Efficient parallel simulation method and system for a very large scale low earth orbit satellite network

Abstract

The application discloses a kind of high-efficiency parallel simulation method and system of super large scale low-orbit satellite network, to carry out high-efficiency parallel simulation while being able to simulate low-orbit satellite network dynamic characteristic.The application can be in the topological partition of simulation process, topological construction and parallel simulation of multiple CPU computing cluster automation;Meanwhile, a kind of satellite network topology division mechanism proposed in the application can quickly divide network topology into multiple simulation processes, reduce inter-process synchronization overhead, improve parallel simulation efficiency;And a kind of conservative synchronization mechanism for satellite dynamic network is proposed, which can ensure the topological consistency between multiple simulation processes, ensure causality, and promote the correct operation of parallel simulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network simulation, and particularly relates to an efficient parallel simulation method and system for a super-large low-orbit satellite network. BACKGROUND

[0002] As an important part of future B5G / 6G global communication, large-scale low-orbit satellite networks have the advantage of providing seamless connectivity for global users, and are not affected by space, time and geographical limitations. With the decline in satellite manufacturing and launch costs, the number of low-orbit satellites deployed has grown rapidly, driving a new round of development of low-orbit mega-constellation networks. Taking Starlink as an example, it has launched more than 4,000 near-earth orbit satellites, and is expected to expand to more than 30,000. The broad prospects of low-orbit satellite constellations have driven research in constellation design, protocol design, mobility management, and other aspects.

[0003] Due to the highly dynamic nature of satellite networks, global coverage, and other characteristics, the deployment and testing costs are high, making it difficult to conduct large-scale on-orbit verification or semi-physical simulation. Therefore, before a new technology of satellite networks is actually deployed and applied, using network simulation for verification becomes the best choice. Network simulation not only simulates the impact of complex space environments on constellations, but also verifies the effectiveness and feasibility of different technical solutions in a shorter time. Through a large number of high-fidelity simulations, researchers can deeply analyze the performance of different protocols under different constellation architectures, optimize system design and improve network efficiency.

[0004] However, the high-speed movement of satellites leads to frequent changes in network topology, intermittent connection interruptions of data links, and frequent switching of satellite-ground links, further increasing the complexity and time-consuming nature of simulation. As the number of satellite nodes in the constellation further increases, the changes in network topology become more intense, and the link state becomes more complex, exponentially increasing the consumption of simulation time resources. Therefore, when simulating large-scale low-orbit satellite constellations, the core challenge is not only to simulate the effects of space environments, but also to support simulation systems for thousands or even tens of thousands of satellite nodes to meet the growing research needs of large satellite constellations. Using a single processor or CPU simulation engine alone cannot meet the large-scale simulation needs of thousands or even tens of thousands of satellite networks, and it is crucial to design an efficient parallel simulation engine and distributed computing architecture. By distributing tasks to multiple processors to utilize more computing resources, the efficiency of simulation execution can be significantly improved.

[0005] After searching the existing literature, we found that Pan Tian et al. published an article titled "A Container-Based Low-Orbit Satellite Network Protocol Test Bed" in the "Journal of Computer" in 2022. The article proposes a protocol simulation test bed for low-orbit satellite networks, which separates the control plane from the data plane. In the control plane, mathematical modeling is used to simulate discrete events for constellation topology changes. In the data plane, real network simulation of satellites and ground nodes is achieved through Docker containers and Linux virtual network devices. Based on the tunneling protocol, the system is distributed and semi-physical simulated, making it capable of horizontal expansion.

[0006] After searching the existing literature, we found that Haibin Song et al. published an article titled "Accelerating Mega-Scale Satellite Network Simulation in NS-3 via MPI-based Parallelization" in the "IEEE International Conference on Communications (ICC)" in 2024. The article studies a method to accelerate the simulation of mega-scale low-orbit satellite networks using MPI parallelization. By implementing stateful traffic drop within the network interface, the article simulates the on-off links between satellites during simulation runtime. In addition, the article divides the satellite constellation into multiple simulation processes using a hierarchical clustering algorithm.

