Method for generating multiplication triad through dual-mode switching under asynchronous Byzantine model
By switching the dual mode to generate multiplication triplets under the asynchronous Byzantine model, using the asynchronous double continuous value Byzantine protocol and the asynchronous completely secret sharing protocol, the problem of inconsistent number of multiplication triplets is solved, and the security of the secure multi-party computing system is improved. Communication efficiency reduces costs.
Patent Information
- Application Number
- CN202411141455.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-26
AI Technical Summary
The existing technology cannot generate multiplication triplets when switching temporary synchronous network models and fully asynchronous network models under the asynchronous Byzantine model, resulting in inconsistent number of multiplication triplets, increasing information leakage and calculation delay, reducing the security of secure multi-party computing systems and Communication efficiency.
The multiplication triple generation method is adopted under the asynchronous Byzantine model. By initializing the maximum execution time and safe pending buffer, the multiplication triple generation method is switched to the temporary synchronization and fully asynchronous network model, and the asynchronous double continuous value is used. Byzantine protocol and asynchronous fully secret sharing protocol ensure consistency and low communication overhead for multiplication triples.
When there are Byzantine nodes or large sending delays, ensuring the consistent number of multiplication triples, reducing the security and communication efficiency impact of secure multi-party computing systems, and generating random shares and multiplication triples through low communication overhead, reducing asynchronous The cost of safe multi-party calculations.
Smart Images

Figure CN119030691A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of secure multi-party computing, and particularly to a method for generating multiplication triples with dual-mode switching under an asynchronous Byzantine model. Background Art
[0002] In a secure multi-party computing system, a robust asynchronous offline phase is the most challenging component because it must efficiently preprocess random shares and multiplication triples and ensure the delivery of the output regardless of whether it is affected by an adversary. The online phase can use these multiplication triples to efficiently evaluate multiplication gates, while random shares are needed to efficiently evaluate randomized gates in the online phase, and clients can also use these shares to mask their private inputs. The network model in the asynchronous Byzantine model can be divided into a temporarily synchronous network model and a fully asynchronous network model. Therefore, it is necessary to study the methods for generating multiplication triples in the two modes.
[0003] In the prior art, the paper "Design and analysis of a distributed ECDSA signing service" proposed and analyzed a new protocol service for providing distributed ECDSA signatures, which works in an asynchronous communication model; cooperates with n parties, and Byzantine corruption is up to f < n / 3; provides guaranteed output delivery; provides a very efficient and non-interactive online signature phase; supports additional key derivation according to the BIP32 standard.
[0004] Currently, generating multiplication triples under the temporarily synchronous network model and generating multiplication triples under the fully asynchronous network model are carried out using different methods in different network models, and it is impossible to switch the methods for generating multiplication triples in the two models. However, the above prior art does not study or solve this problem. When generating multiplication triples in the two models separately, there may be a situation where the number of multiplication triples is inconsistent, increasing the risk of information leakage and calculation delay, and thus reducing the security and communication efficiency of the secure multi-party computing system. Summary of the Invention
[0005] This application provides a method for generating multiplication triples with dual-mode switching under an asynchronous Byzantine model to solve the problem in the prior art that it is impossible to switch the methods for generating multiplication triples under the temporarily synchronous network model and the fully asynchronous network model, which may lead to inconsistent numbers of multiplication triples, increasing the risk of information leakage and calculation delay, and thus reducing the security and communication efficiency of the secure multi-party computing system.
[0006] On the one hand, this application provides a method for generating multiplication triples with dual-mode switching under an asynchronous Byzantine model, including the following steps:
[0007] Step S1, each participant initializes a maximum execution time, an execution round and two safe pending buffers, and sets the two safe pending buffers to be empty, wherein the two safe pending buffers include a first safe pending buffer and a second safe pending buffer.
[0008] Step S2, in the current execution round, each participant executes the multiplication triplet generation method under the temporary synchronization network model; if the participant can generate the multiplication triplet before the maximum execution time, set the current execution round plus one, and repeat step S2.
[0009] Step S3: If any participant fails to generate a multiplication triple before the maximum execution time, each participant executes an asynchronous dual-continuous value Byzantine agreement with the current execution round minus one as input.
[0010] Step S4, when each participant reaches an agreement on generating multiplication triples under the temporary synchronous network model in step S3, each participant starts to execute the multiplication triples generating method under the completely asynchronous network model.
[0011] The temporary synchronous network model and the completely asynchronous network model are respectively a type of network model in the asynchronous Byzantine model.
[0012] In a possible implementation, step S2 includes:
[0013] Step S21, in the current execution round, each participant executes the multiplication triple generation method under the temporary synchronization network model.
[0014] Step S22: If the participating party can generate a multiplication triplet before the maximum execution time and the second safe pending buffer is not empty, the multiplication triplet in the second safe pending buffer is output; if each participating party can generate a multiplication triplet before the maximum execution time and the second safe pending buffer is empty, the multiplication triplet in the first safe pending buffer is copied to the second safe pending buffer, and the multiplication triplet generated by the participating party is copied to the first safe pending buffer.
