Search

CN-116546519-B - Method, device and equipment for selecting network data analysis function

CN116546519BCN 116546519 BCN116546519 BCN 116546519BCN-116546519-B

Abstract

The invention provides a method, a device and equipment for selecting a network data analysis function. The method for the network registration storage function (NRF) network element side comprises the steps of receiving a NWDAF service discovery request sent by a network data analysis function NWDAF consumer, enabling the NWDAF service discovery request to carry a validity period parameter representing the timeliness of the process of discovering a target NWDAF, obtaining the estimated ending time of the process of discovering the target NWDAF according to the validity period parameter, calculating the time value weight of a NWDAF service instance matched with the NWDAF consumer according to the estimated ending time, and selecting a target NWDAF according to the time value weight. According to the scheme, the target NWDAF is found by considering the time factors, and the matched target NWDAF can be dynamically selected for the consumer in real time, so that global overall optimization is achieved.

Inventors

  • Yue Liexiang
  • DENG LINGLI

Assignees

  • 中国移动通信有限公司研究院
  • 中国移动通信集团有限公司

Dates

Publication Date
20260512
Application Date
20220126

Claims (11)

  1. 1. A method for selecting a network data analysis function, applied to an NRF network element of a network registration storage function, the method comprising: Receiving a NWDAF service discovery request sent by a network data analysis function NWDAF consumer, wherein the NWDAF service discovery request carries a validity period parameter representing flow timeliness of a discovery target NWDAF; Obtaining the expected ending time of the process of finding the target NWDAF according to the validity period parameter; Calculating a time value weight of NWDAF service instance matched with NWDAF consumers according to the estimated end time; And selecting a target NWDAF according to the time value weight.
  2. 2. The method for selecting a network data analysis function as claimed in claim 1, wherein, When the NWDAF service discovery request is a single NWDAF analysis subscription, the validity period parameter is T Valid , and the T Valid indicates that the flow of the single discovery target NWDAF exceeds the T Valid , and the NWDAF consumer no longer needs to wait for the response information of the discovery target NWDAF; Or alternatively When the NWDAF service discovery request is continuous NWDAF analysis subscription, the analysis service is called once at intervals d, the validity period parameter is T Valid list={ T Valid +d,T Valid +2d,…,T Valid +nd, and n is a positive integer.
  3. 3. The method of claim 1, wherein obtaining the predicted end time of the process of discovering the target NWDAF based on the validity period parameter comprises: Acquiring the state of each service instance, the historical completion time and the starting reasoning time T start of the service instance; Based on the status of each service instance, the historical completion time, and the starting inference time T start for the service instance, the predicted end time for the flow of discovery objective NWDAF is obtained.
  4. 4. A method of selecting a network data analysis function according to claim 3, wherein obtaining the predicted end time of the process of finding the target NWDAF based on the status of each service instance, the historical completion time, and the starting inference time T start of the service instance, comprises: If the NWDAF last task completion time > last task start time, T end = current time T current , wherein T end is the predicted end time; If the NWDAF last task completion time < last task start time, the predicted end time of the flow of the alternative target NWDAF is obtained by T end =current time T current + (historical average completion time of service instance-start time of service instance), where T end is the predicted end time, or If the number of NWDAF service discovery requests in the current analysis stage Episode is less than the number of NWDAF service discovery requests in the current analysis stage Episode, the expected end time of the flow of the candidate target NWDAF is obtained by using T end =current time T current +the time required for service instance construction to complete, where T end is the expected end time.
  5. 5. The method of claim 1, wherein calculating a time value weight for NWDAF service instances matching NWDAF consumers based on the predicted end time comprises: Acquiring the state of each NWDAF service instance to be matched according to the logic analysis service type; obtaining NWDAF customer lists NC [ j ], and recording the states of NWDAF service examples as NI [ i ], wherein the global M= < NC [ j ], NI [ i ] >, i and j are index numbers of NWDAF service examples and NWDAF customers respectively; For each logically analyzed service type, a time value weight is calculated for the service instance NWDAF that matches the NWDAF consumer.
  6. 6. The method of claim 5, wherein calculating the time value weight of NWDAF service instance matching NWDAF consumers comprises: for each set { NC [ j ], NI [ i ] }, traversing i, j, calculating: if it is Ti Valid >=Tj end ,V(i,j)=-{（P1(Ti Valid - Tj end )+P2(Ci-Cj)+P3(Mi-Mj)+P4(Si-Sj)+….Pn(Xi-Yj)）}; If Ti Valid <Tj end , V (i, j) = - { int_max (constant maximum) }; Wherein Ti Valid is a validity period parameter, tj end is a predicted ending time, (Ti Valid - Tj end ) is a time value matching factor, (Ci-Cj) is a resource matching value factor, (Mi-Mj) is a confidence matching value factor, (Si-Sj) is a calculation power matching value factor, (Xi-Yj) is another matching value factor, ci and Cj are resource matching factors, mi and Mj are confidence matching factors, si and Sj are calculation power matching factors, xi and Yj are other matching factors, P1, P2 and P3 are measurement weights of various value matching factors, and V (i, j) is the total value of the value matching factors.
  7. 7. The method of claim 6, wherein selecting the target NWDAF based on the time value weight comprises: taking NC [ j ], NI [ i ] as vertex coordinates of the bipartite graph and time value weights as edges to form the bipartite graph; and traversing the depth map of the bipartite graph until a perfect match consisting of feasible edges exists in the bipartite graph, and taking NWDAF corresponding to a service instance corresponding to the perfect match as the target NWDAF.
  8. 8. The method of selecting a network data analysis function according to claim 7, wherein performing depth map traversal on the bipartite graph comprises: Determining a path from each NC [ j ] vertex of the bipartite graph to the other unmatched vertex by taking an alternate path, wherein all edges in the path are feasible edges Edge (i, j): NC [ j ] +NI [ i ] =V (i, j); If all the top marks can be found in the path, ending the flow, otherwise, encountering conflict or not finding complete matching, determining NC in the found sub-graph as NC_S, NJ_S, NC in the non-found sub-graph as NC_R, NJ_R, NC, NJ respectively representing the vertex sets of NWDAF consumer and NWDAF service examples, NC_ S, NJ _S representing the vertex set of the consumer and the vertex set of the example already contained in the current sub-graph, NC_ R, NJ _R representing the vertex set of the consumer and the vertex set of the example not contained in the current sub-graph; modifying the coordinates of the vertexes of the found subgraph NC_S by using a greedy algorithm, subtracting alpha from all vertexes of NI [ i ], adding alpha to the top point of the lower NJ_S top mark NC [ j ], [ alpha ] = Min (NC_S [ i ] +NJ_R [ j ] -V (i, j)); And after vertex coordinates are modified, continuing the depth map traversing algorithm, adding a subgraph to the new feasible right side meeting NC_Sj+NJ_Rj=V (i, j), expanding the subgraph, and finally, enabling the weight of the matched graph side to be increased until the complete matching consisting of the feasible sides exists in the bipartite graph.
  9. 9. A selection device of a network data analysis function, applied to an NRF network element of a network registration storage function, the device comprising: A transceiver module, configured to receive a NWDAF service discovery request sent by a consumer of the network data analysis function NWDAF, where the NWDAF service discovery request carries a validity period parameter that indicates a process age of discovering the target NWDAF; the processing module is used for obtaining the expected ending time of the flow of the discovery target NWDAF according to the validity period parameter, calculating the time value weight of the service instance of NWDAF matched with the NWDAF consumer according to the expected ending time, and selecting the target NWDAF according to the time value weight.
  10. 10. A communication device comprising a processor, a memory storing a computer program which, when executed by the processor, performs the method according to any one of claims 1 to 8.
  11. 11. A computer readable storage medium storing instructions which, when run on a computer, cause the computer to perform the method of any one of claims 1 to 8.

