Implementation method and system of intelligent fusion deterministic network based on IPv6

Abstract

The invention relates to a method and a system for realizing an intelligent fusion deterministic network based on IPv6 (Internet Protocol Version 6). The method comprises the following steps: firstly, collecting service flow data of an access network, identifying various high-requirement service flows through feature analysis, and extracting service quality parameters of the high-requirement service flows; then inputting the parameters into a multi-dimensional resource mapping model, and dynamically allocating differentiated virtual network slices for different services according to the real-time state of the network; calculating an optimal forwarding path based on slice constraints and service quality requirements by adopting a reinforcement learning algorithm, and generating an IPv6 segment routing strategy; according to the strategy, IPv6 extension head packaging is carried out on the data packet, a path instruction and a service identifier are embedded, and forwarding is carried out through a special slice plane; and finally, by monitoring the network performance in real time, dynamically comparing a service quality target, and triggering path re-optimization and updating when the deviation exceeds a threshold value. According to the invention, intelligent distribution of network resources and cross-domain path collaborative optimization can be realized, and end-to-end deterministic service quality is effectively guaranteed.

Description

Technical Field

[0001] This invention relates to the field of communication technology, specifically to a method and system for implementing an intelligent converged deterministic network based on IPv6. Background Technology

[0002] With the rapid development of the Industrial Internet, the Internet of Vehicles, and immersive interactive services, traditional IPv4 networks face severe challenges in terms of address space and quality of service (QoS) assurance. While the IPv6 protocol provides ample address resources, its native design lacks deep support for the deterministic transmission requirements of diverse services. Existing network architectures typically employ static resource configuration or simple priority scheduling mechanisms, making it difficult to adapt to the dynamic changes in demanding service scenarios such as real-time interaction and large-scale machine communication. Current networks, when carrying converged services, cannot achieve controllable and manageable end-to-end QoS, particularly lacking an intelligent mechanism for dynamically allocating network resources and optimizing transmission paths in real time based on service needs. Therefore, how to build an IPv6 network capable of intelligently sensing service needs, dynamically allocating resources, and ensuring end-to-end deterministic QoS has become a critical technical problem that urgently needs to be solved. Summary of the Invention

[0003] To address the shortcomings of traditional IPv4 networks in terms of address space and quality of service guarantees, and the lack of deep support for the deterministic transmission requirements of diverse services in the native design of the IPv6 protocol, the present invention aims to provide a method and system for implementing an intelligent converged deterministic network based on IPv6.

[0004] To achieve the aforementioned objectives, the first aspect of this invention provides a method for implementing an intelligent converged deterministic network based on IPv6, the details of which are as follows:

[0005] An implementation method for an IPv6-based intelligent converged deterministic network includes the following steps:

[0006] Collect service flow data from the access network, identify real-time interactive, large-scale machine communication, and ultra-reliable low-latency communication service flows through feature parsing, and extract the service quality parameter requirements for each type of service flow;

[0007] The service quality parameters are input into a multi-dimensional resource mapping model, and virtual network slices with differentiated guarantee capabilities are dynamically allocated to different service types based on the real-time status of the network infrastructure.

[0008] Based on the resource constraints and service quality requirements of virtual network slices, a reinforcement learning algorithm is used to calculate the optimal forwarding path that meets the end-to-end deterministic index, and an IPv6 segmented routing strategy containing explicit node sequences is generated.

[0009] According to the IPv6 segmentation routing policy, the service flow data packets are encapsulated with IPv6 extension headers, embedding path instructions and quality of service identifiers, and then the data is transmitted through the corresponding virtual network slice forwarding plane.

[0010] Monitor network performance parameters in real time and compare them with target quality of service requirements:

[0011] When the performance deviation exceeds the dynamic threshold, the path re-optimization mechanism is triggered, the IPv6 segmentation routing policy is updated and fed back to the forwarding plane for adjustment.

[0012] Preferred options also include,

[0013] Collect service flow data from the access network and identify service flow types through feature parsing, specifically including:

[0014] Employ deep packet inspection technology or machine learning flow classification methods to distinguish between real-time interactive, large-scale machine communication, and ultra-reliable low-latency communication service flows;

[0015] Extract end-to-end service quality data for various service flows, including latency, jitter, packet loss rate, and bandwidth guarantee requirements.

[0016] Preferably, the method for constructing the multi-dimensional resource mapping model includes:

[0017] Establish a three-dimensional mapping relationship that includes time-domain resources, spatial-domain resources, and frequency-domain resources;

[0018] Based on the service quality requirements data of the business flow and the real-time status of the network infrastructure, virtual network slices are dynamically allocated for several types of business.

