Patent Application
20240056431
This invention relates to cryptographic apparatus and systems and more specifically to those aimed at securing communications and transactions, including protecting traffic data that allows tracing related to who communicates and transacts with whom.
Mix networks have been the subject of significant research, commercial activity, and actual use since they were first disclosed by the present applicant in 1979. Such prior art mixing, however, generally performs public-key cryptographic processing on messages as they travel through the network. This is believed to result in significant commercial disadvantages in terms of efficiency and efficacy relative to the present invention. Moreover, use of mix networks has been limited largely to providing unlinkability of communication, whereas aspects of the present invention not only extend mixing's protections to transactions but also offer other types of protections that address broader needs of communication and interaction of users.
Novel cryptographic apparatus and methods disclosed here allow mixing that is believed significantly more efficient in terms of the amount of computation performed in realtime and consequently allow reduction both in overall delay of messages and cryptographic processing capacity. This is of commercial significance at least because it is believed that efficiency has limited the adoption as well as the efficacy of, and relates to the cost of, mixing systems both used and proposed. It is also believed of commercial significance because delay through each of a number of stages of mixing adds up, which tends towards more batches, use of less stages, reduced protections for users, and consequently less attractive offerings to users. The improved efficiency is believed further of commercial significance because it improves the reduced utilization and related hardware cost of cryptographic processing of prior art schemes. Moreover, users are believed sensitive, especially with increasingly popular portable devices, to computation time and energy usage. Also, some prior-art mixing has been based on stronger assumptions about underlying cryptographic primitives and thus may be more vulnerable to cryptanalytic attack.
In brief non-limiting summary some exemplary embodiments of some inventive aspects will now be provided. A so-called “cascade” or series of n ix devices, called here “nodes,” receives b messages, each message associated with a corresponding entity/device here called a “user,” “user device,” or “subscriber.” Included in the messages are encryptions formed by the subscribers. In what may here be called an “unpermuted” phase, each mix node processes these b inputs into what here may be called a “normalized” form, using non-pubic information held in common pairwise by nodes and corresponding subscribers, and typically employing a mathematically commutative property. Then in what may here be called a “permuting” phase, each of the b messages is transformed successively through each of the n nodes, during which the encryption remaining after normalization is removed and each node permutes the order of the b inputs it processes. The output of the cascade contains the h messages sent by potentially identified subscribers but in an order believed unpredictable to any proper subset of non-colluding nodes.
For clarity a particular non-limiting but concrete example will be described using modular exponentiation as the encryption, first without pre-computation and including successive processing in the unpermuted phase. Each of the h messages is sent by a distinct subscriber, each of which has established a different secret key in common with each of the n mix nodes. A subscriber raises its actual message content c to the power computed as the product of the secrets that subscriber shares with all the nodes, in a group where discrete log is at least believed hard. Then, in an unpermuted phase, each node successively normalizes by raising to a computed secret power that both cancels the unique subscriber power and leaves in place a single round secret power that node uses for all messages in the batch. In the permuting phase, each node permutes the b messages it receives from the preceding node in the cascade and raises each message in the batch to a power that cancels that node's single round power, so that the final output batch contains each original message content c unencrypted but re-ordered by the composition of the permutations applied by the successive nodes.
In an exemplary simplified non-limiting pre-computation embodiment described for clarity, the nodes first cooperate to produce a so-called “shared public key” for homomorphic encryption, such as one that includes independent private keys from each node. The nodes then in effect go through the motions of the two real-time phases already described, unpermuted and permuted, but instead of encrypting a message payload they homomorphically combine keys injected at each stage, the same keys that will be used again at the corresponding stages of the realtime processing. The final accumulation of individual keys that apply to a particular position in the overall output of the cascade is then recovered by the nodes cooperating in decryption of the homomorphic encryption for that output position. Inputs for realtime processing can use the same group operation as that of the homomorphic encryption, so that the set of key group elements that are combined by the group operation with the message to encrypt the message payload in the output can be cancelled by combining with the inverse of what is recovered by decrypting the final result of the homomorphic encryption.
In some exemplary non-limiting functionality extensions, believed adaptable for mixing more generally, message types and protocols for their use are disclosed allowing applications such as, encrypted email, untraceable return addresses, polling/anti-spam, credential mechanisms, and payments including so-called “payer anonymity.” The novel unpermuted phase of processing, with pairwise secured communication between subscribers and nodes, allows nodes to optionally at least partly independently verify aspects of user identity and various transaction parameters.
Prior Art Mixing
Mixing was first proposed by the present applicant in a June 1979 University of California, Berkeley technical report that was submitted for publication that year under the same title and then appeared as “Untraceable electronic mail, return addresses, and digital pseudonyms,” 1981 Communications of the ACM, February 1981, vol. 24 no. 2, pp 84-88.
Included in this prior art was the notion of so-called “hybrid” encryption: that keys could be provided by public key techniques that could then be used for non-public key encryption of additional message payload. Some more recent systems have used such conventional keys for subsequent communication through the same set of mix nodes; however, use beyond the original message begins to erode untraceability, since all such uses are linkable by all nodes and traffic patterns link these to the user and the other parties that the user communicates with as well as the times and volumes of communication.
Inventive aspects of the present application retain the unlinkability as in the original proposal, while they may be able to only use group operations for encryption during transmission of messages, instead of any of the typically far more computation/data-intensive full public-key operations.
The approach to identification of users of an informational system that have been taken in the prior art are believed to be roughly divisible in the main between two alternatives: unchecked user identity or monolithic security model for identity checking. It will be appreciated that systems that do not check user identity are subject to various kinds of abuse, including the use of many identities by a single person and reduced accountability; however, what have been called “reputation” and “credential mechanisms” can potentially overcome some of these disadvantages. Unchecked systems are believed to have the unfortunate property that those few who may make the most serious abuses may take the strongest measures outside the system to remain untraceable, whereas most users of the system may be comparatively easier to link to their uses.
In a monolithic identification system, there is mainly a single authority that the user is to satisfy as to the adequacy of the user's identification. One believed kind of disadvantage of such an approach is that a single entity amasses information about a person that may allow that entity to link to aspects of a person's activities beyond the charter of the system, and in case of compromise of that single entity makes the person vulnerable to impersonation. Another type of disadvantage is that the authority may, in some cases, dilute the efficacy of the system by providing exceptions to certain persons. Well known examples include celebrities, witness protection, intelligence operatives, internal personnel, and administrators.
There are variations on the monolithic approach to identity checking involving the legal “structuring” of entities. For instance, identification checking subcontractors may be employed by a monolithic entity. As another example, so-called “federated” identity schemes allow organizations to benefit from identity checking performed by a central authority or by another organization member of the federation.
The believed novel approach to identification taken in some aspects here allows each node to form their own independent identity verification for the user such that only if the user is accepted separately by each of plural nodes is the user accepted by the system. A believed advantage of the approach is that while users may be reluctant to share a lot of identifying data with a single entity, it is believed they may be more willing to share that same amount or even more data when it is divided into separate parts with respective separate entities, especially entities under separate jurisdictions and/or when entities are not to share the data. Another example believed advantage of the approach is that false identities may be harder to inject into a system as the number of entities that would need to be deceived and/or corrupted to create a false identity.
Untraceable communication systems have been considered infrastructure for transaction systems. As just one example, such an approach was proposed for payments and credential mechanisms in, “Achieving electronic privacy,” by the present applicant, Scientific American, August 1992, p. 96-101. In some few examples, mixing has alternatively been considered as a component for a larger system, such as in an election system or in processing payment orders already approved.
Inventive aspects of the present system advantageously integrate transactions with the mixing communication infrastructure. This is believed to, for one thing, potentially increase the amount of traffic in the underlying system and thereby increase the privacy protection offered. As another example, it is also believed to benefit users by the integration of services and adoption by the bundling of services. Moreover, it is believed that there are significant technical economies, such as in terms of key management, computation, infrastructure, system administration, and so forth. More fundamentally, the security model and/or arrangements of the communication system can be applied to the transaction systems as well.
Several related general aspects are disclosed here without any limitation.
Examples of mix cascades may be described for concreteness and clarity using two nodes, for example, but it will readily be understood and appreciated that any number of nodes may be used. Similarly, in some examples every user may send a single message per batch; in other examples, each user may send more than one message in some batches; and more generally, each user may send zero or one or more messages per batch and in a way that differs per batch.
The keys known to pairs made up of a user and a node are referred to as “common” keys or keys that are “common” to the pair. Common keys are optionally used also for authentication of input messages, such as by so-called MAC's. In some examples a subset of nodes may be sufficient to authenticate the MAC's.
For some commutative encryption functions, such as the well-known discrete log, composition of a series of encryptions can be computed almost as efficiently as a single encryption, using knowledge of the public order of the group to reduce the size of the exponent.
More flexible flows are anticipated, such as if the mixing of inputs proceeds in sub-batches, sliding windows, and/or probabilistically. As just one non-limiting example, each node could delay an item from one batch to a next according to some probability distribution associated with that node and/or other factors in system operation. As another such example, each node could delay each message some amount of time chosen from a suitable distribution.
Proof or auditing that the mixes perform their mixing transformations correctly may be provided, for instance, by multiparty computation techniques, as are known in the art.
For an exponentiation-based system, such as without pre-computation, an example is where it is believed that a mix can “prove” that it has raised each input item to the same power in the realtime phase. For instance, a so-called “challenge” may be formed as set of exponents; the mix publishes a value; and a binary choice is made between the mix being required to provide the power that takes that product of powers of the inputs to the published value or the power that takes the product of powers of the outputs to the shared published value. Such a procedure can be repeated for more certainty and the choice of exponents can be reduced and the number of repeats increased. An example reduces the exponents to zero and one, in the case when the number of repeats may be on the order of one hundred.
