order. This will be stored in the structure of the PartialHashtree. Repeat this step for the father node of all hashes until the root hash is reached. The father nodes themselves are not saved in the hash lists -- they are computable.
3. The list of all partialHashtrees finally is the reducedHashtree. (All of the specified hash values under the same father node, except the father node of the nodes below, are grouped in a PartialHashtree. The sequence list of all Partialhashtrees is the reducedHashtree.)
4. Finally, add the timestamp and the info about the hash algorithm to get an Archive Timestamp.
Assuming that the sorted binary ordering of the hashes in Figure 1 is: h2abc < h1, then the reduced hash tree for data group 1 (d1) is:
Figure 3: Reduced hash tree for data object group 2
The pseudo ASN1 for this reduced hash tree would look like: rht2 = SEQ(pht3, pht4, pht5)
with the PartialHashtrees pht3, pht4, and pht5 pht3 = SEQ (h2b, h2c, h2a) pht4 = SEQ (h1) pht5 = SEQ (h3)
Note there are no restrictions on the quantity or length of hash- value lists. Also note that it is profitable but not required to build hash trees and reduce them. An Archive Timestamp may consist only of one list of hash-values and a timestamp or only a timestamp with no hash value lists.
The data (e.g. certificates, Certificate Revocation Lists (CRLs), or Online Certificate Status Protocol (OCSP) responses) needed to verify the timestamp MUST be preserved, and SHOULD be stored in the timestamp itself unless this causes unnecessary duplication. A timestamp according to [RFC3161] is a CMS object in which certificates can be stored in the certificates field and CRLs can be stored in the crls field of signed data. OCSP responses can be stored as unsigned attributes [RFC3126]. Note [ANSI.X9-95.2005], [ISO-18014-2.2002], and [ISO-18014-3.2004], which specify verifiable timestamps that do not depend on certificates, CRLs, or OCSP responses.
An Archive Timestamp shall prove that a data object existed at a certain time, given by timestamp. This can be verified as follows:
1. Calculate hash value h of the data object with hash algorithm H given in field digestAlgorithm of the Archive Timestamp.
Gondrom, et al. Standards Track [Page 15]
RFC 4998 ERS August 2007
2. Search for hash value h in the first list (partialHashtree) of reducedHashtree. If not present, terminate verification process with negative result.
3. Concatenate the hash values of the actual list (partialHashtree) of hash values in binary ascending order and calculate the hash value h' with algorithm H. This hash value h' MUST become a member of the next higher list of hash values (from the next partialHashtree). Continue step 3 until a root hash value is calculated.
4. Check timestamp. In case of a timestamp according to [RFC3161], the root hash value must correspond to hashedMessage, and digestAlgorithm must correspond to hashAlgorithm field, both in messageImprint field of timeStampToken. In case of other timestamp formats, the hash value and digestAlgorithm must also correspond to their equivalent fields if they exist.
If a proof is necessary for more than one data object, steps 1 and 2 have to be done for all data objects to be proved. If an additional proof is necessary that the Archive Timestamp relates to a data object group (e.g., a document and all its signatures), it can be verified additionally, that only the hash values of the given data objects are in the first hash-value list.
5. Archive Timestamp Chain and Archive Timestamp Sequence
An Archive Timestamp proves the existence of single data objects or data object group at a certain time. However, this first Archive Timestamp in the first ArchiveTimeStampChain can become invalid, if hash algorithms or public key algorithms used in its hash tree or timestamp become weak or if the validity period of the timestamp authority certificate expires or is revoked.
Prior to such an event, the existence of the Archive Timestamp or archive timestamped data has to be reassured. This can be done by creating a new Archive Timestamp. Depending on whether the timestamp becomes invalid or the hash algorithm of the hash tree becomes weak, two kinds of Archive Timestamp renewal are possible:
o Timestamp Renewal: A new Archive Timestamp is generated, which covers the timestamp of the old one. One or more Archive Timestamps generated by Timestamp Renewal yield an Archive Timestamp Chain for a data object or data object group.
