CN113938176A - Low-delay service space-based computing method - Google Patents

Abstract

A low-delay service space-based computing method obtains an arrangement result of task computing by acquiring network state information, combining task attributes and utilizing matching of tasks and computing nodes, and then distributes and computes subtasks and converges results. The method can improve the use efficiency of satellite communication and computing resources, reduce the service transmission processing time delay and meet the service quality requirement of low-time-delay service.

Description

Low-delay service space-based computing method

Technical Field

The invention relates to a low-delay service space-based computing method, and belongs to the technical field of space-based computing.

Background

The low-orbit networking constellation is rapidly developed, and the demand for completing military and civil services through the low-orbit constellation is continuously increased, wherein the low-orbit networking constellation comprises satellite low-delay processing transmission services, such as military applications of rapid target extraction, rapid environment sensing and the like and emergency rescue applications of search and rescue and the like. Meanwhile, due to the difference of regions and time zones, constellation operation and earth rotation, the bearing distribution of the satellite communication network is unbalanced, and the traffic load has a time-varying characteristic. A single star may cover both sparsely populated areas, such as polar regions and the ocean, and densely populated and data-intensive areas, such as major cities in developed countries. The use of satellite computing and communication resources is greatly unbalanced, part of the satellite runs under high load, and part of the satellite computing and communication resources are idle.

With the development of the satellite in-orbit processing technology, a certain degree of computing power can be provided, however, due to the limitations of size and power consumption, the space-based computing platform has the problems of limited resources such as weak computing power and small storage space, and a single satellite node is difficult to meet the computing requirement of low time delay, so how to realize the processing and transmission of low time delay services through the cooperation of a plurality of satellite nodes based on the limited space-based computing resources becomes a problem which needs to be solved urgently.

The existing space-based routing algorithm is mainly based on data communication, does not consider data content in the data transmission process, does not perform service processing on data, does not fully utilize computing resources of a space-based computing platform, and is large in data transmission quantity and long in transmission time.

Disclosure of Invention

The technical problem to be solved by the invention is as follows: the method overcomes the defects of the prior art, provides a low-delay service space-based computing method, solves the problem of long service computing time of the traditional method, and meets the requirements of low-delay service.

The technical solution of the invention is as follows:

a low-delay service space-based computing method comprises the following steps:

1) satellite S with task source node_sThe task to be transmitted and processed is a task T (D, R), wherein D is input data to be processed by the task T, and R is output data processed by the task T; decomposing the T into a plurality of parallel subtasks, and completing the task T when each subtask of the T is completed;

2) each satellite in the low-orbit network periodically reports the network state of the satellite to the high-orbit satellite, and the high-orbit satellite periodically broadcasts the network state of the low-orbit satellite;

each satellite of the low-orbit network has 4 inter-satellite links of 2 front and back in the orbit and 2 left and right between the orbits, and the ith satellite S_iThe reported content comprises an available bandwidth set B of four intersatellite links_iMbps and currently available computing resources

Mflops；B_i＝{B_i,1,B_i,2,B_i,3,B_i,4In which B is_i,1For the ith satellite S_iAvailable bandwidth of the first inter-satellite link, B_i,2For the ith satellite S_iAvailable bandwidth of the second inter-satellite link, B_i,3For the ith satellite S_iAvailable bandwidth of the third inter-satellite link, B_i,4For the ith satellite S_iThe available bandwidth of the fourth inter-satellite link;

3) determining a candidate transmission path node set according to the low earth orbit satellite network state, sequencing subtasks according to the subtasks and subtask attributes decomposed in the step 1), then arranging the subtasks on the candidate transmission path node set according to the sequencing, and mapping the subtasks to corresponding satellite nodes;

4) according to the arrangement result, sending input data to be processed of the subtasks to corresponding satellite nodes, loading a processing program by the satellite nodes according to the calculation requirement of the subtasks, processing the received subtask data and sending the processing result to a target node, wherein the processing result is set as a high-priority service and is sent through a reserved bandwidth;

5) and feeding back each subtask to the task source node after the subtask is correctly executed, and for the task which fails to be executed, redistributing the transmission calculation process by the task source node until the calculation result is correct, and converging the subtasks at the target node after all subtasks are correctly executed to complete the transmission calculation of the whole task.

In the step 3), the subtasks are sequenced according to the following method:

let the j-th subtask have a ranking factor of λ_j，

Take lambda_j＝size(D_j)/size(D_max)-c_{task_j}/c_maxWherein D is_jInput data to be processed for the jth subtask, size (D)_j) Size of input data to be processed for jth sub-task, D_maxFor the input data to be processed corresponding to the subtask with the largest size of the input data to be processed in the subtask set, c_{task_j}The amount of computation required to process the jth sub-task, c_maxIs the calculated amount, lambda, of the subtask in the subtask set with the largest required calculated amount_jA task ordering factor;

each subtask is ordered by an ordering factor from large to small.

