Privacy protection based multi-agent communication method and device

Abstract

The application provides a multi-agent communication method and device based on privacy protection, and the method comprises the following steps: a first agent sends a function demand request carrying a function label to a trusted intermediary; after receiving a second temporary anonymous identifier of a second agent corresponding to the function label returned by the trusted intermediary, the first agent sends a communication request to the second agent based on the second temporary anonymous identifier; if a permission verification request returned by the second agent is received, a zero-knowledge proof for verifying the function permission of the first agent is sent to the second agent; and if a permission verification pass response returned by the second agent is received, an encrypted channel between the first agent and the second agent is established. The second agent is discovered through the function label, and the function permission of the first agent is verified through the zero-knowledge proof, so that precise matching between different agents is established without exposing the real identity labels of the agents, and the privacy protection in the communication process of different agents is improved.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a privacy-preserving multi-agent communication method and apparatus. Background Technology

[0002] With the widespread application of multi-agent systems in scenarios such as collaborative computing and distributed task processing, communication and interaction between agents are becoming increasingly frequent. The industry has raised higher requirements for identity security, access control, and data transmission privacy during multi-agent communication.

[0003] Existing agent communication technologies typically use publicly available real-name identifiers for agent retrieval and matching. This approach may lead to the exposure of agent identity information during the discovery and communication process, resulting in poor privacy protection during communication between different agents. Summary of the Invention

[0004] This invention provides a privacy-preserving multi-agent communication method and apparatus to solve the problem that the real identity of agents is easily exposed in the multi-agent communication process in the prior art. It realizes the establishment of accurate matching between different agents while hiding the real identity of the first agent and the second agent, thereby improving privacy protection in the communication process between different agents.

[0005] This invention provides a privacy-preserving multi-agent communication method, applied to a first agent, the method comprising the following steps: The first intelligent agent sends a functional requirement request carrying a functional tag to a trusted intermediary; After receiving the second temporary anonymous identifier of the second agent corresponding to the function tag returned by the trusted intermediary, a communication request is sent to the second agent based on the second temporary anonymous identifier, and the communication request includes the first temporary anonymous identifier of the first agent. If a permission verification request is received from the second agent, a zero-knowledge proof for verifying the function permissions of the first agent is sent to the second agent, and the zero-knowledge proof is used for the function permission verification of the first agent. If a permission verification pass response is received from the second agent, an encrypted channel is established between the agent and the second agent, wherein the encrypted channel is used for data transmission.

[0006] According to a privacy-preserving multi-agent communication method provided by the present invention, the permission verification request includes a function permission type and verification parameters, and the method further includes: Based on the verification parameters and the first random number, commitment data is generated; The commitment data is sent to the second intelligent agent; Receive the second random number returned by the second agent; The zero-knowledge proof is generated based on the second random number, the first random number, and the key corresponding to the function permission.

[0007] According to a privacy-preserving multi-agent communication method provided by the present invention, before receiving the permission verification request returned by the second agent, the method further includes: Receive a challenge value sent by the second intelligent agent, the challenge value being encrypted with the public key corresponding to the first temporary anonymous identifier; The challenge value is decrypted using the private key corresponding to the public key, and a hash operation is performed on the decrypted challenge value to obtain a hash result; The hash result is signed using the encrypted private key and then sent to the second intelligent agent. The hash result is used to verify the identity and legitimacy of the first intelligent agent.

[0008] According to a privacy-preserving multi-agent communication method provided by the present invention, after establishing an encrypted channel with the second agent, the method further includes: Receive a number of data packets sent by the second intelligent agent, at least some of which are disguised data packets; By filtering out the spoofed data packets from a number of the data packets, at least one real data packet is obtained; Parse each of the real data packets to obtain the service data of each of the real data packets; After removing the filler data from the business data, the data is reassembled to obtain the original business data.

[0009] According to a privacy-preserving multi-agent communication method provided by the present invention, at least some of the data packets are of different sizes and / or at least some of the adjacent data packets have different reception time intervals.

[0010] According to a privacy-preserving multi-agent communication method provided by the present invention, establishing an encrypted channel with a second agent includes: Send a request to a trusted intermediary to obtain routing information; Receive the routing node list sent by the trusted intermediary, the routing node list containing the node identifier, load status and threat level of several routing nodes; Based on the second temporary anonymous identifier, the routing node list is sent to the second intelligent agent; Based on the load and threat level of each routing node, negotiate the target routing node with the second intelligent agent; An encrypted channel is established between the target routing node and the second intelligent agent.

[0011] According to a privacy-preserving multi-agent communication method provided by the present invention, the method further includes: Check whether the first temporary anonymous identifier meets the update conditions; Once it is determined that the first temporary anonymous identifier meets the update condition, an update request is sent to the trusted intermediary. Receive the new temporary anonymous identifier returned by the trusted intermediary for the first intelligent agent.

[0012] This invention also provides a privacy-preserving multi-agent communication method, applied to a second agent, the method comprising: Receive a communication request sent by a first intelligent agent, wherein the communication request includes a first temporary anonymous identifier of the first intelligent agent; Based on the first temporary anonymous identifier, send an authorization verification request to the first intelligent agent; After receiving the zero-knowledge proof returned by the first intelligent agent, the functional permissions of the first intelligent agent are verified based on the zero-knowledge proof. After the first agent's functional permissions are verified, a permission verification successful response is sent to the first agent; Establish an encrypted channel with the first intelligent agent.

[0013] The present invention also provides a privacy-preserving multi-agent communication device, comprising: The first request sending module is used for the first intelligent agent to send a functional requirement request carrying a functional tag to a trusted intermediary. The second request sending module is configured to send a communication request to the second intelligent agent based on the second temporary anonymous identifier of the function tag corresponding to the second intelligent agent returned by the trusted intermediary, after receiving the second temporary anonymous identifier of the second intelligent agent returned by the trusted intermediary, wherein the communication request includes the first temporary anonymous identifier of the first intelligent agent. A zero-knowledge proof sending module is used to send a zero-knowledge proof for verifying the function permissions of the first intelligent agent to the second intelligent agent if a permission verification request is received from the second intelligent agent. The zero-knowledge proof is used for the function permission verification of the first intelligent agent. The first encrypted channel establishment module is used to establish an encrypted channel with the second intelligent agent if it receives a permission verification success response returned by the second intelligent agent, wherein the encrypted channel is used for data transmission.

