A Method for Predicting and Classifying the Quality of Space-Air-Ground Power Communication Links Based on Multi-Source Information Fusion

Abstract

This invention discloses a method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion. First, a deterministic capability classification system based on power service SLA requirements is constructed, which includes quantified multi-level indicator thresholds. Environmental perception data (meteorological, ephemeris) and network measurement data are fused, and a predictive model is used to derive the expected link performance within a future time window. The expected link performance is then matched level by level with a predefined classification system to dynamically determine the dynamic deterministic level of the link within the future time window. When this dynamic level is lower than the required level for the service, an early warning and rerouting preparation are triggered, achieving proactive pre-emptive maintenance of network quality.

Description

Technical Field

[0001] This invention relates to the technical field of integrated air-space-ground networks, specifically to a method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion. Background Technology

[0002] In integrated air-space-ground networks, wireless links (especially satellite and UAV communication links) are affected by various factors such as atmospheric environment, geographical location, and node movement, resulting in drastic fluctuations in channel quality. To achieve deterministic transmission of power services, the network orchestration system needs to anticipate changes in link quality in order to plan reliable paths for high-priority services in advance. Currently, this mainly relies on network measurement techniques to obtain the real-time status of the links.

[0003] Existing solutions primarily rely on active measurement (such as Ping and in-band network telemetry) and passive measurement (such as NetFlow). The controller obtains link metrics such as latency, jitter, and packet loss rate by periodically sending probe packets or collecting device statistics; this is a post-event evaluation method. Therefore, the shortcomings of existing technologies are at least as follows: First, there is a lag; by the time a deterioration in link quality is measured, services may already be affected. There is no way to provide early warnings for sudden degradation of satellite links caused by rainfall or drones being affected by strong winds.

[0004] Second, the information source is singular, relying solely on network layer data and ignoring the physical environment and equipment status. For example, it is unaware that an impending heavy rain will severely impact Ka-band satellite links, nor can it predict the brief interruption caused by the upcoming switching of low-Earth orbit satellites.

[0005] Third, the assessment is static. The assessment of link quality is usually real-time and static, and it is impossible to give a deterministic level for a future time period, which makes it difficult for the network orchestration system to make forward-looking decisions. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention aims to solve the technical problems of existing link quality assessment methods being lagging and having a single information source. It provides a method that can predict link quality in advance and dynamically adjust its deterministic service level. Furthermore, the method of this invention integrates environmental perception data and network measurement data, uses a predictive model to assess future link quality, and dynamically adjusts its deterministic level.

[0007] A method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion, comprising a predefined classification system for deterministic performance capabilities, the method including the following steps: S1 Multi-source heterogeneous data acquisition; Construct a data acquisition layer and simultaneously acquire space-based / environmental data, airborne / environmental sensing data, and ground-based / network performance data; S2 Construct and train the link quality prediction model; After spatiotemporal alignment, the data collected in step S1 is input into the time series prediction model, and the time series prediction model outputs the predicted values ​​of key link performance indicators within a future time window; S3 Dynamic Deterministic Level Assessment: The predicted value output from step S2 is compared and assessed in real time with the predefined deterministic capability grading system in this method. During the assessment, the system matches the predicted performance feature vector with the threshold matrix level by level to determine the dynamic deterministic level of the link in the future time window. S4 Warning and Rerouting Trigger: Based on the dynamic determinism level of the link obtained in step S3, if it is lower than the SLA requirement of the service currently carried on the link, the system immediately generates a warning message and sends the message to the upper-layer orchestration controller.

[0008] It should be noted that the space-based / environmental data in step S1 is obtained from the meteorological service center, including rainfall intensity forecasts and thunderstorm paths; and from the satellite control system, including low-orbit satellite ephemeris, overpass time windows, and attitude adjustment plans. The air-based / environmental perception data is obtained from sensors mounted on the UAV, including real-time environmental data such as wind speed, temperature, and humidity. The ground-based / network performance data is obtained from the network measurement plane, including real-time and historical link performance data, including end-to-end latency, jitter, bit error rate, and received signal strength indication data.

[0009] It should be noted that in step S2, the future time window is the predicted value of the key performance indicators of the link within the next 5 minutes or 10 minutes; wherein, the predicted value includes expected latency, expected jitter, expected bit error rate, and expected interruption probability.

[0010] It should be noted that in step S3, the hierarchical system is pre-built based on the SLA requirements of the power business and includes a hierarchical definition rule base and a multi-dimensional indicator threshold matrix.

[0011] It should be noted that the rule base defines service commitments and applicable scenarios at five deterministic levels, from Level 0 to Level 4; the threshold matrix quantifies these levels into boundary values ​​for key indicators such as end-to-end latency, jitter, packet loss rate, and time synchronization accuracy; during evaluation, the predicted performance feature vector is matched with the threshold matrix level by level to determine the dynamic deterministic level of the link within the future time window.

[0012] It should be noted that in step S4, the orchestration controller initiates rerouting calculation and resource reservation in advance based on the received early warning information, plans a backup path, and completes the smooth migration of critical services before the original link deteriorates.

