The present disclosure generally relates to methods and systems for implementing digital signatures.
In many networks efficient use and testing of message authenticity is an important requirement to establish trust in the network. Often the network is comprised of many communicating (signing) nodes (devices/users) some of which are computationally-constrained such that the ability of those computationally-constrained nodes to sign and verify messages is greatly limited. The needs for efficient generation of signatures and for fast verification of the signed messages are critical to enhance message authenticity in the network.
Batch-verification of signed messages is useful, as a successful verification of a single signature ensures the authenticity of many messages. Conversely, a failed verification of a single message during batch-verification indicates that at least one message was ill-signed. In such a case, performing verification on each one of the batch verified messages enables pin-pointing the source of the rogue messages.
The present disclosure will be understood and appreciated more fully from the following detailed description, taken in conjunction with the drawings in which:
The present invention will be understood and appreciated more fully from the following detailed description, taken in conjunction with the appendix in which:
Appendix A is description of a standard Fiat-Shamir signature method; and
Appendix B is a proof by induction of security of the present method.
In one embodiment, a first signature template is received, the first signature template being one of a signature template of a first message or a null template, the first signature template comprising at least the following fields: an aggregation depth field, a message identifier, one of the first message or a result of applying a one way hash function to the first message, a bit vector, an aggregated square random integer mod N, a signature of the first message. A second signature template is created based on the first signature template, the second signature template created as follows: increment the aggregation depth of the first signature template, determine a unique message identifier for a second message, determine a second bit vector, determine an second aggregated square random integer mod N, and calculate a new signature for the second message. Related methods, apparatus, and systems are also disclosed.
Reference is now made to
System 10 is shown and described herein for the sake of example, to illustrate a typical configuration in which such digital signatures may be used, but is not meant to limit the application of such signatures to the configuration described herein.
In the pictured embodiment, a signing device, such as a signer 20 transmits data over a network 30 to a receiving device 40. Receiving device 40 may comprise a general-purpose computer or other computing device, such as, but not limited to, a hand held computing device, such as a tablet or smart phone.
In the example shown in
The signed message 50 which is signed may comprise a result of a one way function, such as a hash function (e.g., SHA-1, MD2, etc.), of the (unsigned) message 52. In cases where the message 52 is a large message comprising a large number of bits, the hash of the message 52 will typically be signed instead of the message 52. The distinction between a large number of bits and a small number of bits will be system dependent. For example, a video file, which may be considered a “large file”, may be a file several gigabytes in size, while an MP3 audio file, which may be considered a “small file”, may be a file less than 5 megabytes in size. Since the hash of the message is a unique value for the bit value of the message, the signature 54 will still be unique for the message 52. Accordingly, it is assumed below that the message, msg will be hashed by a hash function, h, so that MSG=h(msg), and MSG will be signed (the usage of “MSG”, “msg”, and “h(msg)” here is meant to be local, for this particular example, and without bearing on the description below).
A second processor 42 associated with the receiving device 40 receives signed message 50 via a second interface 44. The second processor 42, uses a public key 46, which is stored in a second memory 48 (used by the second processor 42) in order to verify the signature 54 of the message. Accordingly, the receiving device 40 may be referred to, herein below, as the “verifier”.
Typically, both the processor 22 and the second processor 42 comprise general-purpose computer processors, which may be programmed in software to carry out the functions that are described herein. The software may be downloaded to the either of the processors in electronic form, over a network, for example. Alternatively or additionally, the software may be provided on tangible, non-transitory storage media, such as optical, magnetic, or electronic memory media. Further alternatively or additionally, some or all of the described processing functions may be performed by special-purpose or programmable digital logic circuits.
As noted above,
An auditor 65, which may be thought of as a computerized system administrator, is also present on the network 30. The auditor 65 is a trusted third party which will receive a copy of the template 56 of all signed messages. That is to say that at the time when the signer 20 transmits the signed message 50 (comprising the message 52, signature 54, and the template 56) via an interface 28 over network 30 to the receiving device 40, the signer 20 also transmits the template 56 to the auditor. When the signer 20 needs a template in order to sign a message, such as message 52, the signer 20 may request a template from the auditor 65. The auditor 65 provides templates, such as template 56, to the signer 20. In other words, the auditor 65 both aggregates templates (e.g. template 56) used by signers (e.g. signer 20) and distributes templates (e.g. template 56) for use by signers (e.g. signer 20).
