Search

CN-122001775-A - Quantum key demand prediction method and quantum key demand prediction system

CN122001775ACN 122001775 ACN122001775 ACN 122001775ACN-122001775-A

Abstract

The application discloses a quantum key demand prediction method and a quantum key demand prediction system. The method comprises the steps of obtaining historical key data, network state data, service log data and external event data of a quantum key distribution network, and predicting target quantum key demands of the quantum key distribution network according to the quantum key distribution network, the historical key data, the network state data, the service log data and the external event data based on a pre-trained quantum key demand space-time prediction model. In this way, based on the quantum key demand space-time prediction model which is trained in advance, according to the quantum key distribution network, historical key data, network state data, service log data and external event data are fused, the target quantum key demand which can be matched with the actual demand of quantum encryption communication is predicted, a reliable data base is provided for a pre-negotiation strategy, and further the reliability of encryption service is guaranteed.

Inventors

  • WANG DAWEI

Assignees

  • 中电信量子信息科技集团有限公司

Dates

Publication Date
20260508
Application Date
20251218

Claims (10)

  1. 1. A quantum key demand prediction method, the method comprising: Acquiring historical key data, network state data, service log data and external event data of a quantum key distribution network; And predicting a target quantum key requirement of the quantum key distribution network according to the quantum key distribution network, the historical key data, the network state data, the service log data and the external event data based on a pre-trained quantum key requirement space-time prediction model.
  2. 2. The method of claim 1, wherein predicting the target quantum key demand of the quantum key distribution network based on the pre-trained quantum key demand spatiotemporal predictive model from the quantum key distribution network, the historical key data, the network state data, the traffic log data, and the external event data comprises: constructing a space-time diagram according to the quantum key distribution network, the historical key data, the network state data, the service log data and the external event data; And inputting the space-time diagram into a pre-trained quantum key demand space-time prediction model, and predicting the target quantum key demand of the quantum key distribution network.
  3. 3. The method of claim 2, wherein said constructing a time-space diagram from said quantum key distribution network, said historical key data, said network status data, said traffic log data, and said external event data comprises: Taking a password service node of the quantum key distribution network as a node of the space-time diagram, taking an interaction relation of the password service node of the quantum key distribution network as an edge of the space-time diagram, and taking a link utilization rate corresponding to the edge of the space-time diagram as an initial weight coefficient of the edge of the space-time diagram; determining target feature data of the node of the time-space diagram according to the historical key data, the network state data, the service log data and the external event data; and constructing the space-time diagram according to the nodes, the edges of the space-time diagram and the target characteristic data of the nodes.
  4. 4. A method according to claim 3, wherein said inputting the time-space diagram into a pre-trained quantum key demand space-time prediction model predicts a target quantum key demand of the quantum key distribution network, comprising: Performing space feature extraction processing on the space-time diagram to enable a current node of the space-time diagram to aggregate target feature data of neighbor nodes of the current node so as to predict target nodes with business relations with the current node; Performing time feature extraction processing on the time-space diagram, and predicting quantum key demand required by the communication between the current node and the target node; And determining a target quantum key requirement according to the current node, the target node and the quantum key requirement.
  5. 5. The method according to claim 1, wherein the method further comprises: inputting the target quantum key requirement into a preset verification model; simulating the quantum key distribution network to distribute the quantum key according to the target quantum key requirement based on the preset verification model; And under the condition that the quantum key distribution meets the preset condition, the target quantum key requirement is sent to the quantum key distribution network, wherein the preset condition comprises that the link delay rate of the quantum key distribution network is smaller than a delay rate preset threshold, the key generation success rate is larger than a success rate preset threshold and/or the link cache utilization rate is smaller than a utilization rate preset threshold.
  6. 6. The method of claim 5, wherein the method further comprises: re-predicting a target quantum key requirement of the quantum key distribution network based on a pre-trained quantum key requirement spatiotemporal predictive model in case the quantum key distribution does not meet the preset condition, and/or And sending preset prompt information to the target object.
  7. 7. The method according to claim 1, wherein the method further comprises: Determining an actual quantum key requirement according to the target quantum key requirement and a quantum key buffer margin of a link corresponding to the target quantum key requirement; and determining a key pre-negotiation strategy of the quantum key distribution network according to the actual quantum key requirement.
  8. 8. The method according to any one of claims 1 to 7, wherein the quantum key demand spatiotemporal prediction model comprises a spatiotemporal module comprising a graph convolution unit and a temporal convolution unit connected to the graph convolution unit; the graph convolution unit is configured to perform space feature extraction processing on a space-time graph, so that a current node of the space-time graph aggregates target feature data of neighbor nodes of the current node to predict a target node with a business relationship with the current node; The time convolution unit is configured to perform a time feature extraction process on the time space diagram, and predict a quantum key requirement amount required for the current node to communicate with the target node.
  9. 9. The method of claim 8, wherein the quantum key demand spatiotemporal predictive model includes a fully connected layer connected to the spatiotemporal module; The fully connected layer is configured to determine a target quantum key requirement from the current node, the target node, and the quantum key requirement.
  10. 10. A quantum key demand prediction system, the system comprising a data preprocessing module and a quantum key demand prediction module, the quantum key demand prediction module comprising a pre-trained quantum key demand spatiotemporal prediction model; The data preprocessing module is configured to acquire historical key data, network state data, service log data and external event data of the quantum key distribution network; The pre-trained quantum key demand spatiotemporal predictive model is configured to predict a target quantum key demand of the quantum key distribution network based on the quantum key distribution network, the historical key data, the network state data, the traffic log data, and the external event data, such that the quantum key distribution network adjusts a current key pre-negotiation policy based on the target quantum key demand.