[0007] After searching the existing literature, we found that Su Hailong et al. published an article titled "A Parallel Discrete Event Simulation Engine for the Low-Earth-Orbit Satellite Constellation Networks" in the "China Communications" in 2024. The article introduces a synchronization algorithm based on pre-computed empty messages and a synchronization algorithm based on pre-computed window network simulators to address the issue of traditional single-process simulation engines being unable to handle complex simulations involving a large number of entities.

[0008] After searching the existing literature, it was found that Ke Wang et al. published an article entitled "An Efficient Topology Partitioning Algorithm for System-Level Parallel Simulation of Mega Satellite Constellation Communication Networks" in Computer Networks in 2024. The article designs a parallel simulation platform based on the combination of STK, OPNET and Proxmox VE, proposes a resource evaluation algorithm to take CPU and memory resource consumption as indicators to measure the performance of topology partitioning, and proposes an intelligent topology partitioning algorithm based on load balancing to improve the system-level parallel simulation performance of super large satellite constellation communication networks.

[0009] In summary, these works have made beneficial exploration for parallel simulation of satellite networks, but there are limitations such as too complex simulation configuration process, high demand for computing resources, low support for satellite network simulation scale, lack of dynamic characteristics simulation capability for satellite networks, and low parallel simulation acceleration benefit. SUMMARY

[0010] The purpose of the application is to overcome the shortcomings of the prior art, and the application provides an efficient parallel simulation method and system for super large low-orbit satellite networks to realize efficient network simulation based on discrete events on super large constellation networks with high computing resource requirements.

[0011] Technical solution: In order to achieve the above purpose, the technical solution adopted by the application is:

[0012] In a first aspect, the application provides an efficient parallel simulation method for super large low-orbit satellite networks, comprising the following steps:

[0013] (1) According to the configuration information input by the user, including topology configuration, traffic model and basic simulation configuration, the simulation time is divided into several time slots, and the dynamic change information of the satellite network in each time slot is predicted, including the topology structure and the link delay and the satellite-ground link switching condition;

[0014] (2) According to the topology prediction information and the specified number of simulation processes, the low-orbit satellite network topology is divided into the specified number of simulation processes, and the load balancing of the simulation task among the simulation processes is realized;

[0015] (3) The simulation process constructs the topology of the satellite network according to the simulation process topology division information and the network topology prediction file;

[0016] (4) The parallel conservative synchronization algorithm is used to update the satellite network dynamic scene based on time slots, to promote the correct operation of the simulation, to ensure the topology consistency of each simulation process, and to ensure that the execution order of the simulation events conforms to the causality order.

[0017] Further, step (1) specifically comprises the following steps:

[0018] (1-1) Generating an orbit parameter file including satellites, for calculating the orbit position and motion state of the satellites at a specific time, predicting the network topology at the initial time of the simulation, the inter-satellite link connection information file, and the inter-satellite link connection information file;

[0019] (1-2) According to the time slot value and the simulation time, the inter-satellite link change information and the inter-satellite link connection change information in each time slot are predicted using a general orbit mechanics model, and are supplemented to the corresponding file.

[0020] Further, in step (2), the low-orbit satellite network topology is divided into a specified number of simulation processes, specifically comprising the following steps:

[0021] (2-1) The satellites with intra-orbit links are divided into a set of binary node pairs in ascending order of satellite node number, and each satellite belongs to at most one set of binary node pairs; for each ground station, it is divided into the set of satellites in which the longest inter-satellite link connection time is located;

[0022] (2-2) All sets are sorted in descending order of the number of nodes in the set; for each set, it is divided into the simulation process with the least number of nodes in the simulation process and the number of nodes in the current divided simulation process is updated until all sets are divided;

[0023] (2-3) For the nodes not divided into simulation processes, they are divided into the simulation process with the least number of nodes in the simulation process, and the number of nodes in the current divided simulation process is updated until all nodes are divided.

[0024] Further, in step (3), in the topology construction of the multi-simulation process under the satellite network, each simulation process installs all the required protocols and applications for real nodes, and only installs IP protocols for virtual nodes; periodic topology update events are set, including inter-link channel delay update and inter-satellite link connection switching events; for a simulation process, the real nodes are the nodes divided into the simulation process, and the virtual nodes are other nodes.

