A link dynamic control method for large-scale satellite network simulation platform

Abstract

The present disclosure belongs to the field of satellite network simulation, and in particular, relates to a link dynamic control method for a large-scale satellite network simulation platform. It is characterized in that the method includes: the eBPF program uses the device index as a key to search for the simulation start time of the current device from the eBPF Map; determines whether the simulation has started, and if not, re-searches the simulation start time of the current device; if the simulation has started, calculates the timestamp based on the system time and the simulation start time, and uses the timestamp and the device index as keys to search for the link status information of the current device at the current moment from the eBPF Map; the eBPF program executes the link control logic based on the eBPF according to the link status information. The present invention reduces the system resource overhead during formal simulation, can support large-scale dynamic link status control with finer time granularity, and will not affect the simulation logic of satellite nodes during formal simulation.

Description

Technical Field

[0001] The present invention belongs to the field of satellite network simulation, and particularly relates to a link dynamic regulation method for a large-scale satellite network simulation platform. Background Art

[0002] As a new type of network system, satellite network has broad application prospects in both military and civilian scenarios due to its characteristics of providing high coverage and low latency services. In order to develop related technologies of satellite network, such as inter-satellite routing technology, satellite-ground routing technology, and satellite network protocols, a verification platform is needed to verify the feasibility of new technologies related to satellite network.

[0003] Currently, there are two implementation methods for the verification platform: a real hardware verification platform and a simulation software verification platform. Due to the problems of high cost of satellite network equipment and difficulty in modifying its working logic after hardware solidification, building a real hardware satellite network technology verification platform based on real satellite network equipment has the disadvantages of being uneconomical and inflexible. Therefore, a more economical and flexible network simulation verification platform based on software implementation is more favored by researchers. Considering the characteristics of large scale and high dynamics of low-earth orbit satellite network, that is, the number of satellite nodes and inter-node links in the satellite network is very large (thousands or even tens of thousands), and the link states (link on / off, link delay, link bandwidth) of these links change frequently over time, the simulation verification platform should have the ability to simulate large-scale and high-dynamic networks.

[0004] Currently, there are three technical routes for the satellite network simulation verification platform based on software implementation, namely pure physical model simulation implementation, physical model calculation event combined with discrete event software simulation implementation, and physical model calculation event combined with network function virtualization technology simulation implementation. Since the satellite routing equipment hardware needs to operate stably in the complex space environment, before the satellite routing equipment hardware enters space, it is necessary to test the functions and physical characteristics of the satellite routing equipment hardware in the running process, that is, connect the satellite routing equipment hardware to the satellite network simulation verification platform to complete the hardware-in-the-loop simulation verification. To complete the hardware-in-the-loop simulation verification, the satellite network simulation verification platform needs to have the function of data exchange with physical devices. Among the current implementations of satellite network simulation platforms, only the technical route of physical model calculation event combined with network function virtualization technology simulation implementation can meet this requirement, so this route has gradually become the mainstream implementation method of satellite network simulation verification platforms.

[0005] The existing work on the technical route of the simulation verification platform based on physical model calculation events combined with network function virtualization technology reduces the simulation overhead of satellite nodes by using lightweight virtualization technology, and to a certain extent alleviates the resource overhead caused by the large scale of satellite networks. However, in dealing with the problem of high dynamicity of satellite network links, there are three limitations in the dynamic link regulation scheme given by the above existing work. The first limitation is that the dynamic link regulation logic runs in the formal simulation stage. This means that when the satellite network simulation verification platform is in formal simulation, the system not only has to bear the resource overhead caused by the virtual satellite network simulation logic, but also needs to bear the additional resource overhead caused by the dynamic link regulation logic. And this overhead will consume more system resources as the time granularity is divided finer and there are more link state events. The second limitation is that the dynamic link regulation logic is real-time. This means that within the time of one time granularity, the link states of all links need to be updated, otherwise it will affect the simulation of the next time granularity. Each time the existing work uses the Linux tc tool to regulate the link, a new process needs to be created. Executing a large number of link regulation operations in a short time means creating a large number of new processes in a short time, which is undoubtedly very time-consuming. Therefore, it is impossible to handle a large number of link regulations when the time granularity is small. The third limitation is introduced because of the use of lightweight virtualization technology, that is, the problem that some logic is difficult to parallelize due to the global lock brought by sharing the operating system kernel. In particular, in the network simulation environment, the requests for link regulation and the requests for route distribution in the kernel share the same processing logic. Therefore, a large number of link regulation requests in a short time are not only difficult to parallelize, but also will occupy the processing of route distribution in normal simulation. Based on the above three limitations, the existing work cannot meet the simulation requirements of large-scale and high-dynamic satellite network links, and there is an urgent need for a more efficient link regulation method to meet the simulation requirements of large-scale and high-dynamic satellite networks. Summary of the Invention