[0014] The present invention also provides a privacy-preserving multi-agent communication device, comprising: A request receiving module is used to receive a communication request sent by a first intelligent agent, wherein the communication request includes a first temporary anonymous identifier of the first intelligent agent; The permission verification request sending module is used to send a permission verification request to the first intelligent agent based on the first temporary anonymous identifier; The permission verification module is used to verify the functional permissions of the first intelligent agent based on the zero-knowledge proof after receiving the zero-knowledge proof returned by the first intelligent agent. The permission verification pass response sending module is used to send a permission verification pass response to the first intelligent agent after the first intelligent agent's function permission verification is passed; The second encrypted channel establishment module is used to establish an encrypted channel with the first intelligent agent.

[0015] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the privacy-preserving multi-agent communication method as described above.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the privacy-preserving multi-agent communication method as described above.

[0017] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the privacy-preserving multi-agent communication method as described above.

[0018] This invention provides a privacy-preserving multi-agent communication method and apparatus. A first agent sends a function request carrying a function tag to a trusted intermediary. The first agent receives a second temporary anonymous identifier corresponding to the function tag from the trusted intermediary, enabling accurate retrieval of the second agent without relying on its real identity, thus avoiding the risk of identity information exposure at the source. Subsequently, the first agent sends a communication request to the second agent based on the second temporary anonymous identifier, and the communication request includes its own first temporary anonymous identifier. Upon receiving an authorization verification request from the second agent, the first agent generates a zero-knowledge proof to verify its own function permissions and sends it to the second agent. If the first agent receives a successful authorization verification response from the second agent, it indicates that its function permissions have been verified. This method of verifying function permissions through zero-knowledge proofs reduces the risk of the second agent leaking data to unauthorized agents, further ensuring accurate matching between agents. The encrypted channel established after successful authorization verification ensures the privacy and security of subsequent data transmission. Therefore, the entire process achieves accurate matching and secure communication between different agents while hiding the real identities of the first and second agents, improving privacy protection during communication between different agents. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is one of the flowcharts of the privacy-preserving multi-agent communication method provided by the present invention.

[0021] Figure 2 This is the second flowchart of the privacy-preserving multi-agent communication method provided by the present invention.

[0022] Figure 3 This is the third flowchart of the privacy-preserving multi-agent communication method provided by the present invention.

[0023] Figure 4 This is a schematic diagram of the structure of the privacy-preserving multi-agent communication device provided by the present invention.

[0024] Figure 5 This is a schematic diagram of another privacy-preserving multi-agent communication device provided by the present invention.

[0025] Figure 6This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0027] The following is combined with Figures 1 to 3 This invention describes a privacy-preserving multi-agent communication method. Figure 1 This is one of the flowcharts illustrating the privacy-preserving multi-agent communication method provided by the present invention, such as... Figure 1 As shown, the method includes the following steps: Step 101: The first intelligent agent sends a functional requirement request carrying a functional tag to the trusted intermediary.

[0028] in, Figure 1 The privacy-preserving multi-agent communication method shown can be applied to the first agent. The first agent is an intelligent execution entity with data processing and communication capabilities, which can initiate cross-agent cooperation requests.

[0029] The trusted intermediary is a centrally authenticated service node that integrates trusted directory services and key management services, serving as the trusted scheduling center for multi-agent communication. Specifically, a first agent can send a function request carrying a function tag to the trusted intermediary by sending the request to a trusted directory service node within the intermediary. The trusted directory service node is a sub-service node within the trusted intermediary, and each node stores a mapping table of "temporary anonymous identifier - function tag - encrypted communication address".

[0030] Functional tags are identifiers used to characterize the service capabilities, business types, or processing permissions possessed by an intelligent agent, such as "data computation," "model inference," and "resource scheduling." Functional requirement requests are used to declare the types of collaborative capabilities required by the first intelligent agent to the trusted directory service node.

[0031] Step 102: After receiving the second temporary anonymous identifier of the second agent corresponding to the function tag returned by the trusted intermediary, a communication request is sent to the second agent based on the second temporary anonymous identifier.

[0032] The communication request includes the first temporary anonymous identifier of the first agent.

[0033] Optionally, at least three trusted directory service nodes are deployed in the trusted nodes, and a Byzantine fault-tolerant protocol can be used to ensure consistency. Each trusted directory service node stores a mapping table of "temporary anonymous identifier - function tag - communication address (encrypted)," where the communication address uses 256-bit Advanced Encryption Standard (AES), and the key is held only by the trusted directory service node and the corresponding agent. The trusted directory service node searches the mapping table according to the function tag and filters out matching temporary anonymous identifiers. In some application scenarios, the matching temporary anonymous identifier can be encrypted using the public key of the first agent and returned. After the first agent decrypts it, it obtains the second temporary anonymous identifier for the second agent.

[0034] The second agent is a collaborative agent possessing the capabilities represented by the functional label. The second temporary anonymous identifier is an identifier used by the second agent for communication addressing without revealing its real identity. The first temporary anonymous identifier is an identifier used by the first agent for communication addressing without revealing its real identity. The first agent initiates communication based on the second temporary anonymous identifier, enabling agent addressing without relying on real identity information.

[0035] The method of sending a communication request to the second agent based on the second temporary anonymous identifier can be as follows: 1. Send the communication request and the second temporary anonymous identifier to a trusted intermediary. The trusted intermediary queries the communication address corresponding to the second temporary anonymous identifier stored in the trusted directory service node, and then sends the communication request to the second agent; or the trusted intermediary returns the second temporary anonymous identifier and the communication address of the second agent at the same time. The first agent directly sends the communication request to the second agent based on the communication address of the second agent.

[0036] Step 103: If a permission verification request is received from the second agent, the zero-knowledge proof used to verify the functional permissions of the first agent is sent to the second agent.

[0037] The permission verification request is used by the second intelligent agent to verify whether the first intelligent agent possesses the corresponding business function permissions. Zero-knowledge proofs are used for the first intelligent agent's function permission verification. Zero-knowledge proofs can prove that the first intelligent agent possesses the corresponding function permissions without disclosing the first intelligent agent's privacy information or the plaintext of the permission credentials.

[0038] Step 104: If a permission verification success response is received from the second agent, establish an encrypted channel with the second agent.

[0039] The encrypted channel is used for data transmission. It is a secure communication channel constructed using symmetric or asymmetric encryption algorithms, used for business data transmission between the first and second intelligent agents to prevent data from being stolen, tampered with, or eavesdropped on during transmission.