As an example of proofs believed suitable for a pre-computation system, a subset of the messages, such as in the input and/or output can be selected by a multi-party secure method as are known and those messages opened all the way through the cascade. This can be done at the pre-computation stage to detect deviation from protocol at that stage. Unpermuted phase processing can, for example, be checked by senders because they can inspect the order-preserving transformations of their inputs in the unpermuted phase processing. During a permuting phase, nodes can use commitments they already published in an audited pre-computation phase as part of their proofs. For instance, a pre-computation publication of quotients for each pair of input and output item that are in the same position in the ordering (i.e. first input to first output, second input to second output, etc.) allows checking of the mix processing from the published data and data communicated between nodes.
It may be desired to have so-called “forward secrecy,” a term that is believed applied to a variety of properties in the art. Persistent shared-keys are believed best transformed in an irreversible manner for each transmission, so as to prevent earlier messages from being compromised by any subsequent leak of key state. By conducting key agreement protocols from time to time, which is believed also desirable, even compromised persistent keys only reach forward till the next agreement protocol. Furthermore, by including persistent state in with the agreement protocols, an adversary capable of breaking such agreement protocols may be thwarted.
The non-permuting phase can be so-called “pipelined,” by dividing the input into separate portions that are each processed by a different ordering of the nodes, as will readily be understood. This lets each node process at more or less a so-called high “duty cycle,” as there are no major forced time delays during which a node cannot be processing. Pipelining of the permuting phase in a similar manner, as will be understood, it is believed would divide the input into portions that could, depending on parameters, be linked to corresponding portions in the output—thereby reducing unlinkability. Pipelining the mix phase fora first half of the nodes and then providing the output of this as input to a pipeline for a second half of the nodes is believed to improve unlinkability over a single pipeline. This can be generalized to pipelines of distinct or overlapping nodes with whatever multiplicity.
In some further non-limiting examples of pipelining of the permuting phase, the division into separate portions may not be public but instead known to each respective node. For example, each node can be assigned a “random” or “unpredictable” subset of the inputs to start the corresponding sequential processing. In some examples, the nodes can perform a multiparty computation to assign these portions without subsets of nodes knowing more than their own subset partition. For instance, one example type of multiparty computation to accomplish this can use the system itself to process inputs provided by each node and the position of those in the lexicographically sorted output, known and recognizable by a node, would be that node's portion.
As for the pre-computation phase, pipelining may be applied in a variety of ways. For instance, multiple pre-computations can be conducted in parallel and in advance of when they would actually be deployed. Other examples will be described later, such as with reference to
There may be various input structures, such as buffers, routers, forwarders, aggregators, front-ends and the like that feed early parts of the processing; nevertheless, the input device may be considered the first node of the unpermuted phase processing without any loss of generality. Similarly, there may be various output buffering, forwarding, or delivery structures and processes that feed the results of the second phase of processing to other entities; nevertheless, the last node of the unpermuted phase processing may be considered the output device without any loss of generality. In some examples where a live phase uses non-sequential processing before the permuting phase, a front-end processing capability may provide information to the nodes about the identity of the entity supplying a message input and the input slot of the mix that message will occupy. The nodes may then supply the resulting contributions for aggregate application by the entity performing front-end processing. Further examples are described with reference to
In some yet still further non-limiting examples mentioned, nodes do computation in advance of the processing of input, as what generally may be referred to here as a “pre-computation,” ideally so that the realtime processing of input is sped and/or otherwise enhanced and/or more efficient use is made of computing resources. As just one example, a pre-processing can provide each node with enough data that the commutative operations can be what is sometimes regarded as “non-cryptographic” or “non-complexity-based”; as an example, in the absence of, for instance, an exponentiation with secret exponent as the commutative operation such examples can instead use a group operation in the group by secret values each ideally appearing in a form visible to an attacker only once. This is related to what is sometimes referred to as a “one time pad” type of arrangement, or a system where among other things the number of key values revealed is the same as the total number of values known to the adversary. It will be understood by those of skill in the cryptographic art that such key values can be arrived at in principle by a multiparty computation between the nodes.
It is believed that a pre-computation may in some examples process messages through nodes in the same order as the live phase or, in other examples, in the opposite order as the live phase. When processed in the same order, the result of the homomorphic operation may be applied at the end of the live phase; when processing in the reverse order, the result of the homomorphic operation may be applied at the beginning of the live phase, subsequent to normalization.
The input to the backwards run may be taken as a known constant vector, for instance a random vector, or what is believed more conveniently, the identity vector. The last node in the forward run, the first in the backwards run, permutes this vector using the inverse of the permutation that will be used in the forward run, which it effectively retains at least the ability to reconstruct. This node chooses and retains at least the ability to reconstruct a vector of random values and multiplies each of the permuted values by, for clarity, what may be regarded as the multiplicative inverse of the respective random value retained. The penultimate node in the forward order, adjacent this already described node, takes this resulting pre-computation vector as input from the already described node and processes its elements in this same way, with its own choice of retained values and permutation, so that they can be provided to the node that is its predecessor in the forward order. This cascade processing proceeds in this reverse order, as will readily be understood, through to and including the first node of the forward-order cascade. From this point to the overall input vector, as the reverse premix, each node successively processes the resulting vector from its predecessor in reverse order in a similar manner, ideally with essentially independent random values retained, but with the identity permutation.
The pre-computations just described may be encrypted using a so-called partially homomorphic cryptosystem, such as for instance the well known El Gamal system. In this example, the nodes can compute a so-called “shared” public key, such as by each successively raising the result of such key generation by the preceding node to its own secret-key power; the result is the agreed generator raised to the product of the private keys of the nodes, as will readily be understood by those of skill in the cryptographic art. The final vector of values encrypted with this shared key can be decrypted by the nodes each successively applying the inverse of the secret power to each element of the vector. The elements of this decrypted vector are then combined by the croup operation with the respective input items of the input vector before it is processed in the forward direction in the manner already described above.
Sending through this example pre-computed system, instead of exponentiating, a user includes, using the group operation, a group element corresponding to each node. The element can, for instance, be cryptographically dependent on a secret key common to the user and the node, much as in the systems without pre-computation already described. Each of these elements will be understood to be cancelled by the corresponding node during the non-permuting processing; however, a new element, from the pre-computation phase, will be substituted by the node to replace each element as it is cancelled in the non-permuting processing. Thus, the user-supplied elements will all be cancelled and the pre-computed elements will all be included as the realtime processing moves forward from non-permuting through to permuting. Since the element that was included from the final vector of values encrypted with the shared key is the inverse of the product of all the elements from the pre-processing for a particular input, and these will be multiplied in during the forward direction processing for that input as it appears in corresponding positions in the intermediate vectors, kept together under the permutations, it is believed as will be appreciated that all the pre-processing terms will cancel, leaving the original content element in the output vector.
In some settings it is believed advantageous to allow users to send many messages but to allow messages that have a unique occurrence per user to be recognized as such. One example use, for instance, would be in elections or statements of opinion on a particular issue, where a user statement would have more weight if it were known to be the unique statement of a user related to the particular issue, even though which user made the statement remains hidden. An example way to achieve such functionality sends pairs of messages, such as even and odd positions in the input vector, one message providing the uniqueness for the payload contained in the other message. Such a uniqueness message can be processed by the nodes using a fixed transformation per user in the non-permuting phase; the changing transformations described elsewhere here for the non-permuting phase would still be used for the payload carrying message of the pair.
In some settings it may be desired to provide robustness in case some mix nodes fail or otherwise leave the system. In one example approach to such robustness nodes use so-called “verifiable secret sharing” to provide shares of their keying material so that in their absence a quorum of other nodes can recover the keying material, as is known. So-called “secret sharing” techniques or those proposed earlier by the present applicant using various combined encryptions to achieve similar but more general results can be applied.
In some settings it may be desired for the nodes to selectively trace certain messages back to their origin. Such tracing is possible if the nodes retain suitable records or otherwise access to earlier processing.
What will be called here “shared-key homomorphic cryptosystems” are known. In such systems each of a number of parties can cooperate to create what may be called here a “shared public key,” which is in effect information that defines an encryption operation that at least each party can use to encrypt an input value. Encryptions formed with the shared public key operation can be combined with a group operation, typically addition or multiplication, such that the parties can cooperate to decrypt the combined encryptions and the resulting cleartext is equivalent to the input values combined by the group operation.
This homomorphic property may be shown as h(r1)h(r2) equals h(r1r2), where anyone can apply h to encrypt using the shared public key; but cooperation of the parties can invert h or decrypt by computation of the inverse h−1.
An illustrative example, without any limitation whatsoever and for concreteness and clarity, is what is believed to be a multiplicative shared-key homomorphic cryptosystem, based on the encryption operation of the well known so-called “El Gamal” cryptosystem. This cryptosystem uses a multiplicative group where discrete log is assumed hard and encrypts each value using a random value and public key.
It is believed that such example systems can be implemented in whatever group where the so-called property “discrete log is hard” holds, which is currently believed to include for example arithmetic modulo a large prime, such as a one thousand or two thousand bit prime, particularly a so-called “safe” or even a “strong” prime. It is further believed that not all elements of such groups are generators but that the structure of the cyclic group allows elements chosen from the whole group uniformly to hide other such elements in products; alternatively, however, elements could be restricted to a subgroup where all elements are generators, such as the quadratic residues.
The following example shows for clarity the use of such a system by two parties, each encrypting its own secret value, forming the product of the encryptions, and decryption of the product by cooperation of the parties; however, extension to any number of parties and any number of factors will readily be understood:
(gxyagxyb, rxgarygb)=(gxy(a+b), rxryga+b)
The following claim like language is copied from the application from which priority is claimed as mentioned above:
(a) Message transport apparatus for delivering messages from an input device to an output device, comprising: An input device receiving plural messages, each such message identified as related to at least one of a set of common keys; Plural first phase processing devices, each privy to plural common keys and each applying a transformation to messages received responsive to a corresponding one of the common keys and including the transformation related to a key corresponding to at least a second phase processing device; Plural second phase processing devices, each receiving at least some messages processed by the first phase processing devices and each further processing messages by re-arranging the order of messages substantially unpredictably and applying transformations responsive at least to values known to respective second-phase processing devices; and Such that the order of the items received by the input device substantially different from the order of items output by the output device, and the linking between items in the input and items in the output hidden from those without knowledge of the values known to the devices.