Gondrom, et al. Standards Track [Page 16]
RFC 4998 ERS August 2007
o Hash-Tree Renewal: A new Archive Timestamp is generated, which covers all the old Archive Timestamps as well as the data objects. A new Archive Timestamp Chain is started. One or more Archive Timestamp Chains for a data object or data object group yield an Archive Timestamp Sequence.
After the renewal, always only the last (i.e., most recent) ArchiveTimeStamp and the algorithms and timestamps used by it must be watched regarding expiration and loss of security.
ArchiveTimeStampChain and ArchiveTimeStampSequence have the following ASN.1 Syntax:
ASN.1 ArchiveTimeStampChain and ArchiveTimeStampSequence
ArchiveTimeStampChain ::= SEQUENCE OF ArchiveTimeStamp
ArchiveTimeStampSequence ::= SEQUENCE OF ArchiveTimeStampChain
ArchiveTimeStampChain and ArchiveTimeStampSequence MUST be ordered ascending by time of timestamp. Within an ArchiveTimeStampChain, all reducedHashtrees of the contained ArchiveTimeStamps MUST use the same Hash-Algorithm.
The initial Archive Timestamp relates to a data object or a data object group. Before cryptographic algorithms that are used within the most recent Archive Timestamp (which is, at the beginning, the initial one) become weak or their timestamp certificates become invalid, Archive Timestamps have to be renewed by generating a new Archive Timestamp.
In the case of Timestamp Renewal, the content of the timeStamp field of the old Archive Timestamp has to be hashed and timestamped by a new Archive Timestamp. The new Archive Timestamp MAY not contain a reducedHashtree field, if the timestamp only simply covers the previous timestamp. However, generally one can collect a number of old Archive Timestamps and build the new hash tree with the hash values of the content of their timeStamp fields.
The new Archive Timestamp MUST be added to the ArchiveTimestampChain. This hash tree of the new Archive Timestamp MUST use the same hash algorithm as the old one, which is specified in the digestAlgorithm
Gondrom, et al. Standards Track [Page 17]
RFC 4998 ERS August 2007
field of the Archive Timestamp or, if this value is not set (as it is optional), within the timestamp itself.
In the case of Hash-Tree Renewal, the Archive Timestamp and the archived data objects covered by the Archive Timestamp must be hashed and timestamped again, as described below:
1. Select a secure hash algorithm H.
2. Select data objects d(i) referred to by initial Archive Timestamp (objects that are still present and not deleted). Generate hash values h(i) = H((d(i)). If data groups with more than one document are present, then one will have more than one hash for a group, i.e., h(i_a), h(i_b).., h(i_n)
3. atsc(i) is the encoded ArchiveTimeStampSequence, the concatenation of all previous Archive Timestamp Chains (in chronological order) related to data object d(i). Generate hash value ha(i) = H(atsc(i)). Note: The ArchiveTimeStampChains used are DER encoded, i.e., they contain sequence and length tags.
4. Concatenate each h(i) with ha(i) and generate hash values h(i)' = H (h(i)+ ha(i)). For multi-document groups, this is: h(i_a)' = H (h(i_a)+ ha(i)) h(i_b)' = H (h(i_b)+ ha(i)), etc.
5. Build a new Archive Time Stamp for each h(i)'. (Hash-tree generation and reduction is defined in Section 4.2; note that each h(i)' will be treated in Section 4.2 as the document hash. The first hash value list in the reduced hash tree should only contain h(i)'. For a multi-document group, the first hash value list will contain the new hashes for all the documents in this group, i.e., h(i_a)', h(i_b)'.., h(i_n)')
6. Create new ArchiveTimeStampChain containing the new Archive Timestamp and append this ArchiveTimeStampChain to the ArchiveTimeStampSequence.
