Large-scale low-orbit satellite network service volume prediction method based on region mapping

Abstract

The invention provides a large-scale low-orbit satellite network service volume prediction method based on regional mapping, which comprises the following steps of: dividing the time of each day into a plurality of fixed-length time periods, and dividing the earth surface into a plurality of earth surface service regions; through a bilinear regression weighted prediction model and historical business volume data, obtaining all predicted business volumes between the first surface business area and the second surface business area in a certain time period of a certain day; in a certain time period of the certain day, acquiring a first service volume sharing weight of a first satellite to a first earth surface service area, and acquiring a second service volume sharing weight of a second satellite to a second earth surface service area; and according to all the predicted service volumes, the first service volume sharing weight and the second service volume sharing weight, calculating and acquiring predicted service volumes of the first surface service area and the second surface service area covered by the first satellite and the second satellite in the same time period. The method has the advantages of being simple, high in efficiency and accurate in prediction.

Description

Technical Field

[0001] The present invention relates to the field of satellite communication networks, and in particular to a method for predicting the traffic volume of a large-scale low-orbit satellite network based on regional mapping. Background Art

[0002] As a new generation of aerospace information infrastructure, Low Earth Orbit (LEO) satellite networks, with their significant advantages such as low orbital altitude, short transmission latency, and strong global coverage, are becoming a key solution for achieving ubiquitous connectivity in remote areas such as oceans, aviation, and polar regions. Compared to traditional geosynchronous and medium Earth orbit satellite networks, LEO satellite constellations significantly increase communication bandwidth and system capacity through large-scale networking. They can effectively support high-value services such as high-definition video transmission, IoT data backhaul, and emergency communications, demonstrating their irreplaceable role in scenarios such as smart ports, ocean navigation, and disaster relief. Therefore, in recent years, major communications powers and organizations around the world have planned their own large-scale low-orbit satellite network plans. The scale of the satellite network refers to the number of networking satellites, such as the Starlink plan consisting of 42,000 satellites planned and constructed by the US SpaceX company, the OneWeb constellation plan consisting of more than 7,000 satellites planned and constructed by the European Telecommunications Satellite Company, the Kuiper constellation plan consisting of more than 3,000 satellites planned and constructed by the US Amazon company, and the "GW" constellation plan consisting of 13,000 satellites planned and constructed by my country's StarNet Company.

[0003] The high-speed motion of low-orbit satellites causes network topology to dynamically change on a minute-by-minute basis. Individual user terminals frequently switch satellite nodes during ongoing communications. This dynamic nature necessitates continuous reconfiguration of network routing. Relying solely on real-time state awareness for route calculations incurs significant signaling overhead and processing delays. If inter-satellite traffic volume could be predicted in advance and dynamically reserved, switching paths pre-calculated, and edge cache resources optimized based on the predicted traffic volume, routing switching latency could be significantly reduced and inter-satellite link utilization improved. However, the high-speed motion of nodes in low-orbit satellite networks presents significant challenges for predictive model construction: the extremely short visibility window between satellites and ground users results in a highly fragmented temporal and spatial distribution of traffic. This disrupts the periodic patterns of user behavior within the coverage area of fixed base stations, which is characteristic of traditional terrestrial networks. Consequently, accurate predictions of inter-satellite traffic volume cannot be directly made using historical traffic data.

[0004] The present invention discloses a traffic prediction method based on regional mapping in large-scale low-orbit satellite networks. The method first accurately predicts the traffic between Earth regions, and then reasonably maps the predicted traffic between different regions to the traffic between different satellites, thereby achieving the effect of accurately predicting the traffic between nodes in large-scale low-orbit satellite networks. Summary of the Invention

[0005] In order to solve the problems existing in the background technology, the present invention proposes a large-scale low-orbit satellite network traffic prediction method based on regional mapping.

[0006] A method for predicting traffic volume of a large-scale low-orbit satellite network based on regional mapping, comprising the steps of:

[0007] S100, dividing the time of each day into a plurality of fixed-length time periods, and dividing the earth's surface into a plurality of surface business areas;

[0008] S200, obtaining all predicted traffic volumes between a first surface traffic area and a second surface traffic area within a certain time period on a certain day through a bilinear regression weighted prediction model and historical traffic volume data;

[0009] S300: obtaining a first traffic sharing weight of a first satellite for a first surface service area, and obtaining a second traffic sharing weight of a second satellite for a second surface service area, within a certain time period on the certain day;

[0010] S400. Calculate and obtain the predicted traffic volume between the first satellite and the second satellite for the first surface service area and the second surface service area respectively covered in the same time period based on all the predicted traffic volumes, the first traffic volume sharing weight, and the second traffic volume sharing weight.

[0011] Based on the above, including step S500 and repeating steps S200-S400, the predicted traffic volumes of all areas covered by all satellites corresponding to any time period can be obtained.