(a1) The apparatus of claim a, including a third phase without permutations and processing related to the third phase using keys common to recipients of messages.
(a2) The apparatus of claim a-a1 [to be read throughout these claims as the present claim is multiply dependent on all claims from claim “a” to claim “a1,” inclusive, with method/apparatus/system language adapted as may be understood], including:
A pre-computation phase during which said processing devices develop keying values responsive to permutations of their respective inputs and invertible transformation of such inputs later decrypted by cooperation of the processing devices; and where the first phase and the second phase processing are responsive to the values developed during the pre-computation phase and the second phase responsive to the permutations developed during the pre-computation phase; and the computation performed during the combination of the first and second phases is substantially reduced by the pre-computation phase.
(a3) The apparatus of claim a-a2, comprising: the first and second phase processing devices arranged so that at least one first first-phase processing device is substantially the input device and at least one last second-phase processing device is substantially the output device for at least some messages.
(a4) The apparatus of claim a-a3, such that each of plural messages sent by the same user are unlinkable in the output.
(a5) The apparatus of claim a-a4, including at least one first phase processing device [sic] and a second phase processing device.
(a6) The apparatus of claim a-a5, including the pre-computation phase including keying values transformed by the permutations used in the second phase.
(b1) A cryptographic system for a set of nodes substantially to hide the correspondence between at least a set of inputs supplied to the nodes by subscribers and a responsive set of outputs developed by the nodes, comprising the steps of: Nodes apply a first permutation in a first processing for a first batch; Nodes apply a permutation related to the first permutation in at least a second processing for a second batch; Nodes transform first messages related to the first processing; and Nodes transform second messages related to the at least second processing.
(b2) A cryptographic system for a set of nodes substantially to hide the correspondence between at least a set of inputs supplied to the nodes by subscribers and a responsive set of outputs developed by the nodes, comprising the steps of: Nodes apply a first permutation in a precomputation for a first batch; Nodes apply a permutation at least related to the first permutation in at least a second precomputation for a second batch; Messages are transformed by the nodes in a first processing using the first permutation; and At least indications are processed by the nodes in at least a second processing using the second permutation related to delivery of the output of the first processing.
(c) A cryptographic method for a set of nodes substantially to hide the correspondence between at least a set of inputs supplied to the nodes by subscribers and a responsive set of outputs developed by the nodes, comprising the steps of: Establishing common first keys by nodes that are in common with subscribers; Accepting by at least one node of subscriber submitted encrypted input items including encryption with keying responsive to the common first keys;
An unpermuted phase wherein the common first key aspects of items are replaced with node second keys known to respective nodes; A permuting phase wherein said node second keys are cancelled and the items re-arrange; and So that the relative arrangement of the input and output items is substantially unpredictable to those without knowledge of the common first keys, the permutations, and node second keys.
(c1) The method of claim c, including: A pre-computation phase developing values responsive to permutations of respective inputs and invertible transformation of such inputs, where the non-permuting phase and the permuting phase processing are responsive to the values developed during the pre-computation phase; and the computation performed during the combination of the non-permuting and permuting phases is substantially reduced because of the pre-computation performed in the pre-computation phase.
(c2) The method of claim c-c1, including: each payload message paired with a uniqueness message; the first phase processing of the uniqueness messages the same across instances; the uniqueness messages in the input being enforced as unique per user; and so that the uniqueness message in the output are unique per user and are substantially unlinkable to the input uniqueness messages.
(c3) The method of claim c-c2 including: payload messages being associated with a re-blinded form of at least a master credential per user.
(c4) The method of claim c-c3 including: payload messages including provision for transfers of value between accounts where authentication is included in payloads for source accounts.
(c5) The method of claim c-c4 including: payload messages including provision for an exception to be raised if a withdrawal transaction fails to be consummated and the exception being linked to a corresponding input withdrawal message type.
(c5.1) The method of claim c5 including: transfer payloads resulting in confirming messages.
(c6) The method of claim c-5.1 including untraceable return addresses.
(c7) The method of claims c-c6 including: a third phase encrypting messages with the respective common key of the intended recipient.
(d1) Apparatus or system claims a-c, including plural users forming messages supplied for input processing.
(d2) Apparatus or system claims a-d1, including proofs by nodes that the processing is correct.
(d3) Apparatus or system claims a-d2, including forward secrecy on at least some keys in common between nodes and subscribers.
(d4) Apparatus or system claims a-d3, including authentication checkable by nodes using at least some keys in common between nodes and subscribers.
(d5) Apparatus or system claims a-d4, including a broadcast of messages to nodes during the non-permuting phase.
(e) A mix method or apparatus comprising plural entities receiving encrypted input items in a first order and producing as output at least differently encrypted items in a second order, such that the relationship between the orderings is at least difficult to predict without secrets of at least a quorum of the plural entities.
(e1) The mix of claim c including: each payload message paired with a uniqueness message so that the uniqueness message in the output is unique per user and substantially unlinkable to the input uniqueness messages.
(e1.1) The mix of claim e1 including: so that the user can substantially choose the uniqueness message.
(e1.2) The mix of claim e1.1 including: so that the user can substantially choose the uniqueness message, (e1.2) The mix of claim e1.2 including: so that the uniqueness message includes a count.
(e1.3) The mix of claim e including: a payload message paired with a designator so that the designator in the output substantially unlinkable to the inputs and substantially unpredictable to users and substantially repeatable by any user supplying the same designator in the input.
(e2) The mix of claim a including: payload messages being associated with a re-blinded form of at least a master credential per user.
(e3) The mix of claim a including: payload messages including provision for transfers of value between accounts where authentication is included in payloads for source accounts.
(e4) The mix of claim e including: payload messages including provision for an exception to be raised if a withdrawal transaction fails to be consummated and the exception being linked to a corresponding input withdrawal message type.
(e5) The mix of claim e including: encrypting messages with the respective common key of the intended recipient.
(e6) The mix of claim e including: providing an authentication of the sender to the recipient of the message.
(e7) The mix of claim a including: decrypting messages by nodes with the respective common keys of senders.
(e8) The mix of claim e including: at least some of the output items associated with persistent pseudonym information that remains the same across at least some output items from the same sender.
(f1) A cryptographic system for a set of nodes substantially to hide the correspondence between at least a set of inputs supplied to the nodes by subscribers and a responsive set of outputs developed by the nodes, comprising the steps of: Nodes apply a permutation in processing a first hatch; Nodes apply the same permutation in processing at least a second batch; Messages are transformed by the nodes related to the first processing using the permutation; and Messages are transformed by the nodes related to the at least second processing using the same permutation.
(f2) A cryptographic system for a set of nodes substantially to hide the correspondence between at least a set of inputs supplied to the nodes by subscribers and a responsive set of outputs developed by the nodes, comprising the steps of: Each of multiple nodes applies in a first order a first permutation in processing a first batch from plural senders; Messages of the first batch so processed delivered to respective recipients; Each of the nodes in the reverse of the first order applies an inverse of the first permutation in processing at least a second batch; and At least indications related to the respective messages and destinations processed as the second batch and provided to at least a portion of the respective senders.
(g) A multi-server private information retrieval system, the improvement comprising: each of multiple servers each shares two separate keystreams with each user device; at least one of the first keystreams is supplied by the user device to at least one of the servers; each server combines items inverted responsive to a corresponding first keystream of the user device by a group operation to produce a pre-output; each server combines the pre-output with the respective second keystream from the user device by a group operation to produce a respective encrypted individual server output; servers cooperate to combine individual encrypted server outputs by related group operations to produce a combined server output;
and the combined server output is supplied to the user device.
(g1) The retrieval system of claim G, including; a server hub broadcasting all items to multiple server front-end processors; and at least one of the server front-end processors combining, using a group operation, items broadcast, inverting items corresponding to said first keystream known to that front-end processor.
(g2) The retrieval system of claim G or G1, including said user device publishes untraceably a list of feed identifiers and receives a position indication.
(g3) The retrieval system of claim G, G1 or G2, including: a user device encrypting a data item, sending it for posting and obtaining an identifier for the posted encrypted item; the user device sending a message to at least one desired recipient of the data item, the message including keying information to allow decryption of the item and the identifier for the posted encrypted item; and the recipient using a respective device to send a keystream to a selected server with a desired position selected responsive to the identifier received.
(h) In a mixing system, the improvement comprising: an establisher entity forming mix-stage keys and providing each of plural mix nodes with a respective different one of the keys as a payload of a corresponding establishing mixing, one such establishing mixing for each mix-stage key; the mixes in each establishing mixing cooperating to mix the keys so that a corresponding mix receives the respective mix-stage key and the key, and the correspondence between sender and key, substantially kept from other mixes; a sender entity, optionally distinct from establisher entity, successively processing, where the processing is selected from the group including encrypting or decrypting, a payload using the mix-stage keys, as a layer and including along with the layer a fingerprint of the respective mix-stage key used with each such successive layer and providing the result to at least a first mix; at least a first mix receiving at least a partially such successively processed payload and the at least first mix locating the mix-stage key received earlier by matching the fingerprint of the key with the associated key previously received and then using the located key to process, where the processing is selected from the group including encrypting or decrypting, the layer and forward the result as an output; and at least a second mix receiving at least the output of at least an earlier mix and locating the mix-stage key received previously by matching the fingerprint of the key with the associated key previously received and then using the located key to process, where the processing is selected from the group including encrypting or decrypting, the layer and to forward the result to a receiving party.
(i) In a mixing system, the improvement comprising: nodes creating a coordinated component for a message during an unpermuted phase that includes values representing unique cryptographic functions of the sender identity, so that the sender identity results in a unique value unpredictable to other than the nodes; nodes process the created message along with plural components during a permuting phase and optionally including respective values representing substantially cryptographic functions of the recipient of the message, so that the resulting pseudonym is a substantially unique value corresponding to the sender recipient pair and substantially unpredictable other than to the nodes; and nodes processing the created message along with the other component(s) and revealing the combination of created values in output of the mixing system, whether public or delivered privately to a recipient.