[0040] In the above scheme, the first agent sends a function request carrying a function tag to a trusted directory service node. It then receives a second temporary anonymous identifier corresponding to that function tag from a trusted intermediary, enabling accurate retrieval of the second agent without relying on its real identity, thus avoiding the risk of identity information exposure at the source. Subsequently, the first agent sends a communication request to the second agent based on the second temporary anonymous identifier, which includes its own first temporary anonymous identifier. Upon receiving a permission verification request from the second agent, the first agent generates a zero-knowledge proof to verify its own function permissions and sends it to the second agent. If the first agent receives a permission verification pass response from the second agent, it indicates that its function permissions have been verified. This method of verifying function permissions through zero-knowledge proofs reduces the risk of the second agent leaking data to unauthorized agents, further ensuring accurate matching between agents. The encrypted channel established after successful permission verification ensures the privacy and security of subsequent data transmission. Therefore, the entire process achieves accurate matching and secure communication between different agents while hiding their real identities, improving privacy protection during communication between different agents.

[0041] In one possible embodiment, the privacy-preserving multi-agent communication method further includes: The first agent sends a pre-registration request to a trusted intermediary, which includes the agent's application materials. The agent then receives the trusted intermediary's returned real identity identifier, first temporary anonymity identifier, category of functional permissions, and corresponding key for each functional permission.

[0042] The application materials may include at least one of the following: agent hardware information, domain, functional permission scope, and superior control node information. The trusted intermediary can verify the hardware's legitimacy by connecting the hardware information to the hardware manufacturer's database. The trusted intermediary can also verify the agent's permissions based on its domain and functional permission scope; that is, the trusted intermediary confirms the agent's permission scope based on the joint signature of trusted nodes within the domain. The superior control node information provides node endorsement for the agent; trusted nodes verify the agent's node endorsement through the data signature submitted by the superior control node. In other words, the trusted intermediary performs triple verification of the agent through hardware information verification, permission qualification verification, and node endorsement.

[0043] After successful verification, the trusted intermediary generates dual identity identifiers using a combination of "asymmetric encryption algorithm + hash algorithm". For example, the trusted intermediary first generates 256-bit original identity information based on the first agent's unique hardware identifier (such as CPU serial number or MAC address) and its assigned unique code using a hash algorithm (Secure Hash Algorithm, SHA). The hash algorithm could be SHA-3. Then, an asymmetric encryption algorithm (Rivest-Shamir-Adleman, RSA) is used to encrypt the original identity information, generating the agent's true identity identifier (serving as the agent's unique trusted credential within the system, stored only in the trusted intermediary's encrypted database and not transmitted externally). Specifically, the asymmetric encryption algorithm could be RSA-4096. The first temporary anonymous identifier of the first intelligent agent is dynamically generated using a combination of random number generation and symmetric encryption. For example, a 512-bit random number is first generated using a pseudo-random number generator based on chaos theory (such as a chaotic map), and then the random number is encrypted using the AES-256 symmetric encryption algorithm (the key is uniformly distributed by a trusted intermediary and updated periodically) to generate the first temporary anonymous identifier. Simultaneously, a one-to-one mapping relationship is established between the real identity identifier and the first temporary anonymous identifier. This mapping relationship is stored only in the encrypted storage unit within the trusted intermediary, and a fragmented storage method is used (fragment keys are kept separately by different trusted nodes, requiring multi-node collaborative verification to decrypt the mapping relationship). The real identity identifier of the first intelligent agent is entered into the trusted intelligent agent's whitelist. The whitelist uses a chained storage structure, and each whitelist entry is accompanied by a dual digital signature from the trusted intermediary and the endorsing node to prevent tampering with the whitelist. Simultaneously, the whitelist is synchronized in real-time to the first intelligent agent's local cache (the local cache uses encrypted storage, and the key is independent of the temporary anonymous identifier key) for rapid initial local verification.

[0044] In trusted intermediaries, key management can employ a dual strategy of trusted Key Management Service (KMS) and dynamic rotation. In some application scenarios, the trusted intermediary deploys a KMS to assign a unique key set to each agent, including the first agent. Each agent's key set includes a temporary anonymous key, a communication encryption key, and a key for signing. The node where the KMS is located is referred to as the key management node. Keys are generated using a master key derivation mechanism and are stored in the trusted intermediary's hardware security module; they cannot be exported. The key rotation cycle is dynamically adjusted based on the threat level. For example, the default cycle is 7 days, shortened to within 12 hours when a threat is detected. Key rotation is achieved through encrypted key transmission and multi-node verification in the encrypted storage unit of the trusted node. For instance, a new key is encrypted by the key management node using the corresponding agent's public key and sent. After decryption, the agent must return a verification digest of the new key to the key management node. Simultaneously, the key management node synchronizes the new key to at least three trusted nodes in the aforementioned encrypted storage unit to ensure the key is not lost. The key management node can also manage the public parameters required for zero-knowledge proofs and can update these parameters periodically.

[0045] In one possible embodiment, the privacy-preserving multi-agent communication method further includes: A trusted intermediary receives a registration request from a second agent, which includes a second temporary anonymous identifier, a functional tag, and an encrypted communication address. The intermediary verifies the legitimacy of the second agent's second temporary anonymous identifier. Upon successful verification, the intermediary associates the second agent's second temporary anonymous identifier, functional tag, and encrypted communication address and stores them in multiple distributed trusted directory service nodes. It can also send a registration success response to the second agent. Byzantine Fault Tolerance (BFT) is used to ensure data consistency; if one trusted directory service node fails, it automatically switches to a backup node to provide service.

[0046] In one possible embodiment, before receiving the permission verification request returned by the second agent, the method further includes: receiving a challenge value sent by the second agent, the challenge value being encrypted with the public key corresponding to the first temporary anonymous identifier; decrypting the challenge value using the private key corresponding to the public key, and performing a hash operation on the decrypted challenge value to obtain a hash result; signing the hash result using the private key and sending it to the second agent, the hash result being used for the identity legitimacy verification of the first agent.

[0047] The challenge value is random data randomly generated by the second agent to verify the legitimacy of the first agent's identity. The encryption public key and encryption private key are asymmetric key pairs bound to a temporary anonymous identifier and are used only for authentication. For example, after receiving a communication request, the second agent obtains the encryption public key associated with the first agent's first temporary anonymous identifier from a trusted intermediary. Then, the second agent uses a true random number generator to generate a 128-bit random challenge value, encrypts the challenge value using the encryption public key associated with the first temporary anonymous identifier, and sends it to the first agent.

