A Multi-Agent Asynchronous Distributed Key Generation Method Based on Lightweight Cryptography

By employing hash-based asynchronous verifiable secret sharing and non-interactive threshold boosting techniques, this approach addresses the issues of high computational overhead in key generation and difficulty in verifying public key shares in multi-agent systems. It achieves efficient and unbiased multi-agent key generation, making it suitable for secure multi-agent collaborative scenarios.

CN123122800APending Publication Date: 2026-11-17BEIHANG UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611141193.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-30
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing asynchronous distributed key generation schemes for multi-agent systems suffer from problems such as high computational overhead, difficulty in verifying public key shares, and the possibility of malicious nodes submitting inconsistent public key shares. These schemes cannot simultaneously meet the requirements of lightweight computing, high-threshold reconstruction, asynchronous Byzantine robustness, and scalability of multi-agent systems.

Method used

It employs hash-based asynchronous verifiable secret sharing with integrity, non-interactive threshold lifting, post-lock verification mechanism, and high-threshold verification script aggregation. It verifies share consistency through hash digest and utilizes multi-valued asynchronous Byzantine fault-tolerant consensus and online error correction technology to generate efficient threshold public and private keys.

Benefits of technology

It enables batch generation of low-threshold random secret shares in an asynchronous Byzantine environment, reducing computational overhead, generating consistent and unbiased high-threshold public and private keys to meet high throughput requirements, and ensuring the correctness and consistency of key generation in the presence of malicious nodes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122800A_ABST
    Figure CN123122800A_ABST
Patent Text Reader

Abstract

This invention discloses a multi-agent asynchronous distributed key generation method based on lightweight cryptography, comprising: generating a uniformly extractable random share based on an intact asynchronous verifiable secret sharing within a multi-agent committee; then generating a high-threshold private key share through non-interactive threshold boosting; subsequently locking the public key share set through reliable broadcasting and multi-valued asynchronous Byzantine fault-tolerant consensus; then generating a hash challenge based on the locked set and aggregating high-threshold verification scripts; finally, filtering valid public key shares and interpolating to output the threshold public key, and passively replenishing unlocked broadcast instances through constant rounds when valid shares are insufficient. This invention achieves efficient threshold public-private key generation in a multi-agent asynchronous environment through hash-based intact asynchronous verifiable secret sharing, non-interactive threshold boosting, post-locking verification mechanism, and high-threshold verification script aggregation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of information security technology, specifically relating to a multi-agent asynchronous distributed key generation method based on lightweight cryptography. Background Technology

[0002] In multi-agent systems, multiple agents typically need to collaborate on security tasks such as authentication, signing, decryption, random number generation, and key management in open networks, cross-domain networks, or cloud-edge-device collaborative environments. To avoid single-point key leakage, distributed key generation technology uses multiple multi-agent nodes to collaboratively generate threshold public and private keys, ensuring that no single multi-agent node can independently obtain the complete private key, thereby improving the system's privacy protection and attack resistance.

[0003] In asynchronous network models, messages may be arbitrarily delayed or transmitted out of order, and some multi-agent nodes may be statically corrupted by Byzantine adversaries. Malicious multi-agent nodes can read their own input, forge or discard messages to be sent, and attempt to influence public key output or infer private key shares. Therefore, multi-agent asynchronous distributed key generation methods not only need to ensure that all honest multi-agent nodes eventually output a consistent public key, but also need to ensure that private key shares are not leaked, that the output public key is unbiased, and that a certain proportion of malicious nodes can be tolerated.

