LEO satellite optical network routing load balancing method based on satellite-ground mixed flow detour

Abstract

The invention discloses an LEO satellite optical network routing load balancing method based on satellite-ground mixed flow detour, and relates to the technical field of satellite-ground satellite optical networks. The invention provides an on-satellite redundancy sensing load balancing algorithm and a satellite-ground flow detouring algorithm based on distributed long distance. According to the on-satellite redundancy sensing load balancing algorithm, a plurality of equivalent paths exist in inter-satellite links, and network congestion of the links or nodes is relieved through redundancy; then, a congestion index is introduced, a distributed long circuitous path is designed to replace a part of satellite-ground links, and a rapid distributed routing protocol is adopted to rapidly respond to data forwarding and the congestion path, so that effective flow distribution is realized; simulation experiment results show that in high-throughput satellite-ground communication, along with the increase of the total business volume, compared with other algorithms, the balancing scheme has better comprehensive performance in the aspects of blocking rate, packet loss rate and throughput, cascade congestion can be effectively relieved, and load balancing of the network is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of satellite-to-ground satellite optical networks, and in particular to a LEO satellite optical network routing load balancing method based on satellite-to-ground mixed traffic detour. Background Art

[0002] With the increasing demand for information resources, satellite optical communication technology [1-2] Satellite optical network is a kind of communication technology that uses laser beam as carrier. [3] , achieving high-bandwidth inter-satellite and satellite-to-ground data transmission. Laser communications can provide higher data rates, lower latency and greater capacity. [4] , thus reducing the burden on the ground network. Satellite optical networks provide stable and reliable communication services. With global coverage, they bring convenience to remote areas and hard-to-reach areas such as oceans, supporting the development of various industries. Unlike traditional terrestrial wireless network protocols, the routing protocol of low earth orbit (LEO) satellite optical networks must adapt to the dynamic changes of network topology and the instability of links. [5] Routing complexity becomes more significant when LEO satellites change their coverage areas and serve different numbers of devices accordingly, which ultimately leads to dynamic imbalance and unpredictable traffic load on the entire optical network. At the same time, the uneven distribution density of global hotspots also makes some nodes in the satellite network face idle or congested states, thereby reducing the utilization of satellite resources and causing imbalance in network load.

[0003] Regarding the study of satellite routing algorithms, Liu XM et al. [6] A low complexity routing algorithm (LCRA) is proposed. Each satellite dynamically selects the best next hop based on link congestion and logical address relationship. Although this scheme does not achieve the best effect due to the lack of a global view, it greatly reduces the control signaling overhead. [7] Aiming at the influence of inter-satellite link properties on the topological structure, a new time-varying topology model of low-orbit satellite network is proposed to avoid the loss of topological information when selecting the transmission path and reduce the unreliability of the routing path. The results show that the algorithm is superior to the existing algorithms in terms of packet loss rate, end-to-end delay and throughput.

[0004] In view of the uneven load distribution and periodic satellite motion, LEO satellite traffic overload, Jiang Linyan et al. [8] A GEO cooperative relay-assisted load balancing scheme is proposed. This scheme uses Stackelberg game to optimize the strategies of target nodes and relay nodes, which can maximize the average signal-to-noise ratio of all mobile terminals that need to switch. [9]By analyzing the traditional satellite scheduling method, the large amount of data in the return process through the low earth orbit (LEO) satellite network leads to poor timeliness of remote sensing images. Therefore, an edge computing virtual link is established between the computing node and the user. The Ford-Fulkerson algorithm is used to obtain the maximum flow of the topology with virtual links to improve the total capacity of multi-node information backhaul in the orbital satellite network remote sensing scenario, and it is verified in multiple scenarios. In order to solve the problem of uneven traffic distribution, the literature

[10] The global traffic balancing strategy (GTB) and the local traffic balancing strategy (LTB) have been proposed. Although GTB can cope with local congestion, it is slow and has high communication overhead, while LTB is prone to local lag. The hybrid load balancing strategy combines global and local perspectives to achieve independent decision-making on traffic information and can quickly adapt to traffic changes.

[11] A load balancing routing algorithm with traffic pre-diversion (TPLB) is proposed to delay the congestion time in the network through the traffic pre-diversion mechanism based on load prediction. Liu Shuyang et al.

[12] An adaptive time-scale load balancing (ATLB) routing algorithm based on SDN is proposed by modeling the link history traffic based on the empirical mode decomposition method. As LEO satellites change their coverage area and serve different numbers of devices accordingly, the routing complexity becomes more significant, which ultimately leads to a dynamic imbalance and usually predictable traffic load on the entire constellation.

[13] Since the user switching target selection will significantly affect the communication service quality and the load status of the satellite network, a multi-time slot load balancing scheme is introduced to provide users with a user switching target selection strategy. In the distributed scheme, the main load balancing method is distributed detour: the satellite makes decisions independently to detour part of the traffic from the default route to the alternative route. Most of the current distributed routing algorithms use the greedy shortest path to improve the high robustness and responsiveness of satellite optical networks at the expense of global performance.

[14] A selective split load balancing strategy (SSLB) is proposed, which improves the utilization of each satellite and alleviates congestion in low-latitude areas through the selective iterative Dijkstra algorithm (SIDA).