[0012] Based on the above, in step S100, the length of the time period is on the order of minutes.

[0013] Based on the above, in step S100, the size of the surface service area is divided according to the density of the surface service volume.

[0014] Based on the above, in step S200, the historical business volume data includes two groups of data. The first group of data is the business volume data C in the same time period t within the adjacent p days before the dth day. ij (d-1,t),C ij (d-2,t),…,C ij(dp,t), the business volume forecast result is Where α is the prediction parameter obtained through training; the second set of data is the business volume data C in the adjacent q time periods before the d-th time period t ij (d,t-1),C ij (d,t-2),…,C ij (d, tq), the business volume forecast result is Where β is the prediction parameter obtained through training.

[0015] Based on the above, the traffic volume C between the two regions is calculated according to the traffic volume prediction results of the two sets of data. ij (d,t):

[0016] C ij (d,t)=λC1+(1-λ)C2+ε

[0017] Among them, λ is the combination coefficient and ε is the error correction term.

[0018] Based on the above, during the t-th time period on the d-th day, the service sharing weight of satellite s for the covered surface service area r is for:

[0019]

[0020] T s =T·a / L T

[0021]

[0022] L=(a+b+c) / 2

[0023]

[0024] Where T is the duration of time period t, S r (d, t) is the set of all satellites covering the surface service area r in the time period t on the d day; when the boundary of the coverage area of satellite s is tangent to the boundary of the surface service area r before the coverage of the surface service area r begins, the length of the line connecting the sub-satellite point of satellite s and the center of the surface service area r is b; when the boundary of the coverage area of satellite s is tangent to the boundary of the surface service area r after the coverage of the surface service area r ends, the length of the line connecting the sub-satellite point of satellite s and the center of the surface service area r is c; a is the distance between the two sub-satellite points in the two tangent states.

[0025] Based on the above, in the time period t on the dth day, the total surface service area covered by satellite m is R m (d, t), the total surface service area covered by satellite n is R n(d, t), the predicted traffic volume between satellite m and satellite n for the surface service area i and surface service area j respectively covered for:

[0026]

[0027] in, is the service sharing weight of satellite m for the surface service area i covered, The service sharing weight for the surface service area j covered by satellite n.

[0028] Based on the above, in the time period t on the dth day, for all surface service areas covered by satellite m and all surface service areas covered by satellite n, the total predicted traffic volume C between satellite m and satellite n is mn (d,t) is:

[0029]

[0030] Compared with the existing technology, the present invention has outstanding substantial features and significant progress. Specifically, when predicting the traffic volume between nodes in a large-scale low-orbit satellite network, the present invention divides the network into multiple time periods of fixed length and divides the earth's surface into multiple regions. It combines historical traffic volume data and uses a bilinear regression weighted prediction model to accurately predict the traffic volume between different regions. Then, based on the set of areas covered by the satellite, the traffic volume between the regions is mapped to the traffic volume between satellite nodes, ultimately achieving the effect of accurately predicting the traffic volume between satellite nodes. Not only can it utilize regular historical traffic volume data, but the prediction method is also simpler and more efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic block diagram of the process of the present invention.

[0032] Figure 2 It is a schematic diagram of the present invention for calculating satellite weights based on the satellite coverage duration and the distance from the sub-satellite point to the center of the surface service area.

[0033] Figure 3 It is a schematic diagram of the set of areas covered by satellite nodes within a time period of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.

[0035] like Figure 1 As shown, a large-scale low-orbit satellite network traffic prediction method based on regional mapping includes the following steps: S100, dividing the time of each day into multiple fixed-length time periods, and dividing the earth's surface into multiple surface business areas; S200, obtaining all predicted traffic between a first surface business area and a second surface business area within a certain time period on a certain day through a bilinear regression weighted prediction model and historical traffic data; S300, obtaining a first traffic sharing weight of the first satellite for the first surface business area, and obtaining a second traffic sharing weight of the second satellite for the second surface business area within the certain time period on the certain day; S400, calculating and obtaining the predicted traffic between the first satellite and the second satellite for the first surface business area and the second surface business area respectively covered within the same time period based on all the predicted traffic, the first traffic sharing weight and the second traffic sharing weight.

[0036] Specifically:

[0037] Divide each day into fixed-length time periods. Suppose we want to predict the traffic volume between satellite m and satellite n in time period t on day d, and record it as C mn The length of the time period can be selected as needed. It should not be too long or too short. Too long will lead to inaccurate business volume forecasting, while too short will increase the business volume forecasting overhead. Generally, it is sufficient to be in the order of minutes.