[0019] Preferably, the step of using a reinforcement learning algorithm to calculate the optimal forwarding path specifically includes:

[0020] With end-to-end latency limit and tolerable jitter as constraints, an IPv6 explicit routing sequence that meets the requirements of deterministic transmission is generated based on multi-agent collaborative decision-making. The IPv6 explicit routing sequence includes a segmented routing node list.

[0021] Preferably, the encapsulation of the service flow data packet with an IPv6 extension header includes:

[0022] Insert a segmented routing header after the IPv6 header, explicitly specifying the sequence of path nodes in the IPv6 header;

[0023] Add a custom option field, which carries the service level identifier and slice binding identifier, for intermediate nodes to perform low-latency switching and priority scheduling.

[0024] Preferably, the real-time monitoring of network performance parameters and triggering path re-optimization includes:

[0025] Based on the actual statistical values ​​of latency, throughput, and packet loss rate collected within a sliding time window, the relative deviations from the service quality target are calculated:

[0026] When the deviation exceeds the dynamic adaptive threshold for multiple consecutive periods, a path re-optimization process based on network state prediction is triggered, updating the IPv6 segmentation routing policy and feeding it back to the data plane.

[0027] It should be noted that multiple cycles refer to: collecting actual statistical values ​​of latency, throughput, and packet loss rate based on a sliding time window (e.g., 100ms / window), calculating the relative deviation of each parameter from the service quality target using a preset algorithm (e.g., deviation rate calculation formula); setting a continuous monitoring cycle (e.g., 5 consecutive cycles, each cycle synchronized with the sliding time window), and determining whether the deviation exceeds a dynamic adaptive threshold in each cycle (this threshold is dynamically adjusted according to network load, e.g., 10% for light load and 8% for heavy load). If the deviation in each of the 5 consecutive cycles exceeds the corresponding threshold, and there is no temporary network jitter interference (excluded by jitter detection algorithm), then a path re-optimization process based on network state prediction is triggered. The path node sequence in the IPv6 segmented routing policy is updated through a routing algorithm, and the updated policy is fed back to the data plane in real time to execute the new route.

[0028] This invention also provides an IPv6-based intelligent converged deterministic network system, applicable to any one of the IPv6-based intelligent converged deterministic network implementation methods described above, comprising:

[0029] The service identification module is used to collect service flow data from the access network, identify service flow types through feature parsing, and extract the service quality parameter requirements for various service flows.

[0030] The resource mapping module is used to dynamically allocate virtual network slices to several service types based on the service quality parameter requirements and real-time network status through a multi-dimensional resource mapping model.

[0031] The path calculation module, based on virtual network slicing constraints and quality of service requirements, uses reinforcement learning algorithms to calculate the optimal forwarding path and generate IPv6 segmented routing strategies.

[0032] The encapsulation and forwarding module is used to encapsulate data packets with IPv6 extended headers according to the routing policy, embed path instructions and quality of service identifiers, and transmit data through the corresponding virtual network slice forwarding plane.

[0033] The monitoring and optimization module is used to monitor network performance in real time. When the performance deviation exceeds the threshold, it triggers path re-optimization, updates the IPv6 segmentation routing policy, and feeds back to the encapsulation and forwarding module.

[0034] Preferably, the service identification module includes a deep packet inspection unit and a machine learning classification unit, used to distinguish different service types through protocol feature analysis and traffic behavior pattern identification.

[0035] Preferably, the resource mapping module includes a multi-dimensional resource modeling unit, which is used to establish resource mapping relationships in the time domain, spatial domain, and frequency domain, and to dynamically allocate virtual network slices according to the real-time network status.

[0036] Preferably, the monitoring and optimization module includes a performance acquisition unit and a dynamic decision-making unit, which are used to collect network performance data based on a time window and trigger a path re-optimization process based on the deviation analysis results.

[0037] Compared with the prior art, the present invention has at least the following technical effects:

[0038] This invention provides a method and system for implementing intelligent converged deterministic networks based on IPv6. Through an innovative technical architecture, it effectively solves the fundamental limitations of traditional networks in carrying diverse services. The system can dynamically allocate differentiated virtual network slices according to real-time service requirements. By accurately mapping time, space, and frequency domain resources, it provides tailored resource guarantees for service flows of different priorities, completely changing the problem of low resource utilization in the static resource allocation mode of traditional networks. Employing reinforcement learning algorithms to intelligently calculate the optimal forwarding path, it significantly improves network resource utilization efficiency while meeting strict end-to-end latency and jitter constraints. The generated explicit IPv6 routing sequences can effectively avoid network congestion points, ensuring deterministic transmission of critical services. The intelligent converged deterministic network system based on IPv6 in this invention establishes a complete real-time monitoring and dynamic optimization mechanism. By continuously comparing actual performance with the target value of the Service Level Agreement (SLA), it immediately triggers a path re-optimization process once a performance deviation is detected. This closed-loop control mechanism enables the network to have continuous self-optimization capabilities. This invention constitutes an organic whole capable of intelligently sensing business needs, dynamically allocating network resources, and ensuring transmission performance in real time. Ultimately, it achieves true end-to-end deterministic quality of service assurance under the IPv6 protocol framework, providing reliable basic network support for application scenarios with stringent network performance requirements, such as the Industrial Internet and the Internet of Vehicles. Attached Figure Description

