Resource allocation system and method for 5G-A Internet of Vehicles network slices

Abstract

The invention relates to a resource allocation system for 5G-A Internet of Vehicles network slices, which belongs to the technical field of intelligent traffic systems and comprises a network slice architecture module, a management and control plane module, a programmable module, a safety and fault-tolerant module, a performance monitoring and optimizing module, a mobility management module, a resource arbitration engine and a cross-domain collaborative interface. And the network slice architecture module is used for dividing Internet of Vehicles communication into a plurality of virtual slices based on a 5G-A network slice technology. According to the resource allocation system and method oriented to the 5G-A Internet of Vehicles network slices, accurate control over end-to-end performance indexes is achieved through the slice definition template library and a QoS dynamic mapping mechanism, the problem that key service performance is degraded due to one-step resource allocation of a traditional network is solved, distributed agent collaboration based on the MAPPO algorithm, and the resource allocation efficiency is improved. LSTM load prediction and a heuristic time delay equalization algorithm are combined, so that the system can respond to business demand changes at a millisecond level, and the resource utilization rate is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and specifically to a resource allocation system and method for 5G-A vehicle-to-vehicle network slicing. Background Art

[0002] With the in-depth application of 5G-A technology in the field of Internet of Vehicles, the differentiated demands for network resources from new services such as autonomous driving, vehicle-road collaboration, and remote diagnosis are becoming increasingly significant.

[0003] Traditional Internet of Vehicles architecture faces three core challenges: First, the performance conflict of heterogeneous services. Autonomous driving requires ultra-low latency and reliability, while in-vehicle video services require hundreds of Mbps of bandwidth. The traditional fixed resource allocation model is difficult to balance resource competition among multiple slices, which often leads to performance degradation of key services; second, the ability to adapt to dynamic scenarios is insufficient. The cross-domain switching of slices caused by high-speed movement of vehicles (such as base station switching every 100ms) and periodic load fluctuations during peak hours in the morning and evening make it difficult for traditional static resource scheduling mechanisms to meet real-time requirements, and are prone to sudden increases in latency and waste of resources; third, there are bottlenecks in security isolation and collaborative efficiency. Existing network slicing technology relies on traditional encryption algorithms and is difficult to resist security threats such as identity forgery and DDoS attacks unique to the Internet of Vehicles; the configuration of cross-operator network slices lacks standardized collaborative interfaces, resulting in long service interruptions when vehicles move across regions, seriously affecting user experience.

[0004] To address the above issues, existing technologies use heuristic algorithms or single-agent optimization to schedule resources, but have not formed a systematic solution for multi-module collaboration. In complex business scenarios, there are still defects such as insufficient scheduling accuracy and disconnection between security mechanisms and resource allocation. Therefore, it is very necessary to build a solution that integrates differentiated QoS guarantees, intelligent resource scheduling, active security protection, and seamless cross-domain collaboration through the deep integration of multi-agent reinforcement learning and dynamic slicing management technology. Therefore, a resource allocation system and method for 5G-A vehicle network slicing are proposed to solve the above problems. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a resource allocation system and method for 5G-A vehicle network slicing, which has the advantages of providing reliable network infrastructure support for emerging services such as autonomous driving, and solves the problem of resource allocation in the mixed multi-service scenario of the vehicle network.

[0006] To achieve the above objectives, the present invention provides the following technical solutions: a resource allocation system for 5G-A vehicle network slicing, comprising: a network slicing architecture module, a management and control plane module, a programmability module, a security and fault tolerance module, a performance monitoring and optimization module, a mobility management module, a resource arbitration engine, and a cross-domain collaboration interface;

[0007] The network slicing architecture module is used to divide the Internet of Vehicles communication into multiple virtual slices based on 5G-A network slicing technology. Each slice is independently configured and supports large bandwidth, low latency, high reliability and strong security isolation requirements;

[0008] The network slicing architecture module consists of a slice instantiation and cycle management module, a resource allocation and isolation engine, a slice definition and template management module, a QoS guarantee and performance adaptation module, and a slice status monitoring interface.

[0009] The analysis slice instantiation and cycle management module is responsible for the full lifecycle management of slices, from creation, activation, modification, to destruction. It supports rapid on-demand instantiation (such as creating in-vehicle safety communication slices in milliseconds) and manages slice version iterations (such as configuration updates of software-defined network (SDN) controllers), ensuring that resources do not interfere with each other when multiple slices coexist.

[0010] The management and control plane module provides unified slice management functions, including slice creation, deletion, parameter adjustment, dynamic resource scheduling and monitoring, and supports real-time response to changes in the needs of Internet of Vehicles applications.