[0025] Further, in step (3), the topology construction of the multi-simulation process under the satellite network specifically comprises the following steps:

[0026] (3-1) The simulation process installs the nodes and links of the entire network topology. Based on the network topology division information and simulation process number of the network node, it is determined whether the node is a real node or a virtual node in this process. The inter-satellite links and satellite-to-ground links are modeled based on the P2P link model. If the nodes at both ends of the link belong to different simulation processes, an MPI message transceiver interface is added to the channel interface at both ends of the link, and the inter-process information transmission event is set to be triggered through the MPI message transceiver interface.

[0027] (3-2) The simulation process installs the protocol stack and applications. For real nodes, all protocols and applications required for simulation are installed, while for virtual nodes, only the IP protocol is installed. Each simulation process uses the same scheme to allocate IP addresses to all node interfaces and ensures the consistency of node number, interface number, and IP address in all simulation processes.

[0028] (3-3) Each simulation process sets periodic topology update events, including inter-link channel delay updates and satellite-to-ground link connection switching events. The first trigger time of the event is the topology dynamic change time slot value, and the event occurrence period is the topology dynamic change time slot value.

[0029] (3-4) The simulation statistics module of each simulation process is responsible for collecting the log information generated by the real nodes in the simulation process. Each simulation process forms a separate data statistics file.

[0030] Furthermore, in step (4) of the parallel conservative synchronization algorithm for satellite network dynamic scenarios based on time slot updates, each simulation process calculates the parallel simulation time and executes all events in the internal event queue whose timestamps are less than or equal to this time in parallel; when a topology update event occurs, the minimum channel delay value is recalculated and the synchronization state is entered to maintain the topology consistency within each simulation process; the parallel simulation time is the time of the next update of topology information in the simulation, the minimum value among the minimum timestamps of the next events of each simulation process and the minimum channel delay across the links between each simulation process.

[0031] Furthermore, the parallel conservative synchronization algorithm for satellite network dynamic scenarios based on time slot updates in step (4) specifically includes the following steps:

[0032] (4-1) The simulation process calculates the parallelizable simulation time, and each simulation process executes in parallel all events in the internal event queue whose timestamps are less than or equal to this time:

[0033] T gt =min(T) Next T LBTS +LA)

[0034] Where T gtTo assign a time, each simulation process executes all events less than or equal to this timestamp in parallel; T Next T represents the time of the next topology information update in the simulation. LBTS LA is the minimum of the next event timestamps in each simulation process, and LA is the minimum channel delay across the links between simulation processes.

[0035] (4-2) The simulation process execution event stamp is the topology update event T. Next At time 1, the simulation process executes all topology update events and updates T. Next The values ​​and LA values ​​ensure that the execution order of the simulation process conforms to causality, after which it enters a synchronization state:

[0036] T Next =T Next +T s

[0037] LA update =min(LA, LA′)

[0038] Where T s For topology-dynamically changing time slot values, LA update LA is the current value that needs to be updated. LA is the minimum channel delay of the links between each simulation process before the topology update event is executed. LA′ is the newly calculated minimum channel delay of the links between each simulation process after the topology update event is executed.

[0039] (4-3) The timestamp of the next event in the simulation process is greater than T. gt Alternatively, after the local event queue has finished executing, it enters a waiting and synchronization process. During this process, the simulation process transmits and receives inter-process data packets through the MPI message interface. Once all simulation processes have finished sending and receiving synchronization messages, T is recalculated. gt Confirm the next time that parallel execution can be granted;

[0040] (4-4) Each simulation process follows the time T granted during the parallel operation phase described above. gt The method drives parallel simulation and maintains topological consistency and causality within each simulation process through the topology update time LA update mechanism and the waiting and synchronization mechanism; when the event queues of all simulation processes have been executed, the parallel simulation exits and the simulation is completed.

[0041] (4-5) Summarize the data statistics files generated by each simulation process to obtain the total data statistics file.

[0042] Secondly, the present invention provides a highly efficient parallel simulation system for ultra-large-scale low-Earth orbit satellite networks, comprising:

[0043] A topology prediction module is configured to divide simulation time into time slots according to user input configuration information including topology configuration, traffic model and basic simulation configuration, and predict dynamic change information of each time slot satellite network, including topology structure and link delay and satellite-ground link switching;

[0044] A topology division module is configured to divide low-orbit satellite network topology into a specified number of simulation processes according to topology prediction information and the specified number of simulation processes, and realize load balancing of simulation tasks among simulation processes.