[0048] Hash operations are used to ensure the integrity and irreversibility of response data, while signature operations are used to prove that the response data was legitimately issued by the first agent.

[0049] The first agent decrypts the challenge value using the private key corresponding to the public key, then performs a SHA-3 hash operation on the decrypted challenge value to obtain the hash result. The hash result is then signed with the private key and returned to the second agent. The second agent verifies the signature validity using the public key corresponding to the first temporary anonymous identifier, and re-hashes the challenge value, comparing it with the hash result returned by the first agent. If they match, the first agent's authentication is confirmed to be successful.

[0050] In the above scheme, verifying the identity of the first intelligent agent before verifying its permissions can reduce attacks by unauthorized intelligent agents on the second intelligent agent.

[0051] In one possible embodiment, the permission verification request includes a function permission type and verification parameters. The method further includes: generating commitment data based on the verification parameters and a first random number; sending the commitment data to a second intelligent agent; receiving a second random number returned by the second intelligent agent; and generating a zero-knowledge proof based on the second random number, the first random number, and the key corresponding to the function permission.

[0052] Functional permission types include, but are not limited to, communication permissions and node management permissions. Verification parameters can be publicly available parameters based on the discrete logarithm problem. , , For large prime numbers, for The original root.

[0053] For example, taking a function permission type of communication permission (such as data access permission or function call permission) as an example, the second intelligent agent needs to verify whether the first intelligent agent has communication permission. A zero-knowledge proof scheme can be used: the second intelligent agent sends a permission verification request to the first intelligent agent, which includes the function permission type of data read permission and verification parameters; the first intelligent agent first generates a first random number. Then calculate the commitment data. ,in, ; and will commit data Send to the second agent; the second agent generates a second random number. And send it to the first intelligent agent; the first intelligent agent computes the zero-knowledge proof. ,in, , This is the private key corresponding to the functional permissions, which is unknown to the second agent. Zero-knowledge proofs will be used. Send to the second agent; the second agent verifies. Whether it is valid or not, if it is valid, it proves that the first intelligent agent has the corresponding permissions. During this process, the second intelligent agent is always unable to obtain the private key x corresponding to the functional permissions and the real identity information of the first intelligent agent.

[0054] In the above scheme, zero-knowledge proofs are constructed by using two layers of random numbers and committed data, which further enhances the privacy and anti-attack capabilities of the permission verification process.

[0055] After successful authorization verification, the two agents establish an end-to-end encrypted channel based on temporary anonymous identifiers. First, a session key is negotiated through a trusted intermediary, such as using an Elliptic Curve Diffie-Hellman Ephemeral (ECDHE) key exchange protocol. During the negotiation, only their respective temporary anonymous identifiers are used as communication endpoint identifiers; no real identity information is transmitted. After the session key negotiation is completed, both parties use advanced encryption algorithms to encrypt and transmit communication data. The encrypted message header contains only the temporary anonymous identifier (excluding traceable information such as IP address and port number). Notably, a one-time session identifier is used during channel establishment; a new session identifier is established for each communication and destroyed immediately after the session ends, preventing long-term channel tracking. In one possible embodiment, a communication mode protection based on dynamic routing technology can be employed, that is, a multi-path dynamic switching routing strategy can be used, with multiple trusted and authenticated routing nodes distributed in different physical locations. The method for establishing an encrypted channel with the second agent can include: sending a routing information retrieval request to a trusted intermediary; receiving a list of routing nodes from the trusted intermediary, the list containing node identifiers, load information, and threat levels of several routing nodes; sending the routing node list to the second agent based on a second temporary anonymous identifier; negotiating a target routing node with the second agent based on the load information and threat level of each routing node; and establishing an encrypted channel with the second agent through the target routing node.

[0056] Routing nodes are securely certified nodes responsible only for forwarding encrypted data. When negotiating target routing nodes, priority should be given to routing nodes with low load and no threats to improve communication stability and security.

[0057] For example, when communication is established, both agents can obtain a list of available routing nodes through a trusted intermediary. Then, the first and second agents can select initial routing nodes based on load balancing and threat awareness strategies, for example, prioritizing routing nodes with a load below 60% and no abnormal threats. Optionally, during communication, the routing nodes are dynamically switched at random time intervals (e.g., every 10-30 seconds), with a new routing node renegotiated through the trusted intermediary during the switch. In some application scenarios, if multiple routing nodes are required for communication between the first and second agents, the overlap between the new and original routing nodes should not exceed 30%; simultaneously, the routing nodes only forward encrypted data, do not parse the data content or record communication node information, and the forwarding log is automatically destroyed after only 24 hours. In the above scheme, the security and reliability of the encrypted channel establishment process are improved by selecting the optimal routing node based on load and threat awareness.

[0058] In one possible embodiment, after establishing an encrypted channel with the second agent, the method further includes: First, several data packets sent by the second intelligent agent are received, at least some of which are spoofed data packets. Second, the spoofed data packets are filtered out to obtain at least one genuine data packet. Then, each genuine data packet is parsed to obtain its service data. Finally, the service data containing padding is de-padded and reassembled to obtain the original service data.

[0059] Disguised data packets are data packets with no actual business meaning sent by the second agent to obfuscate communication characteristics and prevent traffic analysis. Padding data is random data added by the second agent to standardize data packet length and avoid length feature leakage. Depadding is used to remove random padding data, and reassembly is used to restore segmented transmitted business data to complete original business data.

[0060] In some application scenarios, when sending data, the second intelligent agent can segment the business data and fill the segmented data packets with interference data. It can also periodically send spoofed data packets (containing randomly encrypted data in the same format as the real data packets), with the spoofed data packets accounting for 20%-30% of the real data packets, further obfuscating the metadata characteristics.

[0061] The above solution protects against traffic feature analysis and communication tracing attacks by filtering out spoofed data packets, removing padding, and reassembling them, while ensuring the integrity of business data.

[0062] In one possible embodiment, at least some of the data packets are of different sizes and / or at least some of the receiving time intervals between two adjacent data packets are different.

[0063] The different data packet sizes are achieved through random padding and fragmentation mechanisms at the sending end, while the different data packet reception time intervals are achieved through time jitter and random interval transmission mechanisms at the sending end. The first intelligent agent ignores the differences in time intervals and lengths during reception, and only performs validity checks on the business data.