[0005] Since the current traditional distributed detour method is used for satellite-to-ground communications, it is easy to cause local lags and trigger cascade congestion. In order to alleviate these problems, a new solution to the above problems needs to be proposed. Summary of the invention

[0006] The purpose of the present invention is to build a LEO satellite optical network scenario model from the perspective of satellite-ground traffic balance, comprehensively consider the on-satellite and satellite-ground characteristics of the LEO satellite optical network, design an on-satellite load balancing algorithm using on-satellite redundant paths, and propose a LEO satellite optical network routing load balancing solution based on satellite-ground mixed traffic detours. This solution uses long-distance detours (LTD) and distributed detours to provide adaptive load balancing to achieve effective traffic distribution and relief of cascade congestion. Finally, the algorithm in this paper is simulated and compared with the traditional algorithm to solve the technical problems raised in the background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: a LEO satellite optical network routing load balancing method based on satellite-ground hybrid traffic detour, comprising at least the following steps:

[0008] S1: Determine the LEO satellite optical network model, adopt the low earth orbit satellite optical network scenario model, that is, LOSN, and the space segment of the LOSN adopts the polar orbit Walker constellation model;

[0009] S2: Building an on-board redundancy-aware load balancing algorithm, using on-board resource redundancy to reduce network congestion and improve network throughput. The on-board redundancy-aware load balancing algorithm is the RALB algorithm.

[0010] S3: Build a satellite-to-ground traffic detour multi-path routing algorithm, which is a DMRA-SGT algorithm that combines the shortest path and the long-distance detour path. The congestion distribution index is obtained based on network status information, and the detour multi-path algorithm selects a long detour path according to the fast distributed routing protocol.

[0011] Furthermore, there are multiple routes between two nodes in the LOSN, which means that there is redundancy in the transmission resources of the LOSN, and this redundancy is used to avoid potential traffic bottlenecks in the network;

[0012] The RALB algorithm balances traffic load and takes link bottlenecks into consideration. The RALB algorithm is an optimization method based on the Dijkstra algorithm. By modifying the topology processing method of the Dijkstra algorithm, heavy-load links are deleted and priority is given to light-load links to improve the utilization of network resources. The link removal process can also reduce the calculation time of the Dijkstra algorithm.

[0013] Furthermore, in order to ensure that the traffic of links at different locations is as balanced as possible, the RALB algorithm uses the current traffic load of each link as a weight when calculating the service path from the source node to the destination node, so as to find the working path with the minimum traffic load;

[0014] Since the topology of LOSN is highly symmetrical and there are multiple equivalent paths between the source node and the destination node, the RALB algorithm will give priority to the path with the smallest load. When there are multiple paths with the same load, the RALB algorithm will select the path with the least hops to reduce the end-to-end propagation delay. In this way, the RALB algorithm can keep the network traffic load balanced as much as possible when calculating the shortest path for service routing.

[0015] Furthermore, the algorithm flow of the RALB algorithm includes at least the following steps:

[0016] S2.1: Initialize the system topology after the algorithm starts;

[0017] S2.2: then obtain the service demand set path and link available bandwidth;

[0018] S2.3: Detect whether the link is overloaded;

[0019] S2.4: When there is a heavy load on a link, delete the heavy load link to reconstruct the topology, calculate multiple service paths with the same load, select the one with the least hops, and enter the service path set;

[0020] S2.5: directly enter the service path set when there is no link heavy load;

[0021] S2.6: After passing through the service path set, link heavy load detection is performed again. When the link heavy load does not exist, the process ends. When the link heavy load exists, the process returns to S2.2 and cycles again according to the steps.

[0022] Furthermore, the DMRA-SGT algorithm adopts traffic packet transmission to replace part of the services from the default route to a non-core detour path that deviates from the core satellite area and has a lower load, so as to achieve effective traffic distribution and cascade congestion relief. The DMRA-SGT algorithm first calculates the detour path. After the detour path is calculated, the routing tables of the detour path and the shortest path are synchronously updated to avoid instantaneous loops, and finally the next hop path is selected.

[0023] Furthermore, the calculation of the circuitous path comprises at least the following steps:

[0024] Firstly, the link weight attributes and network status are initialized according to the satellite optical network topology, and the RALB algorithm is used to obtain the shortest path for load balancing.

[0025] According to the core satellite area and link congestion status, the congestion distribution index H is introduced. ij =L i +A j , based on the current satellite status L i and adjacent satellite A iDynamic adjustment H ij ;

[0026] In the network G = (V, E, W) and H ij After the update, the link weight attribute W' is recalculated:

[0027] W'=W+H·ηW(E)

[0028] Where: η is the weight adjustable factor, the network topology is updated to obtain G = (V, E, W'), and then the shortest path is deleted, and the RALB algorithm is run to obtain the circuitous path, V is the traversed satellite node, and H is the congestion distribution index;

[0029] In this algorithm, the initial traffic of proportion χ∈(0,1) adopts the default shortest path, while the remaining (1-χ) traffic passes through the long-distance circuitous path. The traffic factor χ dynamically adjusts the traffic distribution ratio according to the bandwidth of each path and the current congestion status.

