Patent Application
20170104598
The present invention generally relates to a method of providing a hash value for a piece of data, where the hash value provides for a time-stamp for the piece of data upon verification. The invention also relates to a method of deriving a time stamp for a piece of data. The method further relates to a server of a signing authority for issuing a time stamp signature, to an electronic device arranged to derive a one-time signing key and providing a hash value for a piece of data, to a network node and to computer programs for implementing the methods.
“Big data”, cloud, and the Internet of Things (IoT) are examples of the rapidly expanding area of distributed data networks and acquisition of distributed data. Data generated at a plurality of source nodes is collected for processing and/or analysis. An example of the source nodes includes sensor networks that perform measurements and provide measurement data, e.g., in home automation data networks or industrial processing data networks. A further example includes servers in a data center generating event log records, e.g. for operational security.
The operation of data networks, such as above examples, relies upon the integrity of the data received from the distributed data sources and the control processes. This means that as data is collected, it has to be possible to verify that the data has not been tampered with since the data left the source node. Furthermore, the data source has to be authentic. This means that an indicated source, e.g., a source node indicated by the received data or a data packet including the data, is the actual originator of the data.
Depending on operational security requirements, it is not sufficient that only the intended recipient collecting the data can verify aspects of integrity and authenticity. Rather, it is required that third parties can audit the data exchange between the source nodes and the collecting node. Conventional techniques for authenticating the data source implement public-key cryptography, e.g., using a Public Key Infrastructure (PKI) with signatures on all data exchanged between the nodes.
However, generating signatures is resource consuming in minimalistic source nodes (also referred to as “low-end devices”) as sensors. Furthermore, the impact of signatures on bandwidth and/or storage is disproportionally large compared to the data to be exchanged (e.g., since the nodes have to be prepared for an audit, a large number of signatures have to be stored for relatively long time periods in the nodes). Moreover, signatures verifiable by a PKI are known to be cumbersome to establish and maintain over time, especially if many sources of data have to be distinguished, i.e., identified by means of different certificates.
Other conventional techniques, e.g. below referred to as QI-KSI, implement Merkle trees. Aggregating hash values of the exchanged data in a Merkle tree is efficient, since the “root” of the Merkle tree provides a compressed digest of all individual hash values, so that the Merkle tree reduces storage requirements. However considerable effort is needed to arrange for the keys in each leaf of the tree to be used for authentication.
Ahto Buldas, Andres Kroonmaa and Risto Laanoja have disclosed some principles in “Keyless Signatures' Infrastructure: How to Build Global Distributed Hash-Trees”, below referred to as [1], in “Efficient Quantum-Immune Keyless Signatures with Identity”, below referred to as [2], in “Efficient implementation of Keyless Signatures with Hash Sequence Authentication”, below referred to as [3], and in “Security Proofs for the BLT Signature Scheme”, below referred to as [4]. Ahto Buldas and Sven Laur have disclosed some principles in “Knowledge-Binding Commitments with Applications in Time-Stamping”, below referred to as [5].
Every time the client wants to authenticate himself, a value zk needs to be recomputed from zn, as will be further described in this disclosure. This may be a problem if n is large and there is no capacity to store or re-compute the whole hash chain. The solution to this problem is the technique called “hash sequence traversal”. One such technique was proposed by D. Coppersmith and M. Jakobsson in their paper [6]. In order to derive zk faster than just sequential hashing from zn to zk, the reversed order of hash chain z0←z1← . . . →zk→ . . . can be derived in average log(n) time if one could keep log(n) of intermediate hash values of the hash sequence.
A short description of the M. Jakobsson and D. Coppersmith technique on the intuitive level can be given as follows. Assume the client can keep the value zn/2, then the derivation of any value zk would require at most n/2 hashes, instead of n. Now let us assume that the client keeps two intermediate values zn/2 and zn1/4. Thus, the elements of the first half of the hash chain zk, for k≦n/2, would require re-computation of at most n/4 hashes. When k becomes larger than n/2, the intermediate value zn1/4 can be removed and a new value zn3/4 is derived linearly in time n1/4 hash operations, so that the elements of the second half of the hash chain zk, for k>n/2, can be calculated in at most n/4 hashes as well. It has been shown that having log(n) intermediate hash values, the total time to derive the reverse-order hash chain is log(n), in average.
It is important that a signing returned by a signing authority corresponds to a one-time signing key, OTSK, zi that has been used by the client. QI-KSI proposes the idea that the client actually needs to use 3 (or more) keys zi, zi+1 and zi+2 and send a parallel request to TTP-SA for time-stamping. The signer will get 3 time stamps S's, but the stamp's time will correspond to only one of zi . . . zi+2. This approach means that the signer can produce one signature per 3 time slots.
Consider that a KSI RH (root hash) is computed for each interval. Several intervals may be needed for a piece of data, as discussed above. However, in a use case when the client needs to sign a stream of data, this might be a performance bottleneck. It is therefore a desire to provide an efficient approach, particularly for streams of data.
The invention is based on the understanding that application of a time fraction hash tree splitting time slots into fractions corresponding to the leafs of the time fraction hash tree enables a corresponding increased amount of pieces of data to be signed.
According to a first aspect, there is provided a method of a client for providing a hash value for a piece of data, where the hash value provides for a time-stamp for the piece of data upon verification. The method comprises collecting the piece of data and deriving a one-time signing key, OTSK, of a OTSK hash chain by applying a time fraction hash tree splitting a time slot corresponding to an index into time fractions such that the time slot is divided into fractions according to the number of leafs of the time fraction hash tree, forming a signing request for the piece of data by applying the OTSK for the fraction for the respective piece of data to calculate hash values of the piece of data, and transmitting the signing request comprising the hash values to a server for deriving a time stamp for the piece of data.
The time fraction hash tree may be splitting a time slot corresponding to an index into K time fractions such that the client creates K sets of OTSKs.
