CN108365886B - A time division multiplexing method for data storage and forwarding in inter-satellite networks - Google Patents

Abstract

The invention provides a time division multiplexing inter-satellite network data storing and forwarding method which can meet the storing and forwarding real-time performance of massive multi-type and multi-address data under the condition of limited satellite storage resources and processing capacity, improve the searching efficiency and reduce the storage occupation. The method comprises the following steps: establishing a network datagram entity data pool, and establishing a virtual memory dynamic preemption data structure according to the entity data pool to form a virtual container; acquiring idle virtual element nodes from the entity data pool, filling inter-satellite data in a storage unit pointed by the virtual element nodes, and injecting the virtual element nodes into the virtual container; performing attribute encapsulation on the virtual container, acquiring flow information of the spatial information network node according to the overflow frequency during encapsulation, and adjusting the capacity of the virtual container through the flow information; and performing mapping search according to the effective element nodes in the virtual container, and forwarding according to the data priority.

Description

A time division multiplexing method for data storage and forwarding in inter-satellite networks

technical field

The invention relates to a time-division multiplexing inter-satellite network data storage and forwarding method, which realizes optimal execution and storage efficiency under the condition of limited processing and storage resources of spacecraft.

Background technique

In an inter-satellite network based on time division links, the links between satellites and between satellites and between satellites and ground stations are intermittently available, and the service data in the network supports store-and-forward when routing between ground nodes and satellite nodes. mechanism.

As a time-division multiplexing spatial network with typical spatial delay tolerant networking (DTN) properties, its link delay will be very large, and some links will be available intermittently. Generally, the round-robin link building rule is adopted. With the interconnection of different aerospace systems, different orbits and types of aircraft and application systems in the inter-satellite link and the satellite-to-ground link form a space-ground integrated space information network based on the principle of maximum and effective utilization of space information resources. Time division multiplexing network The types of information transmitted in the aircraft will not be limited to business information in the field of their own aircraft. When the inter-satellite data frames with multiple destination addresses and multiple data types arrive at the satellite node for routing, the satellite node needs to provide a large amount of storage to adapt to the burst traffic, and provide a high processing capacity to adapt to the search in the massive data pool. Inter-satellite frames for forwarding conditions.

There are two traditional store-and-forward strategies. The first strategy is to open a custom cache for each service data type of each destination address, and customize the storage area size based on the average bandwidth or maximum bandwidth of the service data type. However, when the type of business data increases, the amount of storage occupied will increase exponentially, and due to the randomness of network traffic, data types with small business volume will generate a large amount of invalid storage, and data types with large business volume will cause buffer overflow. The second strategy is to store the inter-satellite data frames in the storage area in the order of arrival, and use pointers to traverse and search during routing. However, when the amount of data in the storage area is large, the processor power is required to be very high.

Due to the extremely harsh space environment in which satellite systems are located, satellites generally provide limited platform carrying capacity. Under the conditions of network congestion or limited links, the storage resources and processing capabilities of satellite nodes greatly restrict the execution of aerospace integration missions. In the process of integrating the space-based network and the ground-based network, how to fully exploit the satellite resources and optimize the storage and forwarding efficiency of massive data in the inter-satellite network has become an urgent problem to be solved.

SUMMARY OF THE INVENTION

The invention provides a time-division multiplexing inter-satellite network data storage and forwarding method, which can meet the real-time nature of storage and forwarding of massive multi-type and multi-address data under the condition of limited satellite storage resources and processing capabilities, improve search efficiency, reduce Storage occupancy reduces the probability of inter-satellite frame loss caused by network congestion.

A time-division multiplexing inter-satellite network data storage and forwarding method, comprising:

establishing a network datagram entity data pool, and establishing a virtual memory dynamic preemption data structure according to the entity data pool to form a virtual container;

Obtain an idle virtual element node from the entity data pool, fill the storage unit pointed to by the virtual element node with inter-satellite data, and inject the virtual element node into the virtual container;

Attribute encapsulation is performed on the virtual container, and according to the frequency of overflow during encapsulation, the flow information of the spatial information network node is obtained, and the capacity of the virtual container is adjusted by the flow information;

Map search is performed according to valid element nodes in the virtual container, and forwarding is performed according to data priority.

Further, when establishing the network datagram entity data pool, the size of the data pool is customized according to the actual capability of the onboard computer.

Further, when the virtual container is initialized, the queue linked list of the specified destination address of the specified service type is empty by default, only contains the head pointer and the tail pointer, and occupies a very small memory.

Further, when performing attribute encapsulation on the virtual container, the upper limit of the virtual container with different service type attributes is set according to the service flow characteristics of the time-division multiplexing network.

Further, the upper limit of the virtual container is controlled by sending instructions on the ground.