[0038] The earth's surface is divided into multiple surface service areas (hereinafter referred to as areas), and the traffic volume between area i and area j in time period t on day d is C ij (d, t). Different principles can be used to divide the surface area according to actual conditions. In this embodiment, the surface is divided into spherical quadrilaterals according to longitude and latitude. Different regions can be divided into regions of different sizes. For example, the traffic volume in the ocean is relatively small, so it can be divided into larger regions; the traffic volume in densely populated areas is relatively large, so it can be divided into smaller regions.

[0039] The bilinear regression weighted prediction model is used to calculate the total traffic volume between area i and area j in time period t on day d, which is C ij (d, t) is used for prediction. The weighted bilinear regression in this embodiment means that the C ij (d, t) is predicted based on the two groups and C ij (d, t) has a strong correlation with the historical data to predict it:

[0040] The first set of data is the business volume data C in the same time period t within the p consecutive days before the dth day (the value of p can be selected according to the prediction accuracy requirements) ij(d-1,t),C ij (d-2,t),…,C ij (dp,t), the prediction result is The parameter α d-1 , α d-2 ,…,α d-p The model is obtained by training historical data;

[0041] The second set of data is the business volume data C in the q time periods before the time period t on the dth day (the value of q can be selected according to the prediction accuracy requirements) ij (d,t-1),C ij (d,t-2),…,C ij (d,tq), the prediction result is The parameter β t-1 , β t-2 ,…,β t-q The model is obtained by training historical data.

[0042] After predicting the business volume based on the above two sets of data, their weighted sum is taken as the final prediction result, and its value is C ij (d, t) = λC1 + (1-λ)C2 + ε, where the combination coefficient λ and the error correction term ε are obtained through training. This method can effectively utilize historical traffic data to achieve the traffic volume C between any two regions. ij (d,t) is accurately predicted.

[0043] For the set S of all satellites covering the surface area r in the d-th day time period t r (d, t), according to the coverage time T of each satellite s over the area s , and the average distance L from the subsatellite point to the center of the region s , determine the weight of satellites in sharing the traffic in the area In a period of time, there may be many satellites covering a surface area. The business volume in the area needs to be connected to the satellite network through these satellites. Therefore, the predicted business volume in the area needs to be allocated to each satellite according to a certain ratio. Figure 2 Demonstrates the calculation of a satellite's weight when distributing traffic. Calculation method. During a time period, a satellite travels a certain distance in orbit, and its coverage area also traverses a certain range on the ground. The satellite's trajectory is an arc on a sphere, but since time periods on the order of minutes are short, for ease of illustration, it can be approximated as a straight line. The ground coverage area is also approximated as a plane. The ground area is a spherical quadrilateral. For convenience, its circumscribed circle is used as the area range in the analysis process. (Since this embodiment predicts traffic volume rather than performing rigorous computational derivation, using the circumscribed circle as the surface traffic area range significantly reduces computational complexity, but only slightly overestimates the prediction results without causing significant deviations. Furthermore, since the purpose of traffic volume prediction is to pre-select routing and reserve bandwidth resources, a slightly overestimated prediction result can provide a certain margin for reserved bandwidth, which is not only acceptable but also beneficial.) During a time period, satellite coverage of an area includes at most three stages: pre-coverage, during coverage, and post-coverage, but may also include only one or two of the three stages. Before coverage of the surface service area r begins, when the boundary of the coverage area of satellite s is tangent to the boundary of the surface service area r, the length of the line connecting the sub-satellite point of satellite s (the intersection of the line connecting the center of the earth and the satellite on the earth's surface) and the center of the surface service area r is recorded as b; after coverage of the surface service area r ends, when the boundary of the coverage area of satellite s is tangent to the boundary of the surface service area r, the length of the line connecting the sub-satellite point of satellite s and the center of the surface service area r is recorded as c; and the length of the line connecting the two sub-satellite points in the above two tangent states is recorded as a, then the three lines form a triangle. In this embodiment, two indicators are calculated:

[0044] One is the coverage time T of satellite s over region r s . Let the length of the time period be T, then Figure 2 It can be seen that the coverage time of satellite s to area r is T s =T·a / L T , where L T The total length of the line connecting the two sub-satellite points corresponding to the start and end times of time period T;

[0045] The second is the average distance L from the subsatellite point to the center of region r s We only need to consider the average distance of the satellite coverage area, which is equal to the average length of the line from the center of the area r to any point on the edge a. Let half of the perimeter of the triangle be L = (a + b + c) / 2, and the area of the triangle is Then the average distance L s The value of

[0046] The set of all satellites covering the surface area r in the d-th time period t is Sr (d, t), the T of each satellite in the set can be obtained using the above method. s and L s Therefore, the traffic sharing weight of each satellite is

[0047] After obtaining the satellite's traffic sharing weight, all areas R covered by satellite m and satellite n in time period t on day d are m (d,t) and R n (d, t), calculate the inter-satellite traffic forecast result C based on regional mapping mn (d,t).