[0013] The beneficial effects of this invention are as follows: First, it shifts from a passive to a proactive approach, elevating network operations and maintenance from post-fault recovery to pre-fault prevention, significantly improving network resilience and service reliability.

[0014] Second, the assessment is more accurate. By integrating environmental perception data, it makes up for the blind spots of pure network measurement, making the quality assessment of wireless links, which are greatly affected by the environment, more accurate and comprehensive.

[0015] Third, it supports deterministic scheduling by providing dynamic, timestamped deterministic levels, which provides key data support for upper-layer orchestration systems to achieve forward-looking and deterministic business scheduling. Detailed Implementation

[0016] The following embodiments are based on this technical solution and provide detailed implementation methods and specific operation processes, but the protection scope of this invention is not limited to these embodiments.

[0017] It should be noted that the time-series prediction model of this invention can be a Long Short-Term Memory (LSTM) network, which can be used to effectively capture the long-term evolution trend of link performance; it can also use a more computationally efficient Gated Recurrent Unit (GRU) to meet the real-time requirements of edge computing nodes; or it can employ an advanced Transformer model, utilizing its self-attention mechanism to deeply explore the complex intrinsic relationships between multi-source heterogeneous data, thereby improving prediction accuracy in complex environments. Using any of the above-mentioned time-series models does not affect the implementation of this invention.

[0018] A method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion, comprising a predefined classification system for deterministic performance capabilities, the method including the following steps: S1 Multi-source heterogeneous data acquisition; Construct a data acquisition layer and simultaneously acquire space-based / environmental data, airborne / environmental sensing data, and ground-based / network performance data; S2 Construct and train the link quality prediction model; After spatiotemporal alignment, the data collected in step S1 is input into the time series prediction model, and the time series prediction model outputs the predicted values ​​of key link performance indicators within a future time window; S3 Dynamic Deterministic Level Assessment: The predicted value output from step S2 is compared and assessed in real time with the predefined deterministic capability grading system in this method. During the assessment, the system matches the predicted performance feature vector with the threshold matrix level by level to determine the dynamic deterministic level of the link in the future time window. S4 Warning and Rerouting Trigger: Based on the dynamic determinism level of the link obtained in step S3, if it is lower than the SLA requirement of the service currently carried on the link, the system immediately generates a warning message and sends the message to the upper-layer orchestration controller.

[0019] It should be noted that the space-based / environmental data in step S1 is obtained from the meteorological service center, including rainfall intensity forecasts and thunderstorm paths; and from the satellite control system, including low-orbit satellite ephemeris, overpass time windows, and attitude adjustment plans. The air-based / environmental perception data is obtained from sensors mounted on the UAV, including real-time environmental data such as wind speed, temperature, and humidity. The ground-based / network performance data is obtained from the network measurement plane, including real-time and historical link performance data, including end-to-end latency, jitter, bit error rate, and received signal strength indication data.

[0020] It should be noted that in step S2, the future time window is the predicted value of the key performance indicators of the link within the next 5 minutes or 10 minutes; wherein, the predicted value includes expected latency, expected jitter, expected bit error rate, and expected interruption probability.

[0021] It should be noted that in step S3, the hierarchical system is pre-built based on the SLA requirements of the power business and includes a hierarchical definition rule base and a multi-dimensional indicator threshold matrix.

[0022] It should be noted that the rule base defines service commitments and applicable scenarios at five deterministic levels, from Level 0 to Level 4; the threshold matrix quantifies these levels into boundary values ​​for key indicators such as end-to-end latency, jitter, packet loss rate, and time synchronization accuracy; during evaluation, the predicted performance feature vector is matched with the threshold matrix level by level to determine the dynamic deterministic level of the link within the future time window.

[0023] It should be noted that in step S4, the orchestration controller initiates rerouting calculation and resource reservation in advance based on the received early warning information, plans a backup path, and completes the smooth migration of critical services before the original link deteriorates. Example

[0024] The following steps are used to implement the air-space-ground power communication link quality prediction and classification assessment method based on multi-source information fusion of the present invention: Step 1: Multi-source heterogeneous data acquisition.

[0025] Build a data acquisition layer to simultaneously collect the following three types of data: First, we obtain space-based / environmental data, such as rainfall intensity forecasts and thunderstorm paths from meteorological service centers, and low-orbit satellite ephemeris, overpass time windows, and attitude adjustment plans from satellite control systems.

[0026] Secondly, there is airborne / environmental perception data, which obtains real-time environmental data such as wind speed, temperature, and humidity from sensors mounted on the drone.

[0027] Finally, there is the ground / network performance data, which obtains real-time and historical link performance data from the network measurement plane, including end-to-end latency, jitter, bit error rate, received signal strength indication (RSSI), etc.

[0028] Step 2: Build and train the link quality prediction model.