The method may comprise sending a signing request to a signing authority for a plurality of pieces of data, wherein each piece of data may be assigned a respective index consecutively by using OTSKs with time-forwarded OTSK indexes. The method may comprise, for the stream of pieces of data obtaining a time stamp for each piece of data, collecting the pieces of data and deriving OTSKs of a OTSK hash chain. The forming of signing requests may comprise forming a stream of signing requests for the pieces of data by applying the OTSKs with time-forwarded indexes and fractions for the respective piece of data to calculate hash values of the respective pieces of data. The signing of requests may comprise transmitting the stream of signing requests comprising the hash values to a server for deriving time stamps for the pieces of data, respectively.
The method may comprise deriving OTSKs of signer's OTSK hash chain by a one-way function of a secret key of the signer and a function of an index of the OTSK, and providing the hash value fur the piece of data by a hash function including the piece of data and the derived OTSK. The function of the index may be the index itself. The one-way function may be a cryptographic message authentication code function. The one-way function may be a hash function.
The method may comprise registering the provided hash value.
The method may comprise providing a one-time proof of knowledge of the one-time signing key to a signing authority entity without revealing the one-time signing key. The one-time proof of knowledge may include a hash path for the hash value, and the providing of the one-time proof of knowledge to the signing authority may include sending the index of the one-time signing key, the hash value and the hash path.
According to a second aspect, there is provided a method of a server of a signing authority for issuing a time stamp signature. The method comprises receiving a signing request for a piece of data formed by application of a one-time signing key, OTSK, for a fraction corresponding to an index for time fractions defined by a time fraction hash tree splitting a time slot for the respective piece of data to calculate hash values of the piece of data where the time slot is divided into fractions according to the number of leafs of the time fraction hash tree, deriving a time stamp for the piece of data including a hash path of the time fraction hash tree as a sub-tree of hash tree of the OTSK, and transmitting the time stamp for the piece of data.
The signing requests for a plurality of pieces of data, respectively, may be received for one time slot, wherein the maximum number of signing requests for a time slot may be the number of fractions defined by the time fraction hash tree.
The signing request may be checked for client's signature before the deriving of a time stamp. The checking of the signing request may include determining whether the client has provided a one-time proof of knowledge about a one-time signing key to a server without revealing the one-time signing key. The determining may comprise checking whether the hash corresponds to a one-time signing key for a root hash included in a public certificate associated with the identity, checking whether an index corresponding to the hash path from the one-time signing key to the root hash corresponds to a correct time slot, and determining it to be proven that the electronic device is in possession of the correct one-time signing key when the checks are fulfilled. The checking of the signing request may comprise accessing a certificate matching the identity and a root hash for the first hash, checking validity of the certificate, and verifying whether the hash path for the first hash is correct, wherein if the certificate is not valid or the hash path cannot be verified to be correct, the server omits further actions for that signing request.
The receiving of a signing request may be a part of a sequence of signing requests with time-forwarded OTSKs indexes such that a delay time before the actual time-stamping procedure is provided so that there is enough time for receiving the signing requests, preparing them and synchronize the time with the signature infrastructure.
The server may be a server of a Trusted Third Party Signing Authority.
The signature infrastructure may be a Keyless Signature Infrastructure.
The method may further comprise transmitting a time stamp to the client, wherein a signer is enabled to reveal a signature of the piece of data including an identity, an index of the one-tune signing key, the OTSK and the time stamp for enabling verification of the time stamp for the piece of data.
The receiving of the signing request may comprise receiving a message from an electronic device, the message including a hash, an identity associated with the client and a hash path of the hash.
The method may include verification of a time stamp for a piece of date by determining whether a hash of the message is a leaf of the time stamp hash tree, determining whether the aggregate hash path corresponds to the correct identifier of the server to the server of the signature infrastructure entity, and determining whether the aggregate hash path and calendar hash path correspond to a correct calendar root hash for a time corresponding to the index.
The signing request may comprise a hash value of the piece of data, wherein the hash value is formed using a time-forwarded one-time signing key, OTSK, wherein the time-forwarded OTSK comprises a time-forwarded index, and the deriving of the time stamp may comprise queuing the signing request, pushing the hash value to a signature infrastructure entity at the time-forwarded time, and receiving the time stamp from the signature infrastructure entity.
According to a third aspect, there is provided an electronic device comprising a. processor arranged to implement a functional module for deriving a one-time signing key and providing a hash value for a piece of data by a hash function including the piece of data and the derived one-time signing key, wherein the functional module is arranged to perform the method according to the first aspect.
The electronic device may be a wireless device.
According to a fourth aspect, there is provided a server comprising a processor arranged to implement a functional module for deriving a time stamp, wherein the functional module is arranged to perform the method according to the second aspect.
According to a fifth aspect, there is provided a network node comprising a server according to the fourth aspect.
According to a sixth aspect, there is provided a computer program comprising instructions which, when executed on a processor of an electronic device, causes the electronic device to perform the method according to the first aspect.
According to a seventh aspect, there is provided a computer program comprising instructions which, when executed on a processor of a server, causes the server to perform the method according to the second aspect.
The above, as well as additional objects, features and advantages of the present invention, will be better understood through the following illustrative and non-limiting detailed description of preferred embodiments of the present invention, with reference to the appended drawings.
Certain embodiments disclosed herein relate generally to the technical field of security, and more particularly to the field of hash functions. For the easier understanding of the contribution by the invention, an introduction to mechanisms for providing hash functions used for time stamping is demonstrated below. For further facilitating the reading of this disclosure, commonly used abbreviations are listed below.
Abbreviation Explanation
KSI stands for the “Keyless Signature Infrastructure. This section is based on open sources, such as publications of papers [1-5] mainly on the cryptographic e-print web-based database where the authors describe different aspects of the KSI. The term “keyless signature” that is used in references [1-5] could be seen as slightly misleading. Another term, in the field of cryptography, that could be used instead is a “time stamp” of a given hash value, from the client's perspective. For consistent terminology with earlier work, we will nevertheless stick with the term KSI in the text below.
Another notion that we will use is the Hash Path (HP), which is an ordered sequence of hash values that helps to derive the root hash, starting from a chosen leaf value.
Thus, the path can be written as the sequence of 4 hash values {h3;h2(L);h01(L);h4567(R)}, and having the sequence of “L-R” marks one can compute the root hash explicitly.