[0015] Step S23, set the current execution round to be plus one, and repeat step S2.
[0016] In a possible implementation, step S3 includes:
[0017] Step S31, if any participant does not generate a multiplication triplet before the maximum execution time, each participant takes the current execution round minus one as input and executes an asynchronous dual-continuous value Byzantine protocol. After the asynchronous dual-continuous value Byzantine protocol stops, the participant obtains the output round from the asynchronous dual-continuous value Byzantine protocol.
[0018] Step S32, if the output round is equal to the current execution round minus two, discard the multiplication triplets in the two safe pending buffers at the same time; if the output round is equal to the current execution round minus one, discard the multiplication triplets in the first safe pending buffer and output the multiplication triplets in the second safe pending buffer; if the output round is equal to the current execution round, output the multiplication triplets in the two safe pending buffers.
[0019] Step S33, after step S32 is completed, each participant reaches an agreement on generating a multiplication triple under the temporary synchronization network model.
[0020] In a possible implementation, in step S2, the method for generating multiplication triples under the temporary synchronization network model includes:
[0021] Step S211, each participant generates a random share by calling the Shamir sharing algorithm.
[0022] In step S212, each participant generates a random double share by calling the double Shamir sharing algorithm and the double secret recovery algorithm.
[0023] Step S213: Generate a multiplication triple based on the random share and the random double share and call a batch secret recovery algorithm.
[0024] In a possible implementation, in step S4, the method for generating multiplication triples under the fully asynchronous network model includes:
[0025] Step S41, under the completely asynchronous network model, each participant generates a first random share by calling an asynchronous completely secret sharing protocol and a multi-valued Byzantine protocol.
[0026] In step S42, each participant calculates the secret and sets the evidence according to the first random share, and generates a multiplication triple by calling an asynchronous complete secret sharing protocol and a multi-valued Byzantine protocol.
[0027] In a possible implementation, step S41 includes:
[0028] Step S411, under the fully asynchronous network model, each participant calls an asynchronous fully secret sharing protocol to share a preset number of random secrets.
[0029] In step S412, each participant obtains the same preset number of asynchronous complete secret sharing protocols as other participants by calling a multi-valued Byzantine protocol, and obtains a first share in these asynchronous complete secret sharing protocols.
[0030] Step S413: Each participant calculates the Vandermonde matrix.
[0031] In step S414, each participant calculates a first random share, a commitment and evidence of the first random share based on the first share and the Vandermonde matrix.
[0032] Step S415: Each participant obtains a first random share.
[0033] Step S42 includes:
[0034] Step S421: Each participant calculates a secret based on the first random share.
[0035] Step S422, setting evidence of the participating parties according to the secret.
[0036] Step S423, under the fully asynchronous network model, each participant calls an asynchronous fully secret sharing protocol to share the secret, and broadcasts the evidence to all participants.
[0037] In step S424, each participant obtains the same preset number of asynchronous complete secret sharing protocols as other participants by calling a multi-valued Byzantine protocol, and obtains a second share in these asynchronous complete secret sharing protocols.
[0038] Step S425: Each participant calculates a second random share based on the second share.
[0039] Step S426: Each participant obtains a multiplication triplet based on the second random share.
[0040] In a possible implementation, step S412 includes the following steps:
[0041] Step S4121, set a first index set, and set the first index set to an empty set.
[0042] Step S4122: When each participant receives a secret share corresponding to a random secret in an asynchronous perfect secret sharing protocol, the index of the asynchronous perfect secret sharing protocol is added to the first index set.
[0043] Step S4123, when the first index set contains a preset number of elements, each participant calls a multi-valued Byzantine protocol and uses the first index set as input.
[0044] Step S4124, when the multi-valued Byzantine agreement is terminated, each participant obtains an identical first index set containing a preset number of elements and obtains the first share.
[0045] In a possible implementation, step S421 includes the following steps:
[0046] Step S4211: Each participant divides the first random shares into two equal groups, and calculates the secret based on the two groups of first random shares.
[0047] Step S4212, calculate zero-point evidence.
[0048] Step S4213, calculate and generate meta-evidence.
[0049] Step S4214, calculate the product proof using the product proof protocol.
[0050] In a possible implementation, step S424 includes the following steps:
[0051] Step S4241, set a second index set, and set the second index set to an empty set.
[0052] Step S4242: When a participant receives a secret share corresponding to a secret in an asynchronous perfect secret sharing protocol, the index of the asynchronous perfect secret sharing protocol is added to the second index set.
[0053] Step S4243, when the second index set contains a preset number of elements, each participant calls a multi-valued Byzantine protocol and uses the second index set as input.
[0054] Step S4244, when the multi-valued Byzantine agreement is terminated, each participant obtains an identical second index set containing a preset number of elements and obtains the second share.