40B and 40C are combination block diagram and flowchart of additive split mixing in accordance with aspects and teachings of the present invention, with
Turning to
A description adequate to allow those of skill in the art to make and use will now be provided but it will be understood that simplifications and restrictions may be included for clarity but without any limitation whatsoever.
Turning now to
The dotted rectangle indicates the mix hardware device extent. It may also be considered to be an example of the boundary of the mix network, such as in terms of a block diagram, with the inputs and outputs outside the boundary as already mentioned. The internal structure of the mix configuration is indicated as including an unpermuted section and a permuted section, as will be understood as an example. The input vector is shown on the left and the output vector on the right, as labeled. Similar conventions will be used to some extent in subsequent diagrams, as will be appreciated.
The message content, c, is shown indexed by a natural number, chosen for clarity as if the first user is user number one, the second, two, and so forth; but of course it will be understood that in general the users participating in a particular cascade may be whatever subset of subscribers and they may appear in whatever order, such as the order in which they send their input in. Thus, c(1) may be considered for clarity and concreteness to be as an example the message content, which is sometimes also referred to as the “payload,” of the message of the first user to send a message in for this cascade instance.
The input items, making up the left column as mentioned, are each raised to a power that is shown as the product of two applications of the function k. The first application includes the “x” as the second argument; the second application of k includes “y” as the second argument. The first parameter shown in both cases is the subscriber index as already mentioned.
Accordingly, k(1,x) is the key held as a secret in common between subscriber one and mix node x; and k(2,x) known to subscriber two and mix node x; similarly, k(4,y) for example, is that known to the fourth subscriber and node y.
The processing by node x is indicated by the arrows passing through the box labeled x. Thus, as will readily be appreciated, this node knows the keys it has in common with the subscribers, who are identified in this unpermuted phase, as indicated by the horizontal arrows, and the node cancels these factors, such as by multiplying by the multiplicative inverse of them, as will readily be understood. The node also applies the same key, x, to all the messages, as mentioned, thereby leaving the messages in a normalized form.
The processing by node y is similar, using the keys known to y. Accordingly, y raises the messages to the inverse power corresponding to the keys in common with the respective users and also includes the power y known to it. This leaves the content, c(i), sent by the subscribers but raised to both the x and they power.
Next the vector of such messages is processed, in this example, by node x, in the permuting phase. There are two operations performed by a node on the messages during this phase: permuting and decrypting. Accordingly, x chooses a permutation, ideally one that is hard for others to predict, such as a random permutation, for instance chosen independently and uniformly from all permutations of the number of messages used. The decrypting can be by raising the elements to the power that is the inverse of the key x. These two operations can be performed in whatever order with the same result.
The final operation of the permuting phase, in this example with two nodes, is the similar permuting and decrypting by the second node, as will be seen. Of course, any number of nodes can be used in the mix cascade in general, as is known and will be readily understood is applicable generally to all descriptions provided in the present application.
The result output by the mix, shown as the column on the right as mentioned, is the message content items or payloads sent by the subscribers, but in an order that represents the composition of the permutations chosen by each successive node.
As an example, for further concreteness as may be appreciated, the first input item shown, in the upper left, c(1){circumflex over ( )}k(1,x)k(1,y), represents the message content c(1) raised to a power that is the product of two exponents: k(1,x), the power common to the user and node x, and k(1,y), the power common to the same user and node y. Following this same message along, in the unpermuted phase, it appears in the first position of the intermediate vector between nodes x and y in the form: c(1){circumflex over ( )}k(1,y)x, showing that x has removed the common key and replaced it with the power x. Similarly, the output from y, the second intermediate vector, includes this message in the form: c(1){circumflex over ( )}xy, where the common key between user one and y has been cancelled and the power y injected to normalize the message. Following the message through processing by x, which as indicated includes the permutation that takes one into three, the message can be seen in the third position in the output vector as of the form c(1){circumflex over ( )}y. And, similarly, following it to the output vector, through processing by y, it appears in the fifth position as c(1).
Turning to
The input received by the mix, as described in box 220, includes encryption, such as commutative encryption in the examples, that includes keys held in common by a subscriber and a node. The next step shown 250 is the unpermuted phase, during which nodes know the subscriber or subscriber index and normalize by canceling the common keys while introducing the node keys, as already described with reference to
Turning now to
The example shown uses two nodes, x and y, for clarity; however, use of any number of nodes will readily be understood from the description, as already mentioned more generally. The example performs the precomputation in the forward directions, but other directions may be used, such as the reverse direction, as will be understood.
In the pre-computation, the first node x begins, by applying the homomorphic encryption in the unpermuted precomputation phase, shown in the upper left. Each of the five example unpermuted input what will be called here “slots” receives a random group element, shown as r(i,x), where i ranges over the input slot indices, as will be seen.
The homomorphic function, h(i), allows for the accumulation of elements by the group operation, in this case shown multiplicatively, as is well known. For instance, h(i)h(j) is equal to k(ij). The vector of random values or keys known to x, r(i,x), is the vector of homomorphic intermediate values shown intermediate to x and y. The next values included in or what may be called here “injected into” the homomorphic encryptions are the keys from node y, shown as r(i,y). Thus, the homomorphic encryption is of the component-wise product of the vector of keys from x and the vector of keys from y.
Next node x performs the first permutation of the permuted pre-computation. This rearrangement is indicated by the paths of the respective arrows within the rectangle labeled by x. The output of this, shown intermediate to the similar operation by y, is thus the permutation of the homomorphic encryption vector input with the component wise inclusion of a second vector of keys known to x, the s(i,x). The indices of the input vector can be seen to have been re-arranged in accordance with the arrows and the new vector included subsequently with its unpermuted indices, for clarity.
Then node y performs similar transformations, as just described for x, but with permutation and keys known toy. The vector of products under the homomorphic encryption is thus permuted again and has additional factor from y, the s(i,y).
The homomorphic encryption layer, shown as h( ), can as is well known be removed by cooperation of the nodes who have shared the secret key of the homomorphic system. In the well-known El Gamal homomorphic encryption system (its additive group operation being shown here multiplicatively for clarity), for example, the nodes cooperate to compute a public power of the agreed generator, by each node successively applying its secret exponent. Then, for the corresponding decryption of the homomorphic system, the nodes apply the inverse of their secret exponents successively to remove the encryption on each element of the vector. This is indicated by the vector passing unpermuted through the rectangles of each of the two nodes in the upper right corner. This computation or decryption operation, may, as will be appreciated, be conducted well in advance or, for instance, once the realtime phase completes successfully in some examples.
The realtime portion of the mixing is shown beginning with input items submitted by subscribers, as indicated on the left of the lower portion of the figure. Instead of the exponentiation already described with reference to
The elements included by the subscribers are, again similar to the system already described with reference to
These inputs to the realtime mix are shown then being processed by x in this first part of the unpermuted phase. The vector that x injects into the message vector at this point is the r(i,x) that were the first elements include in the homomorphic unpermuted phase already described. Also, x removes or what may be called here “cancels,” such as by applying the inverse group operation, the value of the respective k(i,x), the keys in common with x and the subscriber i. Thus, the k(i,x) factors are replaced by r(i,x) factors. Again, as will be appreciated has been detailed with reference to
The next processing, by y, is similar, as would be processing by any number of successive nodes. The common key vector is divided out and the corresponding r vector multiplied in (at least in the notation used, but the group operation may actually be addition, as mentioned).
Next, the realtime permuted phase begins, in the example shown for clarity with the first node x. As with the corresponding processing already described with regard to the pre-computation permuted phase, x permutes the input vector with the same rearrangement as in the pre-computation and performs a component wise group operation with the random vector, s(i,x). This leaves four factors in the intermediate value.
The operation performed by the next node, in this example y, is similar: the same permutation is applied to the input vector and the group operation is used to combine component-wise with the same random/key vector used in the pre-computation, in this example s(i,y). Again, any number of nodes will readily be understood, two having been shown as merely an example for clarity.
At this point the cancellation may be accomplished. (In other non-limiting examples, it may be accomplished at other points; this point being believed advantageous and shown for clarity.) It will be appreciated that the output of the homomorphic decryption includes all the factors of each message that remain, apart from the message content. Accordingly, forming the inverse in the group and combining by the group operation, shown as division in this multiplicative notion for clarity, leaves the message content in permuted order as the output vector.
As an example, for further concreteness as may be appreciated, the upper left corner shows x producing an output vector with first component h(r(1,x)) representing injection into the homomorphic system the key r(1,x) for the first slot. Then when y injects r(1,y), the vector contains h(r(1,x)r(1,y)) as first component. Of course, as will readily be understood, this can also be written as h(r(1,x)r(1,y)), because of the homomorphic property of the cryptosystem as already described, but the form shown is believed more readily understood.
Next the component is permuted as shown by x to the third position h(r(1,x)r(1,y)s(3,x)), and the additional key s(3,x) is injected. Once y permutes this to the sixth position (the permutations here being the same as those of
The realtime processing that leads up to this output is shown beginning again in the upper left of the realtime phase as the first component or slot of user input k(1,x)k(1,y)c(1), again with the index being assigned to match the slot number here for clarity in description and notation, but without any limitation whatsoever. (It will be understood that an assignment of senders to slots that is randomized in a way outside the control of less than all the nodes may be advantageous.) This is the product of the input message and the keys in common between this user and the two respective nodes. The output vector from x contains the first element as r(1,x)k(1,y)c(1), where the factor k(1,x) has been replaced on the left by r(1,x). Similarly, the processing by y yields r(1,x)r(1,y)c(1) in the first unpermuted component, with k(1,y) shown replaced in the same position for clarity by r(1,y). Now the permutation by x takes this component to the third position (same permutation as in the pre-computation) and includes s(3,x) as a factor to yield s(3,x)r(1,x)r(1,y)c(1). And finally, processing by y takes this to the fifth position and includes s(5,y) on the left, giving s(5,y)s(3,x)r(1,x)r(1,y)c(1). So, the fifth position of the output vector is the output of message c(1).