It is also worth to note that an LRS is 1-to-1 mapped to the index of the leaf, if the LRS is translated into the binary representation with the following rules: L-R marks closer to the root represent the most significant bits of the index, and closer to the leafs represent the least significant bits of the index; L is translated to the bit value 1 and R is translated to the bit value 0.
Indeed, in the example illustrated in
As a final comment we note that we actually can change the hash function as we move through the tree. In that case an identifier of the hash function used in each merging node has to be encoded into the path.
A basic drawing of the KSI architecture is shown in
The KSI system architecture includes several layers of aggregators, the functionality of each is basically to accept multiple hash values from the children connections, produce the local root hash, and push it to the upper layer for further aggregation. There are physically many aggregators on each layer that are distributed world-wide, but in the
A client or a device may push its own hash value using the entry point called the Gateway. The Core Network (CN) is the last “station” of accumulating hash values into the large Merkle tree, and CN thus computes the aggregation root hash (ARH).
Additionally, CN has an extra modified Merkle tree to add the time stamping to ARH at a given time. The Calendar tree is organized such a way that it includes both the current ARH and the history of all previous ARHs in the past. The result of the Calendar tree is the Calendar Root Hash (CRH).
As the result of aggregation, the client receives back the aggregation hash path (AHP) from the client's hash value to the aggregation root hash, and also the calendar hash path (CHP) from the ARH to the global time stamp hash value that is actually the CRH.
The Core Network beats at regular intervals, for example, say once per second, which means that the CRH and the calendar tree are updated every second. CN also keeps the history of the global time-stamps for each second slot—the combination of ARHs and all historical CRHs in the past.
This way, the client or anyone else can later verify that the combination of a saved aggregation hash path and the calendar hash path at a certain time t lead to the correct value. I.e., LRS of AHP could be served as a static identifier of the Gateway (if the connection of the Gateway to the KSI infrastructure is static), and LRS of CHP is used to verify the time when the hash paths were created.
The global time stamp value CRH can be periodically published in a newspaper so that the Core Network cannot itself to modify the global time stamp back in time.
For verification purposes, the CRH can be received either off-line, or on-line. For off-line use cases one could take the CRH through the printed publications (that may be periodically once per month). For on-line verification use cases, the CRH can be signed by a trusted entity (perhaps, including a certificate), and then it can be downloaded by clients and/or applications at any time for verification purposes.
In the general architecture of KSI the entry point for clients (and/or applications) is the Gateway (GW) that itself can be an aggregation layer, but additionally provides the client's front-end for various services based on KSI's time-stamping infrastructure. This way, the Gateway can be seen as a server-based service engine that should be located close to the customer, or actually be on the customer's side.
The list of possible services that the Gateway may assist with includes: time-stamping of a given hash value, assistance in signing a document, etc. All those services are not really a part of KSI, but some of them are a part of QI-KSI.
KSI returns the aggregation hash path and the calendar hash path. The AHP may be considered as the identifier of the Gateway, since the L-R sequence in the global Merkle tree determines the way the Gateway was connected to the KSI. However, this identifier may be valid if certain conditions are valid:
Later we will see how the identifier is used in the QI-KSI signing model.
QI-KSI stands for “Quantum-Immune KSI”. This section is mainly based on the papers [2] and [4] identified in the introductory part of this disclosure.
QI-KSI is an extension for KSI and provides two hash-based techniques. Hash-based cryptography is, as of today, believed to be quantum immune, so this is the reason for the name “quantum-immune”. QI-KSI proposes the technique for a hash-based authentication, and a hash-based digital signature architecture with the help of KSI's time-stamping service.
This is based on one-time passwords techniques. The client (and/or application) selects a random secret value zn (of size of the hash digest), and generates the hash chain (HC) z0←z1← . . . ←zn as follows:
z
i
=H(zi−1), for i=0, . . . , n−1, and H is a chosen hash function.
The value z0 is then shared with the server side (via some channel) to which the client is intended to authenticate himself.
At any given time, the server holds a value zk (in the beginning, the server holds the value z0). When the client wants to authenticate himself, he uses and sends to the server the value zk+1. The server verifies that H(zk+1)=zk and if the values coincide then the authentication is successful. In the latter case, the server throws away zk and holds zk+1, instead, for future authentications.
This way, one secret key zn can be served for n authentications.
In this scheme, every time the client wants to authenticate himself, the value zk needs to be recomputed from zn. This may be a problem if n is large and there is no capacity to store or re-compute the whole hash chain. The solution to this problem is the technique called “hash sequence traversal”. Such a technique was proposed in [6] by D. Coppersmith and M. Jakobsson. In order to derive zk faster than just sequential hashing from zn to zk, the reversed order of hash chain z0←z1← . . . ←zk← . . . can be derived in average log(n) time if one could keep log(n) of intermediate hash values of the hash sequence. A short description of the M. Jakobsson and D. Coppersmith technique on the intuitive level can be given as follows. Assume the client can keep the value zn/2, then the derivation of any value zk would require at most n/2 hashes, instead of n. Now let us assume that the client keeps two intermediate values zn/2 and zn1/4. Thus, the elements of the first half of the hash chain zk, for k≦n/2, would require re-computation of at most n/4 hashes. When k becomes larger than n/2, the intermediate value zn1/4 can be removed and a new value zn3/4 is derived linearly in time n1/4 hash operations, so that the elements of the second half of the hash chain zk, for k>n/2, can be calculated in at most n/4 hashes as well. It has been shown that having log(n) intermediate hash values, the total time to derive the reverse-order hash chain is log(n), in average.
The model of signing documents in QI-KSI is showed in
In QI-KSI, the signer first creates his own pair of the secret key and the public key. The signer chooses a random secret key zn. Then, the sequence of values (hash chain) z0← . . . ←zn is computed as
z
i
=H(zi+1) for i=0, . . . , n−1, and H is a chosen hash function.
The value z0 is the public key and a relevant certificate for the public key z0 is generated by an external certificate authority (CA). The value z0 is bound with the user's identity in the validity time t0 seconds from a well-known date (number of seconds from the date Jan. 1, 1970 and time 00:00:00) that determines the time after which the certificate is valid. To be more precise, a certificate should include at least the following fields: Cert:={Client's identity records; z0, t0, TTP-SA's identity and identifier}, where the TTP-SA's identifier can be the index of the leaf in the global KSI Merkle tree to which the TTP-SA is statically attached.