[0004] Existing asynchronous distributed key generation schemes typically rely on verifiable secret sharing based on homomorphic commitments to derive public key shares from secret shares and verify their correctness. While these schemes offer good verifiability, homomorphic commitments, zero-knowledge proofs, and related group operations introduce significant computational overhead, especially in batch key generation or high-threshold reconstruction scenarios. This results in substantial overall runtime and computational resource consumption, making them unsuitable for high-throughput multi-agent key management services. Another lightweight approach employs lightweight, complete, asynchronous verifiable secret sharing based on lightweight cryptographic primitives such as hashes to reduce computational costs. However, in high-threshold distributed key generation, these schemes still face challenges such as difficulty in verifying public key shares, susceptibility to bias during challenge generation, and the possibility of malicious nodes submitting inconsistent public key shares. Therefore, existing technologies struggle to simultaneously meet the requirements of lightweight computation, high-threshold reconstruction, asynchronous Byzantine robustness, batch processing capabilities, and scalability for multi-agent systems. Summary of the Invention

[0005] In view of this, the present invention proposes a multi-agent asynchronous distributed key generation method based on lightweight cryptography. It achieves efficient threshold public and private key generation in a multi-agent asynchronous environment by sharing hash-type, integrity-based, asynchronous verifiable secrets, non-interactive threshold boosting, post-lock verification mechanism, and high-threshold verification script aggregation.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a multi-agent asynchronous distributed key generation method based on lightweight cryptography, comprising: Public parameters and multi-agent node initialization phase: Establish a multi-agent committee and disclose the finite field, cyclic group, swarm generator, reconstruction threshold, batch size and security parameters, wherein the total number of committee nodes, the number of malicious nodes tolerated and the reconstruction threshold satisfy a preset asynchronous Byzantine fault tolerance relationship; Randomness sharing phase: Each node in the multi-agent committee distributes its local random secret through asynchronous verifiable secret sharing with hash-type integrity, and reaches a consensus on the fully received sharing instances, extracting a unified random vector share; Non-interactive threshold enhancement phase: Each node, based on a random vector share, uses polynomial order expansion to locally convert shares with low reconstruction thresholds into private key shares and blinding shares with high reconstruction thresholds; Public key share locking phase: Each node calculates the public key share and public blind value based on the private key share and blind share, and broadcasts them. The public key share set is locked through the consensus mechanism, so that all honest nodes output a consistent locked set. Verification script aggregation and public key output stage: After the lock set is formed, a hash challenge is generated. An aggregation verification script is constructed based on the hash challenge and the blinded share. The aggregation verification script is used to verify the legality of each public key share in the lock set. The verified valid public key shares are included in the valid set and the global public key and the public key shares of each node are interpolated and output.

[0007] Preferably, the hash-based asynchronous verifiable secret sharing with integrity uses hash digests instead of homomorphic commitments as the verification basis for share integrity, enabling nodes to verify the consistency of the received shares without performing group exponentiation and zero-knowledge proofs; each node adds the sharing instance index of the successfully output shares to its local set, and when the size of the local set reaches a preset threshold, it triggers multi-valued asynchronous Byzantine fault-tolerant consensus, outputting a unified set of valid indexes; each node then uses a super-invertible matrix to perform a linear transformation on the multiple shares corresponding to the valid index set, mapping multiple low-dimensional share vectors to a unified random vector share.

[0008] Preferably, in the non-interactive threshold enhancement stage, each node decomposes the random vector share into multiple sets of low-order polynomial components. These low-order components together span a high-order polynomial space. The order of the high-order polynomial is determined by the difference between the target reconstruction threshold and the number of malicious nodes tolerating. By calculating the value of this high-order polynomial at its own index, the node converts the secret share of the low security threshold into a private key share and a blinding share under the high security threshold in one go. The private key share is used for the final key generation, and the blinding share is used for information masking in subsequent public verification.

[0009] Preferably, during the public key share locking phase, each node sends the public key share obtained by mapping the private key share to the elliptic curve group and the corresponding public blind value through asynchronous reliable broadcast; asynchronous reliable broadcast ensures that the message sent by any honest node outputs the same content at all honest nodes; after collecting no less than a preset number of reliable broadcast outputs, each node performs multi-valued asynchronous Byzantine fault-tolerant consensus with the node index set corresponding to the output as input, and outputs the locked set.

[0010] Preferably, the hash challenge is a fixed-length value generated by binding the complete information of the lock set with the session unique identifier using a collision-resistant hash function after the lock set has been confirmed by the entire network, so that the challenge value cannot be predicted before the lock set is determined.