Turning to
The first step 450 of the pre-computation, as already described with reference to the example of
The next step 460 is the unpermuted and then the permuted phases of the pre-computation using the shared key to, in the example, develop the homomorphic encryption or otherwise of the cancellation vector, still in encrypted form, such as already described with reference to
In step 470 the combining and/or decryption to obtain the cancellation information, shown in the example of the homomorphism encryption, is before being applied to the messages in order to cancel the various keys included by the nodes that hide the messages during processing. The decrypted cancellation vector can be included at various points in the processing, such as already mentioned. For instance, including it at the end, as in the example described with reference to
As already described with reference to
The realtime processing of the input begins with the non-permuting phase 430, as already detailed in the example of
Finally, box 440 is the realtime permuting processing of the input messages. Each node processes, in the example, by permuting and injecting key material, with the example s key injection. When the cancellation vector is included, these remaining keys are cancelled, leaving the message content as output in the example.
Turning now to
It will again be appreciated that a network topology other than a sequence of hops may result in lower end-to-end delay, for instance, and thereby may be advantageous. It will also be understood that some users may be “closer” in terms of latency to different “entrance” nodes.
With reference to the example of the figure, the first three subscribers are shown supplying their input directly to node x, whereas the last two are shown supplying to node y. After processing by these first-line nodes, the messages are then routed through the other nodes so that all the processing as described, for instance with reference to
Although the example is shown with reference to the pre-computation example, it may be applied to other examples, such as those described with reference to
Turning now to
In the example two nodes are shown for clarity; however, broadcast to any number of nodes will readily be understood. It will be appreciated that a network topology of a star rather than a sequence of hops with the same number of nodes may result in lower end-to-end delay, for instance, and thereby may be advantageous.
Each node receiving the broadcast includes the key information, in the example r(i,x) or r(i,y), and removes the subscriber common keys it knows. However, and as will be appreciated, the common key is advantageously in effect removed also from the other copies when they are re-combined as shown. Thus, accordingly in such examples, the common keys enter in the inverse under the group operation with a multiplicity equal the number of copies, one for each node. For instance, as indicated in the example, the common key divided out by the node appears with multiplicity two, but one copy is shown in the numerator, so the net result is that the inverse appears with multiplicity one; but this is believed what is wanted to cancel the appearance of the factor when recombined as shown.
When the copies are recombined using the group operation, as shown, the result is that the common keys are cancelled and the key matter, r( ) in the example, included by the respective nodes remains. Thus, as will be appreciated, this result is suitable for processing by the realtime permuted phase, shown as a dotted rectangle labeled x,y and including the composition of the permutations used in other examples here.
Very concretely, as may be appreciated for clarity, the third component output by x, k(3,x)k(3,y)c(3)r(3,x)/k(3,x){circumflex over ( )}n, can be written includes k(3,x)k(3,y)c(3)/k(3,x) and so when it is multiplied by the third component of the output vector of y, k(3,x)c(3)r(3,y)/k(3,y), the k( )'s drop out and the result is the desired r(3,x)r(3,y)c(3), the same as shown as the third component of the input to the realtime permuted phase in
Turning now to
Referring specifically to
Referring specifically now to
It will readily be appreciated that variations include combinations of the approaches described with reference to
Turning now to
Referring specifically now to
In the example, each node receives separate authentication information from each subscriber. In some instances, such authentication may for example include cryptographic authenticators, such as for instance code-words, so-called “MAC”s, “digital signatures,” or whatever type of authenticator. These authenticators are shown using the notation: a(<sender>, <node>). For instance, the third line shows sender with identity #88 providing authenticator a(88,x) to node x. In the example the authenticators are removed by a node that is to check them, though this may be optional and may or may not actually result in an efficiency improvement and forwarding all authenticators to all nodes may be advantageous, such as because of simplicity or robustness in the face of changes in processing.
In some examples authenticators may be a single MAC based on the common key between the node and the sender; the authenticator may even be small compared to conventional prior art practice, as the authentication effect is believe to be cumulative across nodes in some examples. For instance, if there are twenty nodes and each authenticator is merely a single eight-bit byte, the combined effect may be considered to be equivalent to a one-hundred-sixty-bit authenticator by those who trust all nodes and an eighty-bit authenticator by those who trust whatever half of the nodes.
Referring now to
Turning to
The nodes establish common secrets with subscribers at least for other purposes as described here, such as with reference to
Referring now specifically to the figure, step 910 is the establishing of the keying information at least by the nodes that is known to pairs made up of a user and a node. Step 940 includes nodes receiving authenticators and checking the validity of the authenticators using the common keys corresponding to the sending subscribers. Then in step 970 nodes may optionally signal other entities, such as for instance other nodes, that a particular authentication has failed to check. In some examples, if authentication fails a message may simply not be forwarded, be replaced by some other information, or not be passed forward. As will be understood, not shown for clarity, is signaling that authentication is successful.
Turning now to
The diagrams show two example phases of mixing, unpermuted and permuted. They apply for instance to the embodiments described with reference to
As used elsewhere here, such as with reference to
Referring now more specifically to
Referring to
In the example illustrated, for clarity and concreteness, two inputs are shown from the same user, #16, one with a payload of c(1a) and the other with payload c(1b). The two inputs from the same user are mixed to different and essentially randomized positions in the output vector; however, because of the tag “T1*” the first field of each is treated as what will here be called a “marker,” shown as “z(1)” and “z′(1),” allowing the pair to be recognized and the two outputs to be grouped together, such as by the concatenation “c(1a) c(1b),” as will be understood. In some examples the outputs may be sorted and these fields may be in the higher-order positions and such pairs or larger sequences may appear as adjacent, particularly if they differ only in their lower order bits as sequence bits and are chosen randomly and of sufficient length to reduce the possibility of collisions, as will be understood. Thus, markers may be said here to “match” if they appear to relate to the same message at least with high likelihood.
Turning now to
Referring to step 1110, a subscriber forms a first message input to the mix that includes a marker that is at least unlikely to be chosen by another user instance, such as by choosing the marker at random. A second message with a related marker is also formed 1130 by a user. The messages are treated as inputs to a mix, in some examples for the same batch, and they are mixed 1150. The resulting messages in the output of the mix are linked 1170 by the related markers, such as by appearing adjacent in sort order as mentioned earlier. Finally, the linked message outputs are assembled and/or treated together 1190.
Turning now to
What are here called “untraceable return addresses,” such as those proposed by the present applicant in the original mix publications mentioned earlier, are known in mixing systems. The present example illustrates how they can be achieved in a novel aspect of the present novel mixing. An example use of such a technique is where the sender of a request wishes to remain anonymous to the recipient of the request, but still wishes to receive the reply.
Referring specifically first to
In the output, the same tag appears again to denote the type of mix, with win the clear, identifying this output instance so that in some examples the decrypted keys can be linked to it and used with it as a first payload as will be described with reference to
Referring now to
When the tag “T3” appears in the input, the mixes know to use the secrets that were encrypted under their respective public keys from the corresponding instance of
It is anticipated that payload content may be encrypted by the second subscriber, such as using keys also supplied by the first subscriber. It is further anticipated that a single first establishing instance can be used for multiple sendings.
Turning now to
The first user sends the return address through a mix 1320, to hide the correspondence between the user and the address, establishing the untraceable return address. Then the first user provides 1340 a second user with the address. Next, the second subscriber who has obtained the return address from the first subscriber uses it 1360 to form a mix input that includes both the return address and an actual message content. Once this input from the second user is mixed 1380 the recipient address and payload appear in the output.
It will be appreciated that among the advantageous properties achieved by the examples shown are believed to be: that the second user cannot learn the identity of the first user; that the first user does not need to anticipate the identity of the second user in Conning the untraceable return address; that no persistent public keys are maintained by the nodes across the various batches; and that no public key operations are performed in sending, mixing or delivering the message from the second subscriber.
Turning to
Referring more specifically to the example of
In another novel aspect, with more general applicability, but shown only here for clarity, instead of factors r(x) and r(y) being included in what is sometimes called the “base,” that is the value being raised to whatever power, factors denoted v(x) and v(y) are shown in the exponent of a modular arithmetic system. Accordingly, these factors are removed by raising to powers corresponding to their inverses during the permuted phase, as will be understood.
The result is the cleartext designator u, already described and to be described further, being output.
Not indicated for clarity as elsewhere here, as will be appreciated, are the additional factors or encryptions inserted during the non-permuting phase, as already described, such as with reference to
It will be appreciated that the user/subscribed being able to choose the designator while keeping confidential which designator which sender chooses but ensuring that no user/subscriber can repeat a designator, allows the designator to be chosen by the user/subscriber, for instance, as the name of an election or online forum or other association of data where such assurances are advantageous. For instance, voting a particular contest is typically to be limited to one vote per user and posting on a forum or group or whatever associated public or limited distribution data under a pseudonym in some circumstances is advantageously done while establishing that the user posts there under no other pseudonyms.
Technically, as will readily be appreciated by those of skill in the art, in some embodiments the users should not be able to choose the u values freely, but rather in a way that limits what can be known about the structure of these values in the number system. Such issues are well known in signature schemes and are solved, for instance, by including padding structure requirements in the number or requiring that the number be an image under other cryptographic functions and the like. A variety of ways are known and anticipated to allow the users to choose the values to encode the election or forum or the like without giving too much control over the structure of such numbers. As a concrete example, is can be defined as the image under a hash function of a unique encoding of text defining the actual election contest or forum or the like.
It will also, however, be appreciated related to
Referring now to
In one example application, users who wish to be in contact with each other only in the case that there is a mutual such interest can send messages with both user identities (e.g., concatenated in a fixed order such as lexicographic). If multiple items are posted as a result, replies can be generated. In some examples, replies can be by untraceable return addresses, such as those described with reference to
Turning now to
Referring now particularly to
It will be understood that the nodes may each “see” a different encryption of the designator, as they remove and/or apply encryption successively, as already mentioned or they may see the input data. Further processing 1530 in a permuted phase by the mix nodes further decrypts the designator. The output of the mix includes 1540 the decrypted designator and a count or the like if applicable.