[0055] The dual-mode switching method for generating multiplication triples under the asynchronous Byzantine model in this application has the following advantages:
[0056] When there are Byzantine nodes or large sending delays in the network, the asynchronous dual-continuous value Byzantine protocol is used to enable the participants to reach a consensus on generating multiplication triplets under a temporarily synchronous network model, avoiding the inconsistency in the number of multiplication triplets. At the same time, it switches to a completely asynchronous network model to generate multiplication triplets, avoiding the impact of the inconsistent number of multiplication triplets on the security and communication efficiency of the secure multi-party computing system.
[0057] Through the multiplication triplet generation method under the fully asynchronous network model, the asynchronous complete secret sharing protocol and the multi-valued Byzantine protocol are combined in both the first random share generation stage and the multiplication triplet generation stage, so that random shares and multiplication triples can be generated with lower communication overhead under the fully asynchronous network model, thereby reducing the cost of asynchronous secure multi-party computing. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0059] Figure 1 A schematic flow chart of a method for generating multiplication triples by dual-mode switching under an asynchronous Byzantine model provided in an embodiment of the present application;
[0060] Figure 2 A flow chart of step S2 of the method for generating multiplication triples by dual-mode switching under an asynchronous Byzantine model provided in an embodiment of the present application;
[0061] Figure 3 A flowchart of step S3 of the method for generating multiplication triples by dual-mode switching under the asynchronous Byzantine model provided in an embodiment of the present application. DETAILED DESCRIPTION
[0062] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0063] like Figures 1 to 3 As shown, the embodiment of the present application provides a dual-mode switching method for generating multiplication triples under an asynchronous Byzantine model, including the following steps:
[0064] Step S1, each participant initializes a maximum execution time, an execution round and two safe pending buffers, and sets the two safe pending buffers to be empty, wherein the two safe pending buffers include a first safe pending buffer and a second safe pending buffer.
[0065] Step S2, in the current execution round, each participant executes the multiplication triplet generation method under the temporary synchronization network model; if the participant can generate the multiplication triplet before the maximum execution time, set the current execution round plus one, and repeat step S2.
[0066] Step S3: If any participant fails to generate a multiplication triple before the maximum execution time, each participant executes an asynchronous dual-continuous value Byzantine agreement with the current execution round minus one as input.
[0067] Step S4, when each participant reaches an agreement on generating multiplication triples under the temporary synchronous network model in step S3, each participant starts to execute the multiplication triples generating method under the completely asynchronous network model.
[0068] The temporary synchronous network model and the completely asynchronous network model are respectively a type of network model in the asynchronous Byzantine model.
[0069] Specifically, multiplication triplets are used in secure multi-party computation to securely calculate the product of two secret shared values, which is extremely important in the privacy computing process. First, based on a pre-generated set of multiplication triplets, each participant calculates some intermediate values by subtracting their own data from these random numbers. Then, each participant exchanges these intermediate values so that they can jointly calculate the final product. In this process, due to the use of random multiplication triplets, the participants cannot infer the original data from the intermediate values, ensuring the privacy of the data. At the same time, because most of the calculations have been completed in advance, this method also reduces the complexity of real-time computing and communication.
[0070] Specifically, in this embodiment, before step S1, the asynchronous Byzantine model is firstly set up as follows: This model includes n=3t+1 participants, of which there are at most t malicious participants, and each participant randomly selects a random number. and publicly, among which is a finite field of order q.
[0071] Specifically, in this embodiment, step S1 includes: each participant P i Initialize a maximum execution time Δ, execute round r = 1 and two safe pending buffers pending1 and pending2, where pending1 is the first safe pending buffer and pending2 is the second safe pending buffer, and set the safe pending buffers pending1 and pending2 to empty sets, that is,
[0072] Exemplarily, step S2 includes:
[0073] Step S21, in the current execution round, each participant executes the multiplication triple generation method under the temporary synchronization network model.
[0074] Step S22: If the participating party can generate a multiplication triplet before the maximum execution time and the second safe pending buffer is not empty, the multiplication triplet in the second safe pending buffer is output; if each participating party can generate a multiplication triplet before the maximum execution time and the second safe pending buffer is empty, the multiplication triplet in the first safe pending buffer is copied to the second safe pending buffer, and the multiplication triplet generated by the participating party is copied to the first safe pending buffer.
[0075] Step S23, set the current execution round to be plus one, and repeat step S2.
[0076] Specifically, in this embodiment, step S21 includes: in the rth round, each participant P i Execute the multiplication triplet generation method under the temporally synchronized network model.
[0077] Step S22 includes: if the participant P i Ability to generate multiplication triples before maximum execution time Δ r , and the safe pending buffer pending2 is not empty, then output the multiplication triple in pending2; if the participant P i Ability to generate multiplication triples before maximum execution time Δ r , and the safe pending buffer pending2 is empty, the triples in the safe pending buffer pending1 are copied to the safe pending buffer pending2, and the multiplication triples Triples r Copied to the safe pending buffer pending1.
[0078] Step S23 includes: setting the execution round r=r+1, and repeating step S2.