[0011] The programmable module supports programming and custom expansion of network functions and is integrated with Internet of Vehicles applications and services through open APIs;

[0012] The security and fault-tolerance module implements data encryption, identity authentication, access control, and redundant backup mechanisms to ensure slice confidentiality, integrity, and fault self-healing capabilities. In the slice access control chain, devices must be registered on the blockchain using a PKI certificate, and zero-knowledge proof is used during verification to ensure privacy.

[0013] The performance monitoring and optimization module monitors slice bandwidth utilization, latency, and packet loss rate in real time, and dynamically adjusts resource allocation to optimize service quality. The performance monitoring and optimization module further includes an LSTM-based prediction submodule that reserves resources in advance to cope with periodic high loads (such as morning and evening peaks).

[0014] Mobility management module: responsible for slice migration when vehicles move across regions (such as redirecting data flows through the SDN controller), and supports geo-fence-triggered resource reallocation;

[0015] Resource arbitration engine: defines priority rules when multiple slice resources conflict (for example, autonomous driving slices can preempt video slice bandwidth) and supports policy customization;

[0016] Cross-domain collaborative interface: Provides standardized interfaces with external operators or edge cloud platforms (e.g., based on ETSI MEC specifications) to enable cross-domain resource pool sharing.

[0017] Furthermore, the resource allocation and isolation engine implements dynamic allocation of physical / virtual resources (such as bandwidth, computing resources, and storage) and hard isolation (such as dividing independent logical links through network function virtualization (NFV) technology) between slices;

[0018] The slice status monitoring interface is responsible for providing a standardized interface (such as the Prometheus indicator interface and gRPC service) and outputting slice status data (such as resource utilization, number of connections, and error rate) in real time.

[0019] Furthermore, the slice definition and template management module provides a standardized slice template library (such as "autonomous driving safety slice template" and "vehicle HD video slice template"), pre-defining QoS parameters, security policies, resource specifications and other attributes;

[0020] The QoS guarantee and performance adaptation module dynamically maps QoS levels (such as the 5QI parameters defined by 3GPP) for different slices, ensuring key indicators such as latency (such as autonomous driving slices ≤ 10ms) and bandwidth (such as high-definition map slices ≥ 500Mbps).

[0021] Furthermore, the management and control plane module consists of a dynamic resource scheduling module, a policy configuration and parameter management module, a monitoring and data analysis module, and a cross-module collaborative interface layer;

[0022] The dynamic resource scheduling module dynamically allocates / reclaims network resources (such as bandwidth, computing power, and storage) based on real-time demand, supporting elastic resource sharing and priority scheduling among slices;

[0023] The policy configuration and parameter management module is responsible for defining resource allocation policies (such as latency priority and bandwidth priority), configuring slice parameters (such as QoS level and isolation level), and supporting dynamic policy loading and version management.

[0024] The monitoring and data analysis module is responsible for collecting slice operation status data (such as resource utilization and fault logs), analyzing the data to generate scheduling optimization suggestions, and providing a visual monitoring dashboard.

[0025] The cross-module collaborative interface layer defines standardized interface protocols (such as RESTful and gRPC), implements data synchronization and command transmission between modules, and supports heterogeneous system integration (such as third-party cloud platforms and vehicle terminal management systems).

[0026] The network slicing architecture module further includes a mobility management module, which is used to predict the coverage area that the vehicle is about to enter based on the vehicle GPS data and network topology information; migrate the slice instance to the target base station in advance through the SDN controller, and reserve resources to ensure seamless switching; and release the network resources occupied by the original area after the switching is completed.

[0027] Furthermore, the management and control plane module uses a multi-agent reinforcement learning (MAPPO) algorithm module to perform distributed agent collaborative optimization of network slice resource allocation strategies;

[0028] The multi-agent reinforcement learning (MAPPO) algorithm module includes a resource scheduling decision module and a policy gradient optimization module;

[0029] Resource scheduling decision module: Each intelligent agent corresponds to a network slice instance and generates resource scheduling decisions based on real-time business load and QoS requirements; Policy gradient optimization module: Dynamically adjusts slice bandwidth, latency and reliability parameters through the policy gradient optimization algorithm to ensure the performance of high-priority services such as autonomous driving and vehicle-road collaboration.