[0045] A topology construction module is configured to construct a topology of multiple simulation processes in a satellite network according to simulation process topology division information and network topology prediction files.

[0046] A simulation synchronization module is configured to promote correct operation of simulation based on a parallel conservative synchronization algorithm of simulation processes in a satellite network dynamic scene updated by time slots, and ensure topology consistency of each simulation process and that simulation event execution sequence conforms to causality sequence.

[0047] Further, the system further comprises a simulation statistics module configured to count log information generated by real nodes in a simulation process by simulation processes, form a separate data statistics file, and aggregate data statistics files formed by each simulation process by a main process to obtain a total data statistics file.

[0048] In a third aspect, the present application provides a computer program product comprising computer programs / instructions, which, when executed by a processor, implement the steps of the efficient parallel simulation method for a super large low-orbit satellite network.

[0049] Beneficial effects: The present application proposes an efficient parallel simulation method and system design for a super large low-orbit satellite network, which can fully utilize more computing resources to accelerate simulation. The present application proposes an efficient LEO mega-constellation parallel simulation architecture aiming to combine low-orbit satellite network topology characteristics and fully utilize computing resources, and is user-friendly. In the architecture, an algorithm for quickly implementing simulation process topology division is also implemented, which balances and allocates simulation tasks among simulation processes. Considering the adaptability to the dynamics of satellite networks, a topology synchronization mechanism and a minimum link delay update mechanism are introduced into the conservative synchronization algorithm mechanism, which guarantees causality of parallel simulation processes and topology consistency of simulation processes. The method and system design can realize automatic efficient parallel simulation of satellite networks in a computing cluster. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 is a parallel simulation satellite network scene diagram provided by an embodiment of the present application.

[0051] Figure 2This is a parallel simulation node protocol architecture and link connection diagram provided in an embodiment of the present invention.

[0052] Figure 3 This is a simulation flowchart provided in an embodiment of the present invention.

[0053] Figure 4 This is a framework diagram of the parallel simulation module provided in an embodiment of the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the embodiments of this invention are described in detail below with reference to the accompanying drawings. These embodiments are implemented based on the technical solutions of this invention, providing detailed implementation methods and specific operating procedures. It should be understood that the specific examples described herein are merely illustrative of this invention, but the scope of protection of this invention is not limited to the following embodiments.

[0055] Combination Figure 1 The scenario shown illustrates an efficient parallel simulation method for ultra-large-scale low-Earth orbit satellite networks disclosed in this invention. This method is used in a discrete event-driven simulator and can run on a shared memory system or a distributed memory system containing one or more CPU cores. The method includes the following steps:

[0056] 1) Based on the user-input topology configuration, traffic model, basic simulation, and other configurations, the simulation time is divided into several time slots. The dynamic nature of the satellites allows for the prediction of the satellite network topology, link delays, and satellite-to-ground link switching dynamics in each time slot. The specific steps are as follows:

[0057] 1-1) Predict and construct the network topology at the initial moment of the simulation, and generate a TLE file containing satellite orbital parameters for calculating the satellite's orbital position and motion state at a specific time; inter-satellite link connection information file and inter-satellite-to-ground link connection information file;

[0058] 1-2) Based on the time slot value and simulation duration, use a general orbital mechanics model (such as SGP4 or other applicable models) to predict the changes in satellite-to-ground links and inter-satellite link connections within each time slot, and supplement them into the corresponding files.

[0059] 2) Based on the topology prediction information and the specified number of simulation processes, the low-Earth orbit satellite network topology is divided into a specified number of simulation processes to achieve load balancing of simulation tasks among the simulation processes. The specific steps are as follows:

[0060] 2-1) Satellites with intra-orbit links are divided into binary node pairs in ascending order of satellite node number. Each satellite belongs to at most one binary node pair set. For each ground station, it is assigned to the set of satellites with the longest inter-satellite-to-ground link connection time during the simulation period.

[0061] 2-2)Sort all sets according to the number of nodes in the set from large to small; for each set, divide it to the simulation process with the least number of nodes in the simulation process and update the number of nodes in the current divided simulation process until all sets are divided;

[0062] 2-3) For the nodes not divided into the simulation process, divide them into the simulation process with the least number of nodes in the simulation process, update the number of nodes in the current divided simulation process until all nodes are divided.