[0064] For example, when the second intelligent agent sends data, it uses a dual processing of "randomization + masquerading" for the communication metadata of the business data (including communication time, data packet size, data packet interval, and communication duration): At the communication time level, through a time jitter mechanism, the actual sending time is randomly selected within a preset communication time window (e.g., ±500ms) to avoid tracking caused by sending at a fixed time. For example, an exponential distribution can be used to randomly generate the interval time (average interval 100ms, fluctuation range ±30ms); At the data packet size level, the actual data packets are processed by "padding + fragmentation": According to the preset data packet size range (e.g., 128B-1024B), data packets smaller than the minimum size are padded with random bytes (the padding bytes are generated using pseudo-random numbers), and data packets larger than the maximum size are fragmented, with the fragment size randomly distributed within the range.

[0065] The above scheme further enhances the anti-traffic analysis and anti-tracking capabilities of the communication process by randomizing the data packet length and sending interval.

[0066] In one possible embodiment, the method further includes: detecting whether the first temporary anonymous identifier meets the update conditions; determining that the first temporary anonymous identifier meets the update conditions, sending an update request to the trusted intermediary; and receiving a new temporary anonymous identifier for the first agent returned by the trusted intermediary.

[0067] Update conditions include, but are not limited to: the expiration of the preset duration, a change in the communication session, and the existence of security risks. The update process is uniformly executed by a trusted intermediary, and the old temporary anonymous identifier becomes invalid immediately after the update.

[0068] For example, the update of the first temporary anonymous identifier can employ a dual mechanism of timed triggering and event triggering. The timed update cycle is dynamically adjusted based on the threat level of the communication system (the threat level is assessed in real time; the cycle is 7 days for low threat level (no abnormal attacks), 2 days for medium threat level (a few abnormal requests), and 12 hours for high threat level (targeted attacks detected). The communication system can be a system that includes a trusted intermediary, several agents including a first agent and a second agent, and several routing nodes.

[0069] Event-triggered updates include three scenarios: "the first agent's authentication failure count exceeds a threshold (e.g., 5 consecutive failures), temporary identifiers may be leaked (e.g., communication channels are hijacked), and the first agent restarts or changes its network environment." Upon triggering, the update process begins immediately. The update process is as follows: The first agent sends an update request to a trusted intermediary. The trusted intermediary generates a new temporary anonymous identifier (the generation method is described above and will not be repeated here), encrypts it using the first agent's public key, and sends it. Simultaneously, it updates the "real identity identifier - temporary anonymous identifier" mapping relationship and synchronizes it to the local caches of each trusted node in the encrypted storage unit and the first agent. The old temporary identifier becomes invalid within 30 minutes of the new identifier taking effect to avoid conflicts between the old and new identifiers. The above scheme reduces the risk of identity association and long-term tracking by dynamically updating temporary anonymous identifiers, thereby further enhancing communication anonymity.

[0070] In one possible embodiment, a distributed behavior monitoring module can be deployed in the communication system. The monitoring indicators include "number of authentication failures (more than 5 times per hour for a single agent is abnormal), communication frequency (deviation from the agent's historical average frequency ±50% is abnormal), number of communication objects (more than 10 new communication objects per hour is abnormal), routing path switching frequency (more than 10 switching in a short period of time (5 minutes) is abnormal), and data transmission volume (deviation from the historical average transmission volume ±100% is abnormal)". The monitoring module adopts a "local preliminary judgment + cloud collaborative analysis" mode. That is, the agent monitors the basic indicators locally in real time, and sends an anomaly report to the cloud monitoring center after detecting an anomaly (the report only contains a temporary anonymous identifier and anomaly indicators, and does not contain other information). The cloud monitoring center combines the anomaly reports from multiple agents and network-wide threat intelligence (such as known attack IPs and attack characteristics) for collaborative analysis to determine whether the anomaly is a malicious attack (such as a distributed denial-of-service attack or an identity probe attack). The analysis results are synchronized to the security policy adjustment module in real time.

[0071] Based on the analysis results of the monitoring module, the security policy adjustment module in the communication system adopts a "rule engine + machine learning" approach to achieve dynamic adjustments. The rule engine presets basic adjustment rules (e.g., when an identity probe attack is detected, shortening the temporary identifier update cycle to 6 hours and increasing the number of zero-knowledge proof verifications; when a Distributed Denial of Service (DDoS) attack is detected, limiting the request frequency of a single agent and increasing the filtering conditions for routing nodes). The machine learning model (using the random forest algorithm, trained based on historical attack data and defense effectiveness) optimizes the adjustment results of the rule engine, balancing security and availability (e.g., when the adjusted communication latency exceeds 100ms, appropriately relaxing some restrictions to ensure communication efficiency). The adjusted policy is encrypted and distributed to each agent through a trusted intermediary. Upon receipt, the policy takes effect immediately and returns an activation confirmation to the trusted intermediary.

[0072] In one possible embodiment, a privacy-preserving multi-agent communication method includes, for example: Figure 2 The following steps are shown: Step 201: Receive the communication request sent by the first intelligent agent.

[0073] The communication request includes the first temporary anonymous identifier of the first agent. Figure 2 The privacy-preserving multi-agent communication method shown can be applied to a second agent.

[0074] Step 202: Send an authorization verification request to the first intelligent agent based on the first temporary anonymous identifier.

[0075] Optionally, the permission verification request is a request made by the second intelligent agent to verify whether the first intelligent agent has the corresponding business function permissions.

[0076] The permission verification request includes the function permission type and verification parameters. The function permission type includes, but is not limited to, communication permissions and node management permissions. The verification parameters can be public parameters based on the discrete logarithm problem. , , For large prime numbers, for The original root.

[0077] Step 203: After receiving the zero-knowledge proof returned by the first intelligent agent, verify the functional permissions of the first intelligent agent based on the zero-knowledge proof.

[0078] For example, taking a function permission type of communication permission (such as data access permission or function call permission) as an example, the second intelligent agent needs to verify whether the first intelligent agent has communication permission. A zero-knowledge proof scheme can be used: the second intelligent agent sends a permission verification request to the first intelligent agent, which includes the function permission type of data read permission and verification parameters; the first intelligent agent first generates a first random number. Then calculate the commitment data. ,in, ; and will commit data Send to the second agent; the second agent generates a second random number. And send it to the first intelligent agent; the first intelligent agent computes the zero-knowledge proof. ,in, , ( (The private key corresponding to the function permission is known only to the first agent) will be used for zero-knowledge proof. Send to the second agent; the second agent verifies. If the condition is met, it proves that the first agent possesses the corresponding permissions. During this process, the second agent is unable to obtain the private key corresponding to the functional permissions. And the true identity information of the first intelligent agent.