[0006] Aiming at the deficiencies of the above existing technologies, the problem to be solved by the present disclosure is to provide a method for regulating the dynamic change of link states for a large-scale and high-dynamic satellite network simulation verification platform based on eBPF, which can solve the problems existing in the dynamic link state regulation scheme given by the existing work and meet the simulation requirements of large-scale and high-dynamic satellite networks.

[0007] In a first aspect, a method for dynamically regulating links for a large-scale satellite network simulation platform is provided. The satellite network simulation platform runs virtual satellite nodes with lightweight virtualization technology, and the virtual satellite nodes are connected by virtual network links provided by Veth devices, including:

[0008] Load the eBPF program into the veth egress mount point of each virtual satellite node;

[0009] The simulation platform converts the discrete event sequence of the link states of all links calculated according to the physical model at a certain event granularity into mapping information with the timestamp combined with the virtual network device index as the key and the link state information as the value, and pre-stores the mapping information and the simulation start time in the eBPF Map;

[0010] Based on the timestamp of the link state when the data packet is sent, look up the link state information of the current device at the current moment in the eBPF Map, and perform link regulation based on the link state information.

[0011] Among them, calculate the timestamp according to the system time and the simulation start time. Calculate the difference between the system time and the simulation start time, and the integer part of the difference divided by the simulation time granularity is the timestamp of the current link at the current simulation moment.

[0012] Among them, store the simulation start time of the device in the eBPF Map of each virtual satellite node.

[0013] Among them, when the link state information of the current device at the current moment is the link disconnection information, the eBPF program will directly return the value TC_ACT_SHOT, and this return value will cause the data packet to be discarded.

[0014] Among them, when the link state information of the current device at the current moment is the link delay information, the eBPF program adds the link delay to tstamp, and the data packet is sent according to the value of tstamp.

[0015] Among them, when the link state information of the current device at the current moment is the link packet loss rate information, the eBPF program generates a pseudo-random number. When this pseudo-random number falls into the interval that conforms to the link packet loss rate for packet loss, the eBPF program will return the value TC_ACT_SHOT, and this return value will cause the data packet to be discarded;

[0016] Among them, when the link state information of the current device at the current moment is the link bandwidth information, obtain the timestamp of the previous packet transmission from the eBPF Map with the device index as the key;

[0017] If the previous packet transmission timestamp does not exist, store the current packet transmission system timestamp in the eBPF Map with the device index as the key;

[0018] If the previous packet transmission timestamp exists, then obtain the timestamp of the previous packet transmission, and calculate the timestamp of the next packet transmission according to the timestamp, the link bandwidth, and the packet length;

[0019] Obtain the system timestamp. When the system timestamp is less than the timestamp of this packet transmission, add the waiting time to the tstamp of the packet structure, and store the system timestamp of this packet transmission in the eBPF Map with the device index as the key;

[0020] When the system time is greater than or equal to the timestamp of this packet transmission, store the system timestamp of this packet transmission in the eBPF Map with the device index as the key.

[0021] In a second aspect, a computer-readable storage medium is provided, on which a computer program / instructions are stored. When the computer program / instructions are executed by a processor, the steps of the above method are implemented.

[0022] In a third aspect, a computing terminal is provided, including a memory and a processor. An executable code is stored in the memory. When the processor executes the executable code, the above method is implemented.

[0023] Compared with the prior art, the present invention can obtain the following technical effects:

[0024] 1. Since all dynamic link regulation work is completed before the formal simulation, the system resource overhead during the formal simulation is reduced.