Description

Quantum key demand prediction method and quantum key demand prediction system Technical Field The invention relates to the technical field of quantum communication, in particular to a quantum key demand prediction method and a quantum key demand prediction system. Background Quantum key distribution (Quantum Key Distribution, QKD) networks are a technology for securing future communications, however, QKD networks are limited by key generation rate and transmission distance, and are difficult to support real-time encryption requirements of high-bandwidth services, and although a key pre-caching mechanism can alleviate the problem to some extent, a matching policy is needed to fall to the ground, so a reasonable key pre-negotiation policy is needed to precisely match actual requirements of quantum encryption communications, and resource mismatch or insufficient supply is avoided. Disclosure of Invention The application provides a quantum key demand prediction method and a quantum key demand prediction system. The embodiment of the application provides a quantum key demand prediction method, which comprises the following steps: Acquiring historical key data, network state data, service log data and external event data of a quantum key distribution network; And predicting a target quantum key requirement of the quantum key distribution network according to the quantum key distribution network, the historical key data, the network state data, the service log data and the external event data based on a pre-trained quantum key requirement space-time prediction model. In this way, the historical key data, network state data, service log data and external event data of the quantum key distribution network are obtained, and the target quantum key requirement of the quantum key distribution network is predicted based on the quantum key requirement space-time prediction model which is trained in advance. The method and the device can predict the target quantum key requirement of the quantum key distribution network according to four multi-mode data including historical key data, network state data, service log data and external event data based on a quantum key requirement space-time prediction model which is trained in advance, provide a reliable data base for a pre-negotiation strategy, and predict multidimensional data such as the historical key data, the network state data, the service log data and the external event data compared with a mode of only depending on data with a single dimension to implement the pre-caching strategy, so that the predicted target quantum key requirement can be matched with the actual requirement of quantum encryption communication, and further ensure the reliability of encryption service. In some embodiments, the predicting the target quantum key requirement of the quantum key distribution network based on the pre-trained quantum key requirement spatiotemporal prediction model from the quantum key distribution network, the historical key data, the network state data, the traffic log data, and the external event data comprises: constructing a space-time diagram according to the quantum key distribution network, the historical key data, the network state data, the service log data and the external event data; And inputting the space-time diagram into a pre-trained quantum key demand space-time prediction model, and predicting the target quantum key demand of the quantum key distribution network. The method comprises the steps of constructing a space-time diagram according to a quantum key distribution network, historical key data, network state data, business log data and external event data, inputting the space-time diagram into a pre-trained quantum key demand space-time prediction model, and predicting target quantum key demands of the quantum key distribution network. Therefore, by constructing the time-space diagram, the abstract quantum key distribution network topology, the scattered historical key data, the network state data, the business log data and the external event data can be structurally integrated, so that the quantum key demand space-time prediction model can capture the space association and time law among nodes based on the topological structure of the time-space diagram and the node characteristic data, and the accurate prediction of the target quantum key demand is realized. In some embodiments, the constructing a time-space diagram from the quantum key distribution network, the historical key data, the network state data, the traffic log data, and the external event data includes: Taking a password service node of the quantum key distribution network as a node of the space-time diagram, taking an interaction relation of the password service node of the quantum key distribution network as an edge of the space-time diagram, and taking a link utilization rate corresponding to the edge of the space-time diagram as an initial weight coefficient of the edge of the space-t