[0039] Figure 1 This is a flowchart of a preferred embodiment of the intelligent converged deterministic network implementation method based on IPv6 of the present invention.

[0040] Figure 2 This is a schematic diagram of the structure of a preferred embodiment of the intelligent converged deterministic network implementation system based on IPv6 of the present invention;

[0041] Figure 3 This is a schematic diagram of a preferred embodiment of an electronic device according to the present invention. Detailed Implementation

[0042] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the invention, but are not intended to limit the invention. It should be noted that all directional indicators (such as up, down, left, right, front, back, upper end, lower end, top, bottom, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of the components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indicator will also change accordingly.

[0043] In this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral part; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0044] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. If the combination of technical solutions is contradictory or impossible to implement, such a combination should be considered non-existent and not within the scope of protection claimed by this invention.

[0045] The present invention will be further described in detail below with reference to the accompanying drawings.

[0046] In existing technologies, traditional IPv4 networks are limited by insufficient address space and lack quality of service (QoS) guarantee mechanisms, making it difficult to meet the diverse business needs of scenarios such as the Industrial Internet and the Internet of Vehicles. Although existing IPv6 networks expand address resources, their native architecture does not fully consider the differentiated transmission requirements of real-time interactive services, large-scale machine communication, and ultra-reliable low-latency communication services. Static resource configuration methods cannot adapt to dynamically changing network loads, and priority scheduling mechanisms cannot achieve end-to-end deterministic transmission, resulting in the inability to effectively guarantee the QoS of high-value services.

[0047] To address the technical problem that existing networks lack the ability to provide differentiated guarantees for diverse services, resulting in low resource utilization efficiency and failing to meet the stringent requirements for deterministic transmission in scenarios such as industrial control and telemedicine, this invention provides a method and system for implementing intelligent converged deterministic networks based on IPv6.

[0048] Figure 1 The illustration shows a method for implementing an intelligent converged deterministic network based on IPv6, which includes the following steps:

[0049] S110. Collect service flow data from the access network, identify real-time interactive, large-scale machine communication, and ultra-reliable low-latency communication service flows through feature parsing, and extract the service quality parameter requirements for each type of service flow.

[0050] S120. Input the service quality parameter requirements into the multi-dimensional resource mapping model, and dynamically allocate virtual network slices with differentiated guarantee capabilities for different service types according to the real-time status of the network infrastructure.

[0051] S130. Based on the resource constraints and service quality requirements of the service flow of virtual network slices, a reinforcement learning algorithm is used to calculate the optimal forwarding path that meets the end-to-end deterministic index and generate an IPv6 segmented routing strategy containing explicit node sequences.

[0052] S140. According to the IPv6 segmentation routing strategy, the service flow data packets are encapsulated with IPv6 extension headers, embedding path instructions and quality of service identifiers, and data is transmitted through the corresponding virtual network slice forwarding plane.

[0053] S150: Monitor network performance parameters in real time and compare them with the target quality of service requirements. When the performance deviation exceeds the dynamic threshold, trigger the path re-optimization mechanism, update the IPv6 segmented routing information and feed it back to the forwarding plane for adjustment.

[0054] It should be noted that when an industrial control terminal accesses the network, the service identification module analyzes the packet payload characteristics to identify it as an ultra-reliable low-latency communication service, extracting its required millisecond-level latency and zero packet loss parameters. The resource mapping module allocates dedicated transmission time slots in the time domain and selects low-hop-count physical links in the spatial domain based on the current network load, forming a dedicated virtual slice. The path calculation module uses latency constraints as boundary conditions and iteratively calculates the optimal node sequence through a reinforcement learning model, generating an explicit routing policy containing three core switches. The encapsulation module adds a segmented routing extension header to the IPv6 header, writes the segment identifier list of the pre-calculated path, and carries a service priority label. The forwarding plane performs low-latency switching according to the extension header instructions. The monitoring module continuously collects end-to-end latency data; when it detects that the latency of a link has increased beyond a threshold, it triggers path recalculation and updates the routing information of subsequent packets.

