An AI agent cross-carrier migration and instance uniqueness management system and method

Abstract

The application discloses an AI intelligent agent cross-carrier migration and instance uniqueness management and control system and method, and the system comprises an identity generation unit, a migration management and control unit, an instance lock control unit and a source end processing unit. The identity generation unit generates a global identity of the intelligent agent in the whole life cycle. The migration management and control unit generates a one-time migration voucher, locks the source end during migration, triggers atomic rollback when migration fails, and prevents state loss. The instance lock control unit ensures that only one legal instance is in an active state at the same time in a commercial environment, covers all running modes such as splitting, copying, mirroring and digital twinning, and prevents instance conflicts. The source end processing unit supports archiving or destroying after migration, and supports reverse recovery when the target end fails. The application solves the problems of unreliable cross-end migration of intelligent agents, multi-end running conflicts and breakpoint non-continuous transmission, and is suitable for the whole life cycle deployment of companion AI, virtual digital people and cross-end intelligent assistants.

Description

I. Technical Field This invention belongs to the field of artificial intelligence system scheduling and trusted computing technology, specifically involving a cross-carrier migration, instance state synchronization and uniqueness control system and method for large model intelligent agents, virtual digital humans, companion AI and cross-terminal intelligent assistants. II. Background Technology As AI agents evolve from single-device to multi-device collaborative operation, the need for agents to move between different platforms (such as mobile phones, in-vehicle systems, cloud computing, and home robots) is becoming increasingly urgent. Existing technologies face the following bottlenecks: First, the migration process is unreliable. Current migration technologies lack atomicity guarantees. If the network is interrupted or the target device crashes during transmission, the agent can easily fall into a lost state ("source device deleted, target device not alive") or a dual operating state ("source device not deleted, target device alive"). Second, instance conflicts. In a distributed environment, the lack of a strong consistency locking mechanism can easily lead to the same agent running simultaneously on different devices (i.e., "clone" conflicts), causing memory errors, duplicate service billing, or logical mutual exclusion. Third, there is a lack of reverse recovery mechanisms. If migration to a high-performance edge (such as the cloud) fails, the existing system cannot automatically and safely fall back to the original edge (such as the mobile device) to continue service, resulting in a broken user experience. Summary of the Invention

[0003] 3.1 Purpose of the Invention The present invention aims to solve the technical problems of state loss, instance conflict and inability to roll back during the cross-carrier migration process of AI intelligent agents, and provides a management and control system based on globally unique identity, one-time migration certificate, atomic rollback and instance locking.

[0004] 3.2 Technical Solution An AI agent cross-carrier migration and instance uniqueness control system includes: an identity generation unit, a migration control unit, an instance locking unit, and a source-end processing unit. The identity generation unit is configured to: generate a globally unique identity identifier bound to the agent and unchanged throughout its entire lifecycle; the migration control unit is configured to: generate a one-time, non-reusable exclusive migration credential, perform a locking operation on the source-end agent instance during the migration process, prohibiting the source-end instance from performing interactive operations during the migration; simultaneously monitor the migration status, and if the migration fails, trigger an atomic rollback operation to restore the source-end agent instance to its state before migration; the instance locking unit is configured to: in a commercial operating environment, only one legitimate instance is active at a time; the source-end processing unit is configured to: after the migration is completed, perform processing operations on the source-end agent instance according to a preset strategy. A method for cross-carrier migration and instance uniqueness control of AI agents includes: an identity generation step: generating a globally unique identity identifier bound to the agent and remaining unchanged throughout its entire lifecycle; a credential generation and migration step: generating a one-time, non-reusable exclusive migration credential, locking the source instance, and performing state transfer; an instance locking step: activating the instance on the target end and ensuring that only one legitimate instance is active at any given time; and a source instance disposal step: disposing of the source instance after successful migration, or performing an atomic rollback if migration fails.

[0005] 3.3 Beneficial Effects 1. By binding a globally unique identity with the entire lifecycle, cross-carrier identity consistency and traceability are achieved.

[0006] 2. Prevent replay attacks and identity theft during the migration process by using a one-time, non-reusable exclusive migration credential.

[0007] 3. By covering all operating modes through the instance locking unit, strong consistency control is achieved, ensuring that only one legitimate instance is active at any given time.

[0008] 4. By employing atomic rollback and reverse recovery mechanisms, the problem of agent loss caused by migration failure is resolved, ensuring service continuity. IV. Description of the attached drawings Figure 1 This is a schematic diagram of the system structure of the present invention.

[0009] Figure 2 This is a schematic diagram of the method flow of the present invention.

[0010] Figure 3 This is a state transition diagram for migration control and atomic rollback in this invention.