[0025] 2. Since the eBPF program calculates the timestamp according to the simulation time granularity during the formal simulation, as long as the link state information associated with the corresponding timestamp is stored in the eBPF Map, the link state change of the corresponding link can show the appearance depicted by the discrete link state events divided by the corresponding time granularity, and it can support large-scale dynamic link state regulation with finer time granularity.

[0026] 3. Since the execution of the eBPF program is embedded in the kernel execution flow and does not affect the satellite node's call to other kernel codes, it does not affect the simulation logic of the satellite node during the formal simulation. Description of the Drawings

[0027] Figure 1 The working process of the satellite network simulation verification platform implemented by the prior art;

[0028] Figure 2 The working process of the satellite network simulation verification platform using the link dynamic regulation method for the large-scale satellite network simulation platform;

[0029] Figure 3 The architecture after the combination of the link dynamic regulation method for the large-scale satellite network simulation platform and the satellite network simulation and verification platform;

[0030] Figure 4It is the overall operation process of the link dynamic regulation method for the large-scale satellite network simulation platform;

[0031] Figure 5 It is the execution process of the eBPF program;

[0032] Figure 6 It is the execution link disconnection regulation logic process of the eBPF program;

[0033] Figure 7 It is the execution link delay control regulation logic process of the eBPF program;

[0034] Figure 8 It is the execution link packet loss rate control regulation logic process of the eBPF program;

[0035] Figure 9 It is the execution link bandwidth control regulation logic process of the eBPF program; Specific implementation mode

[0036] The following describes the best implementation mode of the present invention through embodiments. It should be understood that the specific implementation mode here is used to explain the present invention in detail and should not be construed as a limitation to the present invention. It should be noted that various changes and modifications can be made on the premise of following the principles and core scope of the present invention, and these changes should all be regarded as falling within the protection scope of the present invention. Combining with the attached drawings, the specific implementation steps of the present invention are described in detail.

[0037] Since the satellite routing equipment hardware needs to operate stably in the complex space environment, before the satellite routing equipment hardware enters space, it is necessary to test the functions and physical characteristics of the satellite routing equipment hardware in the running process, that is, to connect the satellite routing equipment hardware to the satellite network simulation and verification platform to complete the hardware-in-the-loop simulation verification. To complete the hardware-in-the-loop simulation verification, the satellite network simulation verification platform needs to have the function of data exchange with the physical device. In the current implementation of the satellite network simulation platform, only the technical route of combining physical model calculation events with network function virtualization technology can meet this requirement. Therefore, this route has gradually become the mainstream implementation method of the satellite network simulation verification platform.

[0038] In the satellite network simulation verification platform of the prior art, the working process is generally as Figure 1 shown:

[0039] The satellite network simulation platform calculates the discrete event sequence of the link states of all links in the satellite network changing with time at a certain time granularity according to the physical model;

[0040] Build virtual satellite nodes based on lightweight isolation mechanisms (such as containers or namespaces);

[0041] Satellite nodes communicate using virtual network links constructed with virtual network devices (veth) of the Linux kernel;

[0042] Start all virtual satellite nodes implemented by software;

[0043] The simulation begins. The simulation platform regulates the links according to the dynamic link regulation logic and conducts the simulation according to the virtual satellite simulation logic.

[0044] In the dynamic link regulation given in the existing work, the dynamic link regulation logic runs in the formal simulation stage, consuming system resources. When the time granularity is small, it cannot handle large-scale link regulation. A large number of link regulation requests in a short period of time are not only difficult to parallelize, but also squeeze the processing of route distribution in normal simulation, unable to meet the link simulation requirements of large-scale and highly dynamic satellite networks.

[0045] To solve the above problems, the present invention provides a link dynamic regulation method for a large-scale satellite network simulation platform that combines physical model calculation events with network function virtualization technology for simulation and verification.

[0046] The eBPF (Extended Berkeley Packet Filter) program is a program that can run securely in the kernel of the Linux system with pre-compilation and sandboxing. It is triggered by events, and the execution of the eBPF program is triggered every time a data packet is sent. In particular, the eBPF map is a special data structure used to store and share data between the eBPF program and the user space.