QI-KSI signing key looks similar to the hash chain in the QI-KSI authentication method, but the difference comes from the meaning of the index k in the notation zk. The values zk, k=1, . . . , n, are used as one-time signing keys (OTSK) each of which can be used only at a certain KSI's time frame that is exactly (t0+k)th second from the global starting time (recall, the Core Network “beats” and produces global time stamp values with the speed once per second).
QI-KSI signing protocol, see e.g. paper [2], is shown in
The verifier may check that zi is actually the i:th pre-image of z0, and the time stamp corresponds to t0+i.
The problem comes for the verifier who wants to check the signature, as he only knows zi but needs to verify it against the public key z0. The verifier does not have those intermediate hash values and thus has to perform i hashes starting from zi and going all the way to z0.
In order to reduce the time and/or complexity of the verification process, the following modification to OTSKs was proposed. This includes building yet another modified Merkle tree on top of OTSKs as shown in
The client's certificate additionally includes the root hash r, and the signature additionally includes the hash path from the OTSK zi to the root r. Thus, the verifier only needs to check that zi participated in the hash root generation, and the L-R sequence of the hash path is translated to the correct index i.
For efficiency reasons, the signer needs to keep the relevant part of the hash-tree, and in the signing process the part of the hash tree that leads to the root hash r may be partly recomputed in log(n) time with log(n) intermediate hash values. That computation also requires the knowledge of other values zi in an efficient way, but then this part may be done with a hash chain traversal technique as demonstrated in U.S. provisional patent application No. 62/118,808.
Synchronization with KSI
It is important that Si returned by TTP-SA corresponds to the OTSK zi that has been used by the client. QI-KSI proposes the idea that the client actually needs to use 3 (or more) keys zi, zi+1 and zi+2 and send a parallel request to TTP-SA for time-stamping. The signer will get 3 time stamps S's, but the stamp's time will correspond to only one of zi . . . zi+2.
All this also means that the signer can produce one signature per 3 time slots, i.e., one signature per 3 seconds. However, as will be demonstrated below under the headline ‘Further Options’, efficiency may be improved also in this sense.
The gist of this disclosure will now be presented, followed by some optional features, and thereafter further disclosure about methods and their implementations. It is readily understood from this disclosure that any combination with the demonstrated features of the KSI concept are applicable.
Consider that KSI RH (root hash) is computed for each interval. Further, assume the intervals to be 1 second (but of course other interval settings are possible). As it was also mentioned in the section discussing synchronization with KSI above, the synchronization that is proposed in QI-KSI makes it possible to the client to make 1 signing per 3 seconds (3 KSI's time slots). However, in a use case when the client needs to sign a stream of data, this might be a performance bottleneck.
The solution demonstrated above has mainly been discussed in view of a general case where an external service (KSI's time stamping, TTP SA service, other external modules that can, but not necessarily, be hash-based) is available once per a time slot (like in KSI, the service is available once per second). However, in the solution demonstrated below, the technique is demonstrated based on the example of the QI-KSI signing scheme for the sake of easier understanding for the reader.
By extending the global KSI tree with a time fraction Merkle sub-tree on the TTP-SA node it is possible for one or more clients/gateways to perform signing for a stream of data items with the average speed faster than the speed of producing time stamps with KSI, i.e. faster than demonstrated with reference to the disclosure of sending sequence of signing requests to TTP-SA using OTSKs with time-forwarded OTSKs indexes above. It is to be noted that the approach demonstrated below may be combined with time-forwarded OTSKs indexes as demonstrated above for efficient provision of OTSK keys, e.g. for performing signatures for a stream of data.
The approach makes it possible for the client to have only one hash chain of OTSKs, where each one-time signing key corresponds to its own fraction of a second. That sub-tree will serve as a KSI's time slot splitter of a time slot into smaller time fractions.
OTSKs of clients thus may be modified in such a way that now the client may create K sets of OTSKs. One set corresponds to and is used within one certain fraction of a second.
For example, if the TTP-SA's time-splitting sub-tree has 4 leafs, then TTP-SA is capable to perform 4 signatures per second, instead of 1 as of the original QI-KSI. The client generates 4 hash chains from 4 secret keys zn(0) . . . zn(3) and uses the OTSKs produced by zn(k) in the signing times t=t0+i+k/4 seconds, for i=1 . . . n, k=0 . . . 3 (for the example of K=4; K may be selected arbitrarily to achieve desired granularity).
For each fraction of a second the TTP-SA returns, as the result of signing request, the same Si, but adds the hash path of the TTP-SA's time fraction sub-tree that is also included by the signer into the final signature.
The verifier may apply:
Note that AHP and CHP parts of Si will be the same for all K signing requests within the same time slot. However, the HP of the time fraction tree will be different. Also note that since TTP-SA enters K hash values to the leafs of the time-fraction tree's HP, the signer or anyone else cannot ignore that HP since then the signature becomes invalid and non-verifiable.
For synchronization purposes, the signer may still use the idea from the QI-KSI design demonstrated above, where for every fraction slot k=0, . . . , K−1, the signer uses 3 consecutive OTSKs zi(k), zi+1(k), zi+2(k). In this case, the average signing speed is 1 signature per 3/K seconds.
A device, application or session of a client may thus be arranged to sign for a stream of data items by a time fraction tree splitting a time slot of a time stamping infrastructure into time fractions.
A server operating a signing authority function may be arranged to receive multiple signing requests of a time slot of a time stamping infrastructure by a time fraction tree splitting the time slot of the time stamping infrastructure into time fractions.
The approach of the fraction tree may be combined with the approach of delayed requests, which will be further demonstrated below.
The TTP-SA server collects signing requests, verifies the client's certificate, and allocates the prepared hashes for the delayed collective time-stamping (with KSI) into a sorted (by time) queue. TTP-SA will push the top of the time fraction sub-tree to KSI infrastructure for time-stamping at a proper time. The hash root of the fraction tree can also be prepared by TTP-SA during the given delay Δ. Thus, the queue of TTP-SA may be just a queue of hash values waiting for their turns to be time stamped.