[0011] Preferably, in the verification script aggregation and public key output phase, each node uses a hash challenge to linearly weight the local blinded share to generate script components and sends them via multicast; after each node collects a sufficient number of script components, it uses online error correction to recover the complete aggregate verification polynomial from the data containing Byzantine errors; the aggregate verification polynomial is composed of multiple sets of recovered sub-polynomials combined in ascending order weights, so that the script components scattered across each node are aggregated into a unified public verification benchmark.

[0012] Preferably, the method for verifying the legality of each public key share in the locked set is as follows: the aggregate verification polynomial is applied to the elliptic curve generator to obtain the baseline verification point, and the hash challenge is applied to the public key share to be verified and the public blind value respectively, and homomorphically combined to obtain the verification point. By comparing whether the baseline verification point and the verification point are consistent, it is determined whether the public key share is located on the same high-order polynomial as the global secret, so that the verification process is completed without exposing any node's private key share; the public key share that passes the verification is included in the set of valid public key shares, and the public key share that fails the verification is discarded.

[0013] Preferably, when the number of shares in the set of valid public key shares is insufficient to meet the number of shares required for reconstruction, passive recovery is initiated: each node independently generates a temporary hash challenge for each unlocked reliable broadcast instance based on the public key share and public blind value of that instance, and verifies the unlocked instances one by one using the same script construction and verification method as the main process, and adds the verified shares to the set of valid public key shares.

[0014] Preferably, in passive recovery, independent temporary hash challenges are generated for different unlocked instances. The verification scripts for each instance are different, and the legality determination of each instance is executed in parallel until the number of valid shares meets the number of shares required for reconstruction, at which point the recovery terminates.

[0015] Preferably, the total number of committee nodes, the number of malicious nodes tolerated, and the reconstruction threshold satisfy the following conditions: the total number of nodes is equal to three times the number of malicious nodes tolerated plus one, and the number of malicious nodes tolerated is not greater than the reconstruction threshold, and the reconstruction threshold is not greater than the total number of nodes minus the number of malicious nodes tolerated minus one; by setting the batch size, each node extracts a component of length equal to the batch size from the random vector share in a single protocol execution, and generates multiple sets of threshold public and private keys in parallel.

[0016] The present invention has achieved at least the following beneficial effects: 1. This invention enables multi-agent committees to generate low-threshold random secret shares in batches in an asynchronous Byzantine environment through hash-based asynchronous verifiable secret sharing with integrity and lightweight randomness sharing.

[0017] 2. This invention uses a non-interactive threshold raising mechanism to convert low-threshold Shamir shares into high-threshold private key shares locally, avoiding communication overhead caused by additional interactions.

[0018] 3. This invention verifies public key shares without relying on homomorphic commitments through a post-lock verification mechanism and a high-threshold verification script aggregation mechanism, thereby generating consistent and unbiased threshold public and private keys.

[0019] 4. The present invention further handles the case of malicious multi-agent nodes submitting invalid public key shares through a constant round passive recovery path, so that the multi-agent committee can still collect a sufficient number of valid public key shares when there are message delays, malicious broadcasts and partial node failures, and finally output the public key and the public key shares corresponding to each multi-agent node.

[0020] Other advantages, objectives, and features of the invention will be set forth in the following description and will be apparent to those skilled in the art in some respects, or may be learned by practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0021] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration: Figure 1 This is an overall flowchart of the multi-agent asynchronous distributed key generation method based on lightweight cryptography of the present invention; Figure 2 This is a flowchart of the lightweight randomness sharing and public key share locking process of the present invention; Figure 3 This is a flowchart of the non-interactive threshold raising and high threshold verification script aggregation process of the present invention. Detailed Implementation