[0079] Exemplarily, step S3 includes:
[0080] Step S31, if any participant does not generate a multiplication triplet before the maximum execution time, each participant takes the current execution round minus one as input and executes an asynchronous dual-continuous value Byzantine protocol. After the asynchronous dual-continuous value Byzantine protocol stops, the participant obtains the output round from the asynchronous dual-continuous value Byzantine protocol.
[0081] Step S32, if the output round is equal to the current execution round minus two, discard the multiplication triplets in the two safe pending buffers at the same time; if the output round is equal to the current execution round minus one, discard the multiplication triplets in the first safe pending buffer and output the multiplication triplets in the second safe pending buffer; if the output round is equal to the current execution round, output the multiplication triplets in the two safe pending buffers.
[0082] Step S33, after step S32 is completed, each participant reaches an agreement on generating a multiplication triple under the temporary synchronization network model.
[0083] Specifically, in this embodiment, step S31 includes: if any participant P i If a multiplication triple is not generated before the maximum execution time Δ, then each participant P i Take the execution round r-1 as input and execute an asynchronous double-continuous value Byzantine agreement. After the asynchronous double-continuous value Byzantine agreement stops, the participant P i The output round R can be obtained from the asynchronous dual-continuous value Byzantine agreement.
[0084] Step S32 includes: if the output round R=r-2, discard the multiplication triplets in the safe pending buffer pending1 and pending2 at the same time; if the output round R=r-1, discard the multiplication triplets in the safe pending buffer pending1 and output the multiplication triplets in the safe pending buffer pending2; if the output round R=r, output the multiplication triplets in the safe pending buffer pending1 and pending2.
[0085] Specifically, in this embodiment, step S4 includes: when in step S3, each participant P i After reaching an agreement on generating multiplication triplets under the temporarily synchronous network model, each participant begins to execute the multiplication triplet generation method under the fully asynchronous network model.
[0086] Exemplarily, in step S2, the method for generating multiplication triples under the temporary synchronization network model includes:
[0087] Step S211, each participant generates a random share by calling the Shamir sharing algorithm.
[0088] In step S212, each participant generates a random double share by calling the double Shamir sharing algorithm and the double secret recovery algorithm.
[0089] Step S213: Generate a multiplication triple based on the random share and the random double share and call a batch secret recovery algorithm.
[0090] Specifically, in this embodiment, step S211 includes: each participant P i Call Shamir's sharing algorithm (t) and receive the j Share Each participant P i Compute locally a Vandermonde matrix M with n rows and n columns, where the element in the lth row and jth column is Each participating party P i computes n random shares of degree t ([a1] i ,[a2] i ,…,[a n i ) T = M([s1] i ,[s2] i ,…,[s n i ); Participating party P l (t + 1 < l ≤ n) calls the secret recovery algorithm (l,[a l i ) and checks whether the share and the corresponding secret are the same and whether the corresponding polynomial degrees are the same. If they are not the same, stop the execution; otherwise, each participating party P i stores the random sharing [a1] i ,[a2] i ,…,[a t+1 i .
[0091] Repeat step S211. If the execution is successful, each participating party P i stores the random sharing [b1] i ,[b2] i ,…,[b t+1 i .
[0092] Step S212 includes: Each participating party P i calls the Shamir sharing algorithm (t) and the Shamir sharing algorithm (2t) and receives the shares from other participating parties P j Each participating party P computes an n×n Vandermonde matrix M locally, where the element in the l-th row and j-th column i Each participating party P computes n random shares of degree t ([r1] i ,[r2] i ,…,[r i n ) i = M([s1] T ,[s2] i ,…,[s i ,…,[s n i ) and n random shares of degree 2t Participating party P l (t + 1 < l ≤ n) calls the secret recovery algorithm (l,[r l i ) and the secret recovery algorithm Recover Secret l , and check whether the shares and the secrets corresponding to the shares are the same and whether the corresponding polynomial orders are the same. If they are not the same, stop the execution; otherwise, each participant P i Storage random double sharing
[0093] Step S213 includes: based on the random sharing [a1] generated above i ,[a2] i ,…,[a t+1 ] i and [b1] i ,[b2] i ,…,[b t+1 ] i , and random double sharing Perform the following calculation: Party P i Calculate [c k ] 2t =[a k ] i ×[b k ] i (k=1,…,t+1); Participant P i Calling the bulk secret recovery algorithm To restore c1-r1,c2-r2,…,c t+1 -r t+1 ;Participant P i Calculate [c k ] i =c k -r k +[r k ] i (k=1,2,…,t+1); Participant P i Output multiplication triplet
[0094] Specifically, the principle of Shamir's sharing algorithm (d) is as follows: Each participant P i Randomly select secret And select a random polynomial f of order d (the above Shamir sharing algorithm (t) is of order t, and the Shamir sharing algorithm (2t) is of order 2t) i (x) makes f i (0) = s i For each participant P j , send share in Indicates that it is to be sent to participant P j About Secrets i D-level sharing.