[0030] Furthermore, the security and fault tolerance module consists of a data encryption module, a slice access control module, a slice access control module, and an active-active redundant architecture;

[0031] Data encryption module: Based on the end-to-end data encryption mechanism of the national secret algorithm, it encrypts and transmits vehicle identity information and control instructions;

[0032] Slice access control module: Slice access control chain based on blockchain technology records and verifies the legitimacy of the identities of all access devices;

[0033] Active-active redundant architecture: Active-active redundant architecture design automatically switches to the backup slice and triggers resource reallocation when a single slice fails.

[0034] Furthermore, the security and fault tolerance module further includes a resource scheduling trigger module: based on security events or risk status, when the security and fault tolerance module detects security threats such as data leakage, malicious attacks (such as DDoS), identity forgery, etc., it triggers the resource scheduling mechanism to reallocate network resources (such as bandwidth, computing power, storage), and the security event response is linked to resource scheduling. For example, when a malicious node is detected, the slice is automatically isolated and resource reallocation is triggered;

[0035] The performance monitoring and optimization module further includes a strategy effect evaluation module: performing multi-dimensional analysis on the real-time data collected by the performance monitoring module (such as bandwidth utilization, latency, packet loss rate, slice throughput, etc.), and comparing the changes in indicators before and after the optimization strategy is implemented.

[0036] A resource allocation method for 5G-Advanced Internet of Vehicles (IoV) network slicing, comprising the following steps:

[0037] S1. Based on 5G-A network slicing technology, the Internet of Vehicles is divided into autonomous driving slices, vehicle-road collaboration slices, and remote diagnosis slices, each with differentiated bandwidth, latency, and reliability parameters.

[0038] S2: Use the multi-agent reinforcement learning (MAPPO) algorithm to dynamically adjust slice resource allocation and optimize mobile user association and slice resource scheduling based on real-time business needs;

[0039] S3: Through the heuristic delay balancing algorithm, network resources are allocated according to the delay ratio of different slices to achieve low latency and high concurrent access;

[0040] S4. Securely isolate the slices and ensure business security through encrypted communication, data integrity verification, and access control policies.

[0041] S5. Real-time monitoring of slice performance indicators, dynamic optimization of resource utilization and triggering of fault recovery mechanisms;

[0042] S6. Reclaim idle resources and reallocate them to high-priority services based on vehicle location changes or slice load status;

[0043] S7. Synchronize network slice configurations of multiple operators through cross-domain collaborative interfaces to ensure service continuity when vehicles move across regions.

[0044] Furthermore, the heuristic delay balancing algorithm in step S3 implements low-latency resource allocation through the following steps:

[0045] S3.1. Prioritize latency based on slice type. Autonomous driving slices are assigned the highest priority, followed by vehicle-infrastructure collaboration slices.

[0046] S3.2. Predict network congestion nodes based on historical latency data and dynamically adjust resource allocation to balance latency.

[0047] S3.3. Use a weighted round-robin mechanism to allocate redundant resource channels for high-concurrency access devices to avoid service interruptions.

[0048] Compared with the existing technology, the technical solution of this application has the following beneficial effects:

[0049] 1. This resource allocation system and method for 5G-Advanced vehicle network slicing achieves precise control of end-to-end performance indicators through a slice definition template library and a QoS dynamic mapping mechanism, solving the problem of key business performance degradation caused by the "one-size-fits-all" resource allocation of traditional networks.

[0050] 2. This resource allocation system and method for 5G-A vehicle network slicing is based on the distributed intelligent agent collaboration of the MAPPO algorithm, combined with LSTM load prediction and heuristic delay balancing algorithm, enabling the system to respond to changes in business needs in milliseconds and improve resource utilization.

[0051] 3. This resource allocation system and method for 5G-A vehicle network slicing integrates blockchain and zero-knowledge proof technology to build a closed-loop security mechanism for registration, verification, and access on the device identity chain, reducing the risk of data leakage; the active-active redundant architecture cooperates with resource reallocation triggered by security events to shorten fault recovery time.

[0052] 4. This resource allocation system and method for 5G-A vehicle network slicing uses slice pre-migration technology based on GPS and network topology to reduce packet loss when vehicles switch across base stations. Combined with dynamic resource adjustment triggered by geo-fences, it effectively solves the problem of service interruption in mobile scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 Schematic diagram of the overall framework of the system of the present invention;

[0054] Figure 2 This is a schematic diagram of the management and control screen module framework of the present invention;

[0055] Figure 3 This is a schematic diagram of the algorithm module framework of the present invention (MAPPO);

[0056] Figure 4 This is a schematic diagram of the safety and fault-tolerant module framework of the present invention. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0058] See also Figure 1 and 3 In this embodiment, a resource allocation system for 5G-A vehicle network slicing includes: a network slicing architecture module, a management and control plane module, a programmability module, a security and fault tolerance module, a performance monitoring and optimization module, a mobility management module, a resource arbitration engine, and a cross-domain collaboration interface.