[0047] As Figure 4 and Figure 5 shown, the link dynamic regulation method for a large-scale satellite network simulation platform provided by the present invention is as follows:

[0048] Step S1: Load the eBPF program to the veth egress mount point of each virtual satellite node;

[0049] Step S2: The simulation platform converts the discrete event sequence of the link states of all links calculated according to the physical model at a certain event granularity into mapping information with the timestamp combined with the virtual network device index as the key and the link state information as the value, and pre-stores the mapping information and the device simulation start time in the eBPF Map;

[0050] Among them, the specific implementation manner of step S2 is:

[0051] The simulation platform reads the discrete event sequence of the link state and converts it into link state information with timestamps and device indexes;

[0052] Among them, the link state information mainly includes: link on / off information, link delay information, link packet loss rate information, and link bandwidth information.

[0053] Load the link state information with timestamps and device indexes into the eBPF Map;

[0054] Load the simulation start time into the eBPF Map for the eBPF program to calculate the timestamp of the current simulation moment of the link;

[0055] Step S3: Based on the timestamp of the link state when the data packet is sent, look up the link state information of the current device at the current moment in the eBPF Map, and perform link regulation based on the link state information.

[0056] Specifically, before the simulation starts, the eBPF program will discard all data packets. After the simulation platform stores the simulation start time in the eBPF Map, the eBPF program will start to work.

[0057] The specific program execution logic is shown in the figure:

[0058] Look up the simulation start time of the current device from the eBPF Map with the device index as the key;

[0059] Judge whether the simulation has started. If not, continue to look up the simulation start time of the current device from the eBPF Map with the device index as the key;

[0060] If so, calculate the timestamp based on the system time and the simulation start time, and look up the link state information of the current device at the current moment from the eBPF Map with the timestamp and the device index as the keys;

[0061] Execute the eBPF-based link regulation logic according to the link state information.

[0062] The eBPF program ends the process.

[0063] Specifically, calculate the difference between the system time and the simulation start time. The integer part of the difference divided by the simulation time granularity is the timestamp of the current simulation moment of the current link. The link state information mainly includes: link on / off information, link delay information, link packet loss rate information, and link bandwidth information.

[0064] It should be noted that every time the simulation platform sends a data packet to the eBPF program, it will trigger the operation of the eBPF program.

[0065] Executing the eBPF-based link regulation logic according to the link state information includes:

[0066] If the obtained link state time information is link on / off information, such as Figure 6As shown, the eBPF program executes the following control logic:

[0067] If the link disconnection information is not obtained, it means the link remains connected, and the next link control logic is executed;

[0068] If the link disconnection information is obtained, the eBPF program will directly return the value TC_ACT_SHOT, and this return value will cause the data packet to be discarded.

[0069] If the obtained link status time information is link delay information, as Figure 7 shown, the eBPF program executes the following control logic:

[0070] The eBPF program will modify the time information tstamp field of the data packet structure sk_buff.

[0071] Among them, the sk_buff (socket buffer) structure is an important data structure in the linux network code, which manages and controls the information of receiving or sending data packets.

[0072] Specifically, the link delay is added to the time information tstamp. After modification, the data packet will be sent according to the value of the time information tstamp, so as to achieve the effect of link delay control.

[0073] If the obtained link status time information is link packet loss rate information, as Figure 8 shown, the eBPF program executes the following control logic:

[0074] After obtaining the link packet loss rate, a pseudo-random number is generated to determine whether to drop the packet.

[0075] When this pseudo-random number falls into the interval where the packet loss conforms to the link packet loss rate, the eBPF program will return the value TC_ACT_SHOT, and this return value will cause the data packet to be discarded.

[0076] If it is not in this interval, the next link control logic is continued.

[0077] If the obtained link status time information is link bandwidth information, as Figure 9 shown, the eBPF program executes the following control logic:

[0078] When the link status information is link bandwidth control, the timestamp of the previous packet transmission is obtained from the eBPF Map with the device index as the key.

[0079] If the previous packet transmission timestamp does not exist, it means that the packet transmission has not started yet. The system timestamp of this packet transmission can be stored in the eBPF Map with the device index as the key and enter the next stage of link control logic.