[0095] Secret recovery algorithm (i,[s i ]) is as follows: Each participant P j Send [s i ] j To the participant P i ; When receiving the shares [s i ] j When the participant P i Randomly select t+1 points for Lagrange interpolation to obtain the polynomial f i ′(x), and check f for all 2t+1 points i ′(j)=[s i ] j If it passes, output f i ′(0). Otherwise, at least one of the 2t+1 points is wrong, and we continue to wait for the remaining participants’ shares and re-execute the secret recovery algorithm.
[0096] Batch Secret Recovery Algorithm ([s1],[s2],…,[s t+1 ]) is as follows: This algorithm aims to recover the secret (s1, s2, …, s t+1 ): Participant P i Compute a local Vandermonde matrix M with n rows and t + 1 columns, where the element in the lth row and jth column is Participant P i Calculate ([y1],[y2],…,[y n ]) T =M([s1],[s2],…,[s t+1 ]); and send [y j ] to the participant P j (j=1,…,n); Participant P i Call the secret recovery algorithm (i, [y i ])Recover secret y i ;Participant P i Call the secret recovery algorithm (i, [y0]) to recover the secrets s1, s2, …, s t+1 .
[0097] Exemplarily, in step S4, the method for generating multiplication triples under the fully asynchronous network model includes:
[0098] Step S41, under the completely asynchronous network model, each participant generates a first random share by calling an asynchronous completely secret sharing protocol and a multi-valued Byzantine protocol.
[0099] In step S42, each participant calculates the secret and sets the evidence according to the first random share, and generates a multiplication triple by calling an asynchronous complete secret sharing protocol and a multi-valued Byzantine protocol.
[0100] Exemplarily, step S41 includes:
[0101] Step S411, under the fully asynchronous network model, each participant calls an asynchronous fully secret sharing protocol to share a preset number of random secrets.
[0102] In step S412, each participant obtains the same preset number of asynchronous complete secret sharing protocols as other participants by calling a multi-valued Byzantine protocol, and obtains a first share in these asynchronous complete secret sharing protocols.
[0103] Step S413: Each participant calculates the Vandermonde matrix.
[0104] In step S414, each participant calculates a first random share, a commitment and evidence of the first random share based on the first share and the Vandermonde matrix.
[0105] Step S415: Each participant obtains a first random share.
[0106] Step S42 includes:
[0107] Step S421: Each participant calculates a secret based on the first random share.
[0108] Step S422, setting evidence of the participating parties according to the secret.
[0109] Step S423, under the fully asynchronous network model, each participant calls an asynchronous fully secret sharing protocol to share the secret, and broadcasts the evidence to all participants.
[0110] In step S424, each participant obtains the same preset number of asynchronous complete secret sharing protocols as other participants by calling a multi-valued Byzantine protocol, and obtains a second share in these asynchronous complete secret sharing protocols.
[0111] Step S425: Each participant calculates a second random share based on the second share.
[0112] Step S426: Each participant obtains a multiplication triplet based on the second random share.
[0113] Specifically, in this embodiment, step S411 includes: in a completely asynchronous network model, each participant P i Call an asynchronous perfect secret sharing protocol to share N random secrets s i1 ,si2 ,…,s iN ,in B is the number of multiplication triplets. In the asynchronous complete secret sharing protocol, for other participants P j (j∈{1,…,n}), P i Send Share To the participant P j , where C ik Represents the participant P i The kth polynomial commitment of ik ] j , Indicates to send to P j Participants P i Secrets ik The t-order Shamir secret share, Indicates to send to P j About Shamir's Secret Share[s ik ] j , evidence.
[0114] Exemplarily, step S412 includes the following steps:
[0115] Step S4121, set a first index set, and set the first index set to an empty set.
[0116] Step S4122: When each participant receives a secret share corresponding to a random secret in an asynchronous perfect secret sharing protocol, the index of the asynchronous perfect secret sharing protocol is added to the first index set.
[0117] Step S4123, when the first index set contains a preset number of elements, each participant calls a multi-valued Byzantine protocol and uses the first index set as input.
[0118] Step S4124, when the multi-valued Byzantine agreement is terminated, each participant obtains an identical first index set containing a preset number of elements and obtains the first share.
[0119] Specifically, in the fully asynchronous network model, any participant cannot expect to wait for all n asynchronous fully secret sharing protocols to end. i In nt different asynchronous complete secret sharing protocols, the participant P receives the secret share corresponding to the random secret. iCall a multi-valued Byzantine agreement and take the index of these completed asynchronous complete secret sharing protocols as input. When the multi-valued Byzantine agreement terminates, each participant will obtain the same nt asynchronous complete secret sharing protocols and get its own share in these nt asynchronous complete secret sharing protocols, that is, the first share.