[0030] Furthermore, after the detour path is calculated, the routing tables of the detour path and the shortest path are synchronously updated to avoid instantaneous loops. In order to achieve rapid response to data forwarding and congested paths during the implementation process, a fast distributed routing protocol (FDR) is designed as the default routing solution for the satellite-to-ground traffic detour algorithm, which calculates the shortest path and long-distance detour path with fewer onboard resources.

[0031] Furthermore, the selection of the next hop path depends on the next hop node state. When a direction is suitable as the next hop, the corresponding next hop path is selected as "supported", otherwise it is "unsupported";

[0032] In order to avoid congestion, the local queue status and the neighboring satellite status are combined to determine the next hop selection, and the normal state and busy state are represented by N and B respectively;

[0033] The state of the ISL buffer is represented by the queue occupancy rate (QOR). When the queue occupancy rate (QOR) is lower than the preset threshold δ q (0%<δ q <100%), the local queue status is set to "normal", otherwise it is set to "busy", which should meet the following conditions:

[0034] Q·(1-δ q )≥C·(N ISL -1)·T prop

[0035] Where Q is the buffer queue size, C is the link capacity, and N is ISL is the number of links, (N ISL -1) is the number of remaining transmission ISLs, T prop is the inter-satellite transmission delay;

[0036] In order to ensure the best hop of adjacent satellites in normal state, the satellite "normal" state and satellite "busy" state are defined as:

[0037]

[0038] P ba (V j ) indicates V j The probability that the best hops are all busy is ISL ≤4:

[0039]

[0040] N B and N N Indicates the number of normal and busy links of the local queue status in the adjacent satellite. The status of the optional hop is determined by combining the local queue status and the adjacent satellite status.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] Aiming at the problem of network congestion caused by high load imbalance of transmission rate in satellite-to-ground satellite optical network, the present invention proposes an on-board redundancy-aware load balancing algorithm and a satellite-to-ground traffic detour algorithm based on distributed long distance; the on-board redundancy-aware load balancing algorithm utilizes the existence of multiple equivalent paths in inter-satellite links to alleviate network congestion of links or nodes through redundancy; then, a congestion index is introduced, a distributed long detour path is designed to replace part of the satellite-to-ground link, and a fast distributed routing protocol is adopted to quickly respond to data forwarding and congested paths, so as to realize effective traffic distribution; simulation experiment results show that in high-throughput satellite-to-ground communication, with the increase of total business volume, compared with other algorithms, the balancing scheme has better comprehensive performance in terms of blocking rate, packet loss rate and throughput, can effectively alleviate cascade congestion, and realize network load balancing. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0044] Figure 1 This is the optical network diagram of the low earth orbit satellite of this issue;

[0045] Figure 2 This is a flow chart of the on-board redundancy-aware load balancing algorithm of the present invention;

[0046] Figure 3 This is a flow chart of the satellite-to-ground traffic circuitous multipath routing algorithm of the present invention;

[0047] Figure 4 This is a schematic diagram of the 288 / 12 / 24 constellation model structure of the present invention;

[0048] Figure 5 A schematic diagram of the motion trajectories of each satellite in the polar-orbiting satellite network of the present invention;

[0049] Figure 6 Schematic diagram of traffic blocking rate under different total traffic volumes of the present invention;

[0050] Figure 7 is a graph of average packet loss rates under different total traffic volumes of the present invention;

[0051] Figure 8 This is a diagram of total throughput under different total traffic volumes of the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0053] The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour includes at least the following steps:

[0054] S1: Determine the LEO satellite optical network model, using Figure 1 The low earth orbit satellite optical network (LOSN) scenario model shown in the figure, the LOSN infrastructure can be divided into the space segment and the ground segment. Usually, the destination gateway is far away from the user terminal, and the data received by the access satellite will be transmitted to the feedback satellite through the ISL, and then sent to the destination through the feedback link. In this model, the space segment of LOSN adopts the polar orbit Walker constellation model [15-18] ;

[0055] It is composed of N=N s ×N o satellites, of which N o is the number of orbitals, N sis the number of satellites in each orbit. Each satellite is connected to four adjacent satellites through up to four bidirectional inter-satellite laser links (ISL). The satellite nodes of LOSN move at high speed relative to the earth, so the connection between the space segment and the ground segment is dynamically changing. Therefore, a virtual node strategy is used to divide LOSN into a series of slices according to the period, and each slice describes the network in a small time period. This strategy can change the routing problem from a moving physical satellite to a stable virtual satellite. The virtual satellite covers different areas and corresponds to the physical satellite. When another physical satellite enters the coverage of the virtual satellite, the relationship does not change, and the previous satellite leaves. Therefore, the satellite topology can be represented as a regular mesh structure, during which the LOSN topology is regarded as an unchanging virtual topology.

[0056] Satellite number v i The orbit number n of the satellite o (1≤n o ≤N o ) and orbital satellite number n s (1≤n s ≤N s ) gives:

[0057] i=N s ·(n o -1)+n s

[0058] In the ground segment, ground stations C are distributed in central cities with relatively concentrated traffic. The area above the ground stations is defined as the core area. When the satellite moves to this area, the satellite that directly communicates with the ground gateway is called the core satellite. The sending satellite transmits data with the ground station through the core satellite. The global information directed graph G = (V, E, W) is introduced, where the set includes satellite node v i and ground node c i , E is the edge set equal to V×V={e ij} satisfies ISL physical visibility, W = {w ij} is the link weight attribute.