Further, the virtual element node is injected into the virtual container to identify the maximum capacity of the data pool and the upper limit of the virtual container. When the upper limit is reached, the data packet is discarded, and an alarm signal is designed to publish the network status to the ground control center; If the upper limit is not reached, but the virtual element nodes of the data pool are exhausted, the data packet will be discarded and the ground control center will be notified.

Further, when forwarding according to the data priority, firstly route the high-priority service datagram, and use a dynamic or static mechanism to determine the destination address of the inter-satellite frame that this star allows to forward at the current moment in the time-division multiplexing network, With the service type and destination address as the index, it is mapped to the corresponding virtual container valid data queue linked list.

Beneficial effects of the present invention:

1. In the time-division multiplexing inter-satellite network node with limited storage resources, the size of the service data container is flexibly determined according to the traffic characteristics of different services in the network, which is conducive to further network optimization in the routing protocol, reducing network congestion and reducing satellite traffic. Occupation of computer storage resources.

2. Convert the operation of the storage entity into the operation of the virtual node. During the entire processing process of the inter-satellite business data in the storage and forwarding node in the time division multiplexing network, the only storage unit of the only element node in the data pool is always occupied. There is copy behavior, which reduces the occupation of satellite computer processor resources and improves the efficiency of network execution.

3. Zero-delay network data forwarding search feature In the same service container, virtual node queues are allocated according to destination addresses. In the forwarding process, according to the destination address output by the routing protocol and the service type determined by priority scheduling, the queue information to be sent can be obtained with zero delay through the mapping operation, which reduces the processing delay of the massive network data of the network nodes and improves the real-time response of the network. sturdiness and reliability.

Description of drawings

Fig. 1 is a schematic diagram of data storage and forwarding;

Fig. 2 is a schematic diagram of service flow adaptation;

Figure 3 is a schematic diagram of data pool initialization;

Fig. 4 is a schematic diagram of service queue initialization;

Fig. 5 is the remote control service queue insertion process;

Fig. 6 is the judgment flow chart of whether there are idle nodes in the data pool;

Fig. 7 is the service queue search process;

Detailed ways

A time division multiplexing inter-satellite network data storage and forwarding method of the present invention includes the following steps:

Step 1: Establish a network datagram entity data pool: In the network layer of the space information network integrating space and earth, a fixed-capacity storage space is opened up for network datagrams. Spacecraft nodes can be divided into backbone network nodes or access network nodes according to their roles in the network, and their storage system architectures and the storage capabilities provided by the adopted memory chips are different. When establishing a data pool, customize the size of the data pool according to the actual capability of the onboard computer. The granularity of the storage unit in the data pool is a delimited network data packet, and is compatible with various network data packet formats. Use linked lists or other data structures to establish virtual mappings for the physical entities of storage units in the data pool to achieve resource management. During initialization, each element in the linked list points to each storage unit in the data pool, and is responsible for maintaining the allocation and recovery mechanism of all storage units.

Step 2, establishing a virtual memory dynamic preemption data structure according to the network datagram entity data pool obtained in step 1: establishing a valid datagram queue linked list for each network address of each service type in the space information network in the spacecraft node, Form a virtual container. When the virtual container is initialized, the queue linked list of the specified destination address of the specified business type is empty by default, only contains the head pointer and the tail pointer, and occupies a very small amount of memory, that is, the resource consumption is extremely low when the virtual container is vacant. When the network layer of the spacecraft node receives the routing relay data message sent by other network nodes, or the spacecraft generates the network interaction information data message by itself, it obtains the idle virtual element node from the data pool, and the element node points to the idle virtual element node. The storage unit entity is filled with inter-satellite data, and the virtual element node is injected into the virtual container by the method of linked list insertion.

Step 3: Perform attribute encapsulation on the virtual container obtained in Step 2. According to the service traffic characteristics of the time-division multiplexing network, the upper limit of virtual containers with different service type attributes is set. The upper limit of the virtual container can be controlled by sending commands from the ground. The process of injecting virtual elements into business virtual containers identifies the maximum capacity of the data pool and the upper limit of virtual containers. When the virtual container reaches the upper limit, the data packet is discarded, and an alarm signal is designed to publish the network status to the subscribing nodes, such as the ground control center; if the virtual container does not reach the upper limit, but the virtual element nodes of the data pool are exhausted, then The network data packets are still discarded and the subscribing nodes are notified.

Step 4: Adjust the capacity of the virtual container according to the frequency of overflow of the virtual container in Step 3, and obtain the traffic experience information of the spatial information network node. If the virtual container of business A continues to be idle, the capacity of the virtual container of the virtual container is reduced, and if the virtual container of business B continues to overflow, the capacity of the virtual container is increased. This process can be intervened based on human judgment on the ground. Design a load balancing mechanism. If a spacecraft node in the space network continues to have no idle virtual elements available, it proves that the storage capacity of the satellite node can no longer meet the most basic traffic requirements, and needs to be considered in network topology chain construction planning and routing planning , the user can use other access satellite nodes to post information and optimize the transmission path.