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] Asynchronous distributed key generation refers to the generation of keys by... A multi-agent committee composed of multiple agent nodes In the middle, at most This invention describes a cryptographic protocol where multiple agent nodes are corrupted by Byzantine forces and network messages can be arbitrarily delayed. The nodes collaboratively generate threshold public and private keys without relying on a trusted distributor. In this invention, the common input for asynchronous distributed key generation includes a finite field. Cyclic groups Group generator Number of multi-agent nodes Malicious node tolerance number Reconstructing thresholds and batch size ;node The private output is a share of the private key. The public output is the global public key. and the public key share of each node Asynchronous distributed key generation should satisfy correctness, consistency, confidentiality, and robustness, meaning that honest nodes output consistent public keys, adversaries cannot obtain additional information about the private keys, and the protocol can terminate in an asynchronous Byzantine environment.

[0024] To achieve the above objectives, this invention combines techniques such as hash-based asynchronous verifiable secret sharing with integrity, reliable broadcasting, multi-valued asynchronous Byzantine fault-tolerant consensus, online error correction, robust multinomial interpolation, Fiat-Shamir hash challenge, non-interactive threshold boosting, and high-threshold verification script aggregation, referencing... Figures 1-3 The technical solution is as follows.

[0025] This invention was first presented at the Multi-Agent Committee The system generates a uniformly extractable random share based on an asynchronous, verifiable secret sharing with integrity; then it generates a high-threshold private key share through non-interactive threshold lifting; next, it locks the public key share set through reliable broadcasting and multi-valued asynchronous Byzantine fault-tolerant consensus; then it generates a hash challenge based on the locked set and aggregates the high-threshold verification script; finally, it filters the valid public key shares and interpolates the threshold public key, and passively replenishes the unlocked broadcast instances through constant rounds when the valid shares are insufficient.

[0026] This invention addresses the scenario of multi-agent distributed key generation in asynchronous networks. The main scheme comprises seven stages: public parameter initialization, lightweight randomness sharing, non-interactive threshold boosting, public key share locking, high-threshold verification script aggregation, public key share verification output, and constant-round passive recovery. The basic components used in this method are introduced first, followed by a detailed explanation of the implementation process according to the stages and steps.

[0027] To achieve the above objectives, the present invention provides the following technical solution: 1. Basic Components 1.1 Asynchronous, verifiable secret sharing with hash-based integrity Asynchronous, verifiable secret sharing with hash-based integrity is used to enable distribution nodes to share secret vectors without using homomorphic commitments. Share with Multiple agent nodes, and make the nodes Obtaining shares This component should satisfy the following requirements: correctness, completeness, confidentiality, and integrity: when the distribution nodes are honest, the shares of all honest nodes lie on the same secret-sharing polynomial; if any honest node outputs, the other honest nodes will eventually output; at most... When a node is corrupted, the adversary cannot obtain additional information about the secret vector; a malicious node cannot make an honest node accept an inconsistent share.

[0028] 1.2 Reliable Broadcasting and Multi-Valued Asynchronous Byzantine Fault-Tolerant Consensus Reliable broadcasting ensures that the same broadcast instance outputs consistent content across all honest multi-agent nodes. Multi-valued asynchronous Byzantine fault-tolerant consensus enables all honest multi-agent nodes to reach a consensus on the index set, thereby ensuring that subsequent hash challenges are generated on the same lock set.

[0029] 1.3 Online Error Correction and Robust Polynomial Interpolation Online error correction is used for errors ranging from containing to... Recover a unique correct polynomial from a dataset of erroneous values. When a cumulative total of at least 100 erroneous values ​​has been received... When there are 1 data point, the node recovers the polynomial required for the verification script or public key interpolation through robust polynomial interpolation.

[0030] 1.4 Hash Challenge Generation The hash challenge generation adopts the Fiat-Shamir method, which generates a challenge vector based on the session identifier and the lock set after the object to be verified is locked.

[0031] 2. Public inputs, secret inputs, and outputs The common input of this method includes a finite field. Cyclic groups Group generator Number of multi-agent nodes Malicious node tolerance number Reconstructing thresholds Batch size Safety parameters and session identifier ,in and Each multi-agent node The secret input is a locally sampled random secret vector. And the local share obtained through an asynchronous, verifiable secret sharing with hash-type integrity. The private output of this method is a node. Private key share The public output is the global public key. and the public key share of each node .