[0059] In LOSN, the laser communication link can be divided into two parts: the free space above the stratosphere and the atmospheric channel below the stratosphere. Signal propagation loss is the main factor affecting the ISL signal-to-noise ratio. The link loss in free space is mainly related to the distance, while in the atmospheric channel it is affected by atmospheric attenuation and atmospheric turbulence effects. This model uses additive Gaussian noise channel and Gamma-Gamma turbulent fading channel [18-19]Modeling of inter-satellite links and satellite-to-ground links Since the bandwidth of ISL is limited, the inter-satellite links close to the core satellite are prone to traffic congestion, which inhibits the growth of network throughput. This phenomenon limits the improvement of the total bandwidth that meets the demand. Therefore, it is very necessary to reduce the service congestion caused by network bottlenecks in the space segment of LOSN.

[0060] S2: Build an on-board redundancy-aware load balancing algorithm to use on-board resource redundancy to reduce network congestion and improve network throughput. The on-board redundancy-aware load balancing algorithm is the RALB algorithm.

[0061] There are multiple routes between two nodes in LOSN, which means there is redundancy in the transmission resources of LOSN. This redundancy is used to avoid potential traffic bottlenecks in the network;

[0062] The RALB algorithm is used to balance traffic load and consider link bottlenecks. The RALB algorithm is an optimization method based on the Dijkstra algorithm. The Dijkstra algorithm is often used to find the shortest path between two nodes in a topology. It calculates each path through the topology and then selects the path with the lowest cost. It works by iteratively calculating the distance of each node in the graph, starting from the source node and reaching the destination node. In a LOSN with a regular mesh topology, the Dijkstra algorithm will lead to the reuse of some links, which will aggravate business congestion and thus limit the improvement of network throughput.

[0063] In order to solve this problem, the topology processing method of the Dijkstra algorithm is modified to delete the heavily loaded links and give priority to the lightly loaded links to improve the utilization of network resources. The algorithm flow chart is as follows: Figure 2 The link removal process can also reduce the computation time of the Dijkstra algorithm.

[0064] In order to ensure that the traffic of links at different locations is as balanced as possible, the RALB algorithm uses the current traffic load of each link as the weight when calculating the service path from the source node to the destination node, so as to find the working path with the minimum traffic load;

[0065] Since the topology of LOSN is highly symmetrical and there are multiple equivalent paths between the source node and the destination node, the RALB algorithm will give priority to the path with the smallest load. When there are multiple paths with the same load, the RALB algorithm will select the path with the least hops to reduce the end-to-end propagation delay. In this way, the RALB algorithm can keep the network traffic load balanced as much as possible when calculating the shortest path for service routing.

[0066] See also Figure 2 , the algorithm flow of the RALB algorithm includes at least the following steps:

[0067] S2.1: Initialize the system topology after the algorithm starts;

[0068] S2.2: then obtain the service demand set path and link available bandwidth;

[0069] S2.3: Detect whether the link is overloaded;

[0070] S2.4: When there is a heavy load on a link, delete the heavy load link to reconstruct the topology, calculate multiple service paths with the same load, select the one with the least hops, and enter the service path set;

[0071] S2.5: directly enter the service path set when there is no link heavy load;

[0072] S2.6: After passing through the service path set, link heavy load detection is performed again. When the link heavy load does not exist, the process ends. When the link heavy load exists, the process returns to S2.2 and cycles again according to the steps.

[0073] S3: Build a detour multi-path routing algorithm for satellite-to-ground traffic. The detour multi-path routing algorithm for satellite-to-ground traffic is a DMRA-SGT algorithm that combines the shortest path and the long-distance detour path. The congestion distribution index is obtained based on the network status information. The detour multi-path algorithm selects the long detour path according to the fast distributed routing protocol.

[0074] Due to limited bandwidth, cascade congestion is prone to occur in the core satellite area that carries high traffic volume in satellite-to-ground communications. The DMRA-SGT algorithm uses traffic packet transmission to replace part of the traffic from the default route to a non-core detour path that deviates from the core satellite area and has a lower load, so as to achieve effective traffic distribution and cascade congestion relief. The DMRA-SGT algorithm first calculates the detour path. After the detour path is calculated, the routing tables of the detour path and the shortest path are synchronously updated to avoid instantaneous loops, and finally the next-hop path is selected.

[0075] The calculation of the detour path includes at least the following steps:

[0076] Firstly, the link weight attributes and network status are initialized according to the satellite optical network topology, and the RALB algorithm is used to obtain the shortest path for load balancing.

[0077] According to the core satellite area and link congestion status, the congestion distribution index H is introduced. ij =L i +A j , based on the current satellite status L i and adjacent satellite A i Dynamic adjustment H ij ;

[0078] In the network G = (V, E, W) and Hij After the update, the link weight attribute W' is recalculated:

[0079] W'=W+H·ηW(E)

[0080] Where: η is the weight adjustable factor, the network topology is updated to obtain G = (V, E, W'), and then the shortest path is deleted, and the RALB algorithm is run to obtain the circuitous path. V is the traversed satellite node, and H is the congestion distribution index. The flowchart is as follows Figure 3 As shown;

[0081] In this algorithm, the initial traffic of the proportion χ∈(0,1) adopts the default shortest path, while the remaining (1-χ) traffic passes through the long-distance circuitous path. The traffic factor χ dynamically adjusts the traffic allocation ratio according to the bandwidth of each path and the current congestion status.