[0080] If the timestamp of the previous packet transmission exists, obtain the timestamp of the previous packet transmission. Based on this timestamp, the link bandwidth, and the packet length, calculate the timestamp for the next packet transmission, i.e., the timestamp for the next packet transmission is equal to the timestamp of the previous packet transmission plus the packet length divided by the link bandwidth.

[0081] After calculating the timestamp for the next packet transmission, obtain the system timestamp and compare the system timestamp with the timestamp for the next packet transmission.

[0082] When the system timestamp is less than the timestamp for the current packet transmission, this means it is not yet time to send the packet. The waiting time can be added to the tstamp of the packet structure, and then the system timestamp of the current packet transmission is stored in the eBPF Map with the device index as the key, and enter the link control logic of the next stage.

[0083] When the system time is greater than or equal to the timestamp for the current packet transmission, it means that the packet can be sent. The system timestamp of the current packet transmission is directly stored in the eBPF Map with the device index as the key, and enter the link control logic of the next stage.

[0084] Through the present invention, the simulation process in the satellite network simulation platform can be improved, such as Figure 3 The architecture after the combination of the present invention and the satellite network simulation and verification platform is shown as follows.

[0085] The satellite network simulation platform calculates a discrete event sequence of the link states of all links in the satellite network changing over time at a certain time granularity according to the physical model, and pre-stores it in the eBPF Map. The virtual satellite nodes are connected and communicate through the virtual links provided by the veth devices. Each virtual satellite node is equipped with an eBPF program that can query the link state information in the eBPF Map.

[0086] Figure 2 The simulation process of the satellite network simulation platform improved based on the present invention is shown in the figure as an embodiment. The process includes:

[0087] S1: The satellite network simulation platform calculates a discrete event sequence of the link states of all links in the satellite network changing over time at a certain time granularity according to the physical model;

[0088] S2: Construct virtual satellite nodes based on the lightweight isolation mechanism;

[0089] In a specific embodiment, the lightweight isolation mechanism can be the container technology Docker or the namespace mechanism.

[0090] Container technology is an isolation technology based on the capabilities of the operating system and is an important solution for resource sharing and isolation in large-scale clusters. The essence of a container is a set of processes that are resource-limited and isolated from each other, and one or more applications can run in the processes.

[0091] Namespaces are a mechanism in the Linux kernel that allows multiple independent processes to run on the same system, and each process has its own independent resources and namespaces, which can isolate and protect different processes.

[0092] In the satellite network simulation platform, the lightweight isolation mechanism has the advantages of more efficient use of system resources, support for continuous payment and deployment, and convenient maintenance.

[0093] S3: Satellite nodes communicate with each other using virtual network links constructed by the virtual network devices (veth) of the Linux kernel;

[0094] Veth is a type of Linux virtual network device that exists in pairs. Each pair of veth devices consists of two endpoints. Data sent from one endpoint can be received by the other endpoint, enabling communication between two virtual satellite nodes.

[0095] S4: Run the dynamic link regulation logic.

[0096] In a specific embodiment, as Figure 4 shown, running the dynamic link regulation logic includes:

[0097] S41: Load the eBPF program into the veth egress mount point of each virtual satellite node.

[0098] A mount point is the entry directory of the disk file system in the Linux system. Through the mount point, users can easily access and manage the data on the storage device, such as operations like reading, writing, and deleting files.

[0099] In this embodiment, the link status information can be queried through the mount point of the virtual satellite node.

[0100] S42: Read the discrete event sequence of the link status of all links calculated by the simulation platform according to the physical model at a certain event granularity, and convert it into mapping information with the timestamp and device index as keys and the link status information as values.

[0101] Among them, the link status information mainly includes: link on / off information, link delay information, link packet loss rate information, and link bandwidth information.

[0102] S43: Store the above mapping information into the eBPF Map.

[0103] S44. Store the simulation start time in the eBPF Map. The eBPF program starts to work and calculates the timestamp.

[0104] Before the simulation starts, the eBPF program discards all data packets. After the simulation platform stores the simulation start time in the eBPF Map, the eBPF program starts to work and performs dynamic link regulation according to the execution logic of the eBPF program of the present invention.