[0059] The network slicing architecture module is used to divide the Internet of Vehicles communication into multiple virtual slices based on 5G-A network slicing technology. Each slice is independently configured and supports large bandwidth, low latency, high reliability and strong security isolation requirements;

[0060] The management and control plane module provides unified slice management functions, including slice creation, deletion, parameter adjustment, dynamic resource scheduling and monitoring, and supports real-time response to changes in the needs of Internet of Vehicles applications.

[0061] The programmable module supports programming and custom expansion of network functions and is integrated with Internet of Vehicles applications and services through open APIs;

[0062] The security and fault-tolerance module implements data encryption, identity authentication, access control, and redundant backup mechanisms to ensure slice confidentiality, integrity, and fault self-healing capabilities. In the slice access control chain, devices must be registered on the blockchain using a PKI certificate, and zero-knowledge proof is used during verification to ensure privacy.

[0063] The performance monitoring and optimization module monitors slice bandwidth utilization, latency, and packet loss rate in real time, and dynamically adjusts resource allocation to optimize service quality. The performance monitoring and optimization module further includes an LSTM-based prediction submodule that reserves resources in advance to cope with periodic high loads (such as morning and evening peaks).

[0064] Mobility management module: responsible for slice migration when vehicles move across regions (such as redirecting data flows through the SDN controller), and supports geo-fence-triggered resource reallocation;

[0065] Resource arbitration engine: defines priority rules when multiple slice resources conflict (for example, autonomous driving slices can preempt video slice bandwidth) and supports policy customization;

[0066] Cross-domain collaborative interface: Provides standardized interfaces with external operators or edge cloud platforms (e.g., based on ETSI MEC specifications) to enable cross-domain resource pool sharing.

[0067] In this embodiment, the IoV is divided into multiple virtual slices, with independent configuration of parameters such as bandwidth, latency, and reliability. This ensures that the Quality of Service (QoS) of different services, such as autonomous driving (low latency) and in-car entertainment (high bandwidth), do not interfere with each other. For example, the autonomous driving slice can allocate fixed, reliable dedicated channels to prevent other services from preempting resources. Historical traffic data (such as morning and evening peaks) can be used to predict load fluctuations and reserve resources in advance. Combined with the resource arbitration engine's priority rules (such as the autonomous driving slice preempting video bandwidth), this achieves a shift from "passive response" to "active optimization," reducing packet loss caused by bursty traffic. Furthermore, through zero-knowledge proofs (ZKPs), privacy protection is achieved (for example, vehicle location verification does not disclose specific coordinates), addressing the pain points of identity forgery and data leakage in the IoV.

[0068] It should be noted that the network slicing architecture module consists of a slice instantiation analysis and cycle management module, a resource allocation and isolation engine, a slice definition and template management module, a QoS guarantee and performance adaptation module, and a slice status monitoring interface.

[0069] The analysis slice instantiation and cycle management module is responsible for the full lifecycle management of slices, from creation, activation, modification, to destruction. It supports rapid on-demand instantiation (such as creating in-vehicle safety communication slices in milliseconds), manages slice version iterations (such as configuration updates of software-defined network (SDN) controllers), ensures that resources do not interfere with each other when multiple slices coexist, supports slice sleep / wake-up mechanisms, and releases idle resources in low-load scenarios (such as in-vehicle entertainment slices during non-peak hours at night).

[0070] The resource allocation and isolation engine implements dynamic allocation of physical / virtual resources (such as bandwidth, computing resources, and storage) and hard isolation between slices (such as dividing independent logical links through network function virtualization (NFV) technology, supporting heterogeneous resource pool management (such as edge cloud resource pool and central cloud resource pool), optimizing deployment locations based on slice requirements (such as prioritizing edge resources for low-latency slices), and providing resource conflict resolution mechanisms (such as priority preemption, where high-priority slices (autonomous driving) can preempt idle resources from low-priority slices (in-car music)).

[0071] The slice status monitoring interface is responsible for providing standardized interfaces (such as Prometheus indicator interface and gRPC service), outputting slice status data (such as resource utilization, number of connections, and error rate) in real time, supporting northbound interface docking with monitoring and data analysis modules, and providing decision-making basis for the management plane (such as triggering expansion warning when the bandwidth utilization of a certain slice is continuously greater than 85%). It is compatible with the southbound interface to collect the underlying hardware status (such as server CPU load and switch port traffic), realizing end-to-end status observability.