[0055] Furthermore, it also includes collecting service flow data from the access network and identifying service flow types through feature parsing. Specifically, this includes using deep packet inspection technology or machine learning flow classification methods to distinguish between real-time interactive, large-scale machine communication, and ultra-reliable low-latency communication service flows, and extracting at least two key service quality parameters from the end-to-end latency, jitter, packet loss rate, and bandwidth guarantee requirements of each type of service flow.

[0056] It should be noted that deep packet inspection (DPI) technology refers to a technique for identifying service types based on protocol characteristics and payload content. Specifically, it can be achieved by parsing application layer protocol fields and payload characteristic patterns in data packets, such as identifying the timestamp field in the RTP header or specific URI patterns in HTTP requests. This technology can accurately identify service flows with known protocol types, providing basic classification information for subsequent resource allocation. Machine learning flow classification methods refer to techniques for training classification models based on traffic behavior characteristics. Specifically, convolutional neural networks or random forest algorithms can be used to perform pattern recognition on traffic statistical characteristics, such as classifying by analyzing temporal features like flow duration, packet size distribution, and sending intervals. This method can adapt to the dynamic classification needs of unknown protocols or encrypted traffic, improving recognition coverage in complex scenarios. Key quality of service (QoS) parameters refer to the core requirements of service flows for network transmission performance, specifically including at least two parameters from end-to-end latency, jitter, packet loss rate, and bandwidth guarantee requirements. By extracting multi-dimensional QoS indicators, differentiated resource allocation can be provided for subsequent virtual network slice allocation.

[0057] Specifically, when a service flow accesses the network, the deep packet inspection unit first performs protocol parsing on the data packets. It matches real-time interactive service characteristics against a predefined protocol feature library, such as identifying SIP signaling characteristics of VoIP sessions or H.264 payload structure for video conferencing. For encrypted traffic or new service flows that cannot be identified through protocol parsing, the machine learning classification unit simultaneously collects traffic statistical features and identifies traffic behavior patterns using a pre-trained classification model, such as identifying industrial sensor data flows based on periodic short message characteristics. After completing service type identification, the parameter extraction module obtains the corresponding quality of service parameters from data packet header fields or traffic statistical results, such as reading the maximum latency tolerance value from the RTP extension header or calculating the average jitter value through a sliding window.

[0058] Furthermore, the construction of the multi-dimensional resource mapping model specifically includes: establishing a three-dimensional mapping relationship that includes time-domain resources, spatial-domain resources, and frequency-domain resources; and dynamically allocating virtual network slices with differentiated bandwidth, latency, and reliability assurance capabilities to different service types based on the service quality requirements of service flows and the real-time status of network infrastructure.

[0059] It should be noted that time-domain resources refer to the scheduling cycle and transmission time slot resources of network devices in the time dimension. Specifically, this can be implemented using a periodic scheduling table based on time-sensitive networking to meet the deterministic latency requirements of real-time services. Spatial-domain resources refer to the physical topology connections between network nodes. This can be implemented by obtaining a link-state database through a software-defined network controller to optimize the spatial distribution of service flow transmission paths. Frequency-domain resources refer to spectrum allocation and channel resources in wireless communication scenarios. This can be implemented using dynamic spectrum sharing technology combined with carrier aggregation mechanisms to improve the transmission efficiency of high-bandwidth services. The three-dimensional mapping relationship, by establishing a coupled analysis model between resources, jointly models service requirement parameters and multi-dimensional resource states to achieve collaborative optimization of cross-domain resources.

[0060] In Industrial Internet of Things (IIoT) scenarios, once an ultra-reliable, low-latency communication service flow is identified, the time-domain resource mapping unit allocates a fixed-period transmission time slot, the spatial-domain resource mapping unit selects a physical path with redundant links, and the frequency-domain resource mapping unit configures a dedicated carrier frequency band. These three types of resources form a virtual network slice through a joint optimization algorithm, enabling the service flow to achieve end-to-end latency guarantees below milliseconds. When the network topology changes, the resource mapping model dynamically adjusts the allocation ratio of resources in each dimension within the slice by collecting parameters such as node load and link congestion status in real time.

[0061] Furthermore, the step of using reinforcement learning algorithms to calculate the optimal forwarding path specifically includes: taking end-to-end latency limits and tolerable jitter as constraints, and optimizing resource utilization efficiency as the objective, generating an explicit IPv6 routing sequence that meets the requirements of deterministic transmission through multi-agent collaborative decision-making, wherein the sequence contains a complete list of segmented routing nodes.