This way, TTP-SA has some time delay during which it can check the request itself (and the certificate), prepare the time fraction sub tree, calculate the root hash of that sub tree, and wait for the correct KSI's time to occur, in order to push the pre-computed hash into KSI for time stamping.
Another solution could be that the possibility for a signer to add a signing request at a fraction of a KSI's time slot t closes shortly before the time t, such that TTP-SA would still have enough time to prepare the root hash of the time fraction tree before its root hash is to be pushed for KSI time-stamping at time t.
The server returns Sis together with the hash paths of its own sub-tree when the signing job for the group of requests is done, and continues to proceed with the next group of signing requests, taken from the local queue, checked and prepared in advance for the next second of the KSI's time slot.
Thus, the client can publish the received stream of signatures with the time delay Δ, and the average speed 1 signature per 1/K second utilizing OTSKs efficiently.
A time stamp service provider applying a hash tree signing system of a hash-tree time-stamping part and a trusted signing authority (TTP-SA) may operate on given intervals for the TTP-SA, and there may be provided a time fraction sub-tree for splitting each interval into K fractions. The signer may create only one OTSK hash chain, where zk corresponds to the time (t0+(k div K))th time interval, e.g. second, and (k mod K)th fraction of the interval.
A device, application or session of a client may thus be arranged to sign for a stream of data items by a time fraction tree splitting a time slot of a time stamping infrastructure into time fractions, and be arranged to transmit signing requests to a signing authority with time-forwarded OTSKs indexes.
A server operating a signing authority function may be arranged to receive multiple signing requests of a time slot of a time stamping infrastructure by a time fraction tree splitting the time slot of the time stamping infrastructure into time fractions, wherein the signing requests comprises time-forwarded OTSKs indexes, and the server is arranged to store them in a queue, and when the KSI time becomes aligned with the time of a request in the queue to push the corresponding hash value to a KSI infrastructure and receive the right time stamping. Any calculations may be pre-calculated for the requests of the queue.
A server operating a signing authority function may be arranged to receive signing requests to a signing authority with time-forwarded OTSKs indexes, store them in a queue, and when the KSI time becomes aligned with the time of a request in the queue to push the corresponding hash value to a KSI infrastructure and receive the right time stamping. Any calculations may be pre-calculated for the requests of the queue.
The generation of further OTSKs, as of the optional approach demonstrated above, may benefits from deriving OTSKs directly by a one-way function, which will be further demonstrated below.
The approach of deriving the OTSKs directly by a one-way function has been shown above to provide an efficient way to provide the OTSK and thus hash value for a piece of data for use in the KSI context as previously known. Below, the applicability for further developments of the KSI approach and novel features thereof will be demonstrated.
We propose that instead of the hash chain z0← . . . ←zn, as discussed above, where zn is the secret key and z0 is the public key, the signer derives any of the OTSKs directly via a one-way function, given that all zis are bound by the root of a Merkle tree on top of the OTSKs sequence {zi}.
Instead of the hash chain z0← . . . ←zn, the signer may derive any of the OTSKs as follows:
z
i
=H(zsk;fi), for i=1, . . . , n, where H is a one-way function,
where zsk is a secret key of the signer, and fi is a function on the index i that generates different values for each i=1, . . . , n. For example, this function can be as simple as fi=i, but it may be a more complex one. As an alternative, the zi:s may be generated as zi=HMAC(zsk;fi). The new scheme is shown in
The signer's certificate then does not need to have z0, but it may comprise: the user's identity records, the root hash value r that combines all OTSKs, the value n that indicates the expiration time for the secret key zsk as well as determines the height of the tree, and the validation time t0, after which the certificate is valid. Since the LRS of the hash path encodes the index of zi uniquely, then this is the way to verify that zi is actually the OTSK that corresponds to the time t0+i.
The root hash thus still can prove that the hash path of OTSK zi was originated from the same secret source. The LRS of the hash path from zi to r determines the time slot when the OTSK zi can be used and be verified against the returned time-stamp.
The use of zsk makes it faster and/or requires less processing/memory resources for the signer to derive the values zi for any i at any time without having to keep O(log(n)) intermediate hash values and spend O(log(n)) of time to derive zi, or calculate all hash values in the hash chain from zn, zn−1, . . . , etc.
Since in this modification the certificate now includes the expiration time (that is equal to t0+n), and the TTP-SA also checks for the validity of the signer's certificate, then this scheme will be as secure as the one proposed in QI-KSI above, and it is more efficient.
After the expiration of the key usage time, the secret key zsk can be thrown away, whereas all created signatures remain valid in time and verifiable.
By this, the signing process is faster and/or requires less storage resources (Cf. M. Jakobsson and D. Coppersmith technique) for the intermediate state, or requires less processing resources (Cf. traditional approach without M. Jacobsson and D. Coppersmith technique).
The above mentioned approach is also applicable for deriving other OTSK values, as will be described below.
Thus, for the QI-KSI signal model, there is provided a method to process for a hash tree infrastructure at predetermined intervals presented derived values of data to obtain a root hash value referred to as time-stamps that may be published such that the presented hash values to be processed depend on previously published root hash values, and a processes to compute and deliver a signature for the data after checking by a trusted signing authority wherein the derived values are the hash or mac of the data using an one-time signing key that is computed by a message authentication function. As demonstrated above, the one-time signing key may be computed as zi=H(zsk;fi), for i=1, . . . , n, where H is a one-way function, where zsk is a secret key of the signer, and fi is a function on the index i that generates different values for each i=1, . . . , n. The function may be fi=i.
A problem may be that TTP-SA does not know the content of the value x—one part of the signing request, in the way that x (the value that the client derives and sends to TTP-SA; see section “QI-KSI Signing Protocol” above) can be anything and be generated by anyone.
While issuing a signature, TTP-SA also does not know if the signing request comes from a legitimate user or from someone else. Hence, the TTP-SA may start to work and later when almost everything is ready find out that the user was not legitimate.
It may thus be desirable to have a solution where the TTP-SA could verify that it is the authorized client who sends the signing request on his name, which would build a better trust in the signature.