[0082] After the detour path is calculated, the routing tables of the detour path and the shortest path are updated synchronously to avoid instantaneous loops. In order to achieve rapid response to data forwarding and congested paths during the implementation process, the Fast Distributed Routing Protocol (FDR) is designed as the default routing solution for the satellite-to-ground traffic detour algorithm, which calculates the shortest path and long-distance detour path with fewer onboard resources.

[0083] The selection of the next hop path depends on the next hop node status. When a direction is suitable as the next hop, the corresponding next hop path selection is "supported", otherwise it is "unsupported";

[0084] In order to avoid congestion, the local queue status and the adjacent satellite status are combined to determine the next hop selection, and the normal state and busy state are represented by N and B respectively;

[0085] The state of the ISL buffer is represented by the queue occupancy rate (QOR). When the queue occupancy rate (QOR) is lower than the preset threshold δ q (0%<δ q <100%), the local queue status is set to "normal", otherwise it is set to "busy", which should meet the following conditions:

[0086] Q·(1-δ q )≥C·(N ISL -1)·T prop

[0087] Where Q is the buffer queue size, C is the link capacity, and N is ISL is the number of links (N ISL -1) is the number of remaining transmission ISLs, T prop is the inter-satellite transmission delay;

[0088] In order to ensure the best hop of adjacent satellites in normal state, the satellite "normal" state and satellite "busy" state are defined as:

[0089]

[0090] P ba (V j ) indicates V j The probability that the best hops are all busy is ISL ≤4:

[0091]

[0092] N B and N N It indicates the number of normal and busy links of the local queue state in the adjacent satellite. The state of the optional hop is determined by combining the local queue state and the adjacent satellite state. The determination rules are shown in Table 1, and the hop selection method is shown in Table 2.

[0093] Table 1 Next hop status

[0094]

[0095] Table 2 Jump selection rules

[0096]

[0097] In order to verify the on-board redundancy-aware load balancing algorithm and the satellite-to-ground traffic detour multipath routing algorithm, the following simulation verification is proposed:

[0098] A polar-orbiting satellite constellation was established by combining Matlab R2018b and satellite simulation software STK 11.6. The constellation includes 288 satellites (constellation parameters 288 / 12 / 24, 1400 km) and 16 ground stations. OPNET Modeler 14.5 was used for simulation verification. The simulation parameters are shown in Table 3.

[0099] Table 3 Simulation parameters

[0100]

[0101] Figure 4 The dynamics of node closeness centrality over time are shown in a simulated network with a Walker-Star constellation of 288 satellites and 16 ground gateways. The constellation consists of 12 orbits, each containing 24 satellites evenly distributed in orbit. Figure 4Assuming that all links have no bit errors, the simulation was run for 3600 seconds at a total traffic volume of 5Tbit / min to 10Tbit / min, and DMRA-SGT, SGC-LB

[21] , RALB and Dijkstra were compared.

[22] Performance between algorithms.

[0102] In order to evaluate network congestion, this paper uses the traffic blocking rate (TBR) to measure the degree of network congestion using different algorithms. Define x p ∈{0,1} indicates whether to select this path. When the total input traffic volume is i, the average blocking rate of this network is defined as a percentage:

[0103]

[0104] The average TBR trend under different total business volumes is as follows: Figure 6As shown. Among the four algorithms, the DMRA-SGT algorithm performs best in the simulation, while the Dijkstra algorithm performs poorly. In contrast, the TBRs of the Dijkstra, RALB, and SGC-LB algorithms are all significantly higher than the DMRA-SGT algorithm. When the total input traffic volume is less than 5.7Tbit / min, the TBR of Dijkstra remains zero, but beyond this point, the network begins to experience traffic congestion, and the TBR begins to grow exponentially with the number of total services. When the total traffic volume reaches the system maximum of 9.9Tbit / min, the TBR of the Dijkstra algorithm reaches a peak of 26.28%. This is because the Dijkstra algorithm needs to store information about the entire network, including the distance and previous nodes of each node, which will result in higher memory consumption when the network is more complex. When the RALB algorithm is used, for a total traffic volume of 6Tbit / min, the network begins to have traffic congestion, and as the total traffic volume rises to 9.9Tbit / min, the TBR reaches a peak of 21.71% of the TBR. However, RALB slightly alleviates congestion by considering link traffic when calculating traffic paths and allocating traffic to avoid using heavily loaded links. SGC-LB directs congested services to the nearest gateway via the shortest path, and because this routing does not consider link load, the network begins to experience traffic congestion at a total traffic volume of 6.6Tbit / min. The ability to relocate congested traffic to another gateway results in a slow increase in the bandwidth utilization rate (BUR) of the network, with SGC-LB reaching a peak TBR of 2.08% at a total traffic volume of 9.9Tbit / min. In contrast, when the DMRA-SGT algorithm is applied, the network begins to experience traffic congestion when the total traffic volume is 6.6Tbit / min, and the network reaches a peak TBR of 0.49% at the highest total traffic volume of 9.9Tbit / min, which is 25.05%, 21.22%, and 1.59% lower than the Dijkstra, RALB, and SGC-LB algorithms, respectively. This is because the DMRA-SGT algorithm avoids passing through high-traffic nodes by detouring, and the total traffic volume increases by 6.6Tbit / min, which causes congestion. The fast distributed routing protocol is used to make the best selection of the detour path. Under the highest simulation total traffic volume specified in this article, the average blocking rate of the network is 0.49%.