[0072] The slice definition and template management module provides a standardized slice template library (such as "autonomous driving safety slice template" and "in-vehicle HD video slice template"), pre-defines QoS parameters, security policies, resource specifications and other attributes, supports user-defined templates (through a graphical interface or API), quickly replicates and deploys similar slices in batches (such as multiple vehicles in a fleet reusing the same slice template), manages template version compatibility (such as slice templates compatible with 5G-A NR and LTE dual-mode terminals), and reduces the cost of cross-generation network deployment;

[0073] The QoS guarantee and performance adaptation module dynamically maps QoS levels (such as the 5QI parameters defined by 3GPP) for different slices, guarantees key indicators such as latency (such as autonomous driving slices ≤10ms) and bandwidth (such as high-definition map slices ≥500Mbps), monitors slice performance fluctuations in real time (such as latency jumps caused by bursty traffic), and achieves performance adaptation by dynamically adjusting scheduling strategies (such as weighted fair queuing (WFQ)). It supports cross-layer QoS linkage (such as coordinated optimization of physical layer MIMO technology and application layer congestion control) and improves end-to-end service quality.

[0074] Overall effect, dynamic slicing and full-link optimization:

[0075] The slice definition and template management module pre-configures a "high-reliability slice template", which includes QoS parameters (latency ≤ 5ms) and resource specifications (redundant deployment of two edge nodes); the slice instantiation and cycle management module quickly creates slice instances based on the template according to the request of the on-board terminal, and triggers the resource allocation and isolation engine to divide independent computing / network resources; the QoS guarantee and performance adaptation module monitors the slice latency in real time, and automatically adjusts the traffic routing to the redundant node when it detects that the latency of an edge node suddenly increases to 8ms; the slice status monitoring interface reports the latency fluctuation data to the monitoring analysis, which recommends optimizing the template's redundancy strategy (such as adding a third backup node) after analysis; the management and control plane module updates the template according to the suggestion and automatically applies the new strategy at the next instantiation.

[0076] In addition, the management and control plane module uses a multi-agent reinforcement learning (MAPPO) algorithm module to perform distributed agent collaborative optimization of network slice resource allocation strategies;

[0077] The multi-agent reinforcement learning (MAPPO) algorithm module includes a resource scheduling decision module and a policy gradient optimization module;

[0078] Resource Scheduling Decision Module: Each agent corresponds to a network slice instance and generates resource scheduling decisions based on real-time service load and QoS requirements. Policy Gradient Optimization Module: Dynamically adjusts slice bandwidth, latency, and reliability parameters using a policy gradient optimization algorithm to ensure performance for high-priority services such as autonomous driving and vehicle-infrastructure collaboration.

[0079] The multi-agent reinforcement learning (MAPPO) algorithm module adds an inter-agent communication protocol (such as exchanging resource status based on the Pub / Sub model) to the resource scheduling decision module; in the policy gradient optimization module, it introduces a mechanism combining offline simulation training with online fine-tuning to improve the robustness of the algorithm.

[0080] See also Figure 2 , the management and control plane module in this embodiment is composed of a dynamic resource scheduling module, a policy configuration and parameter management module, a monitoring and data analysis module and a cross-module collaborative interface layer;

[0081] The dynamic resource scheduling module dynamically allocates / reclaims network resources (such as bandwidth, computing power, and storage) based on real-time demand, supporting elastic resource sharing and priority scheduling among slices;

[0082] The policy configuration and parameter management module is responsible for defining resource allocation policies (such as latency priority and bandwidth priority), configuring slice parameters (such as QoS level and isolation level), and supporting dynamic policy loading and version management.

[0083] The monitoring and data analysis module is responsible for collecting slice operation status data (such as resource utilization and fault logs), analyzing the data to generate scheduling optimization suggestions, and providing a visual monitoring dashboard.

[0084] The cross-module collaborative interface layer defines standardized interface protocols (such as RESTful and gRPC), implements data synchronization and command transmission between modules, and supports heterogeneous system integration (such as third-party cloud platforms and vehicle terminal management systems).

[0085] Scenario example: When the monitoring and data analysis module detects that the latency of an autonomous driving slice in a certain area exceeds the standard, the dynamic resource scheduling module immediately reclaims idle bandwidth from low-priority slices (such as in-vehicle video streaming) to prioritize autonomous driving signal transmission, reducing latency by more than 30%. When the security and fault tolerance module detects that a slice has been attacked by DDoS, it quickly notifies the dynamic resource scheduling module through the interface layer to isolate the attack traffic, and links the programmability module to call the firewall API to block the attack source, allowing the entire response process to be completed quickly.