[0120] Specifically, in this embodiment, step S4121 includes: setting a first index set and will Set to an empty set, that is
[0121] Step S4122 includes: when the participant P i On the participant P j At the end of the asynchronous complete secret sharing protocol, the participant P i You will receive a secret share corresponding to the random secret And set
[0122] Step S4123 includes: when the first index set When contains 2t+1 elements, that is, When the participant P i by As input, call the multi-valued Byzantine agreement. Multi-valued Byzantine agreement is that each participant inputs a different first index set Finally, all participants reach a consensus on the first index set, that is, each participant obtains the same first index set
[0123] Step S4124 includes: when the multi-valued Byzantine agreement is terminated, each participant P i will get the same first index set containing 2t+1 elements And received the participants Share That is the first share.
[0124] Specifically, in this embodiment, step S413 includes: the participant P i Compute the Vandermonde matrix The jth row of the Vandermonde matrix M in
[0125] Specifically, in this embodiment, step S414 includes: based on the first share For k = 1, 2, ..., N, the participant P i Calculate the first random share ([r 1k ] i ,…,[r (t+1)k] i )←M([s 1k ] i ,…,[s (2t+1)k ] i ) T and Calculate the commitment of the first random share and evidence Where l=1,2,…,t+1.
[0126] In this embodiment, step S415 includes: the participant P i Get N(t+1) first random shares
[0127] Specifically, in a possible embodiment, the asymptotic complexity analysis of the first random share generated in step S41 is performed as follows: In the protocol for generating the first random share, a total of Θ(n) instances of the asynchronous complete secret sharing protocol need to be called. In each asynchronous complete secret sharing protocol, the batch size is Ω(nlogn), and the communication cost of n asynchronous complete secret sharing protocols is Ω(κn 3 logn), where κ is the security strength. In addition, a multi-valued Byzantine protocol needs to be called in the first random share generation protocol, and the communication cost of the protocol is Ω(κn 2 ). Therefore, the first random share protocol can be generated with a total communication cost Ω(κn 3 logn) batches to generate N(t+1) first random shares, which is equivalent to the amortized communication cost of each first random share being Ω(κn). Compared with the prior art which uses Ω(κn 2 ) generates random shares with the communication overhead, and the multiplication triplet generation method under the fully asynchronous network model of the present application reduces the communication overhead of generating random shares.
[0128] Exemplarily, step S421 includes the following steps:
[0129] Step S4211: Each participant divides the first random shares into two equal groups, and calculates the secret based on the two groups of first random shares.
[0130] Step S4212, calculate zero-point evidence.
[0131] Step S4213, calculate and generate meta-evidence.
[0132] Step S4214, calculate the product proof using the product proof protocol.
[0133] Specifically, in this embodiment, step S4211 includes: the participant P i Set the first B first random shares of the 2B first random shares to The next B first random shares are set to For each k∈[1,2,…,B], the participant P i Computing Secrets Randomly select secret And choose the t-order polynomial and Used to share secrets and in and
[0134] Step S4212 includes: based on the polynomial and Calculating Zero-Point Evidence
[0135] Step S4213 includes: calculating and generating meta-evidence and where g and h are cyclic groups of order q The generator of .
[0136] Step S4214 includes: calculating the product proof using the following product proof protocol
[0137] calculate and in Computational Challenges in is a collision-resistant hash function; calculate z1=e1+x[a k ] i , z3=e3+x[b k ] i , and Setting up product evidence
[0138] Specifically, in this embodiment, step S422 includes: setting the participant P i The evidence is
[0139] Specifically, in this embodiment, step S423 includes: the participant P i Call the asynchronous perfect secret sharing protocol to share the secret And proof i Broadcast to all participants P j (j∈{1,…,n}). Specifically, each participant P i Call an asynchronous perfect secret sharing protocol to share B secrets In the asynchronous complete secret sharing protocol, for other participants P j (j∈{1,…,n}), P i Send Share and proof i To the participant P j , where C ik Represents the participant P i The kth polynomial commitment of Indicates to send to P j Participants P i The Secret The t-order Shamir secret share, Indicates to send to P j About Shamir's Secret Share[s ik ] j , evidence.
[0140] Exemplarily, step S424 includes the following steps:
[0141] Step S4241, set a second index set, and set the second index set to an empty set.
[0142] Step S4242: When a participant receives a secret share corresponding to a secret in an asynchronous perfect secret sharing protocol, the index of the asynchronous perfect secret sharing protocol is added to the second index set.
[0143] Step S4243, when the second index set contains a preset number of elements, each participant calls a multi-valued Byzantine protocol and uses the second index set as input.
[0144] Step S4244, when the multi-valued Byzantine agreement is terminated, each participant obtains an identical second index set containing a preset number of elements and obtains the second share.
[0145] Specifically, in the fully asynchronous network model, any participant cannot expect to wait for all n asynchronous fully secret sharing protocols to end. i In nt different asynchronous complete secret sharing protocols, the participant P receives the secret share corresponding to the secret. i Call a multi-valued Byzantine agreement and take the index of these completed asynchronous complete secret sharing protocols as input. When the multi-valued Byzantine agreement terminates, each participant will obtain the same nt asynchronous complete secret sharing protocols and get its own share in these nt asynchronous complete secret sharing protocols, that is, the second share.