[0062] It should be noted that the end-to-end latency limit refers to the maximum allowable transmission delay of a service flow from the source node to the destination node. This can be determined through network measurement probes combined with service requirement analysis, and is used to constrain the path selection range. Tolerable jitter refers to the range of arrival time fluctuations of adjacent data packets that a service flow can withstand. Parameter values ​​can be obtained from a service type feature library and used as an evaluation indicator of path stability. Resource utilization efficiency optimization refers to minimizing network resource consumption while meeting service quality requirements. This can be achieved using a weighted calculation model of link bandwidth utilization and node processing load. Multi-agent collaborative decision-making refers to control units distributed in different network domains forming joint decisions through information exchange. This can be achieved using a distributed Q-learning algorithm to coordinate cross-domain paths. The distributed Q-learning algorithm runs in parallel on multiple network domain agents, and its core Q-value update formula is as follows:

[0063] .

[0064] Subscript i: represents the i-th domain agent. In cross-domain path coordination scenarios, the entire network is divided into multiple management domains (such as an Autonomous System), and each domain is managed by an independent agent responsible for path decision-making within its domain;

[0065] Superscript (k): represents the number of iterations or time steps of the algorithm. This represents the Q-value table of agent i at the k-th iteration.

[0066] and : Represent the state and action of agent i, respectively; state This typically includes real-time network information for the domain, such as available bandwidth, link latency, queue status, and hop count to edge nodes. (Action) This represents the decision that agent i can make in the current state, typically choosing the next border router or exit point to forward to.

[0067] α is the learning rate (0 < α ≤ 1). It controls how quickly newly acquired information overwrites old information. α = 1 means that the old Q value is completely replaced by the new estimate; α close to 0 means that almost no learning occurs.

[0068] : Local reward function. This is the function that agent i performs when executing actions. The immediate feedback obtained from its own environment. Its design is key to the algorithm, and it is usually based on the network objective of the local domain, for example: =-(link latency + packet loss penalty) or =Available bandwidth. The reward function guides the agent to optimize performance in this domain.

[0069] γ (Gamma): Discount factor (0 ≤ γ < 1). It determines how much importance the agent places on future rewards. The closer γ is to 1, the more farsighted the agent is, tending to consider long-term cumulative rewards; γ = 0 makes it short-sighted, only concerned with immediate rewards.

[0070] : indicates a new state Below, agent i considers all possible actions. The maximum Q-value estimate. This represents the agent i's opinion on the new state. Starting now, to achieve the best cumulative returns in the future;

[0071] This is a key element in achieving cross-domain coordination;

[0072] (Coordination Signal): This is a signal received from downstream neighboring agents (or the global coordinator). It represents the "evaluation" or "cost estimate" of the path by domain j when a traffic flow moves from the current domain i to the next domain j. For example, it can include estimated latency, congestion levels, or path reliability information for downstream domains;

[0073] λ (Lambda): Coordination factor. It controls the weight of the coordination signal Φ in the overall Q-value update. The larger λ is, the more each agent will consider the impact of its actions on the global end-to-end performance when making decisions, thereby promoting cooperation.

[0074] IPv6 explicit routing sequences refer to a pre-determined complete list of forwarding path nodes. Specifically, data plane forwarding control is achieved by encoding node identifiers in the segmented routing header using SRv6 technology. When a service flow enters the network, the path calculation module constructs a state space based on slice resource constraints and quality of service parameters. A reinforcement learning model, with latency jitter as boundary conditions and link load balancing as the objective function, generates a set of candidate paths through interactive multi-agent training. Each agent corresponds to a control unit in a network domain, iteratively optimizing the global path scheme by exchanging network topology state information and local decision results. The final output IPv6 explicit routing sequence contains a verified list of node identifiers, encoded in the segmented routing header of the data packet, guiding intermediate nodes to forward data packets in a predetermined order.

[0075] Furthermore, the encapsulation of the service flow data packets with an IPv6 extension header specifically includes: inserting a segmented routing header after the IPv6 header, in which the path node sequence is explicitly specified; and adding a custom option field to carry service level identifiers and slice binding identifiers for intermediate nodes to perform low-latency switching and priority scheduling.

[0076] The segmentation routing header is a structure in the IPv6 extension header used to store explicit path information. Specifically, it can be implemented using the SegmentRoutingHeader defined in the SRv6 specification. By sequentially arranging the IPv6 addresses of path nodes in the header, it enables data packets to be transmitted along a predetermined path. The custom options field is a user-defined data area located in the IPv6 extension header. It can be implemented using a type-length-value encoding format, carrying service priority and slice ownership information through predefined identifier encoding rules, providing scheduling decision-making basis for intermediate nodes.