Let H be the new secure hash algorithm ha(1), ha(2), ha(3) are as defined in step 4 above h1' = H( binary sorted and concatenated (H(d1), ha(1))) d1 is the original document from data group 1 h3' = H( binary sorted and concatenated (H(d3), ha(3))) d3 is the original document from data group 3
h2a = H(first data object of data object group 2) ... h2c = H(third data object of data object group 2) h2a' = H( binary sorted and concatenated (h2a, ha(2))) ... h2c' = H( binary sorted and concatenated (h2c, ha(2)))
h2abc = H( binary sorted and concatenated (h2a', h2b', h2c'))
ArchiveTimeStamps that are not necessary for verification should not be added to an ArchiveTimeStampChain or ArchiveTimeStampSequence.
To get a non-repudiation proof that a data object existed at a certain time, the Archive Timestamp Chains and their relations to each other and to the data objects have to be proved:
Gondrom, et al. Standards Track [Page 19]
RFC 4998 ERS August 2007
1. Verify that the first Archive Timestamp of the first ArchiveTimestampChain (the initial Archive Timestamp) contains the hash value of the data object.
2. Verify each ArchiveTimestampChain. The first hash value list of each ArchiveTimeStamp MUST contain the hash value of the timestamp of the Archive Timestamp before. Each Archive Timestamp MUST be valid relative to the time of the following Archive Timestamp. All Archive Timestamps within a chain MUST use the same hash algorithm and this algorithm MUST be secure at the time of the first Archive Timestamp of the following ArchiveTimeStampChain.
3. Verify that the first hash value list (partialHashtree) of the first Archive Timestamp of all other ArchiveTimeStampChains contains a hash value of the concatenation of the data object hash and the hash value of all older ArchiveTimeStampChain. Verify that this Archive Timestamp was generated before the last Archive Timestamp of the ArchiveTimeStampChain became invalid.
In order to complete the non-repudiation proof for the data objects, the last Archive Timestamp has to be valid at the time the verification is performed.
If the proof is necessary for more than one data object, steps 1 and 3 have to be done for all these data objects. To prove the Archive Timestamp Sequence relates to a data object group, verify that each first Archive Timestamp of the first ArchiveTimeStampChain of the ArchiveTimeStampSequence of each data object does not contain other hash values in its first hash value list (than the hash values of the other data objects).
6. Encryption
If service providers are used to archive data and generate Archive Timestamps, it might be desirable or required that clients only send encrypted data to be archived. However, this means that evidence records refer to encrypted data objects. ERS directly protects the integrity of the bit-stream and this freezes the bit structure at the time of archiving. This precludes changing of the encryption scheme during the archival period, e.g., if the encryption scheme is no longer secure, a change is not possible without losing the integrity proof of the EvidenceRecord. In such cases, the services of a data transformation (and by this also possible re-encryption) done by a notary service might be a possible solution. To avoid problems when using the evidence records in the future, additional special precautions have to be taken:
Gondrom, et al. Standards Track [Page 20]
RFC 4998 ERS August 2007
o Evidence generated to prove the existence of encrypted data cannot always be relied upon to prove the existence of unencrypted data. It may be possible to choose an algorithm or a key for decryption that is not the algorithm or key used for encryption. In this case, the evidence record would not be a non-repudiation proof for the unencrypted data. Therefore, only encryption methods should be used that make it possible to prove that archive-timestamped encrypted data objects unambiguously represent unencrypted data objects. All data necessary to prove unambiguous representation should be included in the archived data objects. (Note: In addition, the long-term security of the encryption schemes should be analyzed to determine if it could be used to create collision attacks.)
o When a relying party uses an evidence record to prove the existence of encrypted data objects, it may be desirable for clients to only store the unencrypted data objects and to delete the encrypted copy. In order to use the evidence record, it must then be possible to unambiguously re-encrypt the unencrypted data to get exactly the data that was originally archived. Therefore, additional data necessary to re-encrypt data objects should be inserted into the evidence record by the client, i.e., the LTA never sees these values.
An extensible structure is defined to store the necessary parameters of the encryption methods. The use of the specified encryptionInfoType and encryptionInfoValue may be heavily dependent on the mechanisms and has to be defined in other specifications.
encryptionInfo contains information necessary for the unambiguous re-encryption of unencrypted content so that it exactly matches with the encrypted data objects protected by the EvidenceRecord.