[0146] Specifically, in this embodiment, step S4241 includes: setting the second index set Y i , and i Set to an empty set, that is
[0147] Step S4242 includes: when the participant P i On the participant P j The asynchronous complete secret sharing protocol receives the secret share corresponding to the secret and proof i , participant P i Verify the received proof i If the proof i Correct, Party P i Save Share And set Υ i ←Υ i ∪{j}.
[0148] Step S4243 includes: when the second index set Y i When there are 2t+1 elements in |Υ i |=2t+1, the participant P i With Υ i As input, call the multi-valued Byzantine agreement. The multi-valued Byzantine agreement is that each participant inputs a different second index set Y i , and finally all participants reach a consensus on the second index set, that is, each participant obtains the same second index set Υ.
[0149] Step S4244 includes: when the multi-valued Byzantine agreement terminates, each participant P i will get the same second index set Υ containing 2t+1 elements, and receive the participant P j The share of (j∈Υ) That is the second share.
[0150] Specifically, in this embodiment, step S425 includes: based on the second share For k = 1, 2, ..., B, the participant P i Calculate the second random share where λ j is the value of the jth Lagrange interpolation polynomial at zero, that is
[0151] Specifically, in this embodiment, step S426 includes: each participant P i Get B multiplication triplets
[0152] Specifically, in a possible embodiment, the multiplication triples generated in step S42 are analyzed by asymptotic complexity as follows: In the protocol for generating multiplication triples, a total of Θ(n) times of generating the first random share protocol are required. In each generating the first random share protocol, the batch size is Set N = Ω(nlogn). In the protocol for generating multiplication triplets, it is necessary to call Θ(n) times to generate the asynchronous complete secret sharing protocol, and the batch size is B = Ω(n 2 logn), the communication cost of n asynchronous complete secret sharing protocols is Ω(κn 4 logn), where κ is the security strength. In addition, a multi-valued Byzantine protocol needs to be called in the protocol for generating multiplication triples, and the communication cost of the protocol is Ω(κn 2 ). Therefore, the protocol for generating multiplication triplets can be implemented with a total communication cost of Ω(κn 4 logn) batch generation B = Ω(n 2 logn)B multiplication triplets, which is equivalent to an amortized communication cost of Ω(κn 2 ). Compared with the prior art method of using Ω(κn 3 ) generates multiplication triples with the communication overhead, and the multiplication triples generation method under the fully asynchronous network model of the present application reduces the communication overhead of generating multiplication triples.
[0153] In the embodiment of the present application, when there are Byzantine nodes or large transmission delays in the network, the asynchronous dual-continuous value Byzantine protocol is used to enable the participants to reach a consensus on generating multiplication triplets under a temporary synchronous network model, thereby avoiding the situation where the number of multiplication triplets is inconsistent. At the same time, it switches to a completely asynchronous network model to generate multiplication triplets, thereby avoiding the impact of the inconsistent number of multiplication triplets on the security and communication efficiency of the secure multi-party computing system.
[0154] Through the multiplication triplet generation method under the fully asynchronous network model, the asynchronous complete secret sharing protocol and the multi-valued Byzantine protocol are combined in both the first random share generation stage and the multiplication triplet generation stage, so that random shares and multiplication triples can be generated with lower communication overhead under the fully asynchronous network model, thereby reducing the cost of asynchronous secure multi-party computing.
[0155] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0156] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A dual-mode switching method for generating multiplication triples under an asynchronous Byzantine model, characterized in that: The following steps are involved: Step S1, each participant initializes a maximum execution time, an execution round and two safety pending buffers, and sets the two safety pending buffers to be empty, where the two safety pending buffers include a first safety pending buffer and a second safety pending buffer; Step S2, in the current execution round, each participant executes the multiplication triple generation method under the temporary synchronization network model; If the participant is able to generate a multiplication triple before the maximum execution time, set the current execution round to be incremented by one and repeat step S2; Step S3, if any participant fails to generate a multiplication triple before the maximum execution time, each participant executes an asynchronous dual-continuous value Byzantine agreement with the current execution round minus one as input; Step S4, when each participant reaches an agreement on generating a multiplication triplet under the temporary synchronous network model in step S3, each participant starts to execute the multiplication triplet generation method under the completely asynchronous network model; The temporary synchronous network model and the completely asynchronous network model are respectively a type of network model in the asynchronous Byzantine model.
2. The dual-mode switching method for generating multiplication triples under the asynchronous Byzantine model according to claim 1 is characterized in that: Step S2 includes: Step S21, in the current execution round, each participant executes the multiplication triple generation method under the temporary synchronization network model; Step S22: if the participant can generate a multiplication triple before the maximum execution time and the second safe pending buffer is not empty, then output the multiplication triple in the second safe pending buffer; if each participant can generate a multiplication triple before the maximum execution time and the second safe pending buffer is empty, then copy the multiplication triple in the first safe pending buffer to the second safe pending buffer, and copy the multiplication triple generated by the participant to the first safe pending buffer; Step S23, set the current execution round to be plus one, and repeat step S2.