By using the hash path of a hash image of the actual OTSK as the proof that the user actually knows the not yet public secret OTSK, the TTP-SA may be better protected from doing work for non-legitimate requestors. The signer's signature may be shorter, and verification faster. The user sends this proof to the TTP-SA along with the signing request, so that the signing authority can verify the legitimacy of the signer before actually producing a signature fingerprint.
The way to generate OTSKs by a user and the usage of such a proof-of-knowledge is not limited by the described use case, and the approach demonstrated here may be combined with any of the other approaches demonstrated in this disclosure. However, the benefit of deriving the OTSK by a one-way function as demonstrated above is evident when used together with this approach of providing one-time proof of knowledge to a signing authority.
By including the path from zi to the root r in the signature, the verifier is enabled to
(a) check in O(log(n)) time that zi has been participated in the computation of the root value r and, by this way, to prove that zi is valid; and
(b) the LRS of the hash path from zi to the root hash r also encodes the index i of the OTSK zi and, thus, the verifier can compare the time t0+i against the KSI's time-stamp's time slot, and that zi was used in the correct time slot.
The hash path from ri's (where ri=H(zi)) to the public root hash r is actually the one-time proof of knowledge (OTPoK) of the secret OTSK zi, without revealing zi.
By the client sending the OTPoK of the secret key to the TTP-SA along with the request, the TTP-SA may (additionally) verify that the OTPoK corresponds to the client's public key r that is stored in the client's certificate. The TTP-SA may add the value of ri into the time-stamping process.
Thus, in the pair zi→ri the value ri “authenticates” to the TTP-SA the value of the OTSK zi.
As a positive side effect, the signer's signature does not need to include the hash path from zi→r, but only the OTSK zi, if TTP-SA includes ri as the part of the time-stamping computation. The verifier sort of “exports” the need to check of the hash path by letting the TTP-SA verify the hash path ri→r before issuing the time-stamp Si. The connection between zi and ri (and, thus, to ri→r) can be verified later on if the TTP-SA pushes the hash of the modified vector {x; userID; ri} into KSI for time-stamping.
After a key generation, the signer has the sequence of OTSKs {z1 . . . zn}, that are valid for signing at times t0+i time slot of KSI. The root hash r of a Merkle tree with the leafs values ris, i=1, . . . , n, where ri=H(zi) for some one-way function H, as demonstrated above. The client's certificate may include the vector: {user ID; r; t0; TTP-SA's KSI identifier}
Signing of a message M at some time t=t0+i, where t0<t≦t0+n, may then comprise the following protocol:
1. Signer computes x=H(H(M); zi) and sends to TTP-SA the vector {i; x; userID; HP ri→r}, where HP indicates hash path.
2. TTP-SA picks the right certificate that matches the pair {userID; r} and checks that it has not been revoked.
3. TTP-SA checks that HP ri→r is correct, and that LRS of that HP is mapping to the index i.
4. TTP-SA sends the hash of {x; userID; ri} to KSI in time t=t0+i, and receives the time-stamping Si, that is AHP (aggregate HP) and CHP (calendar HP).
5. TTP-SA sends Si back to the signer.
6. The signer reveals the signature of the message M as <userID;i;zi;Si>
The verification process is considered successful if:
1. H(x=H(H(M);zi); userID; ri=H(zi)) is the leaf of Si;
2. LRS of Si's AHP leads to the correct TTP-SA's KSI identifier that is bound in the signer's certificate;
3. Si's AHP and CHP lead to the correct CRH for the time t0+i.
Thus, the approach may include that the signing authority receives a proof-of-possession before starting to process the request in the hash tree, computation and issuing of signature. The proof of possession may comprise sending the OTPoK of the secret key to the TTP-SA along with the request. The TTP-SA may verify that OTPoK corresponds to the client's public key that is stored in the client's certificate. The TTP-SA may add the value of ri into the time-stamping process.
A device, application or session of a client may thus be arranged to transmit a proof-of-possession to a signing authority TTP-SA before the TTP-SA starts to process a request in the hash tree, compute and issue of signature. The transmission of the proof of possession may comprise sending the OTPoK of the secret key to the TTP-SA along with the request.
A server operating a signing authority function may be arranged to receive a proof-of-possession from a client, to verify whether the OTPoK corresponds to the client's public key that is stored in the client's certificate, and omit computation and issuing of a signature when the OTPoK does not correspond to the client's public key and compute and issue signature when the OTPoK corresponds to the client's public key. The TTP-SA may add the value of ri into the time-stamping process.
Sending Sequence of Signing Requests to TTP-SA using OTSKs with Time-Forwarded OTSKs Indexes
This part of the disclosure relates to Time Fraction Sub-Trees in Hash Based Time Stamping Services for Faster Streaming of Requests of Services. Consider that KSI RH (root hash) is computed for each interval. Further, assume the intervals to be 1 second (but of course other interval settings are possible).
An issue may be how QI-KSI, as demonstrated above, proposes to synchronize the OTSK zi with the KSI's tune. In the QI-KSI solution demonstrated above, when the signer sends a signing request to the TTP-SA, the client may take a group of OTSKs, for example three consecutive signing keys zi, zi+1 and zi+2, and send 3 signing requests to the TTP-SA simultaneously. The TTP-SA then can choose one out of the given three whose i corresponds to the current KSI's time, and push it to the KSI for time-stamping in the proper time slot. Since the client may reveal OTSK zi+2 only at time t>t0+i+2, this means that the client can produce a stream of signatures with the speed of 1 signature per 3 KSI's time slots (that is 1 signature per 3 seconds, with the above discussed design of the KSI's Core Network).
For the use case where the client needs to sign a stream of data messages, the QI-KSI's way to synchronize the time between the signer and KSI is not optimal. It is a desire to provide a better synchronization solution so that the client does not waste OTSK keys and can perform signatures for a stream of data, e.g. with the speed in average of 1 signing per KSI's time slot (a second).
Consider that the signer is enabled to send the sequence of signing requests to the TTP-SA using OTSKs with time-forwarded OTSKs indexes, and by this the client proposes the delay time before the actual time-stamping procedure, so that there is enough time for TTP-SA to receive the requests, prepare them and synchronize the time with KSI. In the use case where one needs to produce a stream of signatures, this synchronization scheme, as will be demonstrated in further detail below, utilizes the KSI resources particular, KSI's time slots) efficiently, which makes the signing speed to converge to 1 signature per one KSI time slot, in average.