[0086] The network slicing architecture module further includes a mobility management module, which is used to predict the coverage area that the vehicle is about to enter based on the vehicle GPS data and network topology information; migrate the slice instance to the target base station in advance through the SDN controller, and reserve resources to ensure seamless switching; and release the network resources occupied by the original area after the switching is completed.

[0087] In this embodiment, by real-time perception of business needs (such as latency sensitivity of autonomous driving) combined with a priority preemption mechanism (reclaiming bandwidth from video slices), millisecond-level resource reallocation is achieved, which reduces the latency of key services, meets the Internet of Vehicles' need for rapid response to burst traffic, improves overall resource utilization, and is expected to reduce the waste of idle resources.

[0088] See also Figure 1 and 4 ,The security and fault tolerance module described in this embodiment consists of a data encryption module, a slice access control module, a slice access control module, and an active-active redundant architecture;

[0089] Data encryption module: Based on the end-to-end data encryption mechanism of the national secret algorithm, it encrypts and transmits vehicle identity information and control instructions;

[0090] Slice access control module: Slice access control chain based on blockchain technology records and verifies the legitimacy of the identities of all access devices;

[0091] Active-active redundant architecture: Active-active redundant architecture design automatically switches to the backup slice and triggers resource reallocation when a single slice fails;

[0092] The security and fault tolerance module further includes a resource scheduling trigger module: based on security events or risk status, when the security and fault tolerance module detects security threats such as data leakage, malicious attacks (such as DDoS), identity forgery, etc., it triggers the resource scheduling mechanism and reallocates network resources (such as bandwidth, computing power, storage), and the security event response is linked to resource scheduling. For example, when a malicious node is detected, the slice is automatically isolated and resource reallocation is triggered. The "resource scheduling trigger module" is introduced in the security module, and its positioning should be: based on security events or risk status, it triggers response actions at the resource level, forming a closed loop of "security monitoring → risk assessment → resource adjustment", thereby enhancing the dynamic and proactive nature of security protection;

[0093] The performance monitoring and optimization module further includes a strategy effect evaluation module: performing multi-dimensional analysis on the real-time data collected by the performance monitoring module (such as bandwidth utilization, latency, packet loss rate, slice throughput, etc.), and comparing the changes in indicators before and after the optimization strategy is implemented.

[0094] In this embodiment, the data encryption module ensures the confidentiality of vehicle identity information and control command transmission while meeting domestic regulatory compliance requirements and avoiding potential backdoor risks of overseas encryption standards (such as AES). The slice access control module records device identity and access logs through blockchain to ensure that data cannot be tampered with (for example, when a malicious node forges an identity, the historical records on the chain can be traced). It combines zero-knowledge proof (ZKP) to achieve privacy protection. The resource scheduling trigger module directly links security events (such as DDoS attacks) with resource reallocation, forming a closed loop of "detection → judgment → isolation → recovery". The active-active slice instances synchronize data in real time and, combined with the resource reallocation mechanism, ensure the continuity of key services (such as autonomous driving).

[0095] A resource allocation method for 5G-Advanced Internet of Vehicles (IoV) network slicing, comprising the following steps:

[0096] S1. Based on 5G-A network slicing technology, the Internet of Vehicles is divided into autonomous driving slices, vehicle-road collaboration slices, and remote diagnosis slices, each with differentiated bandwidth, latency, and reliability parameters.

[0097] S2: Use the multi-agent reinforcement learning (MAPPO) algorithm to dynamically adjust slice resource allocation and optimize mobile user association and slice resource scheduling based on real-time business needs;

[0098] S3: Through the heuristic delay balancing algorithm, network resources are allocated according to the delay ratio of different slices to achieve low latency and high concurrent access;

[0099] S4. Securely isolate the slices and ensure business security through encrypted communication, data integrity verification, and access control policies.

[0100] S5. Real-time monitoring of slice performance indicators, dynamic optimization of resource utilization and triggering of fault recovery mechanisms;

[0101] S6. Reclaim idle resources and reallocate them to high-priority services based on vehicle location changes or slice load status;

[0102] S7. Synchronize network slice configurations of multiple operators through cross-domain collaborative interfaces to ensure service continuity when vehicles move across regions.

[0103] It should be noted that the heuristic delay balancing algorithm in step S3 implements low-latency resource allocation through the following steps:

[0104] S3.1. Prioritize latency based on slice type. Autonomous driving slices are assigned the highest priority, followed by vehicle-infrastructure collaboration slices.