[0105] The average packet loss rate and total throughput under different total traffic volumes are shown in Figure 7 and Figure 8 .like Figure 7As shown in the figure, the simulation results show that DMRA-SGT has the lowest average packet loss rate compared with Dijkstra, RALB and SGC-LB. When the total traffic volume is 10Tbit / min, the packet loss rates of DMRA-SGT, SGC-LB, RALB and Dijkstra are 3.17%, 5.43%, 6.42% and 14.12% respectively. The fundamental reason for the low average packet loss rate is that the path selected by the DMRA-SGT algorithm has a low probability of congestion, that is, the long-distance detour path can bypass the predicted high traffic area to alleviate the packet loss rate caused by cascade congestion. The SGC-LB algorithm is better than the RALB algorithm that removes the high-load link because it can change the destination to a light-loaded ground gateway with a light-loaded path, thereby allowing the path to be effectively selected arbitrarily when calculating the service path. The packet loss rate of the next-hop satellite is further reduced. However, when calculating the shortest path, the Dijkstra algorithm may select a link that is already congested, causing the data packet to be discarded during transmission, resulting in a much higher packet loss rate than other algorithms. The effectiveness of DMRA-SGT can also be evaluated by the total throughput, such as Figure 8 As shown. The simulation results show that the total throughput of DMRA-SGT, SGC-LB, RALB and Dijkstra algorithms increases gently with the increase of total business volume. The difference is the largest when the total business volume is 9.9Tbit / min. The total throughputs are 9821.6, 9278.7, 8488.8 and 8020.1Gbit respectively. It can be concluded that the total throughput accounts for 99.2%, 93.7%, 85.7% and 81% of the total business volume respectively. In terms of total throughput, the gap between DMRA-SGT and RALB and Dijkstra algorithm is 18.2% and 4.7%. Figures 6 to 8 ,The total throughput of different algorithms is not as obvious as the ,service blocking rate and packet loss rate in the graph. The reason is that the network has a certain redundancy ,under different algorithms, but under high load, the packet loss rate and blocking rate may ,rise sharply.

[0106] In summary:

[0107] Considering the impact of service load in satellite optical networks, the present invention first proposes an on-board redundancy-aware load balancing algorithm, which uses on-board resource redundancy to reduce network congestion and improve network throughput. Then, a DMRA-SGT algorithm that combines the shortest path and the long-distance detour path is proposed. The congestion distribution index is obtained based on network status information, and the detour path algorithm selects the long detour path according to the fast distributed routing protocol. Through simulation results, it is compared and analyzed that the DMRA-SGT algorithm is better than other algorithms in service blocking rate, packet loss rate and total throughput under high-throughput services. Compared with the most widely used Dijkstra algorithm, the blocking rate is reduced by 25.05%, the packet loss rate is reduced by 10.95%, and the total throughput is significantly improved.

[0108] References cited for use in the present invention:

[0109] [1]Toyoshima M.Recenttrends in space laser communications for smallsatellites and constellations[J].Journal ofLightwave Technology,2021,39(3):693-699.

[0110] [2]Kolev DR, Carrasco-Casado A, Trinh PV, et al. Latest developments in the field of optical communications for small satellites and beyond[J]. Journal of Lightwave Technology, 2023, 41(12): 3750-3757.

[0111] [3]KimW H.BER Analysis of IM and BPPM for Satellite-to-Ground LaserCommunications[C] / / 2020International Conference on Electronics, Information, and Communication(ICE-IC),January 19-22,2020,Barcelona,Spain.New York:IEEE,2020:1-2.

[0112] [4]Sodnik Z,Furch B,Lutz H.Optical intersatellite communication[J].IEEE journal of selected topics in quantum electronics,2010,16(5):1051-1057.

[0113] [5] Zhao Shanghong, Peng Cong, Li Yongjun, et al. Progress in key technologies of next-generation satellite optical networks for satellite internet[J]. Laser & Optoelectronics Progress, 2023, 60(7): 0700001.

[0114] Zhao S H,Peng C,Li Y J,et al.Key TechnologyProgress ofNext-GenerationSatellite Optical Network for Satellite Internet[J].Laser&OptoelectronicsProgress,2023,60(7):0700-001.

[0115] [6]Liu X M,Yan X M,Jiang Z Q,et al.Alow-complexity routing algorithmbased on loadbalancing forLEO satellite networks[C] / / 2015IEEE 82ndVehicularTechnology Conference(VTC2015-Fall),September 06-09,2015,Boston,MA,USA.NewYork:IEEE,2015:1-5.