[0048] During a period of time, each satellite will cover multiple areas during its movement, and a portion of the traffic in each covered area will be shared by the satellite. Figure 3 This is an example of calculating the predicted traffic volume between satellite m and satellite n. In the time period t on the dth day, the gray part on the left side of the figure is the area set R covered by satellite m. m (d, t), the gray part on the right is the area set R covered by satellite n n (d,t). For any region i∈R m (d, t) and any region j∈R n (d, t), according to the satellite's weight of sharing the traffic in each area, the predicted traffic volume C between them ij (d,t) has This much is allocated for communication between satellite m and satellite n.

[0049] Consider R m (d,t) and R n For all regions in (d, t), the total predicted traffic volume between satellite m and satellite n can be obtained as

[0050] In the inter-satellite traffic prediction method in a large-scale low-orbit satellite network provided by the present invention, instead of directly predicting the traffic between satellites with strong time-varying characteristics, the inter-regional traffic with strong correlation is first predicted. Then, based on the situation of the satellite coverage area, the predicted inter-regional traffic is mapped to the inter-satellite traffic. This method can significantly improve the prediction accuracy of the inter-satellite traffic and provide a reliable basis for the subsequent dynamic reservation of inter-satellite link bandwidth, pre-calculation of switching paths, etc.

[0051] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A large-scale low-orbit satellite network traffic prediction method based on regional mapping, characterized in that: Including steps: S100, dividing the time of each day into a plurality of fixed-length time periods, and dividing the earth's surface into a plurality of surface business areas; S200, obtaining all predicted traffic volumes between a first surface traffic area and a second surface traffic area within a certain time period on a certain day through a bilinear regression weighted prediction model and historical traffic volume data; S300: obtaining a first traffic sharing weight of a first satellite for a first surface service area, and obtaining a second traffic sharing weight of a second satellite for a second surface service area, within a certain time period on the certain day; S400. Calculate and obtain the predicted traffic volume between the first satellite and the second satellite for the first surface service area and the second surface service area respectively covered in the same time period based on all the predicted traffic volumes, the first traffic volume sharing weight, and the second traffic volume sharing weight.

2. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 1, characterized in that: The process includes step S500 and repeats steps S200 to S400 to obtain the predicted traffic volumes of all the surface service areas covered by all satellites in any time period.

3. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 1, characterized in that: In step S100, the length of the time period is on the order of minutes.

4. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 1, characterized in that: In step S100, the size of the surface service area is divided according to the density of the surface service volume.

5. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 1, characterized in that: In step S200, the historical business volume data includes two groups of data. The first group of data is the business volume data C in the same time period t within the adjacent p days before the dth day. ij (d-1,t),C ij (d-2,t),…,C ij (dp,t), the business volume forecast result is Where α is the prediction parameter obtained through training; the second set of data is the business volume data C in the adjacent q time periods before the d-th time period t ij (d,t-1),C ij (d,t-2),…,C ij (d, tq), the business volume forecast result is Where β is the prediction parameter obtained through training.

6. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 5, characterized in that: Based on the business volume forecast results of the two sets of data, calculate the business volume C between the two regions ij (d,t): C ij (d,t)=λC1+(1-λ)C2+ε Among them, λ is the combination coefficient and ε is the error correction term.

7. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 1, characterized in that: The service sharing weight of satellite s for the surface service area r covered during the tth time period on the dth day is for: T s =T·a / L T L=(a+b+c) / 2 Where T is the duration of time period t, S r (d, t) is the set of all satellites covering the surface service area r in the time period t on the d day; when the boundary of the coverage area of satellite s is tangent to the boundary of the surface service area r before the coverage of the surface service area r begins, the length of the line connecting the sub-satellite point of satellite s and the center of the surface service area r is b; when the boundary of the coverage area of satellite s is tangent to the boundary of the surface service area r after the coverage of the surface service area r ends, the length of the line connecting the sub-satellite point of satellite s and the center of the surface service area r is c; a is the distance between the two sub-satellite points in the two tangent states.

8. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 1, characterized in that: During the time period t on the dth day, the total surface service area covered by satellite m is R m (d, t), the total surface service area covered by satellite n is R n (d, t), the predicted traffic volume between satellite m and satellite n for the surface service area i and surface service area j respectively covered for: in, is the service sharing weight of satellite m for the surface service area i covered, The service sharing weight for the surface service area j covered by satellite n.

9. The method for predicting large-scale low-orbit satellite network traffic based on regional mapping according to claim 8, characterized in that: In the time period t on the dth day, for all surface service areas covered by satellite m and all surface service areas covered by satellite n, the total predicted traffic volume C between satellite m and satellite n is mn (d, t) is:

Images