[0105] S3.2. Predict network congestion nodes based on historical latency data and dynamically adjust resource allocation to balance latency.

[0106] S3.3. Use a weighted round-robin mechanism to allocate redundant resource channels for high-concurrency access devices to avoid service interruptions.

[0107] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0108] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A resource allocation system for 5G-A vehicle network slicing, characterized by: include: Network slicing architecture module, management and control plane module, programmability module, security and fault tolerance module, performance monitoring and optimization module, mobility management module, resource arbitration engine and cross-domain collaboration interface; The network slicing architecture module is used to divide the Internet of Vehicles communication into multiple virtual slices based on 5G-A network slicing technology. Each slice is independently configured and supports large bandwidth, low latency, high reliability and strong security isolation requirements; The network slicing architecture module consists of a slice instantiation and cycle management module, a resource allocation and isolation engine, a slice definition and template management module, a QoS guarantee and performance adaptation module, and a slice status monitoring interface. The analysis slice instantiation and cycle management module is responsible for the full lifecycle management of slices, from creation, activation, modification, to destruction. It supports rapid on-demand instantiation (such as creating in-vehicle safety communication slices in milliseconds) and manages slice version iterations (such as configuration updates of software-defined network (SDN) controllers), ensuring that resources do not interfere with each other when multiple slices coexist. The management and control plane module provides unified slice management functions, including slice creation, deletion, parameter adjustment, dynamic resource scheduling and monitoring, and supports real-time response to changes in the needs of Internet of Vehicles applications. The programmable module supports programming and custom expansion of network functions and is integrated with Internet of Vehicles applications and services through open APIs; The security and fault-tolerance module implements data encryption, identity authentication, access control, and redundant backup mechanisms to ensure slice confidentiality, integrity, and fault self-healing capabilities. In the slice access control chain, devices must be registered on the blockchain using a PKI certificate, and zero-knowledge proof is used during verification to ensure privacy. The performance monitoring and optimization module monitors slice bandwidth utilization, latency, and packet loss rate in real time, and dynamically adjusts resource allocation to optimize service quality. The performance monitoring and optimization module further includes an LSTM-based prediction submodule that reserves resources in advance to cope with periodic high loads (such as morning and evening peaks). Mobility management module: responsible for slice migration when vehicles move across regions (such as redirecting data flows through the SDN controller), and supports geo-fence-triggered resource reallocation; Resource arbitration engine: defines priority rules when multiple slice resources conflict (for example, autonomous driving slices can preempt video slice bandwidth) and supports policy customization; Cross-domain collaborative interface: Provides standardized interfaces with external operators or edge cloud platforms (e.g., based on ETSI MEC specifications) to enable cross-domain resource pool sharing.

2. A resource allocation system for 5G-Advanced Internet of Vehicles network slicing according to claim 1, characterized in that: The resource allocation and isolation engine implements dynamic allocation of physical / virtual resources (such as bandwidth, computing resources, and storage) and hard isolation (such as dividing independent logical links through network function virtualization (NFV) technology) between slices; The slice status monitoring interface is responsible for providing a standardized interface (such as the Prometheus indicator interface and gRPC service) and outputting slice status data (such as resource utilization, number of connections, and error rate) in real time.

3. The resource allocation system for 5G-Advanced Internet of Vehicles network slicing according to claim 2, characterized in that: The slice definition and template management module provides a standardized slice template library (such as "autonomous driving safety slice template" and "vehicle HD video slice template"), pre-defines QoS parameters, security policies, resource specifications and other attributes; The QoS guarantee and performance adaptation module dynamically maps QoS levels (such as the 5QI parameters defined by 3GPP) for different slices, ensuring key indicators such as latency (such as autonomous driving slices ≤ 10ms) and bandwidth (such as high-definition map slices ≥ 500Mbps).

4. The resource allocation system for 5G-Advanced Internet of Vehicles network slicing according to claim 1, characterized in that: The management and control plane module consists of a dynamic resource scheduling module, a policy configuration and parameter management module, a monitoring and data analysis module, and a cross-module collaborative interface layer; The dynamic resource scheduling module dynamically allocates / reclaims network resources (such as bandwidth, computing power, and storage) based on real-time demand, supporting elastic resource sharing and priority scheduling among slices; The policy configuration and parameter management module is responsible for defining resource allocation policies (such as latency priority and bandwidth priority), configuring slice parameters (such as QoS level and isolation level), and supporting dynamic policy loading and version management. The monitoring and data analysis module is responsible for collecting slice operation status data (such as resource utilization and fault logs), analyzing the data to generate scheduling optimization suggestions, and providing a visual monitoring dashboard. The cross-module collaborative interface layer defines standardized interface protocols (such as RESTful and gRPC), implements data synchronization and command transmission between modules, and supports heterogeneous system integration (such as third-party cloud platforms and vehicle terminal management systems).

