Sevilization RAN protocol stack decoupling method and system based on graph theory

Abstract

The invention provides a graph theory-based servitization RAN protocol stack decoupling method and system. The method comprises the following steps of: analyzing a control plane function; based on the control plane function, modeling the control plane, defining a control plane component, and extracting a control plane flow; the component is set as a minimum processing unit of the control surface; establishing a function logic diagram based on the component and the control plane flow; performing service segmentation on the function logic diagram to form a plurality of sub-diagrams; and based on a subgraph segmentation result, performing segmentation design on functions and interfaces in the service development of the RAN control plane, and performing testing. According to the scheme, traditional cloud RAN software with bloated functions and tight coupling can be split into services with cohesive functions and definite boundaries, and the RAN control plane is decoupled into a plurality of service instances, so that a flexible and fine service RAN system is formed, the research and development and upgrading cost of the RAN system is effectively reduced, and the utilization efficiency of computing power resources is improved.

Description

Technical Field

[0001] The present invention relates to the fields of communication network transmission and communication network data processing, and in particular to a graph-theory-based service-oriented RAN protocol stack decoupling method and system. Background Art

[0002] As mobile communication technology gradually extends to more complex and diversified application scenarios, the existing radio access network (RAN) presents problems such as large size and bloated structure when dealing with diversified needs, resulting in a significant increase in the difficulty and cost of early research and development and later upgrades. At the same time, the solidified coupling relationship between protocol layers has also become the main bottleneck for performance optimization, limiting the rapid deployment and application of 5G to vertical industry applications. In this regard, relevant research on service-oriented RAN has been carried out both at home and abroad, but the existing related research is in the theoretical exploration and planning stage, lacking a technical verification system. To solve the above problems, service-oriented RAN breaks the coupling between traditional solidified protocol layers, breaks through service decoupling technology, decouples different protocol functions into independent services or functions, and flexibly arranges and combines according to user needs to achieve personalized and customized user services, with the advantages of agile development and rapid iteration. Based on general computing hardware infrastructure, it is crucial to realize the reconstruction of protocol stack functions and build a flexible and low-cost service-oriented RAN for the upgrade of the next-generation mobile communication architecture.

[0003] In the existing technology, the network slicing algorithm proposed in 2019 adopts different functional segmentation schemes to meet the needs of different business scenarios to make the network flexible; or adopts functional segmentation and network slicing technology to design a variety of standard base stations to perform resource allocation and isolate wireless network slicing mechanisms during the transition period of the wireless side. The existing technology based on functional segmentation and network slicing will have the following problems:

[0004] 1) Different business scenarios require different slicing and functional segmentation strategies, which may make network management and configuration more complicated and require more advanced management tools and technical support.

[0005] 2) Since different slices may have different resource requirements, functional segmentation may lead to uneven resource allocation. Some slices may face resource waste, while other slices may experience resource shortage, thus affecting the resource utilization efficiency of the overall network.

[0006] 3) In order to achieve flexible functional segmentation and slice management, additional protocols and interfaces may need to be introduced. These overheads will affect the overall performance of the network and increase processing and transmission delays.

[0007] 4) Introducing multiple slicing and functional splitting solutions into the network may cause compatibility issues, especially in terms of interoperability between different network devices and systems. It is necessary to ensure that the newly introduced solutions are compatible with the existing network infrastructure. Summary of the invention

[0008] This solution aims to solve the performance bottleneck caused by the solidified coupling relationship between protocol layers in the traditional RAN system, which makes it difficult to meet the diversified user needs of the future network and has difficulties in rapid iteration and upgrading. A service-oriented RAN protocol stack decoupling method and system based on graph theory is proposed. Specifically, the present invention provides the following technical solutions:

[0009] On the one hand, the present invention provides a service-oriented RAN protocol stack decoupling method based on graph theory, the method comprising:

[0010] S1. Analyze control plane functions;

[0011] S2. Based on the control plane function, the control plane is modeled, and control plane components are defined to extract the control plane process; the components are set as the minimum processing unit of the control plane;

[0012] S3. Create a functional logic diagram based on components and control plane processes;

[0013] S4, performing service segmentation on the functional logic diagram to form multiple sub-diagrams;

[0014] S5. Based on the results of subgraph segmentation, the functions and interfaces in the service-oriented development of the RAN control plane are segmented and designed, and then tested.