[0077] Specifically, after completing path calculation, the forwarding node sequentially encodes the generated segmented routing node sequence into the segment list field of the segmented routing header, forming an explicit instruction for the end-to-end transmission path. Simultaneously, a separate options field is created in the extension header, generating a service level identifier based on the service quality level corresponding to the service type, and attaching a unique binding identifier for the current slice resource. During transmission, the encapsulated data packet is processed by intermediate nodes that parse the segmented routing header to obtain the next-hop address. Simultaneously, based on the identifier information in the options field, it matches the preset priority queue and slice resource, achieving fast forwarding and differentiated scheduling.

[0078] Furthermore, the real-time monitoring of network performance parameters and triggering path re-optimization specifically includes: collecting actual statistical values ​​of latency, throughput, and packet loss rate based on a sliding time window, calculating their relative deviation from the service quality target; when the deviation exceeds a dynamic adaptive threshold within multiple consecutive periods, triggering a path re-optimization process based on network state prediction, updating segmented routing information, and feeding it back to the data plane.

[0079] Sliding time window acquisition refers to a method of periodically collecting network performance parameters. This can be implemented using a circular buffer, recording latency, throughput, and packet loss rate data at fixed time intervals to eliminate the impact of instantaneous fluctuations on judgment. Dynamic adaptive thresholding refers to trigger thresholds that are automatically adjusted based on historical deviation data. This can be implemented using an exponentially weighted moving average algorithm, dynamically adjusting the threshold to avoid frequent misjudgments or response delays caused by fixed thresholds. The path re-optimization process based on network state prediction refers to a path adjustment mechanism that predicts future performance changes based on the current network state. This can be implemented using a time series prediction model, generating new segmented routing information based on the prediction results to proactively avoid potential performance degradation.

[0080] The data plane periodically reports latency, throughput, and packet loss rate metrics to the control plane, which are then stored in a sliding time window for smoothing. The control plane compares the statistical values ​​within the window with preset service quality targets and calculates the relative deviation percentage. When the deviation exceeds a dynamic adaptive threshold for three consecutive monitoring periods, a path re-optimization process is triggered. At this point, the prediction model generates network load predictions for the next three time segments based on the current network topology and historical trend data. Based on the prediction results, the path calculation engine regenerates segmented routing sequences that meet end-to-end determinism metrics and sends the updated routing information to forwarding nodes via the southbound interface.

[0081] refer to Figure 2 Secondly, the present invention also provides an IPv6-based intelligent converged deterministic network system, applied to any of the aforementioned IPv6-based intelligent converged deterministic network implementation methods, comprising:

[0082] The service identification module is used to collect service flow data from the access network, identify service flow types through feature parsing, and extract the service quality parameter requirements for various service flows.

[0083] The resource mapping module is used to dynamically allocate virtual network slices to different service types based on the service quality parameter requirements and real-time network status through a multi-dimensional resource mapping model.

[0084] The path calculation module is used to calculate the optimal forwarding path based on virtual network slice constraints and quality of service requirements, and to generate IPv6 segmented routing strategies using reinforcement learning algorithms.

[0085] The encapsulation and forwarding module is used to encapsulate data packets with IPv6 extended headers according to the routing policy, embed path instructions and quality of service identifiers, and transmit data through the corresponding virtual network slice forwarding plane.

[0086] The monitoring and optimization module is used to monitor network performance in real time. When the performance deviation exceeds the threshold, it triggers path re-optimization, updates the routing policy, and feeds back to the encapsulation and forwarding module.

[0087] Furthermore, the service identification module includes a deep packet inspection unit and a machine learning classification unit, used to distinguish different service types through protocol feature analysis and traffic behavior pattern identification. The service identification module performs deep analysis on inbound traffic, uses protocol feature matching and machine learning models to identify service types, and extracts key parameters such as latency and jitter. The resource mapping module establishes a three-dimensional resource model based on the identification results and dynamically allocates slice resources in conjunction with the network topology. The path calculation module uses slice resources as constraints and iteratively calculates the optimal path using a reinforcement learning algorithm to generate an IPv6 segmented routing policy containing an explicit node list. The encapsulation and forwarding module encapsulates data packets in headers according to the policy, adds routing instructions and quality of service (QoS) tags, and ensures that intermediate nodes forward packets along predetermined paths. The monitoring and optimization module continuously collects transmission latency and packet loss rate; when performance deviates from the target value, it recalculates the path and updates the forwarding policy.

[0088] Furthermore, the deep packet inspection unit refers to a functional module that identifies service types by parsing protocol features in the packet header and payload content. Specifically, it can be implemented using rule-based protocol parsing technology. This unit extracts fixed field features of known protocols, such as port numbers, protocol identifiers, or specific payload formats, to perform preliminary classification of service flows such as real-time interactive traffic and large-scale machine communication traffic, overcoming the limitation of traditional methods in being unable to identify encrypted or disguised traffic.