[0063] 3) The simulation process constructs the topology of the satellite network according to the simulation process topology division information and the network topology prediction file. The specific steps are as follows:

[0064] 3-1) The simulation process installs the nodes and links of the entire network topology, and judges whether the node is a real node or a virtual node in the process according to the network node belonging to the network topology division information and the simulation process number; the inter-satellite link and the satellite-ground link are modeled based on the P2P link model, if the two end nodes of the link belong to different simulation processes, then add MPI message receiving and sending interfaces on the two end channel interfaces of the link, and set the inter-process information transmission event to trigger through the MPI message receiving and sending interface;

[0065] 3-2) The simulation process installs the protocol stack and application, as shown in Figure 2 , for real nodes, install all protocols and applications required for simulation, for virtual nodes, only install IP protocol; each simulation process uses the same scheme to allocate IP addresses for all node interfaces, and ensures the consistency of node number, interface number and IP address in all simulation processes;

[0066] 3-3) Each simulation process sets a periodic topology update event, including link channel delay update, satellite-ground link connection switching and other events, the first trigger time of the event is the topology dynamic change time slot value, and the event occurs periodically with a topology dynamic change time slot value;

[0067] 3-4) The simulation statistics module of each simulation process is responsible for statistics of the log information generated by the real nodes in the simulation process during the simulation process, and each simulation process forms a separate data statistics file saved in the memory space of the simulation process.

[0068] 4) The simulation process based on the satellite network dynamic scene of time slot update parallel conservative synchronization algorithm, promotes the correct operation of the simulation, ensures the consistency of the topology of each simulation process, and the execution order of the simulation event conforms to the causality order. The specific steps are as follows:

[0069] 4-1) Simulation processes calculate the parallel simulation time, each simulation process executes all events with time stamp less than or equal to the time in parallel:

[0070] T gt = min(T Next , T LBTS + LA)

[0071] where T gt is the grant time, each simulation process executes all events with time stamp less than or equal to the time in parallel. T Next is the time of next topology update in the simulation, T LBTS is the minimum of next event time stamp of each simulation process, and LA is the minimum channel delay of links between simulation processes;

[0072] 4-2) Simulation processes execute events at time T Next , the time stamp of topology update event, and update the value of T Next and LA, to ensure that the execution order of simulation processes is causal, and then enter the synchronization state:

[0073] T Next = T Next + T s

[0074] LA update = min(LA, LA')

[0075] where T s is the topology dynamic change time slot value, LA update is the current value to be updated, LA is the minimum channel delay of links between simulation processes before executing topology update events, and LA' is the newly calculated minimum channel delay of links between simulation processes after executing topology update events;

[0076] 4-3) If the next event time stamp of a simulation process is greater than T gt or the local event queue is executed, the simulation process enters the waiting and synchronization process. In this process, the simulation process transmits and receives messages through the MPI message interface to transfer inter-process data packets. When all simulation processes synchronize message transmission and reception, T gt is recalculated to confirm the grant time for the next parallel execution;

[0077] 4-4) Each simulation process drives the parallel simulation according to the grant time T gt of the above parallel running stage, and maintains the topology consistency and causality in each simulation process through the topology update time LA update mechanism and the waiting and synchronization mechanism. When the event queue of all simulation processes is executed, the parallel simulation is exited, and the simulation is completed.

[0078] 4-5) Main process string running data statistics summary script Recycle the data statistics file formed by each simulation process, and summarize it into a total data statistics file.