[0032] 3. Steps of this method Phase 1: Initialization of common parameters and multi-agent nodes: including steps (1) to (2), initialization of the multi-agent committee and initialization of password parameters.

[0033] Step 1: Multi-agent committee initialization. The system establishes a multi-agent committee. , which includes Multi-agent nodes The system sets a tolerance limit for malicious nodes. and demand To support asynchronous Byzantine fault tolerance. Each node obtains its own index. Session identifier And communication authentication information.

[0034] Step 2: Initialize password parameters. The system exposes a limited field. Cyclic groups and generator And set a reconstruction threshold Batch size and safety parameters If a high threshold output is required, set it to... This is used for subsequent non-interactive threshold enhancement.

[0035] Phase 2: Lightweight randomness sharing phase: including steps (3) to (6), random secret sampling, low-threshold verifiable secret sharing, consensus of legal index set, and random vector share extraction.

[0036] Step 3: Random secret sampling. Each node Based on target length Sampling random secret vector ,in The amount of blinding required is determined by the batch size and the threshold increase.

[0037] Step 4: Low-threshold secret sharing. Each node Initiating an asynchronous, verifiable secret sharing with hash-based integrity, using itself as the distribution node, Share with other nodes. Any node. From node Output share in the sharing instance Then, record the index of the sharing instance. .

[0038] Step 5: Consensus on the set of valid indexes. Nodes Add the index of the shared instance whose share has been output to the local collection. .when At that time, node by For input, perform multi-valued asynchronous Byzantine fault-tolerant consensus, with all honest nodes outputting the same set. .

[0039] Step 6: Extract random vector shares. Node According to the set share Extracting random vector shares from superinvertible matrices and cut off the front Each element serves as local input for subsequent key generation.

[0040] Phase 3: Non-interactive threshold boosting phase: includes steps (7) to (10), share grouping, boosting polynomial construction, local share calculation, private key share and blind share extraction.

[0041] Step 7: Share Grouping. Node Low threshold random share Divided into Groups, respectively denoted as Each group corresponds to one Shamir shared his polynomials.

[0042] Step 8: Construct the lifting polynomial. The system defines the lifted high-threshold polynomial according to the following formula: Step 9: Calculate the high-threshold share locally. Node Without interacting with other nodes, directly calculate its own high-threshold share according to the following formula: Step 10: Extract the private key share and the blinding share. Node Will The former The elements are denoted as and set ; after The elements are denoted as This is used for blinding the subsequent verification script.

[0043] Phase 4: Public Key Share Locking Phase: Includes steps (11) to (13), public key share calculation and broadcasting, broadcast output collection, and lock set formation.

[0044] Step 11: Calculate and broadcast the public key share. Node Calculate public key shares and publicly blinded share And transmit the tuple via reliable broadcast. .

[0045] Step 12: Collect broadcast output. Node Each reliable broadcast output received That is, record index When the cumulative number of indexes reaches At that time, the node inputs the index set into the multi-value asynchronous Byzantine fault-tolerant consensus.

[0046] Step 13: Form a locked set. Consensus outputs a unified index set. Then, the honest nodes form a lock set. This is used to indicate that the set was fixed before the challenge was generated.

[0047] Phase 5: High-threshold verification script aggregation phase: including steps (14) to (17), challenge vector generation, verification script share calculation, online error correction and recovery, and verification script aggregation.

[0048] Step 14: Generate challenge vectors. Each honest node generates a challenge vector based on its session identifier. and locked sets Calculate the challenge vector: Step 15: Calculate the share of the verification script. Node For each calculate : Step 16: Online error correction and recovery script polynomial. Node multicast and upon receiving at least After grouping the data, perform online error correction and recovery. .

[0049] Step 17: Aggregate high-threshold verification scripts. The node generates the final verification script according to the following formula: Phase 6: Public Key Share Verification and Output Phase: Includes steps (18) to (20), including share verification, valid set maintenance, and public key interpolation output.