[0089] The machine learning classification unit refers to a functional module that identifies business types by analyzing traffic behavior patterns. Specifically, it can be implemented using time series modeling methods based on traffic statistical features. This unit collects dynamic features such as data packet arrival intervals, burst traffic durations, and transmission directions to construct a traffic behavior pattern recognition model, solving the adaptive classification problem of unknown traffic types in complex business scenarios.

[0090] Furthermore, the resource mapping module includes a multi-dimensional resource modeling unit, used to establish resource mapping relationships in the time, spatial, and frequency domains, and to dynamically allocate virtual network slices based on real-time network conditions. The resource mapping module refers to a device for dynamically allocating network resources, specifically implemented using a multi-dimensional resource modeling unit. By establishing resource mapping relationships in the time, spatial, and frequency domains, it dynamically allocates virtual network slices for different service types based on real-time network conditions.

[0091] Furthermore, the monitoring and optimization module includes a performance acquisition unit and a dynamic decision-making unit, which are used to collect network performance data based on a time window and trigger a path re-optimization process based on the deviation analysis results.

[0092] The performance acquisition unit refers to the data monitoring component deployed on network nodes. It can be implemented using distributed probes or traffic mirroring technology to acquire real-time statistical values ​​of latency, throughput, and packet loss rate. The dynamic decision-making unit is a processing module with data analysis capabilities. It can be implemented using a sliding window algorithm combined with a dynamic threshold calculation model to assess the deviation between network performance and service objectives and trigger optimization mechanisms. The time window refers to the continuous monitoring time interval. It can be implemented using a fixed-duration or adaptively adjusted sliding window to balance the real-time nature of data acquisition with computational resource consumption.

[0093] The aforementioned modules can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device described in this application can be divided into different functional units or modules to complete all or part of the functions described above.

[0094] Thirdly, Figure 3 This invention provides an electronic device, which can be a server. The electronic device includes a processor, a memory, and a communication interface connected via a system bus. The processor provides computing and control capabilities. The memory can be implemented using any type of volatile or non-volatile storage device or a combination thereof, including but not limited to: disks, optical disks, EEPROMs, EPROMs, SRAMs, ROMs, magnetic storage, flash memory, and PROMs. The memory provides an environment for the operation of the operating system and computer programs stored within it. The communication interface is a network interface used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps of the IPv6-based intelligent fused deterministic network implementation method described in the above embodiments.

[0095] The present invention also provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps described in the various embodiments above. This computer-readable storage medium includes, but is not limited to, ROM and RAM. , disks, and floppy disks.

[0096] The present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps described in the various embodiments above.

[0097] This invention provides a method and system for implementing intelligent converged deterministic networks based on IPv6. Through an innovative technical architecture, it effectively solves the fundamental limitations of traditional networks in carrying diverse services. The system can dynamically allocate differentiated virtual network slices according to real-time service requirements. By accurately mapping time, space, and frequency domain resources, it provides tailored resource guarantees for service flows of different priorities, completely changing the problem of low resource utilization in the static resource allocation mode of traditional networks. Employing reinforcement learning algorithms to intelligently calculate the optimal forwarding path, it significantly improves network resource utilization efficiency while meeting strict end-to-end latency and jitter constraints. The generated explicit IPv6 routing sequences can effectively avoid network congestion points, ensuring deterministic transmission of critical services. The intelligent converged deterministic network system based on IPv6 in this invention establishes a complete real-time monitoring and dynamic optimization mechanism. By continuously comparing actual performance with the target value of the Service Level Agreement (SLA), it immediately triggers a path re-optimization process once a performance deviation is detected. This closed-loop control mechanism enables the network to have continuous self-optimization capabilities. This invention constitutes an organic whole capable of intelligently sensing business needs, dynamically allocating network resources, and ensuring transmission performance in real time. Ultimately, it achieves true end-to-end deterministic quality of service assurance under the IPv6 protocol framework, providing reliable basic network support for application scenarios with stringent network performance requirements, such as the Industrial Internet and the Internet of Vehicles.