[0015] Preferably, in S2, in a single-unit wireless access network, the components constitute a centralized unit control plane; in a service-based wireless access network, the components constitute network elements, and the network elements constitute a centralized unit control plane;

[0016] The process described above represents the process of parameter configuration required to complete the service, network and interruption.

[0017] Preferably, in S3, the functional logic diagram is constructed as follows:

[0018] S31. Extract components of the control plane, extract processes from the protocol, split each process into one or more components, and establish an undirected graph with the components as vertices;

[0019] S32. Components are regarded as vertices of an undirected graph. Components are connected to form edges based on the process. The number of consecutive appearances of the relationship in the process represents the weight of the edge between vertices, thereby forming a functional logic graph.

[0020] Preferably, in S32, the edge represents the corresponding relationship between different components.

[0021] Preferably, in S4, the specific manner of service segmentation is:

[0022] S41, determine the number of sub-graphs N that need to be segmented, and input the original functional logic graph G;

[0023] S42, the functional logic graph G is represented by a compressed sparse row format to obtain G csr ;

[0024] S43, for Figure G csr The adjacent edges of all vertices in are reordered in descending order according to the edge weights;

[0025] S44, for the reordered graph, delete the minimum adjacent edges of all vertices to obtain a deleted subgraph G'; and at the same time, record the information of the deleted adjacent edges in a priority queue Q; the priority queue Q is sorted in descending order of weight;

[0026] S45, calculating connected components based on the subgraph G' to obtain the number of subgraphs N';

[0027] S46, determine the size relationship between N and N'; if they are equal, end; if N is greater than N', return to step S44; if N is less than N', remove the adjacent edge with the largest weight in the priority queue Q and add it to the subgraph G', and then return to S45.

[0028] Preferably, in S44, the minimum adjacent edges of all vertices are deleted by connected component calculation, specifically:

[0029] S441. Initialize two signature values ​​v for each vertex in 、v out , and use this as the ID value of the vertex;

[0030] S442, propagate the maximum signature value along the edge; for each directed edge (u, v), where u is the source vertex and v is the target vertex, if the v of the target vertex out The value is greater than the u of the source vertex out value, then update the u of the source vertex out The value is v of the target vertex out Value; if the source vertex u in The value is greater than the target vertex's v in value, then update the target vertex v in The value is the source vertex u in Value; repeat the iteration until there is no vertex v in or out The value changes again;

[0031] S443. If the signatures of the source vertex and the target vertex are different, the edge is deleted as the minimum neighboring edge.

[0032] Preferably, the number N of subgraphs to be segmented is set by configuration, and N is used as a segmentation granularity requirement to control the target degree of decoupling.

[0033] Preferably, in S43, the reordering method is:

[0034] G csr It includes row array, edge array and value array. During reordering, the edge array and value array are reordered in descending order according to the weights of the adjacent edges of the vertices.

[0035] Preferably, in S5, if the test result does not meet the requirement, the segmentation parameter value is adjusted, and the segmentation parameter value at least includes the number N of sub-graphs that need to be segmented.

[0036] On the other hand, the present invention also provides a graph-theory-based service-oriented RAN protocol stack decoupling system, the system comprising:

[0037] A modeling module, used to model the control plane based on the analysis of the control plane function, define the control plane components, and extract the control plane process; the components are set as the minimum processing unit of the control plane;

[0038] The logic diagram building module builds a functional logic diagram based on components and control plane processes;

[0039] A service segmentation module, used to perform service segmentation on the functional logic diagram to form multiple sub-diagrams;

[0040] The design module performs segmentation design on the functions and interfaces in the service-oriented development of the RAN control plane based on the results of the subgraph segmentation.