Referring to
Turning to
Referring more specifically now to the example of
Referring to
Turning to
Referring specifically now to the figure, subscriber #75 uses tag “T6” to indicate transmission of what will be understood by those of skill in the art may be referred to as a “credential on a pseudonym.” This is made up of the two payloads shown for clarity: the first is the pseudonym itself; the second is the credential “power” or “signature” on the pseudonym.
At each unpermuted step, the nodes apply the same function fused to create the base pseudonym in processing the first component, as already described with reference to
During each permuted step, the nodes apply the x″′ and so forth to the first payload component and the second payload is decrypted using x and y, as in already described examples as will be understood. What is believed accordingly to be provided to the recipient is a pair of values, the second being the secret q power of the first.
The value q in the example is the secret exponent of the organization or entity issuing the credential, as will be understood by those of skill in the an. This issuer in the example makes public a power of an already public number, in the example shown, the x′″ and y′″ power of the public generator g. Thus, user #75 can form c, as already mentioned, and then apply c to this public value to obtain the second component shown transmitted.
The recipient organization would of course typically wish to verify that the two components received are in fact related by the exponent of the issuer, q. One way this can be accomplished, as will readily be apparent to those of skill in the art, is through an interactive proof with the issuer. In such a proof, the pair can even be hidden from the issuer, by so-called “blinding,” as will also be understood.
It is believed that in typical example applications the recipient may wish to what is sometimes referred to as “link” the various pseudonyms that arrive from the same user. One example way to accomplish this, as will be readily understood, is to include a further component in the transmissions and for it to be some form of authenticator of the user, such as a pass-phrase or signature or the like. In other believed advantageous examples, as will also be appreciated, a credential may be shown that is unique to the combination of issuer and user. Accordingly, for instance, the recipient chooses a unique exponent key for the user and issues the user a credential with it and then expects to see that power on the same base as already described for the other credential received in the transmission.
Turning now to
Referring to
Referring to
Turning to
Referring more specifically to
It will be understood by those of skill in the cryptographic transaction art that types of authenticators other than pre-images under a public one-way function may be employed for such authentication purposes. Examples include, public key digital signatures, undeniable signatures, hash function trees or chains, etc. It is believed that a one-way function may be more economical than a signature and adequate under some assumptions about the system; these functions will be used here in the description for concreteness and clarity but without any limitation whatsoever. In some examples, not shown for clarity, more than one such image or pre-image under the one-way function may fit in a single payload, and so more than one underlying operation can be performed per message. Also, as will be appreciated the choice of one-way function may encode denomination and/or currency.
An administrator of the accounts on the output side of the sending system is anticipated. This may be a entity separate from the nodes, related to the nodes, and/or with whatever multiplicity. It is anticipated that the account administrator, considered a single separate entity for clarity, maintains a list of accounts. Each account consists of at least some of: an identifier, an owner authentication, a balance a denomination/currency, and/or an indication of status such as open or closed. In some examples the identifier is the image under the one-way function (or the public key) related to the values used in establishing the account; the pre-image (or signatures) are then the authenticators used to move value from the account, whether to other accounts or to the subscribers, both as will be described. What will be referred to here as a “denomination” is the amount of value in the account. It is well known that a fixed binary denomination scheme, such as 2, 4, 8, and so forth units of value is efficient and can be enhance untraceability. Other denomination schemes are anticipated as are accounts with arbitrary value possibilities, such as integer number of cents, millicents, or even real number values.
The account(s) on the input side of the system, as will be understood, may be part of any known or existing type of financial or value system, such as bank accounts, digital currencies, or certificates of ownership. Moreover, as will readily be appreciated, whatever multiplicity and variety of such accounts may be provided. The accounts may be administered and/or owned and/or controlled by whatever entity or entities, such as the administrator of the output side, already mentioned, one or more nodes separately or jointly, and/or other entities for this purpose.
Referring now more specifically to
In some examples it may be required that the account exist before it can be funded and thus the transaction already described with reference to
Referring now to
In some applications of payments, withdrawals may be made directly by at least certain entities, such as shops, without the privacy of the protocol example described, as will be understood.
Also, in some applications the exception provision may be replaced by the use of a non-permuting tail technique and certified transactions that will be described with reference to
Turning now to
Referring to
The nodes process 2015 the request, such as by considering it a payload and passing it through the un-permuting and permuting phases already described, in order to hide the particular use requesting the particular account. Once the message appears in the mix output, the account is created 2020 with the corresponding identifier and optionally with the requested denomination. Any collision can result in no account being created and/or an exception, as already described with reference to
Referring to
Referring to
Turning to
Referring specifically now to
As will be understood, various conditions that may not hold can be assumed and/or dealt with in various ways. For instance, (a) the source account may have no value; (b) the destination account may have a different denomination; (c) the destination account may already be funded; and/or (d) the destination account may not exist. Various strategies for dealing with these error situations may readily be conceived. For instance: non-existent accounts may be created (d); denominations of accounts adjusted (b); or the whole transaction cancelled (a) or (c). An exception could be generated, such as already described with reference to
Referring specifically next to
Referring finally now to
As just one example variation, it will be understood that the initiator of a transfer or change transaction may wish to be notified in case it succeeds or in other examples if it fails. One way to accomplish this is using the non-permuting tail techniques that will be described with reference to
Turning to
The transactions are initiated by receipt of a request, or depending on the perspective, the sending of a request, as mentioned. The request includes 2220 designation of at least one source account and at least one destination account. Authentication is supplied 2240 for the at least one source account. The payload mixed includes 2240 identifiers for the at least one destination account. The step, accomplished after the mixing, of checking 2260 the authentication for the at least one source account, which as mentioned earlier may be assumed to allow ready determination of the corresponding account identifiers(s). Finally, value is transferred 2280 from the at least one source account to the at least one destination account, assuming the conditions for the transfer, such as those described with reference to
Turning to
Referring specifically now to
Processing recovers the payload, in a manner roughly similar for example to that for tag type T1 already described with reference to
An additional unpermuted processing by nodes, what may here be called a “tail,” is shown. The tail decrypts the received payload in stages, by the corresponding nodes successively removing the corresponding protection, as will be understood; however each also includes the k(,) respective encryption for the recipient already identified as #21. Thus, the message content can be delivered to the recipient, identified in the example output in cleartext, in a form that the recipient can decrypt, by re-constructing the k(,), and thereby recover the message content c(1).
Referring now to the exemplary embodiment of
In the tail processing shown, which is non-permuting, each successive node replaces its d( ) by a corresponding k(,) for the recipient #21. This is delivered to the recipient who is able to decrypt the k(,) factors and thereby recover the “#40” that identifies the sender.
As just one example to illustrate uses for a tail that includes the sender identity the option for a “certified” transfer between accounts has been mentioned earlier with reference to
As already mentioned with reference to
Turning now to
Relative to the example already described with reference to
As the messages travel through the stages of the three phases, the pair consisting of these two payloads travels together, as can be seen. In the first phase, which is unpermuted, the k(,) and k′(,) are exchanged for xy and x′y′, respectively. Then, in the permuting phase, the x,y are removed but the x′,y′ are replaced by x″,y″. This allows the next phase, the unpermuted tail, to deliver to the intended recipients using the cleartext recipient identifier revealed by the decrypted first payload; but with message content, c( ) that remains protected and at this stage by the application of k″(,) with the recipient identity as the first parameter.
Turning now to
Relative to the example already described with reference to the upper portion of
Turning now to
Relative to the example already described with reference to the lower portion of
The results of the homomorphic decryption, from
Turning to FIG. 27ABC, combination block diagrams and flowcharts of mixing with output delivery are shown in accordance with aspects and teaching of the present invention.
Referring specifically now to
The next three steps are the unpermuted, permuted, and unpermuted processing, such as for example described with reference to
Referring to
Referring finally now to
Turning to FIG. 28ABC, combination block diagram and cryptographic schematic diagrams of mixing with coordinated instances are shown in accordance with aspects and teaching of the present invention.
Referring more particularly to
The permutations realized by the respective nodes, x, y, and z, are shown as p(x), p(y), and p(z), respectively. The nodes use these in the optional pre-computation and in the processing, as already described.
Referring more particularly now to
Referring inure particularly to
The pre-unpermuted phase and the post-unpermuted phase may again each separately be considered optional in such an embodiment. Some example uses include for confirmation of transactions, such as those already described with reference to FIG. 19ABC, FIG. 20ABC, FIG. 21ABC, and
Taming to
Referring more particularly to
Referring more particularly now to
Turning now to
Referring to
Referring finally to
Turning to
Referring more specifically to the figure, two payloads are shown on the L.H.S. of the upper portion, making up the main portion of initial transmission already mentioned. The first payload contains an embedded tag, T10.2, that will label the components on the R.H.S. of this initial mix.
The two values following the tag each correspond to a respective destination for a copy of the multicast second payload. The if operator is intended to indicate that a so-called hash or other compression of the full untraceable return address, as already described with reference to
The output of the upper portion, on the R.H.S., indicates the copying into instances, each to be delivered using a corresponding untraceable return address. It will be appreciated that the t( ) values remain in place and will be replaced by each respective node during the unpermuted phase of the lower portion LHS.
In the example shown, the particular untraceable return address instance depicted for clarity is, arbitrarily chosen for the illustration, the one related to user #98. The full w, in
Turning now to
Referring more particularly to the figure, box 3220 is the sending, by the originating sender, of at least one message that includes both some kind of indication of the untraceable return addresses and some content to be copied to the corresponding recipients. Next, box 3240 shows that the mixing disasociates or unlinks the untraceable return addresses from the sender. This step is optional but believed desirable. For one thing, it may hide the number of copies sent; for another, it may hide the repeated use of a set of untraceable return addresses.
The next step in processing by the mix nodes is shown in box 3260 as including copying the content, typically still in protected form, such as already described with reference to
Turning to
The sender, user #77 in the example, chooses a key e, ideally in a way that is hard or infeasible for others to guess, as is well known. The sender then forms the transmission shown on the left hand side, including the tag T20 for this type of message. The next component is the public generator, g, raised to this key power, g{circumflex over ( )}e, encrypted for node x, with a single shared key k(x,1) in this example of this transaction type. Because of the large message stream cipher of this figure, the stream cipher encryption of c(2), is shown above the box; the short arrow between the unpermuted and permuted phase is intended to suggest this first stream encryption directly above it. The stream cipher is denoted by the plus symbol “+” for clarity, as will readily be understood, as by any suitable operation, such as for instance the often used exclusive-OR.