Step 5: Obtain valid element nodes in the virtual container according to Steps 3 and 4, and perform a fast mapping search. When routing network data packets at the network layer, data packets are forwarded according to the data priority determined by service characteristics. First, route high-priority service datagrams, and use dynamic or static mechanisms to determine the destination address of the inter-satellite frames that the local satellite is allowed to forward in the time-division multiplexing network at the current moment. With the service type and destination address as the index, it can be quickly mapped to the corresponding virtual container valid data queue linked list.

For example, if the number of service types is set to M, the number of destination addresses of network data packets allowed to be forwarded by this satellite is X, and the number of virtual element nodes in the data pool is L, the maximum number of searches is M×X. However, if the traditional sequential storage strategy is adopted, the number of searches required is M×X×L, which results in a much larger processing and delay cost than this method.

After obtaining a network packet data element node from the virtual container valid data queue list, encapsulate or modify the link layer protocol frame header on the physical storage unit mapped by the virtual element node, and insert the element node into the network data packet The output queue list is waiting to be sent to the inter-satellite or ground access station through the physical layer. The network routing function extracts virtual element nodes from the virtual container effective data queue list according to the priority and inserts them into the output queue list according to the first-in, first-out rule, until the bandwidth of the time-division multiplexing network is fully occupied.

After the complete data packets in the physical storage units mapped by the virtual element nodes in the output queue list are written to the physical layer in sequence, each virtual element node in the output queue list is recycled to the idle resource manager of the data pool, waiting for the next A use of inter-satellite service data entering the network layer. From the network service data packet entering the network layer, being stored in the data pool and managed by the virtual container, encapsulating the link protocol, inserting it into the network data packet output queue, and writing the full life cycle of the physical layer, the network service data packet The document always occupies the only storage unit mapped by the unique virtual element node in the data pool, and there is no copying behavior, which greatly improves the processing efficiency of the onboard computer.

When the datagram search is performed according to the optimal routing plan, the qualified forwarding frame can be directly found according to the index formed by the service type and the destination address, which reduces the search time. It realizes the real-time storage and forwarding of massive multi-type and multi-address data, improves search efficiency, reduces storage occupation, and reduces the probability of inter-satellite frame loss caused by network congestion according to the dynamic adjustment of virtual containers and the status feedback of routing planning. .

The present invention will be further introduced below in conjunction with the accompanying drawings.

The present invention provides the storage and forwarding mechanism of the time division multiplexing network based on dynamic allocation and service address index, realizes the network characteristic according to the priority, the processing characteristic of zero copy, the search characteristic of zero delay, including the establishment of data pool and service queue. , service queue management, service search and forwarding process.

(1) Establishment of data pools and business queues

Each satellite node in the space network opens up a data pool according to the traffic it carries. The data pool can be described by two aspects: storage entities and virtual nodes.

The storage entity is a memory block with continuous addresses, and the organization of fixed-length frames in the storage area is implemented in the form of an array. Each virtual node is an element in the doubly linked list, and the attribute information contained in the element is the position in the storage entity corresponding to the virtual node, which is implemented in the form of a pointer. When the data pool is established, each virtual node in the data pool is inserted into the free queue linked list, as shown in Figure 3.

After the data pool is established, virtual containers need to be established for each business. To facilitate the search process, a linked list of valid data queues is established for each address of each service. Each valid data queue linked list is empty in the initial state, and each queue maintains the length information of its own queue, and each service type also maintains the saturation state of its own container. As shown in Figure 4.

(2) Business queue management

When the satellite application layer generates data that needs to be sent to the inter-satellite and writes it to the network layer, the following operations are performed.

(1) Obtain a virtual node at the end of the queue from the doubly linked list of the idle queue.

(2) Obtain the address pointer of the storage entity unit to which the virtual node is attached from the attribute information of the virtual node.

(3) Fill the data sent to the inter-satellite into the data field of the fixed-length frame of the storage entity unit.

(4) Determine the type and destination address of the data service sent to the inter-satellite.

(5) Insert the virtual node into the queue linked list corresponding to the address corresponding to the service.

Taking the remote control service as an example, the insertion process of the service queue is shown in FIG. 5 .

When the information that needs to be forwarded to other satellites is received from the inter-satellite, and enters the network layer for routing, the operation modification in step 3) is performed to directly fill the fixed-length inter-satellite frame into the storage entity unit.

In the process of obtaining virtual nodes from the data pool and inserting them into the valid data service queue, it is necessary to determine whether the size of the remote control service container exceeds the limit and whether there are idle nodes in the data pool.