[0029] The multi-source data collected in step 1 is spatiotemporally aligned and then input into a time-series prediction model. This model can be a Long Short-Term Memory (LSTM) network to effectively capture the long-term evolution trend of link performance; it can also be a more computationally efficient Gated Recurrent Unit (GRU) to meet the real-time requirements of edge computing nodes; or it can adopt the current advanced Transformer model, utilizing its self-attention mechanism to deeply explore the complex intrinsic relationships between multi-source heterogeneous data and improve prediction accuracy in complex environments.

[0030] The model outputs predicted values ​​of key performance indicators (KPIs) for the link within a future time window (e.g., the next 5 minutes or 10 minutes), including: expected latency, expected jitter, expected bit error rate, and expected outage probability.

[0031] Step 3: Dynamic Determinism Level Assessment.

[0032] The predicted KPIs output in step 2 are compared and evaluated in real time with the predefined deterministic capability grading system in this method. This grading system is pre-constructed based on power business SLA requirements and includes a grading definition rule base and a multi-dimensional indicator threshold matrix. The rule base defines the service commitments and applicable scenarios for five deterministic levels, from Level 0 to Level 4; the threshold matrix quantifies these levels into boundary values ​​for key indicators such as end-to-end latency, jitter, packet loss rate, and time synchronization accuracy. During evaluation, the system matches the predicted performance feature vector with the threshold matrix level by level to determine the dynamic deterministic level of the link within a future time window.

[0033] For example, suppose a link currently measures a latency of 5ms, meeting Level-3 requirements. However, the prediction model outputs that "in the next 5 minutes, due to heavy rain covering the link path, the expected latency will rise to 150ms." Based on the rule base, the system determines that the deterministic level of this link will automatically downgrade from Level-3 to Level-1 (bounded latency guarantee) within the next 5 minutes.

[0034] Step 4: Warning and rerouting trigger.

[0035] When the future link grade assessed in step 3 is lower than the SLA requirements of the services currently carried on that link, the system immediately generates an early warning message and sends this message (including the degraded link ID, predicted grade, and degraded time period) to the upper-layer orchestration controller. Based on the early warning message, the orchestration controller initiates rerouting calculations and resource reservations in advance, plans an alternative path, and completes the smooth migration of critical services before the original link degrades.

[0036] For those skilled in the art, various corresponding changes and modifications can be made based on the above technical solutions and concepts, and all such changes and modifications should be included within the protection scope of the claims of this invention.

Claims

1. A method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion, comprising a predefined classification system for deterministic performance capabilities, characterized in that, The method includes the following steps: S1 Multi-source heterogeneous data acquisition; Construct a data acquisition layer and simultaneously acquire space-based / environmental data, airborne / environmental sensing data, and ground-based / network performance data; S2 Construct and train the link quality prediction model; After spatiotemporal alignment, the data collected in step S1 is input into the time series prediction model, and the time series prediction model outputs the predicted values ​​of key link performance indicators within a future time window; S3 Dynamic Deterministic Level Assessment: The predicted value output from step S2 is compared and assessed in real time with the predefined deterministic capability grading system in this method. During the assessment, the system matches the predicted performance feature vector with the threshold matrix level by level to determine the dynamic deterministic level of the link in the future time window. S4 Warning and Rerouting Trigger: Based on the dynamic determinism level of the link obtained in step S3, if it is lower than the SLA requirement of the service currently carried on the link, the system immediately generates a warning message and sends the message to the upper-layer orchestration controller.

2. The method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion as described in claim 1, characterized in that, The space-based / environmental data in step S1 is obtained from the meteorological service center, including rainfall intensity forecasts and thunderstorm paths; and from the satellite control system, including low-orbit satellite ephemeris, overpass time windows, and attitude adjustment plans. The air-based / environmental perception data is obtained from sensors mounted on the UAV, including real-time environmental data such as wind speed, temperature, and humidity. The ground-based / network performance data is obtained from the network measurement plane, including real-time and historical link performance data such as end-to-end latency, jitter, bit error rate, and received signal strength indication.

3. The method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion as described in claim 1, characterized in that, In step S2, the future time window is the predicted value of the key performance indicators of the link within the next 5 minutes or the next 10 minutes; wherein, the predicted value includes expected latency, expected jitter, expected bit error rate, and expected interruption probability.

4. The method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion according to claim 1, characterized in that, In step S3, the hierarchical system is pre-built based on the SLA requirements of the power business and includes a hierarchical definition rule base and a multi-dimensional indicator threshold matrix.

5. The method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion according to claim 4, characterized in that, The rule base defines service commitments and applicable scenarios for five deterministic levels, from Level 0 to Level 4; the threshold matrix quantifies these levels into boundary values ​​for key indicators such as end-to-end latency, jitter, packet loss rate, and time synchronization accuracy; during evaluation, the predicted performance feature vector is matched with the threshold matrix level by level to determine the dynamic deterministic level of the link within the future time window.

6. The method for predicting and classifying the quality of air-space-ground power communication links based on multi-source information fusion according to claim 1, characterized in that, In step S4, the orchestration controller initiates rerouting calculation and resource reservation in advance based on the received early warning information, plans a backup path, and completes the smooth migration of critical services before the original link deteriorates.