In some embodiments, the auditor 65 may be implemented locally on each of a plurality of signers 20. In such a case, each auditor 65 will receive templates, such as template 56, from the network and store, in an internal storage 68 on the signer 20, the templates being intended for use when the signer 20 needs a template 56 for signing a message 52. Additionally each auditor 65 may periodically perform full verification (as described below) on messages which exceed some threshold. For example, once a chain of signed messages exceeds ten messages in length (as discussed below, with reference to
A successful verification of only a subset of all messages signed according to embodiments described herein in a network provably assures, with a desired level of confidence, that no ill-signed messages have been injected into the network. Conversely, if one or more of the messages fails verification, the presence of rogue signatures indicates that the network is apparently compromised. The presence of signatures which fail verification may serve as a trigger to some corrective action based, in part, on the identification of where the rogue signature was introduced. Accordingly, if a signature received by the receiving device 40 may undergo full verification, as instructed by the auditor 65.
Further, a feature of the system 10 is that the (aggregated) signature of any message maintains its size; it does not grow with the number of signatures (of different signers and their respective messages) which are reflected in the signature.
For efficient processing, the processor 22 may comprise dedicated hardware logic circuits, in the form of an application-specific integrated circuit (ASIC), field programmable gate array (FPGA), or full-custom integrated circuit, or a combination of such devices. Alternatively or additionally, some or all of the functions of the processor 22 may be carried out by a programmable processor, such as a microprocessor or digital signal processor (DSP), under the control of suitable software. The software may be downloaded to the processor in electronic form, over a network, for example. Alternatively or additionally, the software may be stored on tangible storage media, such as optical, magnetic, or electronic memory media. The second processor 42 may similarly be implemented in either hardware or software or a combination of hardware and software components of the types described above.
Reference is now made to
A message, such as message 205B may be uniquely identified by its source (i.e. the index value, i, of the user where the message originated and was signed) and its time stamp (i.e. typically the time and date when a particular message was signed). It is also appreciated that the unique identification information for a particular message (e.g. 205A, 205B, 205C, and so forth) typically does not indicate the message's destination. It is also appreciated that, in principle, there is no limitation to the nature of the message, in that any information which can be expressed as a string of bits and bytes may be signed according to embodiments of the signature method described herein. It is further appreciated that the information concerning a previously signed message and its signer, used in signing a later message, typically will not provide the signer of the later message with useful information about the previously signed message. By way of example, information pertaining to previously signed message 205B is not available in any useful form to a signer of message 205C, because of the one-way nature of the hash function used in generating the signature, as well as the difficulty presented by the discrete logarithm problem, as will be apparent to a person skilled in the art, and based on the discussion below. It is also noted that a message and its signature might be in more than one chain of messages. By way of example, message 205B is involved in chains 215, 220, and 230.
Some chains of messages in
Reference is now additionally made to
The exemplary template 300 of
When there are no templates available, the signer uses an initial template (i.e. a “null template”) provided by the auditor 65 (
The null template has the same fields as the template 300 described above in Table 1. The values of the field in the null template are described below, in Table 2.
It will be appreciated that the values of AsRmuid and Sigmuid have both been set to 1 as a matter of convenience, as the ensuing calculations are thereby simplified, as will be seen below. In principle, any integer values may be used for AsRmuid and Sigmuid except 0.
As was noted above, a previous message which is used as the basis for signing a current message will have been signed with a previous template. As such, the current message will, accordingly be signed with a different template provided by the auditor 65 (
As noted above, the signer has a signed index value, i, indicating the signer's position in the chain of signers, as discussed above with reference to
If the Template Validity test is failed, then the signer i may report the failure to the auditor 65 (
Table 3, below, provides a list of terminology used herein.
IDX(i,j) = 0; j = (1, . . . , K)
IDX(i,j) = T(j); j = (1, . . . , K)
If the null template is not used, then a new template CAST (Current Aggregated Signature Template) is created by the signer 20 (
AgD=DAgD+1
muid=a combination of the signer identifier, i, and the current message identifier, as explained above in Table 1.