5. The network slicing architecture module further includes a mobility management module, which is used to predict the coverage area that the vehicle is about to enter based on the vehicle GPS data and network topology information; migrate the slice instance to the target base station in advance through the SDN controller, and reserve resources to ensure seamless switching; and release the network resources occupied by the original area after the switch is completed.

6. The resource allocation system for 5G-Advanced Internet of Vehicles network slicing according to claim 1, characterized in that: The management and control plane module uses a multi-agent reinforcement learning (MAPPO) algorithm module to perform distributed agent collaborative optimization of network slice resource allocation strategies; The multi-agent reinforcement learning (MAPPO) algorithm module includes a resource scheduling decision module and a policy gradient optimization module; Resource scheduling decision module: Each intelligent agent corresponds to a network slice instance and generates resource scheduling decisions based on real-time business load and QoS requirements; Policy gradient optimization module: Dynamically adjusts slice bandwidth, latency and reliability parameters through the policy gradient optimization algorithm to ensure the performance of high-priority services such as autonomous driving and vehicle-road collaboration.

7. The resource allocation system for 5G-Advanced Internet of Vehicles network slicing according to claim 1, characterized in that: The security and fault tolerance module consists of a data encryption module, a slice access control module, a slice access control module and an active-active redundant architecture; Data encryption module: Based on the end-to-end data encryption mechanism of the national secret algorithm, it encrypts and transmits vehicle identity information and control instructions; Slice access control module: Slice access control chain based on blockchain technology records and verifies the legitimacy of the identities of all access devices; Active-active redundant architecture: Active-active redundant architecture design automatically switches to the backup slice and triggers resource reallocation when a single slice fails.

8. The resource allocation system for 5G-Advanced Internet of Vehicles network slicing according to claim 1, characterized in that: The security and fault tolerance module further includes a resource scheduling trigger module: based on security events or risk status, when the security and fault tolerance module detects security threats such as data leakage, malicious attacks (such as DDoS), identity forgery, etc., it triggers the resource scheduling mechanism to reallocate network resources (such as bandwidth, computing power, storage), and the security event response is linked to resource scheduling. For example, when a malicious node is detected, the slice is automatically isolated and resource reallocation is triggered; The performance monitoring and optimization module further includes a strategy effect evaluation module: performing multi-dimensional analysis on the real-time data collected by the performance monitoring module (such as bandwidth utilization, latency, packet loss rate, slice throughput, etc.), and comparing the changes in indicators before and after the optimization strategy is implemented.

9. A resource allocation method for 5G-A vehicle network slicing, characterized in that: The following steps are involved: S1. Based on 5G-A network slicing technology, the Internet of Vehicles is divided into autonomous driving slices, vehicle-road collaboration slices, and remote diagnosis slices, each with differentiated bandwidth, latency, and reliability parameters. S2: Use the multi-agent reinforcement learning (MAPPO) algorithm to dynamically adjust slice resource allocation and optimize mobile user association and slice resource scheduling based on real-time business needs; S3: Through the heuristic delay balancing algorithm, network resources are allocated according to the delay ratio of different slices to achieve low latency and high concurrent access; S4. Securely isolate the slices and ensure business security through encrypted communication, data integrity verification, and access control policies. S5. Real-time monitoring of slice performance indicators, dynamic optimization of resource utilization and triggering of fault recovery mechanisms; S6. Reclaim idle resources and reallocate them to high-priority services based on vehicle location changes or slice load status; S7. Synchronize network slice configurations of multiple operators through cross-domain collaborative interfaces to ensure service continuity when vehicles move across regions.

10. The resource allocation method for 5G-Advanced Internet of Vehicles network slicing according to claim 8, characterized in that: The heuristic delay balancing algorithm in step S3 implements low-latency resource allocation through the following steps: S3.

1. Prioritize latency based on slice type. Autonomous driving slices are assigned the highest priority, followed by vehicle-infrastructure collaboration slices. S3.

2. Predict network congestion nodes based on historical latency data and dynamically adjust resource allocation to balance latency. S3.

3. Use a weighted round-robin mechanism to allocate redundant resource channels for high-concurrency access devices to avoid service interruptions.

Images