[0079] The implementation of the satellite network parallel simulation method of the embodiment of the application can be based on a discrete event simulator NS-3. NS-3 is a simulator for simulating network communication behavior and evaluating the performance of network protocols, and is widely used in academic research and engineering practice. In the parallel simulation method of the embodiment of the application facing the satellite network, a plurality of new mechanisms are introduced: (1) Unlike the traditional manual simulation process division topology mechanism, the load balancing division mechanism based on the characteristics of the low-orbit satellite network is adopted in the embodiment of the application, which can reduce the communication between simulation processes, evenly distribute the simulation tasks in each simulation process, improve the utilization efficiency of computing resources, and speed up the entire simulation speed; (2) The inter-satellite link and the satellite-ground link of the satellite network are modeled based on the P2P channel, and the MPI related interface supporting parallel simulation is added, which can automatically realize the topology construction in parallel simulation and the message passing between simulation processes; the parallel and distributed statistics and recycling mechanism of simulation statistics data is established, which can realize automatic and efficient global statistics and data management; (3) Unlike the traditional conservative synchronization algorithm which can only face the static network topology with fixed channel delay, the topology update mechanism based on time slots is introduced in the embodiment of the application, all simulation processes enter the topology update synchronization stage at the beginning of each time slot, and the minimum link delay value between processes is updated at the same time, which ensures the internal topology consistency of all simulation processes at the initial time of each time slot in parallel simulation, does not violate the causal order, and can adapt to the dynamic scenarios of the satellite network.

[0080] Next, a specific simulation example will be described by way of example. In this example, it is assumed that 1578 low-orbit LEO satellites of the StarLink constellation of SpaceX company are simulated, and the satellite orbit parameters are shown in Table 1.

[0081] Table 1 StarLink constellation configuration information

[0082]

[0083] Other main parameters include that the ground stations are set to the top 100 cities of global population, the traffic model is randomly generated between the ground stations, the traffic size is 1 Gbps, the UDP, OSPF and IPv4 protocols are enabled, the simulation time is set to 100 s, the topology dynamic update time slot is 1 s, the computing cluster is an AMD server equipped with 64 CPUs @ 2.09 GHz, 4 simulation processes are used, and each simulation process corresponds to one CPU.

[0084] The method flow of the embodiment is as follows Figure 3As shown, the method comprises the following steps:

[0085] (1) Read the user-configured simulation parameters, including the constellation configuration parameters (constellation configuration, number of orbits and satellites, altitude, inclination, etc.), ground station parameters (number and location, etc.), traffic model, and simulation basic parameters (time parameters such as simulation time and topology dynamic update time slot, and the number of processes running the simulation, etc.).

[0086] (2) Generate TLE files for satellites, ISL files for inter-satellite links, and two intra-orbit links and two inter-orbit links for each satellite. Use the SGP4 dynamic model to generate satellite-ground link connection information and on-off information for the satellite-ground link within the simulation time.

[0087] (3) According to the inter-satellite link connection information, divide the satellite node pairs that exist in the same orbit link set according to the satellite node number from small to large, and each satellite belongs to at most one node pair set; divide the ground station node to the set in which the satellite of the longest link time within the simulation time is located.

[0088] (4) Distribute the sets to the simulation processes with the number of nodes in the simulation process as the load, and balance the number of nodes in the simulation process; finally, distribute the unassigned nodes to the simulation process with the lightest load until all nodes are assigned. The number of nodes in the last four simulation processes is 421, 420, 421, and 422, respectively, and the distribution of the number of nodes is balanced.

[0089] (5) Each simulation process constructs a topology according to the topology prediction information file and the topology division file, including node and link installation, where the nodes at both ends of the link are not in the same simulation process, then add an MPI message receiving interface at both ends of the link; install the protocol stack and application, install all protocols and applications required for simulation of the real node assigned to this simulation process, including application layer traffic transmission application, UDP and OSPF protocols and IPv4 protocol, and virtual nodes only install IPv4 protocol, not install application.

[0090] (6) Enable the topology update module in each simulation process, trigger topology update periodically, the initial trigger time is 1s, the trigger period is 1s, and enable the data statistics module, only real node related information in the simulation process is counted.

[0091] (7) The simulation process executes the events in the local event queue with a timestamp not greater than the granted time in parallel; if the timestamp of the executed event is the same as the topology update time, the time includes the topology update event, including link delay update, satellite-ground link switching, etc., and the simulation process updates the inter-process minimum link delay and the next topology update time.

[0092] (8) The simulation process performs the event within the granted time, enters the synchronization stage, transmits and receives inter-process messages through the MPI interface, re-calculates the new granted time when all the message transmission and reception ends, and the synchronization stage ends.