The penultimate component on the left has the addressee identifier as payload, p(#55). The final component contains as payload two exemplary alternate forms of keying material: the first is simply the secret exponent e itself; the other example option is a vector of the actual keys used to successively encrypt c(2), as will be explained further. The former is believed faster for the sender to compute but more time consuming for the recipient; the latter faster for the recipient but more time consuming for the sender. These last two components travel through the unpermuted and permuted and final unpermuted phases much as with output delivery, as already explained such as with reference to
The first node, x, receives the data from the sender and transforms it and forwards the result on to the second node. More particularly, the third from last component already described is decrypted by the first node x, after which x applies the example secret exponent x to it and uses the result as input to m that yields a value with the same size as c(2) that can be combined with c(2) as an additional stream cipher sequence. This transformation is indicated by the short arrow at the interface between the first and second halves of the permuted phase and relates to the result of the transformation shown above the box, as mentioned. The curly braces “{ }” are used to indicate that these quantities are ideally encrypted by keys unique to each adjacent pair of nodes in sequence. The node x further transforms the third from last component by raising to a further secret power shown as before transferring to the next node in sequence, y. This is believed to allow y to produce the next key stream but to keep y from being able to re-produce the key stream used by x.
The second node, y, receives the data front its predecessor and forwards on to its successor essentially as any subsequent node in the sequence would, as already described for x. This is also indicated by the ellipsis in the figure. For one thing the third from last component is encrypted, in this case using key y, and this is used to generate a key sequence to further stream encrypt the c(2) payload. For another, the third from last component is further encrypted by using y′ before being sent on, as already mentioned.
Once the messages are delivered to the recipient, user #55 in the example, the final component reveals the keying material to user #55 so it can recover c(2). In the first example option, more particularly, the value c is received. In this case, #55 can apply ƒ to e to recover the first key stream and raise the first public value g{circumflex over ( )}x (shown above the stream encrypted values) to the e power in order to obtain the seed for the second key stream and then raise the second public value to the e power to get the seed for the next stream. Removing the streams, such as by exclusive-OR as is well known, the message content c(2) is obtained by #55. In the second example option, the seeds for the various streams are simply computed by #77 instead of #55 and formed into the vector n(i) for #55 to use to recover c(2).
Turning now to
Referring more particularly to the figure, box 3420 is the creation of the key or keys by the sending subscriber. Then box 3440 shows the encryption of the content by the user with at least a part of the key matter; also, the sender transforms the seed so that it will not be readily computable by the first node.
Each node in sequence performs at least the encryption of the message portion of the mixing cascade, as described in box 3460. One operation related to this is to transform the seed received from the previous stage to produce the seed used for the encryption at this stage. Another operation, which may be optional but believed desirable, is to hide the value used for the encryption at this stage in what is forwarded on to the next node.
Referring finally to box 3480, after the cascade has successively encrypted the message, the recipient subscriber receives the seed from the send via the mix, as already described earlier, and uses it to re-construct the seeds and corresponding streams and to remove these from the message in order to recover the cleartext message.
Turning to
Referring more specifically to the figure, the left side shows the first or posting phase (as indicated by the numeral one reversed from the black disc). The user, in the example shown as #18, has what is denoted for clarity as a, an item of data, such as for example but without limitation chosen from the group including for instance a photograph, a sound clip, a video clip, a scan, a software program, a digital document, any or all in whatever format. The user “uploads” the data item, such as by sending it to a designated server or service over a computer network, but in an encrypted form. The example form of encryption shown for clarity uses an encryption function, e, with a first argument z that is a unique key of the user and the second argument that is the data item to encrypt, as will readily be understood.
Obtained by the sending user #18, in the example, is an identifier from a conveniently dense space of such identifiers, such as a sequence number or the like. For clarity and so as to fit on the page, a very few data items are shown in the illustration; typically, there might be from thousands to millions of such items. The identifier for the particular example uploaded data item is shown as the number three.
Referring now to the second phase, mixing with pre- and post-unpermuted phases is shown, such as has already been described, for instance with reference to
(As an illustrative example, for clarity but without any limitation, user #55 may, for instance, have received the message including some text, t, and an indication that there is a photo that can be obtained. The message may, not shown explicitly for clarity, contain and/or reference some kind of compact indication of the nature of the photo, such as an icon or low-res image. It may also typically be referred to in the text portion of the message. In any event, user #55 may not immediately try to obtain the photo, but rather may do so after some delay, such as after reading the text message.)
Referring to the third and final phase, where user #55 seeks to obtain the particular data item without revealing which of the data items that user has selected and will receive. To accomplish this, in outline, the user first forms a request, sends it to the servers x and y, two instead of a larger number being shown for clarity as will be appreciated, who operate according to it, providing a response to the user shown, and then the user is able to recover the encrypted data item and decrypt it using the already mentioned received key z. The request includes an indication of the user identity #55.
The request also includes what may be called here a “sent selection string,” shown as a literal binary string in square brackets, which is encrypted using a key shared with a selected one of the servers, in the example x. The sent selection string in the example illustrated is seven bits long, one bit for each of the identifiers in this as-already-mentioned small example for clarity. Each non-selected server recreates a corresponding selection string using its respective shared key with the identified user. Thus, the non-selected server y generates the seven bit string in the example, “0100100,” from the shared key in an agreed manner replicable by the user, as shown: applying k(y,55,i) for the ith bit of the string.
The user, as will be appreciated, also computes the exact same generated selection strings and then uses these to determine the sent selection string. This is done so that the bitwise-sum of all the selection strings, one per server, is all zero except that there is a one in the position corresponding to the identifier of the desired data, in this example the number three received. Put differently, the selection strings sum bitwise to the zero vector apart from the single set bit in the position of the desired data item.
As indicated in the figure, each bit of a selection vector is combined with the entire corresponding data item, which is shown for clarity using the symbol “,” This in the example group, without limitation for clarity, flips every bit of the data item if the selection bit is one and does not flip any of the bits if the selection bit for that item is zero; put differently, all the unselected items are forced to appear with even parity and the selected item with odd parity. Thus, every item—except the single one desired by the user—cancels, since each set bit of it enters the exclusive-or sum an even number of times; however, the third data item in the example is the result, since each of its bit positions with a set bit results in a set bit in the component-wise sum of all the vectors. Those data items that are inverted are shown with a check symbol, “0,” and are to be flipped; those without the check are not to be flipped. It will be appreciated that only the data item q(3) has an odd number of checks/flips, and so it is the only item not cancelled and appears in the sum in its entirety unmodified.
The bitwise exclusive-or sum, the group used without limitation in the example, is formed by the servers; however, included in the vector each server contributes is an ideally fresh key sequence generated from the shared key, as indicated by the k′ function. So user #55 also computes these key sequences and bitwise exclusive-or's them into the received vector, so that the user-generated and node generated pairs cancel each other, and the user recovers the encrypted data item, e(z,a), which the user then decrypts with the key z as already described as received.
Turning to
Referring specifically now to
Referring now to
The underlying technique for this is similar to that already described with reference to the third phase of
Not shown for clarity, but as will be understood, an example architecture for use by a node in forming the sum of the possibly many vectors of feeds (or of photos or the like as in the example already described with reference to
Turning now to
Referring specifically now to
Box 3720 indicates that the user supplies a server with the sent selection string to be in effect a substitute for that server's first keystream. Box 3730 describes how each server uses the first keystream, including the sent selection string, to select the items for the group operation that in effect causes all but the desired items to cancel. In the examples, this includes inverting or not inverting based on the whether the selection bit is zero or one. Box 3740 depicts each server combining the pre-output of the server with corresponding second keystream to produce the server outputs, which are each in effect encrypted with the respective shared keys. Box 3750 is the cooperation of the servers in forming the combination under the group operation of the server outputs to produce the combined encrypted server output. Box 3760 is the supply of the combined server output to the user and/or user device, which may for instance be accomplished by the servers forming a ring or tree structure and/or by cooperation with other nodes that combine separately received streams and forward them on to users.
Referring specifically now to
Turning now to FIG. 38ABC, a combination block diagram and flowchart of anonymous selected data and subsets of data and user computation is shown in accordance with aspects and teachings of the present invention. The process steps relate to the description already provided for
Referring to
Box 3830 indicates first that the second user receiving the message from the first user may wait some time after receipt of the message before sending a request, which includes a sent selection string, to a server. The sent selection string includes a desired position selected that is related to the identifier received by the second user.
Referring to
Referring to
Next is the sending of the request, including the sent selection string, to the particular server chosen for this purpose. One example way the server may be chosen, for instance, is that the choice is fixed for all users; another non-limiting type of approach would attempt to load balance and/or provide locality.
Finally box 3880 is first the receiving of the combined output from the servers or entity they cooperate with in forming this. Then the user device exclusive-or's, assuming this group operation for clarity, the received string bitwise with the received vector and the result should be the item, whether encrypted by the sender or not-encrypted by the nodes compiling the various feeds.
Turning to
The cryptographic embodiment described includes three main parts the inputs to the initial column of nodes; the transformation results by each respective node in each respective column; and the outputs provided by the final columns of nodes. Two alternate forms of input are shown, as are two alternate forms of the output; in both cases an alternate is indicted as an option, partly because the non-option case is believed best explained first for clarity.
Referring particularly now to the inputs, they are shown as having originated from users number twenty-three and eighty-five, respectively, as indicated by the user number prefix on the messages. The second component of each message is a sum, in a suitable finite group, such as for instance, an additive group modulo a large prime, for which additive notation is used in this figure for clarity. The second component is shown as a quotient, with the divisor equal the number of initial nodes. It is believed that when this quotient appears as a term three times, the sum that is the resulting group element will be the numerator. Thus, when the outputs of a column are summed per a respective input, two things are believed to occur: the result is the payload c(1); and the additive inverses of the shared keys shown, one per initial node, cancel the shared keys added in by the initial nodes.