[0098] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. A method for implementing an intelligent converged deterministic network based on IPv6, characterized in that, Includes the following steps: Collect service flow data from the access network, identify real-time interactive, large-scale machine communication, and ultra-reliable low-latency communication service flows through feature parsing, and extract the service quality parameter requirements for each type of service flow; The service quality parameters are input into a multi-dimensional resource mapping model, and virtual network slices with differentiated guarantee capabilities are dynamically allocated to different service types based on the real-time status of the network infrastructure. Based on the resource constraints and service quality requirements of virtual network slices, a reinforcement learning algorithm is used to calculate the optimal forwarding path that meets the end-to-end deterministic index, and an IPv6 segmented routing strategy containing explicit node sequences is generated. According to the IPv6 segmentation routing policy, the service flow data packets are encapsulated with IPv6 extension headers, embedding path instructions and quality of service identifiers, and then the data is transmitted through the corresponding virtual network slice forwarding plane. Monitor network performance parameters in real time and compare them with target quality of service requirements: When the performance deviation exceeds the dynamic threshold, the path re-optimization mechanism is triggered, the IPv6 segmentation routing policy is updated and fed back to the forwarding plane for adjustment.

2. The method for implementing an IPv6-based intelligent converged deterministic network according to claim 1, characterized in that, It also includes, Collect service flow data from the access network and identify service flow types through feature parsing, specifically including: Employ deep packet inspection technology or machine learning flow classification methods to distinguish between real-time interactive, large-scale machine communication, and ultra-reliable low-latency communication service flows; Extract end-to-end service quality data for various service flows, including latency, jitter, packet loss rate, and bandwidth guarantee requirements.

3. The method for implementing an IPv6-based intelligent converged deterministic network according to claim 1, characterized in that, The method for constructing the multi-dimensional resource mapping model includes: Establish a three-dimensional mapping relationship that includes time-domain resources, spatial-domain resources, and frequency-domain resources; Based on the service quality requirements data of the business flow and the real-time status of the network infrastructure, virtual network slices are dynamically allocated for several types of business.

4. The method for implementing an IPv6-based intelligent converged deterministic network according to claim 1, characterized in that, The calculation of the optimal forwarding path using a reinforcement learning algorithm specifically includes: With end-to-end latency limit and tolerable jitter as constraints, an IPv6 explicit routing sequence that meets the requirements of deterministic transmission is generated based on multi-agent collaborative decision-making. The IPv6 explicit routing sequence includes a segmented routing node list.

5. The method for implementing an IPv6-based intelligent converged deterministic network according to claim 1, characterized in that, The encapsulation of service flow data packets with IPv6 extension headers includes: Insert a segmented routing header after the IPv6 header, explicitly specifying the sequence of path nodes in the IPv6 header; Add a custom option field, which carries the service level identifier and slice binding identifier, for intermediate nodes to perform low-latency switching and priority scheduling.

6. The method for implementing an IPv6-based intelligent converged deterministic network according to claim 1, characterized in that, The real-time monitoring of network performance parameters and triggering path re-optimization includes: Based on the actual statistical values ​​of latency, throughput, and packet loss rate collected within a sliding time window, the relative deviations from the service quality target are calculated: When the deviation exceeds the dynamic adaptive threshold for multiple consecutive periods, a path re-optimization process based on network state prediction is triggered, updating the IPv6 segmentation routing policy and feeding it back to the data plane.

7. An IPv6-based intelligent converged deterministic network system, applied to the IPv6-based intelligent converged deterministic network implementation method as described in any one of claims 1-6, characterized in that, include: The service identification module is used to collect service flow data from the access network, identify service flow types through feature parsing, and extract the service quality parameter requirements for various service flows. The resource mapping module is used to dynamically allocate virtual network slices to several service types based on the service quality parameter requirements and real-time network status through a multi-dimensional resource mapping model. The path calculation module, based on virtual network slicing constraints and quality of service requirements, uses reinforcement learning algorithms to calculate the optimal forwarding path and generate IPv6 segmented routing strategies. The encapsulation and forwarding module is used to encapsulate data packets with IPv6 extended headers according to the routing policy, embed path instructions and quality of service identifiers, and transmit data through the corresponding virtual network slice forwarding plane. The monitoring and optimization module is used to monitor network performance in real time. When the performance deviation exceeds the threshold, it triggers path re-optimization, updates the IPv6 segmentation routing policy, and feeds back to the encapsulation and forwarding module.

8. The intelligent converged deterministic network system for IPv6 according to claim 7, characterized in that, The service identification module includes a deep packet inspection unit and a machine learning classification unit, which are used to distinguish different service types through protocol feature analysis and traffic behavior pattern identification.

9. The intelligent converged deterministic network system for IPv6 according to claim 7, characterized in that, The resource mapping module includes a multi-dimensional resource modeling unit, which is used to establish resource mapping relationships in the time domain, spatial domain, and frequency domain, and to dynamically allocate virtual network slices according to the real-time network status.

10. The intelligent converged deterministic network system for IPv6 according to claim 7, characterized in that, The monitoring and optimization module includes a performance acquisition unit and a dynamic decision-making unit, which are used to collect network performance data based on a time window and trigger a path re-optimization process based on the deviation analysis results.

Images