[0093] (9) All simulation processes perform the local event queue, the event with the simulation time less than or equal to 100s is performed, and the parallel simulation ends, and the data statistics files of all simulation processes are recycled and summarized. The running time of all simulation processes is 19316 seconds, compared with the serial running time 69683 seconds, and the simulation efficiency speedup ratio reaches 3.6 times.

[0094] Based on the same inventive concept, the embodiment of the application also discloses a high-efficiency parallel simulation system of a super-large low-orbit satellite network, as shown in the figure, comprising: Figure 4 As shown in the figure, comprising: a topology prediction module, configured to divide the simulation time into a plurality of time slots according to the configuration information input by a user, including topology configuration, traffic model and basic simulation configuration, predict the dynamic change information of the satellite network in each time slot, including the topology structure and the link delay and the satellite-ground link switching condition; a topology division module, configured to divide the low-orbit satellite network topology into a specified number of simulation processes according to the topology prediction information and the specified number of simulation processes, and realize load balancing of the simulation task among the simulation processes; a topology construction module, configured to construct the topology of the multiple simulation processes under the satellite network according to the simulation process topology division information and the network topology prediction file; a simulation synchronization module, configured to promote the correct operation of the simulation based on the parallel conservative synchronization algorithm of the simulation process under the time slot updated satellite network dynamic scene, ensure the topology consistency of each simulation process, and ensure that the simulation event execution sequence conforms to the causality sequence.

[0095] Further, the system further comprises a simulation statistics module, configured to count the log information generated by the real node in the simulation process by the simulation process, form a separate data statistics file, and form a total data statistics file by the main process.

[0096] The embodiment of the application also discloses a computer program product, comprising computer programs / instructions, which realize the steps of the high-efficiency parallel simulation method of the super-large low-orbit satellite network when executed by a processor.

[0097] The above only describes the preferred embodiments of the application, and it should be noted that for those skilled in the art, without departing from the principles of the application, a number of improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the application.