Assume that the possible resynchronization time between the signer's and KSI's clocks can be at most Δ seconds (for instance, let Δ be 5 seconds), including possible delays in communication between the signer and the TTP-SA. When the signer sends a signing request at his local time tsig, he actually may use the time-forwarded OTSK with the index tsig−t0+Δ, prepare and send a signing request to the TTP-SA.
The TTP-SA may have a local queue with incoming signing requests that are already checked for the client's certificate and are waiting for being entered to KSI at the right tune for time-stamping. When the KSI time becomes aligned with the time of the first request in the queue, i.e., when the KSI's time becomes tsig+Δ−1, the TTP-SA pushes the corresponding hash value to the KSI infrastructure and receives the right time stamping.
Thus, the client reveals the stream of signatures with the delay of Δ seconds. In the use case of a stream of signatures, the performance may then converge to the speed 1 signature per KSI time slot.
Note that in the above the clocking is described as being 1 second. However it is readily understood that it is equally applicable to other clocking interval settings.
A trusted signing authority applying a hash tree signing system may have a local queue that comprises signing requests that use a time-forward OTSK that are already verified and which start further processing when the time with the hash tree system gets aligned. The OTSK may be determined by having the index tsig−t0+Δ.
The sending of the sequence of signing requests to TTP-SA using OTSKs with time-forwarded OTSKs indexes benefits from being combined with the approach for deriving the OTSKs directly via a one-way function, and may also benefit from the approach for one-time proof of knowledge for one-time signing keys in hash based signing schemes, as well as with an approach comprising a combination of them.
A device, application or session of a client may thus be arranged to transmit signing requests to a signing authority with time-forwarded OTSKs indexes.
A server operating a signing authority function may be arranged to receive signing requests with time-forwarded OTSKs indexes, store them in a queue, and when the KSI time becomes aligned with the time of a request in the queue to push the corresponding hash value to a KSI infrastructure and receive the right time stamping. Any calculations may be pre-calculated for the requests of the queue.
The method of the client is for providing a hash value for a piece of data, where the hash value provides for a time-stamp for the piece of data upon verification. The method comprises collecting 100 the piece of data and deriving 102 a one-time signing key, OTSK, of a OTSK hash chain by applying a time fraction hash tree splitting a time slot corresponding to an index into time fractions such that the time slot is divided into fractions according to the number of leafs of the time fraction hash tree. A signing request for the piece of data is formed 104 by applying the OTSK for the fraction for the respective piece of data to calculate hash values of the piece of data. The signing request comprising the hash values is transmitted 106 to a server for deriving a time stamp for the piece of data.
The one or more OTSKs may be derived 102 by a one-way function. The respective OTSK is used to provide a hash value using a hash function including the OTSK and the piece of data which is intended for time signing. One or more of the following options may be included in the method: registering the hash value such that the piece of data at least later on may be verified regarding its presence at a time corresponding to an index of the OTSK used for providing the hash value, providing a one-time proof of knowledge, OTPoK, to a signing authority that the OTSK was in possession of the signer at a certain time without revealing the OTSK, and sending a signing request to the signing authority using OTSKs with time-forwarded OTSK indexes such that signing of the hashes of the pieces of data may be queued at the signing authority and the slots corresponding to OTSK indexes can be efficiently used. Further options and variants have been demonstrated above.
The method of the server is for issuing a time stamp signature. The method comprises receiving 150 a signing request for a piece of data formed by application of a one-time signing key, OTSK, for a fraction corresponding to an index for time fractions defined by a time fraction hash tree splitting a time slot for the respective piece of data to calculate hash values of the piece of data where the time slot is divided into fractions according to the number of leafs of the time fraction hash tree. A time stamp is derived 152 for the piece of data including a hash path of the time fraction hash tree as a sub-tree of hash tree of the OTSK, and the time stamp for the piece of data is transmitted 154, e.g. to the client. The signing request may be checked 151 for client's signature before the deriving of a time stamp. The checking of the signing request may include determining whether the client has provided a one-time proof of knowledge about a one-time signing key to a server without revealing the one-time signing key. The determining whether the client has provided a one-time proof of knowledge about the one-time signing key may comprise checking whether the hash corresponds to a one-time signing key for a root hash included in a public certificate associated with the identity, checking whether an index corresponding to the hash path from the one-time signing key to the root hash corresponds to a correct time slot, and determining it to be proven that the electronic device is in possession of the correct one-time signing key when the checks are fulfilled. Alternatively or additionally, the checking of the signing request may comprise accessing a certificate matching the identity and a root hash for the first hash, checking validity of the certificate, and verifying whether the hash path for the first hash is correct. If the certificate is not valid or the hash path cannot be verified to be correct, the server omits 153 further actions for that signing request.
The method may include verification 155 of a time stamp for a piece of date by determining whether a hash of the message is a leaf of the time stamp hash tree, determining whether the aggregate hash path corresponds to the correct identifier of the server to the server of the signature infrastructure entity, and determining whether the aggregate hash path and calendar hash path correspond to a correct calendar root hash for a time corresponding to the index.
A device, application or session of a client may thus be arranged to derive a one-time signing key zi as zi=H(zsk;fi), for i=1, . . . , n, where H is a one-way function, where zsk is a secret key of the signer, and fi is a function on the index i that generates different values for each i=1, . . . , n. The function may be fi=i.
Network 220 may comprise one or more of IP networks, public switched telephone networks (PSTNs), packet data networks, optical networks, wide area networks (WANs), local area networks (LANs), wireless local area networks (WLANs), wired networks, wireless networks, metropolitan area networks, and other networks to enable communication between devices.