[0041] The test module is used to test the designed functions and interfaces.

[0042] On the other hand, the present invention also provides a device, which includes a processor and a memory, and the processor can call computer instructions stored in the memory to execute the graph theory-based service-oriented RAN protocol stack decoupling method as described above.

[0043] Compared with existing technologies, this solution can split the traditional bloated and tightly coupled cloud-based RAN software into functionally cohesive and clearly defined services, and decouple the RAN control plane into multiple service instances through graph partitioning technology based on graph theory and actual engineering evaluation, so as to form a flexible and sophisticated service-based RAN system. This solution not only effectively reduces the R&D and upgrade costs of the RAN system, but also significantly improves the utilization efficiency of computing resources and facilitates the rapid application of networks in vertical industries. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 A schematic diagram of a flow chart of a RAN control plane decoupling method according to an embodiment of the present invention;

[0046] Figure 2 CUCP logical function diagram of an embodiment of the present invention;

[0047] Figure 3 A control plane component connection diagram according to an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of the decoupling result of an embodiment of the present invention;

[0049] Figure 5 The figure is a flow chart of the service segmentation method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the figures in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0051] The core of this solution is to decouple the RAN control plane. By constructing a functional logic diagram of the RAN control plane and applying a graph-theoretic partitioning strategy, the functional boundaries and service interfaces of the decoupled control plane services are designed, thereby achieving flexible deployment and horizontal expansion and reduction of the service-oriented RAN control plane. The corresponding overall technical solution flow chart is as follows: Figure 1 shown.

[0052] Combination Figure 1 The processing flow of the graph-based service-oriented RAN protocol stack decoupling method proposed in this solution is as follows:

[0053] First, the control plane is modeled based on graph theory, the smallest processing unit of the control plane - component is defined, the control plane business process is extracted, and the functional logic connection diagram of the Central Unit Control-plane (CUCP) is constructed, that is, the CUCP functional logic diagram; then, the service segmentation strategy based on graph theory is adopted to divide the control plane service; finally, based on the above analysis, the function and interface design of each service in the control plane service-oriented development are decoupled, and unit testing and system integration testing are performed. In terms of guiding the segmentation design, preferably, based on the CUCP logical function diagram and the control plane segmentation based on graph theory, the function and interface design of each service in the control plane service-oriented development are determined, and then development is carried out, and finally the corresponding unit testing and system integration testing are carried out. In the unitization and system integration testing, preferably, the unitization test refers to the testing of the decoupled functions and interfaces; the integration test refers to the testing of the integrated whole.

[0054] To achieve the above objectives, two processing optimization strategies are provided in this embodiment, namely: CUCP functional logic graph construction and graph-theory-based service segmentation, which are respectively described below.

[0055] Here, the components that define the control plane specifically refer to: based on the CUCP function, various components, network elements, control planes, signaling, processes, etc. involved are analyzed and abstracted to form the vertex information of the corresponding CUCP function logic diagram. The vertices in the constructed diagram are referred to as components, which are defined as the smallest unit of signaling processing.

[0056] 1. Construction of CUCP functional logic diagram

[0057] The established CUCP functional logic diagram of the wireless access network mainly includes components. As the smallest unit of signaling processing, components can constitute network elements or control planes in different network scenarios, and reflect the process, signaling and the relationship between them through components and relationships, such as Figure 2 As shown, Figure 2 Each vertex in represents a single component. Components are the smallest units of signaling processing. In a single wireless access network, components form a centralized unit control plane; in a service-based wireless access network, components form network elements, and network elements form a centralized unit control plane. Signaling is a message used by network elements or components to pass context between each other. The process is the process in which the network and the terminal need to configure parameters in order to complete the service. This process often involves the operation of multiple components and the transmission of signaling between components or network elements.