Claims

1. A highly efficient parallel simulation method for ultra-large-scale low-Earth orbit satellite networks, characterized in that, Includes the following steps: (1) Based on the configuration information input by the user, including topology configuration, traffic model and basic simulation configuration, the simulation time is divided into several time slots, and the dynamic change information of the satellite network in each time slot is predicted, including topology structure, link delay and satellite-to-ground link switching. (2) Based on the topology prediction information and the specified number of simulation processes, the low-orbit satellite network topology is divided into a specified number of simulation processes to achieve load balancing of simulation tasks among simulation processes; (3) The simulation process constructs the topology of multiple simulation processes under the satellite network based on the simulation process topology partitioning information and the network topology prediction file; (4) The simulation process is based on the parallel conservative synchronization algorithm in the dynamic scenario of satellite network with time slot update, which promotes the correct operation of the simulation, ensures the topological consistency of each simulation process, and the execution order of simulation events conforms to the causal order. Step (1) specifically includes the following steps: (1-1) Generate a file containing satellite orbital parameters, which is used to calculate the satellite's orbital position and motion state at a preset time, predict the network topology at the initial moment of the simulation, and generate inter-satellite link connection information files and inter-satellite-to-ground link connection information files. (1-2) Based on the time slot value and simulation duration, use a general orbital mechanics model to predict the changes in satellite-to-ground links and inter-satellite link connections within each time slot, and supplement them into the corresponding files; Step (2) involves dividing the low-Earth orbit satellite network topology into a specified number of simulation processes, specifically including the following steps: (2-1) Satellites with intra-orbit links are divided into binary node pairs in ascending order of satellite node number. Each satellite belongs to at most one binary node pair set. For each ground station, it is assigned to the set of satellites with the longest inter-satellite link connection time during the simulation period. (2-2) Sort all sets in descending order of the number of nodes in the set; for each set, assign it to the simulation process with the fewest nodes in the simulation process and update the number of nodes in the current simulation process until all sets have been assigned. (2-3) For nodes that have not been divided into simulation processes, assign them to the simulation process with the fewest nodes in the simulation process, update the number of nodes in the current simulation process, until all nodes have been divided. The topology construction of multiple simulation processes under the satellite network in step (3) specifically includes the following steps: (3-1) The simulation process installs the nodes and links of the entire network topology. Based on the network topology division information and simulation process number of the network node, it is determined whether the node is a real node or a virtual node in this process. The inter-satellite links and satellite-to-ground links are modeled based on the P2P link model. If the nodes at both ends of the link belong to different simulation processes, an MPI message transceiver interface is added to the channel interface at both ends of the link, and the inter-process information transmission event is set to be triggered through the MPI message transceiver interface. (3-2) The simulation process installs the protocol stack and applications. For real nodes, all protocols and applications required for simulation are installed, while for virtual nodes, only the IP protocol is installed. Each simulation process uses the same scheme to allocate IP addresses to all node interfaces and ensures the consistency of node number, interface number, and IP address in all simulation processes. (3-3) Each simulation process sets periodic topology update events, including inter-link channel delay updates and satellite-to-ground link connection switching events. The first trigger time of the event is the topology dynamic change time slot value, and the event occurrence period is the topology dynamic change time slot value. (3-4) The simulation statistics module of each simulation process is responsible for collecting the log information generated by the real nodes in the simulation process. Each simulation process forms a separate data statistics file. The parallel conservative synchronization algorithm for satellite network dynamic scenarios based on time slot updates in step (4) specifically includes the following steps: (4-1) The simulation process calculates the parallelizable simulation time, and each simulation process executes in parallel all events in the internal event queue whose timestamps are less than or equal to this time: ； in To assign a time, each simulation process executes all events less than or equal to this timestamp in parallel; This is the time for the next update of topology information in the simulation. This is the minimum value among the next event timestamps for each simulation process. The minimum channel delay across the links between simulation processes; (4-2) The simulation process execution event is a topology update event. At that moment, the simulation process has executed all topology update events and updated... Value and The value ensures that the execution order of the simulation process conforms to causality, and then enters the synchronization state: ； in For topologically dynamic time slot values, This is the value that needs to be updated. To determine the minimum channel delay of the links between simulation processes before executing the topology update event, This refers to the minimum channel delay of the links between each simulation process newly calculated after the topology update event is completed; (4-3) The timestamp of the next event in the simulation process is greater than Alternatively, after the local event queue has finished executing, it enters a waiting and synchronization process. During this process, the simulation process transmits and receives inter-process data packets through the MPI message interface. Once all simulation processes have finished sending and receiving synchronization messages, the calculation is recalculated. Confirm the next time that parallel execution can be granted; (4-4) Each simulation process follows the time allocated for the parallel operation phase described above. This method drives parallel simulation and updates the topology at specific times. The update mechanism, along with the waiting and synchronization mechanism, maintains topological consistency and causality within each simulation process; when the event queues of all simulation processes have been completed, the parallel simulation exits, and the simulation is finished. (4-5) Summarize the data statistics files generated by each simulation process to obtain the total data statistics file.

2. A highly efficient parallel simulation system for ultra-large-scale low-Earth orbit satellite networks, characterized in that, The system, used to implement an efficient parallel simulation method for a large-scale low-Earth orbit satellite network according to claim 1, comprises: The topology prediction module is used to divide the simulation time into several time slots based on the configuration information input by the user, including topology configuration, traffic model and basic simulation configuration, and predict the dynamic changes of the satellite network in each time slot, including topology structure, link delay and satellite-to-ground link switching. The topology partitioning module is used to partition the low-Earth orbit satellite network topology into a specified number of simulation processes based on topology prediction information and the specified number of simulation processes, thereby achieving load balancing of simulation tasks among simulation processes. The topology construction module is used by the simulation process to construct the topology of multiple simulation processes under the satellite network based on the simulation process topology partitioning information and the network topology prediction file. The simulation synchronization module is used to implement a parallel conservative synchronization algorithm for satellite network dynamic scenarios based on time slot updates in the simulation process. This ensures the correct operation of the simulation, guarantees the topological consistency of each simulation process, and ensures that the execution order of simulation events conforms to the causal order.

3. The efficient parallel simulation system for a large-scale low-Earth orbit satellite network according to claim 2, characterized in that, It also includes a simulation statistics module, which is used to collect log information generated by real nodes during the simulation process and form a separate data statistics file, and the main process to summarize the data statistics files formed by each simulation process to obtain a total data statistics file.

4. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the efficient parallel simulation method for a large-scale low-Earth orbit satellite network as described in claim 1.

Images