[0079] Step 204: After the first agent's functional permissions are verified, send a permission verification successful response to the first agent.

[0080] Step 205: Establish an encrypted channel with the first intelligent agent.

[0081] In one possible embodiment, a privacy-preserving multi-agent communication method includes, for example: Figure 3 The following steps are shown: Step 301: The first intelligent agent sends a pre-registration request to the trusted intermediary.

[0082] The first intelligent agent sends a pre-registration request to a trusted intermediary, which includes the agent's application materials. These materials may include at least one of the following: the agent's hardware information, its domain, scope of functional permissions, and information about the superior control node. The trusted intermediary can verify the hardware's legitimacy by connecting the hardware information to the hardware manufacturer's database. The trusted intermediary can also verify the first intelligent agent's permissions based on its domain and scope of functional permissions; that is, the trusted intermediary confirms the first intelligent agent's permission scope based on the joint signatures of trusted nodes within the domain. The superior control node information provides node endorsement for the first intelligent agent; trusted nodes verify the first intelligent agent's node endorsement through the data signature submitted by the superior control node. In other words, the trusted intermediary performs triple verification of the first intelligent agent through hardware information verification, permission qualification verification, and node endorsement.

[0083] Step 302: The trusted intermediary returns the real identity identifier, the first temporary anonymity identifier, the function permission type, and the corresponding private key to the first intelligent agent.

[0084] The trusted intermediary generates the first agent's real identity identifier and a first temporary anonymous identifier. After successful verification, the trusted intermediary generates dual identity identifiers using a combination of "asymmetric encryption algorithm + hash algorithm". For example, the trusted intermediary first generates 256-bit original identity information based on the first agent's unique hardware identifier (such as CPU serial number, MAC address) and the unique code assigned to it, using the SHA-3 hash algorithm; then, it uses an asymmetric encryption algorithm (such as Rivest-Shamir-Adleman, RSA) to encrypt the original identity information, generating the agent's real identity identifier (which serves as the agent's unique trusted credential in the system, stored only in the trusted intermediary's encrypted database, and not transmitted externally). The first temporary anonymous identifier of the first intelligent agent is dynamically generated using a combination of random number generation and symmetric encryption. For example, a 512-bit random number is first generated using a pseudo-random number generator based on chaos theory (such as Logistic chaotic mapping), and then the random number is encrypted using a symmetric encryption algorithm (such as AES-256) to generate the first temporary anonymous identifier. The key is uniformly distributed by a trusted intermediary and updated periodically. Simultaneously, a one-to-one mapping relationship is established between the real identity identifier and the temporary anonymous identifier. This mapping relationship is stored only in the encrypted storage unit within the trusted intermediary and uses a fragmented storage method (fragment keys are kept separately by different trusted nodes, requiring multi-node collaborative verification to decrypt the mapping relationship). The real identity identifier of the first intelligent agent is entered into the trusted intelligent agent's whitelist. The whitelist uses a chained storage structure, and each whitelist entry is appended with a dual digital signature from the trusted intermediary and the endorsing node to prevent tampering with the whitelist. Simultaneously, the whitelist is synchronized in real-time to the first intelligent agent's local cache (the local cache uses encrypted storage, and the key is independent of the temporary anonymous identifier key) for rapid initial local verification.

[0085] Step 303: The first intelligent agent sends a functional requirement request carrying a functional tag to the trusted intermediary.

[0086] Step 304: The trusted intermediary returns the second temporary anonymous identifier of the second agent corresponding to the function label to the first agent.

[0087] Step 305: Send a communication request to the second agent based on the second temporary anonymous identifier.

[0088] Step 306: The second agent sends a public key acquisition request to the trusted intermediary.

[0089] Step 307: The trusted intermediary sends the verification public key associated with the first temporary anonymous identifier to the second agent.

[0090] Step 308: Send challenge data to the first intelligent agent.

[0091] The challenge data is encrypted using a verification public key associated with a first temporary anonymous identifier.

[0092] Step 309: After decrypting the challenge data, perform a hash operation to obtain the hash result and send it to the second intelligent agent.

[0093] The second agent uses the encrypted public key corresponding to the first temporary anonymous identifier to verify the validity of the signature, and re-hashes the challenge value. It then compares the hash result received from the first agent with the result received from the first agent. If they match, the first agent's identity verification is confirmed.

[0094] Step 310: After the second agent determines that the first agent's identity verification is successful based on the hash result, it sends an authorization verification request to the first agent.

[0095] In cases where the authentication of the first agent fails, communication with the first agent can be terminated.

[0096] Step 311: The first agent sends a zero-knowledge proof to the second agent.

[0097] Step 312: After the second agent verifies the first agent's functional permission verification based on zero-knowledge proof, it sends a permission verification pass response to the first agent.

[0098] Specifically, if the function permission verification of the first agent fails, the communication with the first agent will be terminated.

[0099] Step 313: The first agent establishes an encrypted channel with the second agent.

[0100] During communication between the first and second intelligent agents, continuous security maintenance can be maintained. This maintenance may include, but is not limited to, periodically or based on events updating temporary anonymous identifiers, monitoring abnormal access behavior, and dynamically adjusting security policies.

[0101] Optionally, the first agent can also determine whether communication with the second agent should continue. The determination can be made by: determining whether the first agent still needs to continue communication; if so, determining that communication between the first and second agents should continue. If the determination result is that communication continues, the communication connection between the first and second agents is re-established; otherwise, communication can be determined to have ended, and communication with the second agent can be stopped.

[0102] The privacy-preserving multi-agent communication device provided by the present invention is described below. The privacy-preserving multi-agent communication device described below corresponds to the privacy-preserving multi-agent communication method applied to the first agent described above. For example... Figure 4 As shown, the privacy-preserving multi-agent communication device 400 can be applied to a first agent. The privacy-preserving multi-agent communication device 400 includes: The first request sending module 401 is used for the first intelligent agent to send a functional requirement request carrying a functional tag to a trusted intermediary. The second request sending module 402 is used to send a communication request to the second intelligent agent based on the second temporary anonymous identifier after receiving the second temporary anonymous identifier of the second intelligent agent corresponding to the function tag returned by the trusted intermediary. The communication request includes the first temporary anonymous identifier of the first intelligent agent. The zero-knowledge proof sending module 403 is used to send a zero-knowledge proof for verifying the function permissions of the first intelligent agent to the second intelligent agent if it receives a permission verification request returned by the second intelligent agent. The zero-knowledge proof is used for the function permission verification of the first intelligent agent. The first encrypted channel establishment module 404 is used to establish an encrypted channel with the second intelligent agent if it receives a permission verification pass response returned by the second intelligent agent, wherein the encrypted channel is used for data transmission.