Signer i, (SNRi) selects a random integer r Mod N, i.e., 0<r<N, and computes the following parameters of CAST:
AsRmuid=(DAsRmuid·r2)mod N
Ēmuid=M(
T=HASH(AsRmuid∥MSGmuid∥Z(
Sigmuid=(r·DSIGmuid·CF mod N) where CF, a correction factor is defined as follows:
CF=Πj=1LCFj where:
CFj=1 if Ējmuid=
CFj=Sj mod N if Ējmuid=1 &
CFj=Sj−1 mod N if Ējmuid=0 &
The signature template, CAST for message of SNRi has now been generated.
It is appreciated that S, and Sj are a set of L secret values randomly selected from a field Z*N, Sj being the jth element in set S. Accordingly, set S is the set from which the private key, comprising the plurality of selected elements Sj, is selected.
Rationale for Correction Factor, CF
When viewing the vector Ēmuid as part of message MSGmuid the similarity to Fiat-Shamir Signature (see Appendix A) is apparent to those of skill in the art. (The message may be considered extended to include some previous-modified Ēmuid).
The correction factor, CF is derived in order to satisfy the verification Ēmuid=Hash(Sig2·VĒ
The current signer of the message uses different signer template data than previous signers.
Thus, the previous signer has a vector E′, which corresponds to the vector Ēdmuid used by the current message such that E′=(e′1, . . . , e′j, e′L) and R′, Sig′, such that, R′2=Sig′2Πj=1LVje′j
On the other hand, the current signer has: E=(e1, . . . , ej, . . . eL), R(=r R′), Sigmuid (also referred to as “Sig”), such that, R2=(r R′)2=Sig2Πj=1LVjej
Since, by definition, Sigmuid=r·Πj=1LCFj·Sig′:
R2=r2R′2=r2·Πj=1LCFj2·Sig′2·Πj=1LVjej should be satisfied.
Let ej=e′j+êj and substitute above; then:
Hence if ej=e′j then êj=0 and CFj=1
Before signer i (SNRi) uses DAST (generated by signer n) in signing the signer's message and generating a CAST, the signer or the signer's trusted agent performs a Template Validity Test (TVT) on DAST as a precondition for DAST's use. The TVT is as follows:
Let the signer of the DAST be SNRn (with a corresponding dmuid field in DAST).
Calculate a K bit vector U=Hash(DAsRdmuid∥DMSGdmuid∥Z(
Test that U(j)=
If U(j)=
A trusted third party (TTP), such as the auditor 65 (
Successful TVT is a sufficient condition for the security proof (by induction, refer to Appendix B) that successful verification of the last message in the batch ensures that all previous messages in the chain are properly signed.
Full Verification
As described above, the verifier receives the following from SNRi: the signature field of the message, Sigmuid; the aggregated square random of the message (an integer Mod(n), as noted elsewhere), AsRmuid; the message itself (or a hash thereof), MSGmuid; and the bit vector associated with the message, Ēmuid.
Full verification for the message of SNRi comprises two steps:
STEP 1: Calculate X=Sigmuid2Πj=1LVjĒ
STEP 2: As in Fiat-Shamir signature verification, calculate a K-bit vector SNRi=hash(AsRmuid∥MSGmuid∥Z(Ēmuid, i)) and test if SNRj=Ēmuid(IDX(i,j)), for all j=1,2, . . . , K.
If both above tests, step 1 and step 2 are true, then:
The following process is used for identification of a rogue signature and its origin, i.e., the point at which it was introduced.
If full verification fails, a verification test is repeated for each previous message in the chain until the verification test for a particular message passes the verification test. The point of the first instance of rogue signature injection is now known; i.e., the earliest incidence in the chain which fails the verification test is the first instance of a rogue signature. In some embodiments, the auditor 65 of
Reference is now made to
The fields in the first signature template or the null template are described above with reference to Tables 1 and 2.
At step 420, the processor 22 creates a second signature template based on the first signature template, the processor determining the following fields in the second signature template based on the first signature template:
Reference is now made to
At step 520 the second processor 42 calculates X=Sigmuid2·Πj=1LVjĒ
Finally, at step 550, the processor tests if SNRj=Ēmuid(IDX(i,j)), for all j=1,2, . . . , K. If:
AsRmuid=X;
SNRi=hash(AsRmuid∥MSGmuid∥Z(Ēmuid, i)); and
SNRj=Ēmuid(IDX(i,j)), for all j=1,2, . . . , K,
then all of the previous messages which are reflected in the signature (as discussed above, with reference to
It is appreciated that software components of the present invention may, if desired, be implemented in ROM (read only memory) form. The software components may, generally, be implemented in hardware, if desired, using conventional techniques. It is further appreciated that the software components may be instantiated, for example: as a computer program product or on a tangible medium. In some cases, it may be possible to instantiate the software components as a signal interpretable by an appropriate computer, although such an instantiation may be excluded in certain embodiments of the present invention.