Cryptographic algorithms and parameters that are used within Archive Timestamps must be secure at the time of generation. This concerns the hash algorithm used in the hash lists of Archive Timestamp as well as hash algorithms and public key algorithms of the timestamps. Publications regarding security suitability of cryptographic algorithms ([NIST.800-57-Part1.2006] and [ETSI-TS102176-1-2005]) have to be considered by verifying components. A generic solution for automatic interpretation of security suitability policies in electronic form is desirable but not the subject of this specification.
Redundancy
Retrospectively, certain parts of an Archive Timestamp may turn out to have lost their security suitability before this has been publicly known. For example, retrospectively, it may turn out that algorithms have lost their security suitability, and that even TSAs are untrustworthy. This can result in Archive Timestamps using those losing their probative force. Many TSAs are using the same signature algorithms. While the compromise of a private key will only affect the security of one specific TSA, the retrospective loss of security of a signature algorithm will have impact on a potentially large number of TSAs at once. To counter such risks, it is recommended to
Gondrom, et al. Standards Track [Page 22]
RFC 4998 ERS August 2007
generate and manage at least two redundant Evidence Records with ArchiveTimeStampSequences using different hash algorithms and different TSAs using different signature algorithms.
To best achieve and manage this redundancy, it is recommended to manage the Archive Timestamps in a central LTA.
Secure Timestamps
Archive Timestamping depends upon the security of normal time stamping. Security requirements for Time Stamping Authorities stated in security policies have to be met. Renewed Archive Timestamps should have the same or higher quality as the initial Archive Timestamp. Archive Timestamps used for signature renewal of signed data, should have the same or higher quality than the maximum quality of the signatures.
Secure Encryption
For non-repudiation proof, it does not matter whether encryption has been broken or not. Nevertheless, users should keep secret their private keys and randoms used for encryption and disclose them only if needed, e.g., in a lawsuit to a judge or expert. They should use encryption algorithms and parameters that are prospected to be unbreakable as long as confidentiality of the archived data is important.
[CCITT.X208.1988] International Telephone and Telegraph Consultative Committee, "Specification of Abstract Syntax Notation One (ASN.1)", CCITT Recommendation X.208, November 1988.
[CCITT.X209.1988] International Telephone and Telegraph Consultative Committee, "Specification of Basic Encoding Rules for Abstract Syntax Notation One (ASN.1)", CCITT Recommendation X.209, 1988.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC3161] Adams, C., Cain, P., Pinkas, D., and R. Zuccherato, "Internet X.509 Public Key Infrastructure Time-Stamp Protocol (TSP)", RFC 3161, August 2001.
Gondrom, et al. Standards Track [Page 23]
RFC 4998 ERS August 2007
[RFC3280] Housley, R., Polk, W., Ford, W., and D. Solo, "Internet X.509 Public Key Infrastructure Certificate and Certificate Revocation List (CRL) Profile", RFC 3280, April 2002.
[RFC3852] Housley, R., "Cryptographic Message Syntax (CMS)", RFC 3852, July 2004.
[ANSI.X9-95.2005] American National Standard for Financial Services, "Trusted Timestamp Management and Security", ANSI X9.95, June 2005.
[CCITT.X680.2002] International Telephone and Telegraph Consultative Committee, "Abstract Syntax Notation One (ASN.1): Specification of basic notation", CCITT Recommendation X.680, July 2002.
[CCITT.X690.2002] International Telephone and Telegraph Consultative Committee, "ASN.1 encoding rules: Specification of basic encoding Rules (BER), Canonical encoding rules (CER) and Distinguished encoding rules (DER)", CCITT Recommendation X.690, July 2002.
[ETSI-TS102176-1-2005] European Telecommunication Standards Institute (ETSI), Electronic Signatures and Infrastructures (ESI);, "Algorithms and Parameters for Secure Electronic Signatures; Part 1: Hash functions and asymmetric algorithms", ETSI TS 102 176-1 V1.2.1, July 2005.
[ISO-18014-1.2002] ISO/IEC JTC 1/SC 27, "Time stamping services - Part 1: Framework", ISO ISO-18014-1, February 2002.