If the inter-satellite network is congested due to an inter-satellite network burst, or the satellite node in the space network is damaged, the inter-satellite frame of a destination address cannot be routed out, which will permanently occupy the storage space of the data pool. If the data exchanged periodically between satellites does not reach the target after the exchange period, it will affect the normal information exchange in the next cycle and needs to be cleared in the network. Therefore, a queue clearing mechanism based on services and addresses is added to the queue management. The service queue can be cleared by sending commands on the ground or by periodically and autonomously initiated by the satellite.

(3) Service search and forwarding

The satellite node searches and forwards the frames in the service queue according to the routing table determined by the routing protocol. The search and forwarding steps are as follows:

(1) Determine the destination address of the data frame that the current time slot is allowed to send to the inter-satellite according to the routing table.

(2) Select the current scheduling priority according to the service priority.

(3) Directly locate the valid data queue according to the service type and destination address.

(4) Check whether the valid data queue is empty, if not, take out the first virtual node according to FIFO.

(5) Obtain the storage unit corresponding to the virtual node, and if the storage unit only contains the data field, encapsulate the frame header and the frame tail of the inter-satellite frame in the storage unit.

(6) Insert the virtual node into the sending queue, and execute the above steps cyclically until the bandwidth of the time slot is full.

(7) When the inter-satellite data frame is sent to the inter-satellite through the physical channel, the virtual node is recycled to the idle queue.

Claims (8)

1. a time division multiplexing inter-satellite network data storage and forwarding method, is characterized in that, comprising: establishing a network datagram entity data pool, and establishing a virtual memory dynamic preemption data structure according to the entity data pool to form a virtual container; Obtain an idle virtual element node from the entity data pool, fill the storage unit pointed to by the virtual element node with inter-satellite data, and inject the virtual element node into the virtual container; Attribute encapsulation is performed on the virtual container, and according to the frequency of overflow during encapsulation, the flow information of the spatial information network node is obtained, and the capacity of the virtual container is adjusted by the flow information; Map search is performed according to valid element nodes in the virtual container, and forwarding is performed according to data priority. 2 . The time division multiplexing inter-satellite network data storage and forwarding method according to claim 1 , wherein the size of the data pool is customized according to the actual capability of the onboard computer when establishing the network datagram entity data pool. 3 . 3. a kind of time-division multiplexing inter-satellite network data storage and forwarding method as claimed in claim 2, is characterized in that, when the virtual container is initialized, the queue linked list of the specified destination address of the specified business type is empty by default, and only comprises the head pointer and tail pointer. 4. A time division multiplexing inter-satellite network data storage and forwarding method as claimed in claim 1, 2 or 3, characterized in that, when performing attribute encapsulation on the virtual container, according to the service flow characteristics of the time division multiplexing network, set the method. Set the upper limit of virtual containers for different business type attributes. 5 . The method for storing and forwarding data in a time-division multiplexed inter-satellite network according to claim 4 , wherein the upper limit of the virtual container is controlled by sending instructions on the ground. 6 . 6. a kind of time division multiplexing inter-satellite network data storage and forwarding method as claimed in claim 5 is characterized in that, virtual element node is injected into virtual container process, identify data pool maximum capacity and virtual container upper limit, when reaching upper limit , the data packet is discarded, and an alarm signal is designed to publish the network status to the ground control center; if the upper limit is not reached, but the virtual element nodes of the data pool are exhausted, the data packet is discarded and the ground control center is notified. 7. A kind of time division multiplexing inter-satellite network data storage and forwarding method as claimed in claim 1 or 2 or 3, it is characterized in that, when the described forwarding is carried out according to data priority, firstly route for high-priority service datagram , using a dynamic or static mechanism to determine the destination address of the inter-satellite frame that the local satellite is allowed to forward in the time division multiplexing network at the current moment, using the service type and destination address as the index, and mapping it to the corresponding virtual container valid data queue linked list. 8. a kind of time division multiplexing inter-satellite network data storage and forwarding method as claimed in claim 1 or 2 or 3, it is characterized in that, the described forwarding according to data priority specifically adopts the following mode: (1) Determine the destination address of the data frame that the current time slot is allowed to send to the inter-satellite according to the routing table; (2) Select the current scheduling priority according to the business priority; (3) Directly locate the valid data queue according to the business type and destination address; (4) Check whether the valid data queue is empty, if not, take out the first virtual node according to FIFO; (5) obtaining the storage unit corresponding to the virtual node, if the storage unit only contains the data domain, then encapsulate the frame header and the frame tail of the inter-satellite frame in the storage unit; (6) Insert the virtual node into the sending queue, and perform the above steps cyclically until the time slot bandwidth is full; (7) When the inter-satellite data frame is sent to the inter-satellite through the physical channel, the virtual node is recycled to the idle queue.

Images