[0103] According to a privacy-preserving multi-agent communication device 400 provided by the present invention, the permission verification request includes a function permission type and verification parameters, and the zero-knowledge proof sending module 403 is further configured to: Based on the verification parameters and the first random number, commitment data is generated; The commitment data is sent to the second intelligent agent; Receive the second random number returned by the second agent; The zero-knowledge proof is generated based on the second random number, the first random number, and the key corresponding to the function permission.

[0104] According to the present invention, a privacy-preserving multi-agent communication device 400 is provided, which further includes an authentication module. Figure 4 (Not shown) Before receiving the permission verification request returned by the second intelligent agent, the authentication module is used to: Receive a challenge value sent by the second intelligent agent, the challenge value being encrypted with the public key corresponding to the first temporary anonymous identifier; The challenge value is decrypted using the private key corresponding to the public key, and a hash operation is performed on the decrypted challenge value to obtain a hash result; The hash result is signed using the encrypted private key and then sent to the second intelligent agent. The hash result is used to verify the identity and legitimacy of the first intelligent agent.

[0105] According to the present invention, a privacy-preserving multi-agent communication device 400 is provided, which further includes a data transmission module. Figure 4 (Not shown) After establishing an encrypted channel with the second agent, the data transmission module is used for: Receive a number of data packets sent by the second intelligent agent, at least some of which are disguised data packets; By filtering out the spoofed data packets from a number of the data packets, at least one real data packet is obtained; Parse each of the real data packets to obtain the service data of each of the real data packets; After removing the filler data from the business data, the data is reassembled to obtain the original business data.

[0106] According to the present invention, a privacy-preserving multi-agent communication device 400 is provided, wherein at least some of the data packets are of different sizes and / or at least some of the receiving time intervals of two adjacent data packets are different.

[0107] According to the present invention, a privacy-preserving multi-agent communication device 400 includes a first encrypted channel establishment module 404 establishing an encrypted channel with a second agent, comprising: Send a request to a trusted intermediary to obtain routing information; Receive the routing node list sent by the trusted intermediary, the routing node list containing the node identifier, load status and threat level of several routing nodes; Based on the second temporary anonymous identifier, the routing node list is sent to the second intelligent agent; Based on the load and threat level of each routing node, negotiate the target routing node with the second intelligent agent; An encrypted channel is established between the target routing node and the second intelligent agent.

[0108] According to the present invention, a privacy-preserving multi-agent communication device 400 is provided, which further includes a temporary anonymous identifier update module. Figure 4 (Not shown in the image) The temporary anonymous identifier update module is also used for: Check whether the first temporary anonymous identifier meets the update conditions; Once it is determined that the first temporary anonymous identifier meets the update condition, an update request is sent to the trusted intermediary. Receive the new temporary anonymous identifier returned by the trusted intermediary for the first intelligent agent.

[0109] Please see Figure 5 Another privacy-preserving multi-agent communication device 500 can be applied to a second agent. This privacy-preserving multi-agent communication device 500 includes: The request receiving module 501 is used to receive a communication request sent by the first intelligent agent, wherein the communication request includes a first temporary anonymous identifier of the first intelligent agent; The permission verification request sending module 502 is used to send a permission verification request to the first intelligent agent based on the first temporary anonymous identifier; The permission verification module 503 is used to verify the functional permissions of the first intelligent agent based on the zero-knowledge proof after receiving the zero-knowledge proof returned by the first intelligent agent. The permission verification pass response sending module 504 is used to send a permission verification pass response to the first intelligent agent after the function permission verification of the first intelligent agent is passed. The second encrypted channel establishment module 505 is used to establish an encrypted channel with the first intelligent agent.

[0110] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6As shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communications interface 620, and the memory 630 communicate with each other via the communication bus 640. The processor 610 can invoke logical instructions in the memory 630 to execute a privacy-preserving multi-agent communication method. This method includes: a first agent sending a function request carrying a function tag to a trusted intermediary; upon receiving a second temporary anonymous identifier of a second agent corresponding to the function tag returned by the trusted intermediary, sending a communication request to the second agent based on the second temporary anonymous identifier, the communication request including the first temporary anonymous identifier of the first agent; if a permission verification request is received from the second agent, sending a zero-knowledge proof for verifying the function permissions of the first agent to the second agent, the zero-knowledge proof being used for verifying the function permissions of the first agent; if a permission verification pass response is received from the second agent, establishing an encrypted channel with the second agent, wherein the encrypted channel is used for data transmission. Alternatively, the method includes: receiving a communication request sent by a first intelligent agent, the communication request including a first temporary anonymous identifier of the first intelligent agent; sending an authorization verification request to the first intelligent agent based on the first temporary anonymous identifier; verifying the functional permissions of the first intelligent agent based on the zero-knowledge proof after receiving a zero-knowledge proof returned by the first intelligent agent; sending an authorization verification pass response to the first intelligent agent after the functional permissions of the first intelligent agent are verified; and establishing an encrypted channel with the first intelligent agent.

[0111] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0112] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the privacy-preserving multi-agent communication method provided by the above methods. The method includes: a first agent sending a function request carrying a function tag to a trusted intermediary; after receiving a second temporary anonymous identifier of a second agent corresponding to the function tag returned by the trusted intermediary, sending a communication request to the second agent based on the second temporary anonymous identifier, the communication request including the first temporary anonymous identifier of the first agent; if a permission verification request is received from the second agent, sending a zero-knowledge proof for verifying the function permission of the first agent to the second agent, the zero-knowledge proof being used for verifying the function permission of the first agent; if a permission verification pass response is received from the second agent, establishing an encrypted channel with the second agent, wherein the encrypted channel is used for data transmission. Alternatively, the method includes: receiving a communication request sent by a first intelligent agent, the communication request including a first temporary anonymous identifier of the first intelligent agent; sending an authorization verification request to the first intelligent agent based on the first temporary anonymous identifier; verifying the functional permissions of the first intelligent agent based on the zero-knowledge proof after receiving a zero-knowledge proof returned by the first intelligent agent; sending an authorization verification pass response to the first intelligent agent after the functional permissions of the first intelligent agent are verified; and establishing an encrypted channel with the first intelligent agent.