[0105] Among them, the calculation method of the timestamp is to calculate the difference between the system time and the simulation start time. The integer part of the difference divided by the simulation time granularity is the timestamp of the current link at the current simulation moment.

[0106] S45. The link regulation ends and the simulation starts.

[0107] S5: Start all virtual satellite nodes implemented by software;

[0108] S6: The simulation starts, and the simulation platform performs the simulation following the virtual satellite simulation logic.

[0109] The difference between the satellite network simulation verification platform implemented based on this solution and the satellite network simulation verification platform implemented by existing work is that the work of the dynamic link regulation logic part is placed before the formal simulation. When the eBPF program controls the link bandwidth and delay according to the EDT traffic shaping mechanism, each execution needs to read the link delay and bandwidth configuration from a storage area called eBPF Map that can be read and written by the eBPF program. The link loaded with the eBPF program will naturally show dynamic changes in the link state that conform to the physical model calculation during the simulation.

[0110] In the process based on the present invention, the virtual links built by the satellite network simulation verification platform have the characteristics of dynamic changes in the link state that conform to the physical model, and there is no need to perform real-time link state regulation during the formal simulation.

[0111] The specific embodiments described above further elaborate on the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A link dynamic control method for a large-scale satellite network simulation platform, wherein the satellite network simulation platform runs virtual satellite nodes using lightweight virtualization technology, and the virtual satellite nodes are connected through virtual network links provided by Veth devices, characterized in that: include: Load the eBPF program to the veth egress mount point of each virtual satellite node; The simulation platform converts the discrete event sequence of link status of all links calculated according to the physical model at a certain event granularity into mapping information with timestamp combined with virtual network device index as key and link status information as value, and pre-stores the mapping information and simulation start time in the eBPF Map; Based on the timestamp of the link state when the data packet is sent, the link state information of the current virtual satellite device at the current moment is searched in the eBPF Map, and link regulation is performed based on the link state information.

2. The link dynamic control method according to claim 1, characterized in that: Calculate the timestamp based on the system time and the simulation start time: The difference between the system time and the simulation start time is calculated. The integer part of the difference divided by the simulation time granularity is the timestamp of the current simulation moment of the current link.

3. The link dynamic control method according to claim 1, characterized in that: The simulation start time of each virtual satellite device is stored in the eBPFMap of the virtual satellite node.

4. The link dynamic control method according to claim 1, characterized in that: When the link status information of the current virtual satellite device is link disconnection information at the current moment, the eBPF program will directly return the value TC_ACT_SHOT, which will cause the data packet to be discarded.

5. The link dynamic control method according to claim 1, characterized in that: When the link status information of the current virtual satellite device at the current moment is the link delay information, the eBPF program adds the link delay to tstamp, and the data packet is sent according to the value of tstamp.

6. The link dynamic control method according to claim 1, characterized in that: When the link status information of the current virtual satellite device is the link packet loss rate information, the eBPF program generates a pseudo-random number. When the pseudo-random number falls into the interval that meets the link packet loss rate, the eBPF program returns the value TC_ACT_SHOT, which causes the data packet to be discarded.

7. The link dynamic control method according to claim 1, characterized in that: When the link status information of the current virtual satellite device is link bandwidth information, the timestamp of the last packet transmission is obtained from the eBPF Map using the device index as the key; If the timestamp of the last packet does not exist, the system timestamp of this packet is stored in the eBPF Map with the device index as the key; If the timestamp of the last packet transmission exists, obtain the timestamp of the last packet transmission, and calculate the timestamp of the next packet transmission based on the timestamp, link bandwidth, and data packet length; Get the system timestamp. When the system timestamp is less than the timestamp of this packet, add the waiting time to the tstamp of the data packet structure, and store the system timestamp of this packet in the eBPF Map with the device index as the key. When the system time is greater than or equal to the timestamp of this packet, the system timestamp of this packet is stored in the eBPF Map with the device index as the key.

8. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to claims 1-7 are implemented.

9. A computing terminal, comprising a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method according to any one of claims 1 to 7 is implemented.

Images