3. The dual-mode switching method for generating multiplication triples under the asynchronous Byzantine model according to claim 1 is characterized in that: Step S3 includes: Step S31, if any participant does not generate a multiplication triple before the maximum execution time, each participant takes the current execution round minus one as input and executes an asynchronous dual-continuous value Byzantine protocol. After the asynchronous dual-continuous value Byzantine protocol stops, the participant obtains the output round from the asynchronous dual-continuous value Byzantine protocol. Step S32, if the output round is equal to the current execution round minus two, the multiplication triples in the two safe pending buffers are discarded at the same time; if the output round is equal to the current execution round minus one, the multiplication triples in the first safe pending buffer are discarded and the multiplication triples in the second safe pending buffer are output; if the output round is equal to the current execution round, the multiplication triples in the two safe pending buffers are output; Step S33, after step S32 is completed, each participant reaches an agreement on generating a multiplication triple under the temporary synchronization network model.
4. The dual-mode switching method for generating multiplication triples under the asynchronous Byzantine model according to claim 1 is characterized in that: In step S2, the method for generating multiplication triples under the temporary synchronization network model includes: Step S211, each participant generates a random share by calling the Shamir sharing algorithm; Step S212, each participant generates a random double share by calling a double Shamir sharing algorithm and a double secret recovery algorithm; Step S213: Generate a multiplication triple based on the random share and the random double share and call a batch secret recovery algorithm.
5. The method for generating multiplication triples by dual-mode switching under the asynchronous Byzantine model according to claim 1 is characterized in that: In step S4, the method for generating multiplication triples under the fully asynchronous network model includes: Step S41, in a completely asynchronous network model, each participant generates a first random share by calling an asynchronous completely secret sharing protocol and a multi-valued Byzantine protocol; In step S42, each participant calculates the secret and sets the evidence according to the first random share, and generates a multiplication triple by calling an asynchronous complete secret sharing protocol and a multi-valued Byzantine protocol.
6. The method for generating multiplication triples by dual-mode switching under the asynchronous Byzantine model according to claim 5 is characterized in that: Step S41 includes: Step S411, under the fully asynchronous network model, each participant calls an asynchronous fully secret sharing protocol to share a preset number of random secrets; Step S412, each participant obtains the same preset number of asynchronous complete secret sharing protocols as other participants by calling a multi-valued Byzantine protocol, and obtains a first share in these asynchronous complete secret sharing protocols; Step S413, each participant calculates the Vandermonde matrix; Step S414, each participant calculates a first random share, a commitment and a proof of the first random share based on the first share and the Vandermonde matrix; Step S415, each participant obtains a first random share; Step S42 includes: Step S421, each participant calculates a secret according to the first random share; Step S422, setting the evidence of the participants according to the secret; Step S423, in a fully asynchronous network model, each participant calls an asynchronous fully secret sharing protocol to share the secret, and broadcasts the evidence to all participants; Step S424, each participant obtains the same preset number of asynchronous complete secret sharing protocols as other participants by calling a multi-valued Byzantine protocol, and obtains a second share in these asynchronous complete secret sharing protocols; Step S425, each participant calculates a second random share based on the second share; Step S426: Each participant obtains a multiplication triplet based on the second random share.
7. The method for generating multiplication triples by dual-mode switching under the asynchronous Byzantine model according to claim 6 is characterized in that: Step S412 includes the following steps: Step S4121, setting a first index set, and setting the first index set to an empty set; Step S4122: when each participant receives a secret share corresponding to a random secret in an asynchronous perfect secret sharing protocol, an index of the asynchronous perfect secret sharing protocol is added to the first index set; Step S4123, when the first index set contains a preset number of elements, each participant calls a multi-valued Byzantine protocol and uses the first index set as input; Step S4124, when the multi-valued Byzantine agreement is terminated, each participant obtains an identical first index set containing a preset number of elements and obtains the first share.
8. The method for generating multiplication triples by dual-mode switching under the asynchronous Byzantine model according to claim 6, characterized in that: Step S421 includes the following steps: Step S4211, each participant divides the first random shares into two groups, and calculates a secret based on the two groups of first random shares; Step S4212, calculating zero-point evidence; Step S4213, calculating and generating meta-evidence; Step S4214, calculate the product proof using the product proof protocol.
9. The method for generating multiplication triples by dual-mode switching under the asynchronous Byzantine model according to claim 6, characterized in that: Step S424 includes the following steps: Step S4241, setting a second index set, and setting the second index set to an empty set; Step S4242: when a participant receives a secret share corresponding to a secret in an asynchronous perfect secret sharing protocol, the index of the asynchronous perfect secret sharing protocol is added to the second index set; Step S4243, when the second index set contains a preset number of elements, each participant calls a multi-valued Byzantine protocol and uses the second index set as input; Step S4244, when the multi-valued Byzantine agreement is terminated, each participant obtains an identical second index set containing a preset number of elements and obtains the second share.