Network node 200 comprises processor 202, storage 203, interface 201, and antenna 201a. These components are depicted as single boxes located within a single larger box. In practice however, a network node may comprises multiple different physical components that make up a single illustrated component (e.g., interface 201 may comprise terminals for coupling wires for a wired connection and a radio transceiver for a wireless connection). Similarly, network node 200 may be composed of multiple physically separate components e.g., a NodeB component and a RNC component, a BTS component and a BSC component, etc.), which may each have their own respective processor, storage, and interface components. In certain scenarios in which network node 200 comprises multiple separate components(e.g., BTS and BSC components), one or more of the separate components may be shared among several network nodes. For example, a single RNC may control multiple NodeB's. In such a scenario, each unique NodeB and BSC pair, may be a separate network node. In some embodiments, network node 200 may be configured to support multiple radio access technologies (RATs). In such embodiments, some components may be duplicated (e.g., separate storage 203 for the different RATs) and some components may be reused (e.g., the same antenna 201a may be shared by the RATs).
Processor 202 may be a combination of one or more of a microprocessor, controller, microcontroller, central processing unit, digital signal processor, application specific integrated circuit, field programmable gate array, or any other suitable computing device, resource, or combination of hardware, software and/or encoded logic operable to provide, either alone or in conjunction with other network node 200 components, such as storage 203, network node 200 functionality. For example, processor 202 may execute instructions stored in storage 203. Such functionality may include providing various wireless features discussed herein to a wireless devices, such as WD 210, including any of the features or benefits disclosed herein.
Storage 203 may comprise any form of volatile or non-volatile computer readable memory including, without limitation, persistent storage, solid state memory, remotely mounted memory, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), removable media, or any other suitable local or remote memory component. Storage 203 may store any suitable instructions, data or information, including software and encoded logic, utilized by network node 200. Storage 203 may be used to store any calculations made by processor 202 and/or any data received via interface 201.
Network node 200 also comprises interface 201 which may be used in the wired or wireless communication of signaling and/or data between network node 200, network 220, and/or WD 210. For example, interface 201 may perform any formatting, coding, or translating that may be needed to allow network node 200 to send and receive data from network 220 over a wired connection. Interface 201 may also include a radio transmitter and/or receiver that may be coupled to or a part of antenna 201a. The radio may receive digital data that is to be sent out to other network nodes or WDs via a wireless connection. The radio may convert the digital data into a radio signal having the appropriate channel and bandwidth parameters. The radio signal may then be transmitted via antenna 201a to the appropriate recipient (e.g., WD 210).
Antenna 201a may be any type of antenna capable of transmitting and receiving data and/or signals wirelessly. In some embodiments, antenna 201a may comprise one or more omni-directional, sector or panel antennas operable to transmit/receive radio signals between, for example, 2 GHz and 66 GHz. An omni-directional antenna may be used to transmit/receive radio signals in any direction, a sector antenna may be used to transmit/receive radio signals from devices within a particular area, and a panel antenna may be a line of sight antenna used to transmit/receive radio signals in a relatively straight line.
WD 210 may be any type of wireless endpoint, mobile station, mobile phone, wireless local loop phone, smartphone, user equipment, desktop computer, PDA, cell phone, tablet, laptop, VoIP phone or handset, which is able to wirelessly send and receive data and/or signals to and from a network node, such as network node 200 and/or other WDs. WD 10 comprises processor 212, storage 213, interface 211, and antenna 211a. Like network node 200, the components of WD 210 are depicted as single boxes located within a single larger box, however in practice a wireless device may comprises multiple different physical components that make up a single illustrated component (e.g., storage 213 may comprise multiple discrete microchips, each microchip representing a portion of the total storage capacity).
Processor 212 may be a combination of one or more of a microprocessor, controller, microcontroller, central processing unit, digital signal processor, application specific integrated circuit, field programmable gate array, or any other suitable computing device, resource, or combination of hardware, software and/or encoded logic operable to provide, either alone or in combination with other WD 210 components, such as storage 213, WD 210 functionality. Such functionality may include providing various wireless features discussed herein, including any of the features or benefits disclosed herein.
Storage 213 may be any form of volatile or non-volatile memory including, without limitation, persistent storage, solid state memory, remotely mounted memory, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), removable media, or any other suitable local or remote memory component. Storage 213 may store any suitable data, instructions, or information, including software and encoded logic, utilized by WD 210. Storage 213 may be used to store any calculations made by processor 212 and/or any data received via interface 211.
interface 211 may be used in the wireless communication of signalling and/or data between WD 210 and network node 200. For example, interface 211 may perform any formatting, coding, or translating that may be needed to allow WD 210 to send and receive data from network node 200 over a wireless connection. Interface 211 may also include a radio transmitter and/or receiver that may be coupled to or a part of antenna 211a. The radio may receive digital data that is to be sent out to network node 201 via a wireless connection. The radio may convert the digital data into a radio signal having the appropriate channel and bandwidth parameters. The radio signal may then be transmitted via antenna 211a to network node 200.
Antenna 211a may be any type of antenna capable of transmitting and receiving data and/or signals wirelessly. In some embodiments, antenna 211a may comprise one or more omni-directional, sector or panel antennas operable to transmit/receive radio signals between 2 GHz and 66 GHz. For simplicity, antenna 211a may be considered a part of interface 211 to the extent that a wireless signal is being used.
In some embodiments, the components described above may be used to implement one or more functional modules used in a collision-blocking method for hash tree based time stamping. The functional modules may comprise software, computer programs, sub-routines, libraries, source code, or any other form of executable instructions that are run by, for example, a processor. In general terms, each functional module may be implemented in hardware and/or in software. Preferably, one or more or all functional modules may be implemented by processors 212 and/or 202, possibly in cooperation with storage 213 and/or 203. Processors 212 and/or 202 and storage 213 and/or 203 may thus be arranged to allow processors 212 and/or 202 to fetch instructions from storage 213 and/or 203 and execute the fetched instructions to allow the respective functional module to perform any features or functions disclosed herein. The modules may further be configured to perform other functions or steps not explicitly described herein but which would be within the knowledge of a person skilled in the art.
The methods according to the present invention is suitable for implementation with aid of processing means, such as computers and/or processors, especially for the case where the processing element 202, 212 demonstrated above comprises a processor handling security functions. Therefore, there is provided computer programs, comprising instructions arranged to cause the processing means, processor, or computer to perform the steps of any of the methods according to any of the embodiments described above and those roughly summarized with reference to
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2016/050858 | 1/18/2016 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 62118808 | Feb 2015 | US |