Description

Method, device and equipment for selecting network data analysis function Technical Field The present invention relates to the field of communications technologies, and in particular, to a method, an apparatus, and a device for selecting a network data analysis function. Background With the complexity of future communication scenarios, the diversification of service demands, and the individualization of service experiences, current 5G networks still lack sufficient intelligence to provide on-demand services and higher network resource utilization efficiency. NWDAF (Network DATA ANALYTICS Function) is a data perception analysis Network element, which is used for automatically perceiving and analyzing a Network based on Network data and participating in Network planning, construction, operation and maintenance, network optimization and operation whole life cycle, so that the Network is easy to maintain and control, the Network resource utilization efficiency is improved, and the service experience of consumers is improved. NWDAF collects network operation data from 5G Network Functions (NF), terminal and network related statistics data obtained from OAM (Operation Administration AND MAINTENANCE, operation and maintenance management system), and application data from third party applications through NEF (network element function unit). The analysis result generated by NWDAF may also be output to 5G network functions, OAM, or third party applications. The 5G network function, OAM, or application layer server may utilize NWDAF's analysis results to perform different optimization operations. The existing NWDAF consumer finds out the NWDAF flow through NRF (Network Repository Function, network warehouse function) at least has the following problems: Due to environmental, resource limitations in practical communication systems, multiple consumers and instances are in a resource gaming relationship with limited competition between time periods. The selection of a certain consumer may change the overall resource amount, thereby affecting the access and performance of other consumers, and possibly causing other consumers to be unable to access effectively. Disclosure of Invention The invention aims to provide a method, a device and equipment for selecting a network data analysis function. The matching target NWDAF can be dynamically selected for the consumer in real time quickly, so that global overall optimization is achieved. In order to solve the technical problems, the technical scheme of the invention is as follows: A method for selecting a network data analysis function, applied to an NRF network element of a network registration storage function, the method comprising: Receiving a NWDAF service discovery request sent by a network data analysis function NWDAF consumer, wherein the NWDAF service discovery request carries a validity period parameter representing flow timeliness of a discovery target NWDAF; Obtaining the expected ending time of the process of finding the target NWDAF according to the validity period parameter; Calculating a time value weight of NWDAF service instance matched with NWDAF consumers according to the estimated end time; And selecting a target NWDAF according to the time value weight. Optionally, when the NWDAF service discovery request is a single NWDAF analysis subscription, the validity period parameter is T Valid, the T Valid indicates that the flow of the single discovery target NWDAF exceeds the T Valid, and the NWDAF consumer no longer needs to wait for the response information of the discovery target NWDAF, or When the NWDAF service discovery request is continuous NWDAF analysis subscription, the analysis service is called once at intervals d, the validity period parameter is T Validlist＝{TValid+d,TValid+2d,…,TValid +nd, and n is a positive integer. Optionally, obtaining the predicted end time of the process of finding the target NWDAF according to the validity period parameter includes: Acquiring the state of each service instance, the historical completion time and the starting reasoning time T start of the service instance; Based on the status of each service instance, the historical completion time, and the starting inference time T start for the service instance, the predicted end time for the flow of discovery objective NWDAF is obtained. Optionally, obtaining the predicted end time of the process of discovering the target NWDAF according to the state of each service instance, the historical completion time and the starting inference time T start of the service instance, including: If the NWDAF last task completion time > last task start time, indicating that the last task has ended, tend=current time Tcurrent; If the NWDAF last task completion time < last task start time, it is stated that the last task has not yet ended, by Tend = current time tcurrent+ (historical average completion time of service instance-start time of service instance), obtaining the predicted end tim