[0011] Figure 4 This is a schematic diagram of the physical form covered by the locking unit in an example of the present invention. V. Detailed Implementation Methods 5.1 Specific Implementation of the Identity Generation Unit When an agent is first created, the identity generation unit generates a globally unique identity identifier based on an asymmetric encryption algorithm or a distributed unique ID algorithm (such as Snowflake). This identifier is bound to the agent's core memory and personality parameters and does not change with the carrier.

[0013] 5.2 Specific Implementation of the Migration Control Unit The migration control unit adopts a "two-phase commit + rollback compensation" mechanism. Phase 1 (Locking and Transmission): A one-time credential is generated, the source instance is locked (writing new memories is prohibited, only reading is allowed), and the context state is transferred. Phase 2 (Commit and Release): The target end verifies the credential and restores the state. If successful, an acknowledgment is sent to the source end; if it fails (e.g., timeout, verification error), the target end destroys the temporary instance, and upon receiving the failure signal, the source end automatically unlocks, restores write permissions, and achieves atomic rollback.

[0014] 5.3 Specific Implementation of the Instance Locking Unit The instance locking unit maintains a global state registry (which can be implemented based on a distributed consensus algorithm such as Raft). States include: active, dormant, migrating, and locked. When any instance attempts to activate, it must request an "active token" from the locking unit. If the locking unit detects that another instance already holds a token, it rejects the new request or forces the old instance into dormancy, ensuring "singleton operation".

[0015] 5.4 The specific implementation of the source-side processing unit supports two modes: "cut" and "copy". In "cut" mode, after successful migration, the source instance is archived as "dormant" or directly destroyed, releasing computing power. In "copy" mode (such as in a clone scenario), the source instance remains in "read-only" or "restricted interaction" state, determined by the user's policy. The "operational anomalies" include, but are not limited to: target instance response timeout, status verification failure, insufficient resources, service crash, and network interruption. When any of the above anomalies is detected, the system automatically triggers a reverse recovery process.

Claims

1. A system for cross-carrier migration and instance uniqueness control of AI intelligent agents, characterized in that, include: The system comprises an identity generation unit, a migration control unit, an instance locking unit, and a source-end disposal unit. The identity generation unit is configured to generate a globally unique identity identifier bound to the agent and remaining unchanged throughout its entire lifecycle. The migration control unit is configured to generate a one-time, non-reusable exclusive migration credential, lock the source agent instance during the migration process, prohibiting the source agent instance from performing interactive operations during the migration, and monitor the migration status. If the migration fails, an atomic rollback operation is triggered to restore the source agent instance to its state before migration. The instance locking unit is configured to ensure that only one legitimate instance is active at a time in a commercial operating environment. The source-end disposal unit is configured to perform disposal operations on the source agent instance according to a preset strategy after the migration is completed.

2. The system according to claim 1, characterized in that, The atomic rollback operation includes: the target destroys the temporary instance, the source unlocks and restores write permissions, and the agent state is rolled back to the most recent consistent snapshot before the migration was triggered.

3. The system according to claim 1, characterized in that, The instance locking unit covers any logical entity that carries the operational state of an intelligent agent.

4. The system according to claim 1, characterized in that, The migration control unit uses a distributed protocol or centralized coordination service to ensure that only one instance of the same agent obtains the activation token at any given time.

5. The system according to claim 1, characterized in that, The source-end processing unit is configured to perform archiving, hibernation, or destruction processing on the source-end agent instance according to a preset strategy after successful migration.

6. The system according to claim 1, characterized in that, It also includes a traceability and evidence storage unit, which is used to generate tamper-proof evidence records for identity generation, migration, locking, and disposal operations.

7. A method for cross-carrier migration and instance uniqueness control of AI intelligent agents, characterized in that, include: Identity generation step: Generate a globally unique identity identifier that is bound to the agent and remains unchanged throughout its entire lifecycle; Credential generation and migration step: Generate a one-time, non-reusable exclusive migration credential, lock the source instance, and perform state transfer; Instance locking step: Activate the instance on the target side and ensure that only one legitimate instance is active at a time; Source instance disposal step: Dispose of the source instance after successful migration, or perform an atomic rollback if migration fails.

8. The method according to claim 7, characterized in that, The atomic rollback operation includes: the target destroys the temporary instance, the source unlocks and restores write permissions, and the agent state is rolled back to the most recent consistent snapshot before the migration was triggered.

9. The method according to claim 7, characterized in that, It also includes a reverse recovery step: when the target instance fails to activate or malfunctions, the system automatically triggers the reverse recovery process, returning control to the source instance so that the agent can continue to provide services on the original carrier.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method of any one of claims 7 to 9.

Images