It is appreciated that various features of the invention which are, for clarity, described in the contexts of separate embodiments may also be provided in combination in a single embodiment. Conversely, various features of the invention which are, for brevity, described in the context of a single embodiment may also be provided separately or in any suitable subcombination.
It will be appreciated by persons skilled in the art that the present invention is not limited by what has been particularly shown and described hereinabove. Rather the scope of the invention is defined by the appended claims and equivalents thereof:
The Signer has signing secret set of mod N values {Si} (i=1, . . . , 64) (e.g., N˜2 kbits). The Signer's public set of Vi's satisfies Si2·Vi=1 mod N.
For message M the Signer picks a random number R mod N, and computes: a K-bit (e.g., 64 bits) vector, E=(e1, e2, . . . e64)=Hash(R2∥M) and Sig=R·Πi=164Sie
The Signer sends [(e1, e2, . . . e64), Sig, M] to a Verifier.
The Verifier uses public {Vi}'s and verifies that received (e1, e2, . . . e64)=Hash(Sig2·Πi=164Vie
The security provided by the standard Fiat Shamir signature scheme relies on finding {S} (i.e., the hardness of factoring N), and robustness of the hash being a one way function.
The signature Sig on a message M originated by user U together with bit vector E satisfies:
E(restricted to indexes of U)=hash(Sig2·VE∥M∥zero(E on indexes of U))
Where V is the vector of inverses of the squares of the secrets.
The following is a proof by induction of security i.e., that passing full verification ensures that all other messages in the chain have been validly signed.
Given that ViS2i=1 modN, the last relation can be rewritten as:
Sig2=Sig2·r2·[(V−1|U)](Ê|U) (3)
Substituting (1) in (2) yields:
R2·r2=Sig2·VE, and substituting (3) in (4) yields: (4)
R2·r2=Sig2·r2·[(V−1|U)](Ê|U)·VE (5)
Dividing both sides by r2 gives:
R2=Sig2·[(V−1|U)](Ê|U)·VE., (6)
It can now be shown that by definition of Ê the term [(V−1|U)](Ê|U)·VE=VE. For j index of U: Êj=1 if Ej=0 and Ej=1 i.e. Vj−1·Vj=1=VjEj, and, similarly, for j index of U: Êj=−1 if Ej=1 and Ej=0, i.e., (Vj−1)−1·1j=Vj=VjEj. Further, for index j of U: Êj=0 if Ej=Ej resulting in (Vj−1)0=1 and VjEi_=VjEj. Thus (6) can be rewriten as R2=Sig2·VE
Knowing that user U conducted the quick validation test and verified that E|U=hash (R2∥M∥zero(E on indexes of U), R2 can be replaced by the value Sig2·VE and accordingly: E|U=hash (Sig2·VE∥M∥zero(E on indexes of U) which is, in fact, the required (full) verification of the message M.
Therefore, it is proven by induction that all the aggregated messages in the chain are correctly signed.