[0058] The specific construction method is as follows:

[0059] from Figure 2 to Figure 3 , is to complete the logical mapping of raw data to vertices and assign relevant weight information to the edges of the graph.

[0060] S1. Extract control plane components from the original single-body wireless access network, extract processes from the protocol, split each process into a combination of control plane components, and establish an undirected graph with the control plane components as vertices, such as Figure 3 As shown in the figure, CF represents each component and W represents the edge weight.

[0061] It can be understood here that extracting control plane components mainly involves analyzing the components, network elements, control planes, signaling, processes and the relationships between them involved in the protocol and extracting them. Figure 2 As shown in the figure, the vertices in the figure represent various components. Components are the smallest units of signaling processing. In a single-type wireless access network, components constitute the centralized unit control plane; in a service-oriented wireless access network, components constitute network elements, and network elements constitute the centralized unit control plane. Signaling is a message used by network elements or components to transfer contexts to each other. The process is the process in which the network and the terminal need to configure parameters in order to complete the service. This process often involves the operation of multiple components and the transmission of signaling between components or network elements.

[0062] S2. Determine the weight of the edge between components. When two components appear consecutively in a process, on the one hand, it indicates that the two components have a high degree of coupling in the process; on the other hand, if these two components appear in two network elements, the number of signaling transmissions may increase. Therefore, the number of consecutive appearances of the control plane components in the process is used to represent the coupling between vertices, that is, the weight between two vertices in the undirected graph. At this time, the undirected graph formed is the CUCP functional logic graph.

[0063] It should be noted here that continuous occurrence means that when extracting vertices and edges, the extracted components and the relationships between components appear frequently and multiple times. Usually, different components (i.e., different vertices) are extracted first, and then the corresponding relationships, i.e., different edges, are extracted, and then the weights of the corresponding edges are assigned based on the number of times the relationships appear. Figure 3 There is no continuous appearance in the graph, and the continuous appearance is reflected by the weight information of the edges in the graph.

[0064] 2. Service segmentation strategy based on graph theory

[0065] In this embodiment, during the service segmentation process, the connection lines between the vertices in the CUCP functional logic graph, that is, the edge data of the vertices in the graph, are first sorted in parallel according to the edge weights. Then, the connection lines with the smallest edge weight of each vertex are removed. The number of subgraphs existing in the graph after the connection lines are removed is calculated through connected component calculation. If the number of subgraphs meets the expected number of required segmentations, the algorithm ends. If the number of subgraphs is less than the expected number of required subgraphs, the above-mentioned connection line removal operation is executed in a loop. If the number of subgraphs is more than the expected number of required subgraphs, the maximum connection edges after deletion are connected in turn to determine whether they meet the requirements. If they do, the algorithm ends. The control plane component corresponding to the vertex in the subgraph is the control plane component included in the expected network element. Since the entire graph is not considered in each step of the algorithm, the algorithm itself cannot automatically determine when to stop removing the connection lines in the graph. Therefore, the number of subgraphs finally formed needs to be used as an input parameter. This parameter can be used as the granularity of segmentation to control the degree of decoupling. The segmentation result of the algorithm is as follows: Figure 4 shown.

[0066] Here, more preferably, the connected component calculation can be set as follows:

[0067] The algorithm operates on a directed graph with unique vertex IDs and computes two signature values ​​for each vertex v, set to v in and v out At the end of the calculation, any of the two values ​​can uniquely identify the strongly connected component to which the vertex v belongs, denoted as SCC. The algorithm is iterated until the algorithm converges (ends). Each iteration is divided into three stages:

[0068] Phase 1: Signature initialization. Initialize the two signature values ​​v for each vertex in , v out The ID value of the corresponding vertex.