[ISO-18014-2.2002] ISO/IEC JTC 1/SC 27, "Time stamping services - Part 2: Mechanisms producing independent tokens", ISO ISO-18014-2, December 2002.
[ISO-18014-3.2004] ISO/IEC JTC 1/SC 27, "Time stamping services - Part 3: Mechanisms producing linked tokens", ISO ISO-18014-3, February 2004.
Gondrom, et al. Standards Track [Page 24]
RFC 4998 ERS August 2007
[MER1980] Merkle, R., "Protocols for Public Key Cryptosystems, Proceedings of the 1980 IEEE Symposium on Security and Privacy (Oakland, CA, USA)", pages 122-134, April 1980.
[NIST.800-57-Part1.2006] National Institute of Standards and Technology, "Recommendation for Key Management - Part 1: General (Revised)", NIST 800-57 Part1, May 2006.
[RFC3126] Pinkas, D., Ross, J., and N. Pope, "Electronic Signature Formats for long term electronic signatures", RFC 3126, September 2001.
[RFC4810] Wallace, C., Pordesch, U., and R. Brandner, "Long-Term Archive Service Requirements", RFC 4810, March 2007.
An Evidence Record can be added to signed data or enveloped data in order to transfer them in a conclusive way. For CMS, a sensible place to store such an Evidence Record is an unsigned attribute (signed message) or an unprotected attribute (enveloped message).
One advantage of storing the Evidence Record within the CMS structure is that all data can be transferred in one conclusive file and is directly connected. The documents, the signatures, and their Evidence Records can be bundled and managed together. The description in the appendix contains the normative specification of how to integrate ERS in CMS structures.
The Evidence Record also contains information about the selection method that was used for the generation of the data objects to be timestamped. In the case of CMS, two selection methods can be distinguished:
1. The CMS Object as a whole including contentInfo is selected as data object and archive timestamped. This means that a hash value of the CMS object MUST be located in the first list of hash values of Archive Timestamps.
2. The CMS Object and the signed or encrypted content are included in the Archive Timestamp as separated objects. In this case, the hash value of the CMS Object as well as the hash value of the content have to be stored in the first list of hash values as a group of data objects.
However, other selection methods could also be applied, for instance, as in [RFC3126].
In the case of the two selection methods defined above, the Evidence Record has to be added to the first signature of the CMS Object of signed data. Depending on the selection method, the following Object Identifiers are defined for the Evidence Record:
The attributes SHOULD only occur once. If they appear several times, they have to be stored within the first signature in chronological order.
If the CMS object doesn't have the EvidenceRecord Attributes -- which indicates that the EvidenceRecord has been provided externally -- the archive timestamped data object has to be generated over the complete CMS object within the existing coding.
In case of verification, if only one EvidenceRecord is contained in the CMS object, the hash value must be generated over the CMS object without the one EvidenceRecord. This means that the attribute has to be removed before verification. The length of fields containing tags has to be adapted. Apart from that, the existing coding must not be modified.
If several Archive Timestamps occur, the data object has to be generated as follows:
o During verification of the first (in chronological order) EvidenceRecord, all EvidenceRecord have to be removed in order to generate the data object.
o During verification of the nth one EvidenceRecord, the first n-1 attributes should remain within the CMS object.
o The verification of the nth one EvidenceRecord must result in a point of time when the document must have existed with the first n attributes. The verification of the n+1th attribute must prove that this requirement has been met.
Ralf Brandner InterComponentWare AG Industriestra?e 41 Walldorf D-69119 Germany
EMail: ralf.brandner@intercomponentware.com
Ulrich Pordesch Fraunhofer Gesellschaft Rheinstra?e 75 Darmstadt D-64295 Germany
EMail: ulrich.pordesch@zv.fraunhofer.de
Gondrom, et al. Standards Track [Page 31]
RFC 4998 ERS August 2007
Full Copyright Statement
Copyright (C) The IETF Trust (2007).
This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights.
This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Intellectual Property
The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79.
Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr.
The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org.
Acknowledgement
Funding for the RFC Editor function is currently provided by the Internet Society.