[0116] [7]Han Z Z,Xu C,Zhao G F,et al.Time-varying topology model fordynamic routing in LEO satellite constellation networks[J].IEEE Transactionson Vehicular Technology,2022,72(3):3440-3454.

[0117] [8]Jiang LY,Cui G F,Liu S J,et al.Cooperative relay assisted loadbalancing scheme based on Stackelberg game for hybrid GEO-LEO satellitenetwork[C] / / 2015 International Conference onWireless Communications&SignalProcessing(WCSP),October 15-17,2015,Nanjing,China.NewYork:IEEE,2015,978:1-5.

[0118] [9]Wang Y Q,Che J B,Wang N Y,et al.Load-balancing method for leosatellite edge-computing networks based on the maximum flow ofvirtual links[J].IEEE Access,2022,10:100584-100593.

[0119]

[10] Duan C,Peng W,Wang B S.GLIB:A Global and Local Integrated LoadBalancing Scheme for DatacenterNetwork[C] / / 202214th International Conferenceon Communication Software andNetworks(ICCSN),June10-12,2022,Chongqing,China,NewYork:IEEE,2022:166-173.

[0120]

[11] Shi W D,Liu J,Liu S Y.Load Balancing RoutingAlgorithm withTraffic Pre-shunting in the LEO Satellite Network[C] / / 2022 IEEE 95thVehicular Technology Conference:(VTC2022-Spring),June 19-22,2022,Helsinki,Finland.NewYork:IEEE,2022:1-5.

[0121]

[12] Liu S,Liu J,Xia B.Adaptive Timescale Load BalancingRoutingAlgorithm for LEO Satellite Network[C] / / 2023 IEEE / CIC InternationalConference on Communications in China(ICCC),August 10-12,2023,Dalian,China.New York:IEEE,2023:1-5.

[0122]

[13] Chen H R,Nie G F,Tian H.A Multi-Slot Load Balancing Scheme forLEO Satellite Communication Handover Target Selection[C] / / 2024IEEE WirelessCommunications and Networking Conference(WCNC),April 21-24,2024,Dubai,UnitedArab.New York:IEEE,2024,979:1-6.

[0123]

[14] Liu J,Luo R Z,Huang T,et al.A load balancing routing strategy forLEO satellite network[J].IEEE Access,2020,8:155136-155144.

[0124]

[15] Wei Y,Li H,Du X.An efficient LEO global navigation constellationdesign based on Walker constellation[C] / / 2020IEEE Computing,Communicationsand IoT Applications(ComComAp),December 20-22,2020,Beijing,China.New York:IEEE,2020:1-6.

[0125]

[16] Ning Y,Yi L,Zhao Y,et al.Load-balancing routing algorithms forservice congestion avoidance in LEO optical satellite networks[J].Journal ofOptical Communications and Networking,2023,15(12):1038-1049.

[0126]

[17] Wang F, Jiang D, Wang Z, et al. Dynamic networking for continuabletransmission optimization in leo satellite networks [J]. IEEE Transactions on Vehicular Technology, 2022, 72(5): 6639-6653.

[0127]

[18] Lee YH,Choi J P.Connectivity analysis of mega-constellationsatellite networks with optical intersatellite links[J].IEEE Transactions onAerospace and Electronic Systems,2021,57(6):4213-4226.

[0128]

[19] Bao Chaoyuan, Cao Yang, Peng Xiaofeng, et al. Performance analysis of RIS-assisted MUD-RF / FSO hybrid system under co-channel interference[J]. Acta Optica Sinica, 2023, 43(22): 2206003.

[0129] Bao CY,Cao Y,Peng XF,et al.Performance Analysis of RIS-AssistedMUD-RF / FSO Hybrid System Under Co-Channel Interference[J].Acta Optica Sinica,2023,43(22):2206003.

[0130]

[20] Cao Yang, Bao Chaoyuan, Peng Xiaofeng, Xing Wenjun. Performance analysis of RIS-assisted FSO-RF hybrid system under co-channel interference[J]. Acta Optica Sinica, 2024, 44(6): 0606001.

[0131] Cao Y,Bao CY,Peng XF,et al.PerformanceAnalysis ofRIS-AssistedFSO-RFHybrid Systems Under Co-Channel Interference[J].Acta Optica Sinica,2024,44(6):0606001.

[0132]

[21] Tao JH,Na ZY,Lin B,et al.A joint minimum hop and earliest arrival routing algorithm for leo satellite networks[J].IEEE Transactions onVehicular Technology,2023,72(12):16382-16394.

[0133]

[22] Ning YX, Yi LT, Zhao YL, et al. Load-balancing routing algorithms for service congestion avoidance in LEO optical satellite networks [J]. Journal of Optical Communications and Networking, 2023, 15(12): 1038-1049.

[0134] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour, characterized by: At least the following steps are included: S1: Determine the LEO satellite optical network model, adopt the low earth orbit satellite optical network scenario model, that is, LOSN, and the space segment of the LOSN adopts the polar orbit Walker constellation model; S2: Building an on-board redundancy-aware load balancing algorithm, using on-board resource redundancy to reduce network congestion and improve network throughput. The on-board redundancy-aware load balancing algorithm is the RALB algorithm. S3: Build a satellite-to-ground traffic detour multi-path routing algorithm, which is a DMRA-SGT algorithm that combines the shortest path and the long-distance detour path. The congestion distribution index is obtained based on network status information, and the detour multi-path algorithm selects a long detour path according to the fast distributed routing protocol.

2. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 1 is characterized in that: There are multiple routes between two nodes in the LOSN, which means that there is redundancy in the transmission resources of the LOSN, and this redundancy is used to avoid potential traffic bottlenecks in the network; The RALB algorithm balances traffic load and takes link bottlenecks into consideration. The RALB algorithm is an optimization method based on the Dijkstra algorithm. By modifying the topology processing method of the Dijkstra algorithm, heavy-load links are deleted and priority is given to light-load links to improve the utilization of network resources. The link removal process can also reduce the calculation time of the Dijkstra algorithm.

3. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 2 is characterized in that: In order to ensure traffic balance of links at different locations, the RALB algorithm uses the current traffic load of each link as a weight when calculating the service path from the source node to the destination node, so as to find the working path with the minimum traffic load; Since the topology of LOSN is highly symmetrical and there are multiple equivalent paths between the source node and the destination node, the RALB algorithm will give priority to the path with the smallest load. When there are multiple paths with the same load, the RALB algorithm will select the path with the least hops to reduce the end-to-end propagation delay. In this way, the RALB algorithm can keep the network traffic load balanced as much as possible when calculating the shortest path for service routing.

4. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 3 is characterized in that: The algorithm flow of the RALB algorithm includes at least the following steps: S2.1: Initialize the system topology after the algorithm starts; S2.2: then obtain the service demand set path and link available bandwidth; S2.3: Detect whether the link is overloaded; S2.4: When there is a heavy load on a link, delete the heavy load link to reconstruct the topology, calculate multiple service paths with the same load, select the one with the least hops, and enter the service path set; S2.5: directly enter the service path set when there is no link heavy load; S2.6: After passing through the service path set, link heavy load detection is performed again. When the link heavy load does not exist, the process ends. When the link heavy load exists, the process returns to S2.2 and cycles again according to the steps.

5. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 1 is characterized in that: The DMRA-SGT algorithm adopts traffic packet transmission and replaces part of the services from the default route to a non-core detour path that deviates from the core satellite area and has a lower load, so as to achieve effective traffic distribution and cascade congestion relief. The DMRA-SGT algorithm first calculates the detour path. After the detour path is calculated, the routing tables of the detour path and the shortest path are synchronously updated to avoid instantaneous loops, and finally the next hop path is selected.

6. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 5 is characterized in that: The calculation of the circuitous path comprises at least the following steps: Firstly, the link weight attributes and network status are initialized according to the satellite optical network topology, and the RALB algorithm is used to obtain the shortest path for load balancing. According to the core satellite area and link congestion status, the congestion distribution index H is introduced. ij =L i +A j , based on the current satellite status L i and adjacent satellite A i Dynamic adjustment H ij ; In the network G = (V, E, W) and H ij After the update, the link weight attribute W' is recalculated: W'=W+H·ηW(E) Where: η is the weight adjustable factor, the network topology is updated to obtain G = (V, E, W'), and then the shortest path is deleted, and the RALB algorithm is run to obtain the circuitous path, V is the traversed satellite node, and H is the congestion distribution index; In this algorithm, the initial traffic of the proportion χ∈(0,1) adopts the default shortest path, while the remaining (1-χ) traffic passes through the long-distance circuitous path. The traffic factor χ dynamically adjusts the traffic allocation ratio according to the bandwidth of each path and the current congestion status.

7. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 6 is characterized in that: After the detour path is calculated, the routing tables of the detour path and the shortest path are synchronously updated to avoid instantaneous loops. In order to achieve rapid response to data forwarding and congested paths during the implementation process, a fast distributed routing protocol is designed as the default routing solution for the satellite-to-ground traffic detour algorithm, and the shortest path and long-distance detour path are calculated using fewer onboard resources.

8. The LEO satellite optical network routing load balancing method based on satellite-ground mixed traffic detour according to claim 7 is characterized in that: The selection of the next hop path depends on the next hop node state. When a direction is suitable as the next hop, the corresponding next hop path is selected as "supported", otherwise it is "not supported"; In order to avoid congestion, the local queue status and the neighboring satellite status are combined to determine the next hop selection, and the normal state and busy state are represented by N and B respectively; The state of the ISL buffer is represented by the queue occupancy rate (QOR). When the queue occupancy rate (QOR) is lower than the preset threshold δ q (0%<δ q <100%), the local queue status is set to "normal", otherwise it is set to "busy", which should meet the following conditions: Q·(1-δ q )≥C·(N ISL -1)·T prop Where Q is the buffer queue size, C is the link capacity, and N is ISL is the number of links, (N ISL -1) is the number of remaining transmission ISLs, T prop is the inter-satellite transmission delay; In order to ensure the best hop of adjacent satellites in normal state, the satellite "normal" state and satellite "busy" state are defined as: P ba (V j ) indicates V j The probability that the best hops are all busy is ISL ≤4: N B and N N Indicates the number of normal and busy links of the local queue status in the adjacent satellite. The status of the optional hop is determined by combining the local queue status and the adjacent satellite status.

Images