[0069] Phase 2: Propagate the maximum signature value along the edge. For each directed edge (u, v), where u is the source vertex and v is the target vertex, if the target vertex's v out The value is greater than the u of the source vertex out value, then update the u of the source vertex out The value is v of the target vertex out Similarly, if the source vertex u in The value is greater than the target vertex's v in value, then update the target vertex v in The value is the source vertex u in This stage is repeated until a fixed point is reached, where there is no vertex v in or out The value changes again.

[0070] Phase 3: Delete edges whose source and target vertices belong to different SCCs. This is determined by: if the signatures of the source and target vertices are different, it can be guaranteed that they do not belong to the same SCC and the edge can be safely deleted.

[0071] These three phases are repeated on a simplified graph with the same vertices but a reduced number of edges. in Value and u out When the values ​​are equal, the algorithm terminates.

[0072] Furthermore, the expected number of sub-graphs required may be set based on empirical values, actual segmentation requirements or user needs, and may be flexibly set through configuration.

[0073] In summary, if Figure 5 As shown in the figure, the processing flow of the graph-based service segmentation algorithm is as follows:

[0074] Step 1: Input the number of subgraphs N to be segmented and the original functional logic graph G to be constructed;

[0075] Step 2: The constructed graph G is represented using the Compressed Sparse Row (CSR) format, which is G csr ;Compressed sparse rows can be implemented by representing graph G in CSR format;

[0076] Step 3: Based on G csr , parallel reordering of the neighboring edges of all vertices in the graph in descending order of edge weights;

[0077] Here, the CSR structure mainly includes row array, edge array, and value array. In the reordering, the edge array and value array are reordered in descending order according to the weights of the adjacent edges of the vertices. Parallelism means that the adjacent edges of all vertices are reordered in a multi-threaded manner based on vertex granularity parallelism.

[0078] Step 4: Based on the sorted subgraph, delete the minimum neighbor edge information of all vertices to obtain the deleted subgraph G', and record the deleted neighbor edge information in the priority queue Q. The priority queue is sorted in descending order of weight;

[0079] Step 5: Based on the subgraph G', execute the connected component algorithm to calculate the number N' in the processed subgraph;

[0080] Step 6: Determine whether N and N' are equal. If they are equal, the algorithm ends. If N is greater than N', loop through steps 4 to 6. If N is less than N', remove the edge with the largest weight from the priority queue Q and add it to the subgraph G'. Then loop through steps 5 to 6 until the algorithm ends.

[0081] After the service segmentation, based on the segmentation results, the corresponding functions and development will be guided in the design and development of the RAN protocol.

[0082] For further optimization, after the development is completed, testing is required, such as unit testing and integration testing. If the test shows that there is a loss in performance and accuracy after the integration test, the corresponding segmentation parameter value can be adjusted to adjust the relevant research and development. Here, the segmentation parameter value mainly refers to the number N of segmentation subgraphs.

[0083] In another embodiment, the present solution can be implemented in a system manner, that is, a graph-theory-based service-oriented RAN protocol stack decoupling system, the system comprising:

[0084] A modeling module, used to model the control plane based on the analysis of the control plane function, define the control plane components, and extract the control plane process; the components are set as the minimum processing unit of the control plane;

[0085] The logic diagram building module builds a functional logic diagram based on components and control plane processes;

[0086] A service segmentation module, used to perform service segmentation on the functional logic diagram to form multiple sub-diagrams;

[0087] The design module performs segmentation design on the functions and interfaces in the service-oriented development of the RAN control plane based on the results of the subgraph segmentation.

[0088] The test module is used to test the designed functions and interfaces.

[0089] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A graph-theory-based service-oriented RAN protocol stack decoupling method, characterized in that: The method comprises: S1. Analyze control plane functions; S2. Based on the control plane function, the control plane is modeled, and control plane components are defined to extract the control plane process; the components are set as the minimum processing unit of the control plane; S3. Create a functional logic diagram based on components and control plane processes; S4, performing service segmentation on the functional logic diagram to form multiple sub-diagrams; S5. Based on the results of subgraph segmentation, the functions and interfaces in the service-oriented development of the RAN control plane are segmented and designed, and then tested.