[0113] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the privacy-preserving multi-agent communication method provided by the above methods. The method includes: a first agent sending a function request carrying a function tag to a trusted intermediary; upon receiving a second temporary anonymous identifier of a second agent corresponding to the function tag returned by the trusted intermediary, sending a communication request to the second agent based on the second temporary anonymous identifier, the communication request including a first temporary anonymous identifier of the first agent; if a permission verification request is received from the second agent, sending a zero-knowledge proof for verifying the function permissions of the first agent to the second agent, the zero-knowledge proof being used for verifying the function permissions of the first agent; if a permission verification pass response is received from the second agent, establishing an encrypted channel with the second agent, wherein the encrypted channel is used for data transmission. Alternatively, the method includes: receiving a communication request sent by a first intelligent agent, the communication request including a first temporary anonymous identifier of the first intelligent agent; sending an authorization verification request to the first intelligent agent based on the first temporary anonymous identifier; verifying the functional permissions of the first intelligent agent based on the zero-knowledge proof after receiving a zero-knowledge proof returned by the first intelligent agent; sending an authorization verification pass response to the first intelligent agent after the functional permissions of the first intelligent agent are verified; and establishing an encrypted channel with the first intelligent agent.

[0114] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0115] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A privacy-preserving multi-agent communication method, characterized in that, Applied to a first intelligent agent, the method includes: The first intelligent agent sends a functional requirement request carrying a functional tag to a trusted intermediary; After receiving the second temporary anonymous identifier of the second agent corresponding to the function tag returned by the trusted intermediary, a communication request is sent to the second agent based on the second temporary anonymous identifier, and the communication request includes the first temporary anonymous identifier of the first agent. If a permission verification request is received from the second agent, a zero-knowledge proof for verifying the function permissions of the first agent is sent to the second agent, and the zero-knowledge proof is used for the function permission verification of the first agent. If a permission verification pass response is received from the second agent, an encrypted channel is established between the agent and the second agent, wherein the encrypted channel is used for data transmission.

2. The method according to claim 1, characterized in that, The permission verification request includes the function permission type and verification parameters, and the method further includes: Based on the verification parameters and the first random number, commitment data is generated; The commitment data is sent to the second intelligent agent; Receive the second random number returned by the second agent; The zero-knowledge proof is generated based on the second random number, the first random number, and the key corresponding to the function permission.

3. The method according to claim 1, characterized in that, Before receiving the permission verification request returned by the second intelligent agent, the method further includes: Receive a challenge value sent by the second intelligent agent, the challenge value being encrypted with the public key corresponding to the first temporary anonymous identifier; The challenge value is decrypted using the private key corresponding to the public key, and a hash operation is performed on the decrypted challenge value to obtain a hash result; The hash result is signed using the encrypted private key and then sent to the second intelligent agent. The hash result is used to verify the identity and legitimacy of the first intelligent agent.

4. The method according to any one of claims 1 to 3, characterized in that, After establishing an encrypted channel with the second agent, the method further includes: Receive a number of data packets sent by the second intelligent agent, at least some of which are disguised data packets; By filtering out the spoofed data packets from a number of the data packets, at least one real data packet is obtained; Parse each of the real data packets to obtain the service data of each of the real data packets; After removing the filler data from the business data, the data is reassembled to obtain the original business data.

5. The method according to claim 4, characterized in that, At least some of the data packets are of different sizes and / or at least some of the data packets have different reception time intervals.

6. The method according to any one of claims 1 to 3, characterized in that, Establishing an encrypted channel with the second intelligent agent includes: Send a routing information retrieval request to the trusted intermediary; Receive the routing node list sent by the trusted intermediary, the routing node list containing the node identifier, load status and threat level of several routing nodes; Based on the second temporary anonymous identifier, the routing node list is sent to the second intelligent agent; Based on the load and threat level of each routing node, negotiate the target routing node with the second intelligent agent; An encrypted channel is established between the target routing node and the second intelligent agent.

7. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Check whether the first temporary anonymous identifier meets the update conditions; Once it is determined that the first temporary anonymous identifier meets the update condition, an update request is sent to the trusted intermediary. Receive the new temporary anonymous identifier returned by the trusted intermediary for the first intelligent agent.

8. A privacy-preserving multi-agent communication method, characterized in that, Applied to a second intelligent agent, the method includes: Receive a communication request sent by a first intelligent agent, wherein the communication request includes a first temporary anonymous identifier of the first intelligent agent; Based on the first temporary anonymous identifier, send an authorization verification request to the first intelligent agent; After receiving the zero-knowledge proof returned by the first intelligent agent, the functional permissions of the first intelligent agent are verified based on the zero-knowledge proof. After the first agent's functional permissions are verified, a permission verification successful response is sent to the first agent; Establish an encrypted channel with the first intelligent agent.

9. A privacy-preserving multi-agent communication device, characterized in that, include: The first request sending module is used for the first intelligent agent to send a functional requirement request carrying a functional tag to a trusted intermediary. The second request sending module is configured to send a communication request to the second intelligent agent based on the second temporary anonymous identifier of the function tag corresponding to the second intelligent agent returned by the trusted intermediary, after receiving the second temporary anonymous identifier of the second intelligent agent returned by the trusted intermediary, wherein the communication request includes the first temporary anonymous identifier of the first intelligent agent. A zero-knowledge proof sending module is used to send a zero-knowledge proof for verifying the function permissions of the first intelligent agent to the second intelligent agent if a permission verification request is received from the second intelligent agent. The zero-knowledge proof is used for the function permission verification of the first intelligent agent. The first encrypted channel establishment module is used to establish an encrypted channel with the second intelligent agent if it receives a permission verification success response returned by the second intelligent agent, wherein the encrypted channel is used for data transmission.

10. A privacy-preserving multi-agent communication device, characterized in that, include: A request receiving module is used to receive a communication request sent by a first intelligent agent, wherein the communication request includes a first temporary anonymous identifier of the first intelligent agent; The permission verification request sending module is used to send a permission verification request to the first intelligent agent based on the first temporary anonymous identifier; The permission verification module is used to verify the functional permissions of the first intelligent agent based on the zero-knowledge proof after receiving the zero-knowledge proof returned by the first intelligent agent. The permission verification pass response sending module is used to send a permission verification pass response to the first intelligent agent after the first intelligent agent's function permission verification is passed; The second encrypted channel establishment module is used to establish an encrypted channel with the first intelligent agent.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the privacy-preserving multi-agent communication method as described in any one of claims 1 to 8.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the privacy-preserving multi-agent communication method as described in any one of claims 1 to 8.

Images