[0050] Step 18: Verify the locked public key share. Node For each in the locked set Check if the following expression is true: Step 19: Maintain a valid set. If the above verification is true, then the node... Will Add to the set of valid public key shares If the above verification fails, the public key share is discarded and the corresponding index is recorded.

[0051] Step 20: Interpolate and output the public key. When At that time, node pair In Perform interpolation on each valid public key share and output... Phase 7: Constant Round Passive Recovery Phase: Includes steps (21) to (24), recovery path triggering, unlocked instance challenge generation, instance verification script construction, and effective share replenishment.

[0052] Step 21: Trigger passive recovery. If the number of verified public key shares in the locked set is less than [a certain percentage]... If there are multiple nodes, then the nodes will wait in parallel for the output of the unlocked reliable broadcast instance. .

[0053] Step 22: Generate a separate challenge for the unlocked instance. For the unlocked index... Nodes according to Generate independent challenge vectors and according to Location in the list of unlocked indexes Choose standard basis vectors .

[0054] Step 23: Construct an instance verification script. Node In the instance tag Next multicast And restore the verification script corresponding to the instance through online error correction. .

[0055] Step 24: Verify and replenish valid shares. The node checks the following formula: If the verification is successful, then join in The above process is executed in parallel for multiple unlocked instances until the number of valid public key shares reaches a certain threshold. Then return to stage 6 to perform interpolation output.

[0056] Through the above embodiments, the advantages and effects of the present invention are: 1) By replacing homomorphic commitment-based secret sharing with hash-based asynchronous verifiable secret sharing that has integrity, the high-cost group operation and proof overhead are reduced, thereby reducing the computational overhead of batch distributed key generation; 2) Through a non-interactive threshold boosting mechanism, multi-agent nodes can obtain high-threshold private key shares from low-threshold shares by relying solely on local computation, reducing additional interactions and enhancing high-threshold reconstruction capabilities; 3) By using a lock-on verification and high-threshold verification script aggregation mechanism, the correctness of the public key share is verified without exposing the private key share, ensuring the consistency and unbiasedness of the output public key; 4) Malicious public key shares are handled by passive recovery path through constant rounds, enabling the multi-agent committee to complete protocol output under asynchronous network, Byzantine node and message delay conditions; 5) This invention is applicable to scenarios such as threshold signature, threshold decryption, random beacon, distributed key management and multi-agent secure collaboration, and can improve system throughput, reduce unit key generation latency and enhance scalability.

[0057] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A multi-agent asynchronous distributed key generation method based on lightweight cryptography, characterized in that, include: Public parameters and multi-agent node initialization phase: Establish a multi-agent committee and disclose the finite field, cyclic group, swarm generator, reconstruction threshold, batch size and security parameters, wherein the total number of committee nodes, the number of malicious nodes tolerated and the reconstruction threshold satisfy a preset asynchronous Byzantine fault tolerance relationship; Randomness sharing phase: Each node in the multi-agent committee distributes its local random secret through asynchronous verifiable secret sharing with hash-type integrity, and reaches a consensus on the fully received sharing instances, extracting a unified random vector share; Non-interactive threshold enhancement phase: Each node, based on a random vector share, uses polynomial order expansion to locally convert shares with low reconstruction thresholds into private key shares and blinding shares with high reconstruction thresholds; Public key share locking phase: Each node calculates the public key share and public blind value based on the private key share and blind share, and broadcasts them. The public key share set is locked through the consensus mechanism, so that all honest nodes output a consistent locked set. Verification script aggregation and public key output stage: After the lock set is formed, a hash challenge is generated. An aggregation verification script is constructed based on the hash challenge and the blinded share. The aggregation verification script is used to verify the legality of each public key share in the lock set. The verified valid public key shares are included in the valid set and the global public key and the public key shares of each node are interpolated and output.

2. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, Asynchronous verifiable secret sharing with hash-based integrity uses hash digests instead of homomorphic commitments as the basis for verifying share integrity, enabling nodes to verify the consistency of received shares without performing group exponentiation and zero-knowledge proofs. Each node adds the shared instance index of the successfully output share to the local set. When the size of the local set reaches a preset threshold, a multi-value asynchronous Byzantine fault-tolerant consensus is triggered, and a unified set of valid indexes is output. Each node then uses a super-invertible matrix to perform a linear transformation on multiple shares corresponding to the legal index set, mapping multiple low-dimensional share vectors to a unified random vector share.

3. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, In the non-interactive threshold enhancement phase, each node decomposes the random vector share into multiple sets of low-order polynomial components. These low-order components together span a high-order polynomial space. The order of the high-order polynomial is determined by the difference between the target reconstruction threshold and the number of malicious nodes that can tolerate it. By calculating the value of this high-order polynomial at its own index, the node converts the secret share of the low security threshold into a private key share and a blinding share under the high security threshold in one go. The private key share is used for the final key generation, and the blinding share is used for information masking in subsequent public verification.

4. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, During the public key share locking phase, each node sends the public key share obtained by mapping the private key share to the elliptic curve group and the corresponding public blind value through asynchronous reliable broadcast; asynchronous reliable broadcast ensures that the message sent by any honest node outputs the same content at all honest nodes; After collecting no less than a preset number of reliable broadcast outputs, each node performs multi-valued asynchronous Byzantine fault-tolerant consensus with the node index set corresponding to the output as input, and outputs a lock set.

5. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, The hash challenge is a fixed-length value generated by binding the complete information of the lock set with the unique session identifier using a collision-resistant hash function after the lock set has been confirmed by the entire network. This makes the challenge value unpredictable before the lock set is determined.

6. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, During the verification script aggregation and public key output phase, each node uses hash challenges to linearly weight its local blinded share to generate script components and sends them via multicast. After collecting a sufficient number of script components, each node uses online error correction to recover the complete aggregate verification polynomial from the data containing Byzantine errors. The aggregate verification polynomial is composed of multiple recovered sub-polynomials combined in ascending order weights, so that the script components scattered across each node are aggregated into a unified public verification benchmark.

7. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, The method for verifying the legality of each public key share in the locked set is as follows: the aggregate verification polynomial is applied to the elliptic curve generator to obtain the baseline verification point. At the same time, the hash challenge is applied to the public key share to be verified and the public blind value respectively, and homomorphically combined to obtain the verification point. By comparing whether the baseline verification point and the verification point are consistent, it is determined whether the public key share is located on the same high-order polynomial as the global secret. This allows the verification process to be completed without exposing any node's private key share. The public key shares that pass the verification are included in the set of valid public key shares, and the public key shares that fail the verification are discarded.

8. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, When the number of shares in the valid public key share set is insufficient to meet the number of shares required for reconstruction, passive recovery is initiated: Each node independently generates a temporary hash challenge for each unlocked reliable broadcast instance based on the public key share and public blind value of that instance, and uses the same script construction and verification method as the main process to verify each unlocked instance one by one, and adds the shares that pass the verification to the valid public key share set.

9. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 8, characterized in that, In passive recovery, independent temporary hash challenges are generated for different unlocked instances. The verification scripts for each instance are different, and the legality determination of each instance is executed in parallel until the number of valid shares meets the number of shares required for reconstruction, at which point the recovery terminates.

10. The multi-agent asynchronous distributed key generation method based on lightweight cryptography according to claim 1, characterized in that, The total number of committee nodes, the number of malicious nodes tolerated, and the reconstruction threshold are satisfied with the following conditions: the total number of nodes is equal to three times the number of malicious nodes tolerated plus one, and the number of malicious nodes tolerated is not greater than the reconstruction threshold, and the reconstruction threshold is not greater than the total number of nodes minus the number of malicious nodes tolerated minus one; by setting the batch size, each node extracts a component of length equal to the batch size from the random vector share in a single protocol execution, and generates multiple sets of threshold public and private keys in parallel.