2. The method according to claim 1, characterized in that In S2, in a single-unit wireless access network, the components constitute a centralized unit control plane; in a service-based wireless access network, the components constitute network elements, and the network elements constitute a centralized unit control plane; The process described above represents the process of parameter configuration required to complete the service, network and interruption.

3. The method according to claim 1, characterized in that In S3, the functional logic diagram is constructed as follows: S31. Extract components of the control plane, extract processes from the protocol, split each process into one or more components, and establish an undirected graph with the components as vertices; S32. Components are regarded as vertices of an undirected graph. Components are connected to form edges based on the process. The number of consecutive appearances of the relationship in the process is used to represent the weight of the edges between vertices, thereby forming a functional logic graph.

4. The method according to claim 3, characterized in that In S32, the edge represents the corresponding relationship between different components.

5. The method according to claim 1, characterized in that In S4, the specific method of service segmentation is: S41, determine the number of sub-graphs N that need to be segmented, and input the original functional logic graph G; S42, the functional logic graph G is represented by a compressed sparse row format to obtain G csr ; S43, for Figure G csr The adjacent edges of all vertices in are reordered in descending order according to the edge weights; S44, for the reordered graph, delete the minimum neighboring edges of all vertices to obtain a deleted subgraph G'; and at the same time, record the information of the deleted neighboring edges in the priority queue Q; The priority queue Q is sorted in descending order of weight; S45, calculating connected components based on the subgraph G' to obtain the number of subgraphs N'; S46, judging the magnitude relationship between N and N'; If they are equal, then end; if N is greater than N', return to step S44; if N is less than N', then remove the adjacent edge with the largest weight in the priority queue Q and add it to the subgraph G', and then return to S45.

6. The method according to claim 5, characterized in that In S44, the minimum adjacent edges of all vertices are deleted by connected component calculation, specifically: S441. Initialize two signature values ​​v for each vertex in 、v out , and use this as the ID value of the vertex; S442, propagate the maximum signature value along the edge; for each directed edge (u, v), where u is the source vertex and v is the target vertex, if the v of the target vertex out The value is greater than the u of the source vertex out value, then update the u of the source vertex out The value is v of the target vertex out value; If the source vertex u in The value is greater than the target vertex's v in value, then update the target vertex v in The value is the source vertex u in Value; repeat the iteration until there is no vertex v in or out The value changes again; S443. If the signatures of the source vertex and the target vertex are different, the edge is deleted as the minimum neighboring edge.

7. The method according to claim 5, characterized in that The number N of subgraphs to be segmented is set by configuration, and N is used as the segmentation granularity requirement to control the target degree of decoupling.

8. The method according to claim 5, characterized in that In S43, the reordering method is: G csr It includes row array, edge array and value array. During reordering, the edge array and value array are reordered in descending order according to the weights of the adjacent edges of the vertices.

9. The method according to claim 1, characterized in that: In S5, if the test result does not meet the requirement, the segmentation parameter value is adjusted, and the segmentation parameter value at least includes the number N of sub-graphs that need to be segmented.

10. A graph-theory-based service-oriented RAN protocol stack decoupling system, characterized in that: The system comprises: A modeling module, used to model the control plane based on the analysis of the control plane function, define the control plane components, and extract the control plane process; the components are set as the minimum processing unit of the control plane; The logic diagram building module builds a functional logic diagram based on components and control plane processes; A service segmentation module, used to perform service segmentation on the functional logic diagram to form multiple sub-diagrams; The design module performs segmentation design on the functions and interfaces in the service-oriented development of the RAN control plane based on the results of the subgraph segmentation. The test module is used to test the designed functions and interfaces.

Images