Referring specifically to the first or initial mixing stage, the three example nodes, x, y and z, are shown each as a respective row. The output of each is shown enclosed in a box for clarity; the arrows indicate that these outputs are provided as inputs to the respective nodes of the next column or stage of mixing, as will be apparent. It will be appreciated that the individual user inputs are permuted in the outputs; however, the same permutation is used for the column. In this initial column example with only two inputs for clarity the non-identity permutation has been chosen. (This can be seen by noticing the subscripts on the payload and the shared keys.) Other terms are added in by the nodes; for clarity, these are shown as a, b, and -a-b, though whatever group elements that cancel may be used. In some example implementations, for instance, each node computes the permutation and the other terms from a seed known to all the initial nodes.
Referring to the second mixing stage, the results of the first stage are received by the respective nodes of the second column from the nodes of the first column, as already mentioned is indicated by the arrows. The processing of what is received by the second and subsequent columns is about the same as that by the initial nodes, apart from the shared key terms being omitted in the later stages. Thus, for clarity again as mentioned, the a, b, and -a-b terms are added in and each node performs the same permutation on the user input positions.
Referring to the outputs, now, it will be appreciated that it is believed all terms apart from the c( ) terms cancel. Again, the reason believed for this is that the various shared keys were re-constructed in their additive inverse form from the quotients in the input and the various additional terms were designed to cancel when respective entries are added for a column, which also applies to the last column. It will be appreciated that this example embodiment corresponds in input and output roughly to the basic mixing already described, with reference for instance to
Some exemplary input options will now be described further. The input option shown on the upper right-hand side of the figure is believed applicable where the initial nodes (or the final nodes in the ease of inverse/ack as described with reference to
Some exemplary output options will now be described further. In the example shown, a particular user or recipient is intended for each of the example messages. In such cases, the shared keys for that user can be included as terms by the respective nodes, as shown. Thus, the payload privacy can be protected as it is delivered to the user or user equipment once received, the user can remove the shared keys, such as by in this example subtracting them.
Examples of such encrypted delivery have already been described, such as with reference to
It will be appreciated that what may be called parallel or coordinated instances of messages, such as those occupying the same slot in two or more mixes that are run more or less at the same time with the same permutations, for instance as already described with reference to
Turning finally now to FIG. 40ABC, a combination block diagram and flowchart of additive split mixing is shown in accordance with aspects and teachings of the present invention. The process steps relate to the description already provided with reference to
Referring specifically now to
In box 420, other inverses are included to cancel unwanted multiplicities of the cleartext, also as already described. Box 430 indicates that the user sends the sum prepared as just described ideally as a single group element, but this may also be as multiple such elements, depending on the coding scheme (for instance, bitwise addition modulo two is also believed an example of a suitable group or direct product of groups.)
Box 4040 suggests that each initial node receives the same sending from the user, such as a single group element, instead of sending the elements separately to each node. However, some known systems, such as those proposed by Rabin and Rivest in a paper entitled “Efficient End to End Verifiable Electronic Voting Employing Split Value Representations” separately encrypt and provide that encrypted value separately to each entry node or the like. Here, nodes each “add in,” that is combine using group operations, the respective shared key or key sequence, in order to hide the payload, advantageously reducing bandwidth upload requirements by a factor believed equal the number of initial nodes.
Box 4050 and 4060 also relate to the known systems just mentioned, as will be understood. In particular values that cancel are added in and the same permutation within a column are used. Which entries in the output correspond to the same, yet unidentified, entry in the input, is known.
Box 4070 represents an optional improvement that reduces bandwidth requirements by again a significant factor. The outputs are summed by the nodes, or in cooperation with them, before being provided to at least some other entities.
Referring to
Referring to
Turning to
Two coordinated components are shown, much as in earlier described embodiments, such as those related to
Referring specifically now to the figure, in the unpermuted phase shown on the left, a second (lower) coordinated instance is introduced by the nodes without a corresponding input or payload from the sending user/subscriber. This component is shown as the product of factors, one supplied by each node, such as successively during this unpermuted phase. The function j(,) is used by each node to compute its factor in the example shown. In one exemplary embodiment this function is a cryptographic function of the first argument using a secret key known to the node whose name appears in the second argument. In the example shown, each node uses the sender subscriber number as the first argument and the resulting product is a function of the subscriber number.
After the permuting phase, the s″ factors have been removed by the nodes, as already explained, and the j′ factors are shown being substituted in for the t factors. The j′ function may be an independent function and/or use an independent key, to create separate factors related to the recipient subscriber. (In other examples, the j′ function may be the same as j, to give the same pseudonym no matter which direction the message is being sent.) The product of the j and j′ factors is shown as u, which is provided to the recipient. In the example shown, this is provided privately by way of the k″ factors; in other examples, it may be provided publicly, even though different recipient subscriber identities may be used to allow the same sender more than one public pseudonym.
Turning to
Referring specifically to the figure, and the first box 4220, mixes create and inject during an unpermuted phase a coordinated component. The values, which in the example are factors, injected into the coordinated component by the mixes include images under a cryptographic function that includes keying material accessible to the respective nodes, to produce a pseudonym that is at least very likely unique and believed ideally hard for other than the cooperation of nodes to recreate or check.
In box 4250, the nodes first perform a permuting phase on the components. Then at least some of the nodes include cryptographic functions of the recipient in the example as factors. Again, the cryptographic function includes keying material accessible to the respective nodes to produce a pseudonym that is at least very likely unique and believed ideally hard for other than the cooperation of nodes to recreate or check.
Box 4270 is the revealing of the created combined value pseudonym to the recipient, and/or to a wider audience.
Turning to
Referring specifically to
Referring specifically now to
Accordingly, as will be appreciated, the payload p shown on the right resulted from the final mixing stage, indicate by the vertical dotted line, that takes the element on its left, the encryption of p with key k(x) into the decryption p. The notation k(x,p) denotes the encryption of p with key where here as will be appreciated x stands for a key provided to node x in the establishing described with reference to
The structure shown to the immediate left of that just described indicates that the key known to node y was used to place a layer of encryption around that just described and concatenated with what may be referred to here as a “fingerprint” or hard to invert or cryptographic one-way function of the key that may be compact but ideally avoids collisions to at least a practical extent. The expression below indicates the essentially the same thing using square brackets to enclose the structure on the right, as will be appreciated. The sequence defined thus begins on the left with the outermost encryption layer formed using the key k(z) as indicated and its concatenated fingerprint.
In operation, the first node in this example, z, receives the data shown on the left of the leftmost dotted line and uses the fingerprint it calculated of the mix-stage key already received in cleartext, during the establishing as already described with reference to
It will be understood that the example shown for clarity includes mixes applying successive decryption of layers of encryption formed by the sender. In other examples, however, the mixes may actually apply encryption using the keys provided; the recipient would in such examples remove such layers of encryption. For instance, a sender may apply only a single layer of encryption to the payload and provide the recipient with the keys, or a suitable cryptographic method to generate them; layers are added by the successive mixes as the payload passes through them, but the recipient is able to remove the layers because of access to the mix-stage keys.
Turning finally now to
Referring first specifically to
Each mixing node is supplied one of the mix-stage keys. In the example already described with reference to
Box 4430 shows the mixes cooperating in performing the separate mixing to provide the mix-stage keys to the respective mixes without revealing the keys to other mixes and without revealing the linking between the establisher entity and keys. Moreover, the mix-stage keys may themselves be unlinkable, to whatever extent, as mentioned.
Referring finally specifically now to
Box 4460 finally shows a mix receiving a message as input and providing a corresponding output responsive to at least a mix-stage key previously reachieved, such as from the establishing already described. The mix uses the fingerprint associated with the encrypted portion of the input to find the corresponding key previously received. What may be called “matching the key” or locating it can be by looking up the fingerprint in a list or hash table or whatever data structure as already mentioned may have been used to save it. The key can be used to decrypt the layer (or in other examples mentioned to encrypt the layer) and then forward the result. One example of such forwarding would be to another mix node, such as might be indicated in a header not shown or mentioned elsewhere for clarity or that may be implicit in a fixed cascade ordering. As another example, the result may be the final payload that is published or includes an optional header indicating to whom it should be delivered.
It will be appreciated that the same channel established, such as described with reference to
Although the present invention has been described and illustrated in respect to exemplary embodiments, it is to be understood that it is not to be so limited, since changes and modifications may be made therein which are within the full intended scope of this invention as described in the background section and claimed.
The terms and expressions which have been employed in the foregoing specification are used therein as terms of description and not of limitation, and there is no intention, in the use of such terms and expressions, to exclude equivalents of the features shown and described or portions thereof, it being recognized that the scope of the invention is defined and limited only by the description in the background section and claims.
The description of the invention and its applications as set forth herein is illustrative and is not intended to limit the scope of the invention. Variations and modifications of the embodiments disclosed herein are possible, and practical alternatives to and equivalents of the various elements of the embodiments would be understood to those of ordinary skill in the art upon study of this patent document. For instance, the first node may combine the two operations since it can know the subscriber corresponding to a message, in some examples. In other examples, a more general mixing is anticipated through which messages travel differing routes and/or not in a single batch cascade and/or spread out over time in an at least partially randomized way. These and other variations and modifications of the embodiments disclosed herein may be made without departing from the scope and spirit of the invention.
This application claims priority to and the benefit of U.S. Provisional Application No. 62/133,456, entitled PRECOMPUTED AND TRANSACTIONAL MIXING,” (EFID 21771858, confirmation number 3691, filed 23:07:33), filed Mar. 15, 2015, the entire contents of which are incorporated by reference herein and as if copied in their entirety.
| Number | Date | Country | |
|---|---|---|---|
| 62133456 | Mar 2015 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 17530583 | Nov 2021 | US |
| Child | 18378276 | US | |
| Parent | 16439863 | Jun 2019 | US |
| Child | 17530583 | US | |
| Parent | 15558267 | Sep 2017 | US |
| Child | 16439863 | US |