Number | Date | Country | Kind |
---|---|---|---|
248237 | Oct 2016 | IL | national |
Number | Name | Date | Kind |
---|---|---|---|
6292897 | Gennaro et al. | Sep 2001 | B1 |
6311271 | Gennaro et al. | Oct 2001 | B1 |
7340602 | Serret-Avila | Mar 2008 | B2 |
7574605 | Tanimoto et al. | Aug 2009 | B2 |
7890763 | Law | Feb 2011 | B1 |
7979706 | Case | Jul 2011 | B1 |
8078877 | Cheon | Dec 2011 | B2 |
8255691 | Yi et al. | Aug 2012 | B2 |
8285996 | McCullagh et al. | Oct 2012 | B2 |
20050050332 | Serret-Avila et al. | Mar 2005 | A1 |
20070028114 | McCullagh | Feb 2007 | A1 |
20080150702 | Neill | Jun 2008 | A1 |
20090327735 | Feng | Dec 2009 | A1 |
Number | Date | Country |
---|---|---|
5183401 | Apr 2013 | JP |
1999040702 | Aug 1999 | WO |
Entry |
---|
Bernhard, David et al.; How Not to Prove Yourself:Pitfalls of the Fiat-Shamir Heuristic and Applications to Helios (2012). |
Boneh, Dan et al.; Aggregate and Verifiably Encrypted Signatures from Bilinear Maps (2003). |
Camenisch, Jan et al.; Batch Verification of Short Signatures (2007). |
Cheon, Jung Hee et al.; A New ID-Based Signature with Batch Verification (2004), eprint.lacr. |
Ferrara, Anna Lisa et al.; Practical Short Signature Batch Verification (2009), Springer. |
Gentry, Craig et al.; Identity-Based Aggregate Signatures (2006), Springer. |
Goldwasser, Shafi: On the (IN)Security of the Fiat-Shamir Paradigm (2003), Proceedings of the 44th Annual IEEE Symposium on Foundations of Computer Science. |
Struik, René; Speed-Ups of Elliptic Curve-Based Schemes, IETF-78, Maastricht, The Netherlands, Jul. 25-30, 2010. |
Kim, Kitae et al.; Batch Verification and Finding Invalid Signatures in a Group Signature Scheme, In International Journal of Network Security, vol. 13, No. 2, pp. 61-70, Sep. 2011. |
Bellare M ., Garay J .A., Rabin T. (1998) Fast batch verification for modular exponentiation and digital signatures. In: Nyberg K. (eds) Advances in Cryptology—Eu Rocrypt'98. Eu Rocrypt 1998. Lecture Notes in Computer Science, vol. 1403. Springer, Berlin, Heidelberg (Year: 1998). |
Lim C.H., Lee P .J. (1995) Server(ProverlSigner)-Aided Verification of Identity Proofs and Signatures. In: Guillou L.C., Quisquater JJ . (eds) Advances in Cryptology—Eu Rocrypt '95. Eu Rocrypt 1995. Lecture Notes in Computer Science, vol. 921. Springer, Berlin, Heidelberg (Year: 1995). |
Yoon H., Cheon J.H., Kim Y. (2005) Batch Verifications with ID-Based Signatures. In: Park C., Chee S. (eds) Information Security and Cryptology—ICISC 2004. ICISC 2004. Lecture Notes in Computer Science, vol. 3506. Springer, Berlin, Heidelberg (Year: 2005). |
Bellare, M., Namprempre, C. & Neven, G. J Cryptol (2009) Security Proofs for Identity-Based Identification and Signature Schemes Journal of Cryptology Jan. 2009, vol. 22, Issue 1, pp. 1-61 (Year: 2009). |
Gennaro, Rosario et al.; How to Sign Digital Streams; revised version Feb. 24, 1998. |
Golle, Phillippe et al.; Authenticating Streamed Data in the Presence of Random Packet Loss; In NDSS, vol. 1, pp. 13-22, 2001. |
Perrig, A. et al.; Timed Efficient Stream Loss-Tolerant Authentication (TESLA): Multicast Source Authentication Transform Introduction; TESLA RFC4082; The Internet Society, Jun. 2005. |
Sridevi, J. et al.; Efficient Multicast Packet Authentication Using Digital Signature; Proceedings published by International Journal of Computer Applications (IJCA) International Conference on Emerging Technology Trends (ICETT) 2011. |
O. Dagdelen, et al., “The Fiat-Shamir Transformation in a Quantum World”, In: Sako K., Sarkar P. (eds) Advances in Cryptology—ASIACRYPT 2013. ASIACRYPT 2013. Lecture Notes in Computer Science, vol. 8270. Springer, Berlin, Heidelberg, 30 pages. |
D. Bernhard, et al., “How not to Prove Yourself: Pitfalls of the Fiat-Shamir Heuristic and Applications to Helios”, In: Wang X., Sako K. (eds) Advances in Cryptology—ASIACRYPT 2012. ASIACRYPT 2012. Lecture Notes in Computer Science, vol. 7658. Springer, Berlin, Heidelberg, 26 pages. |
Abstract of “On the Insecurity of the Fiat-Shamir Signatures with Iterative Hash Functions”, In: Pieprzyk J., Zhang F. (eds) Provable Security. ProvSec 2009. Lecture Notes in Computer Science, vol. 5848. Springer, Berlin, Heidelberg, 9 pages. |
Number | Date | Country | |
---|---|---|---|
20180102903 A1 | Apr 2018 | US |