In step 3), the method for determining the candidate transmission path node set includes:

abstracting low-orbit satellite network into 2-dimensional mesh network, and slave task source node S_sTo the target node S_dAll nodes contained in the path set with the minimum hop count form a candidate transmission path node set, and the total number of the nodes in the set is L.

Set task source node S_sTo the target node S_dSet of candidate transmission path nodes is S ═ S_iAbstracting a low-orbit satellite network into a 2-dimensional mesh network, and in the 2-dimensional mesh network topology, marking links between orbits as an x-axis direction and links in the orbits as a y-axis direction; for a certain communication, a task source node S is set_sThe coordinate is (x)_s,y_s) The target node S_dThe coordinate is (x)_d,y_d)；

If the constellation is a constellation without a reverse seam, the task source node S_sTarget node S_dThe difference between the x-direction communication logic distance and the y-direction communication logic distance is as follows:

if the constellation is a constellation with reverse gaps, the task source node S_sTarget node S_dThe difference between the x-direction communication logic distance and the y-direction communication logic distance is as follows:

wherein N is the number of orbital planes, and M is the number of satellites on one orbital plane;

then the slave S_sTo S_dAt most share

The hop distance of the path with the least number of hops is delta x + delta y; the set of nodes on these paths is the candidate transmission path node set, and the path set with the least number of hops forms the candidate transmission path set.

If the constellation is a constellation without a reverse seam, the path in the x-axis direction in the path with the least hop number is as follows:

(a) if | x_d-x_s|≤N-|x_d-x_sIf the path along the x-axis is:

when x is_s>x_dWhen x_s→x_s-1→x_s-2→…→x_d；

When x is_s≤x_dWhen x_s→x_s+1→x_s+2→…→x_d；

(b) If | x_d-x_s|>N-|x_d-x_sIf the path along the x-axis is:

when x is_s≤x_dWhen x_s→x_s-1→x_s-2→…→1→N→…→x_d；

When x is_s>x_dWhen x_s→x_s+1→x_s+2→…→N→1→…→x_d。

Otherwise if the constellation with the reverse seam is:

when x is_s>x_dWhen x_s→x_s-1→x_s-2→…→x_d；

When x is_s≤x_dWhen x_s→x_s+1→x_s+2→…→x_d。

The constellation without reverse slits and the constellation with reverse slits have the same path in the Y-axis direction, and are both

(a) If y_d-y_s|≤M-|y_d-y_sIf the y-axis direction is:

when y is_s>y_dWhen y is_s→y_s-1→y_s-2→…→y_d；

When y is_s≤y_dWhen y is_s→y_s+1→y_s+2→…→y_d；

(b) If y_d-y_s|>M-|y_d-y_sIf the y-axis direction is:

when y is_s≤y_dWhen y is_s→y_s-1→y_s-2→…→1→M→…→y_d；

When y is_s>y_dWhen y is_s→y_s+1→y_s+2→…→M→1→…→y_d。

In the step 3), the method for arranging the subtasks comprises the following steps:

for each subtask after sorting:

for each satellite node S in the set of candidate transmission path nodes_k1, L, k, S₁Is a source node S_sWhen k is L, S_LIs a target node S_dThe following processes are performed:

is a node S_kThe size of the computing resource allocated to the jth sub-task,

is a node S_kThe size of available computing resources;

for all slave source nodes S_sTo node S_kThe maximum available bandwidth in the set of candidate transmission paths of step 2), the maximum available bandwidth being determined by the available bandwidth of the inter-satellite link in step 2),

for the slave source node S_sTo node S_kAllocating communication bandwidth of the jth sub-task on the candidate transmission path,

to place the jth sub-task in the node S_kReduced latency for execution, switch being the source node S_sTo node S_kTotal delay, size (D) of the satellite node data exchange on the transmission path_j) The size of input data to be processed for the jth subtask, Task being the set of all subtasks, c_{task_j}The amount of computation required to process the jth sub-task;

if it is not

The jth subtask is scheduled to the node S_kAnd updating the system state as follows:

task- { subtask j }

For all nodes S_sTo node S_kNode S on the maximum bandwidth path_i：

Otherwise, the jth subtask is arranged to the source node to be executed, and the system state is updated as follows:

task- { subtask j }

Compared with the prior art, the invention has the beneficial effects that:

(1) the invention combines the routing and the service calculation distribution, combines the data transmission delay, the task calculation delay and the result convergence delay, globally considers the timeliness of the task execution, shortens the information generation period and reduces the overall service delay.

(2) The invention abstracts the low-orbit constellation into the self-organizing mesh network, and parallelizes and distributes the satellite computing task by combining the distributed computing thought so as to improve the timeliness and fault tolerance of task execution.

(3) The invention jointly considers the communication load and the calculation load, carries out distribution according to the network state and the demand, can realize the load balance of the space-based nodes and improve the calculation efficiency and the communication efficiency.

Drawings

FIG. 1 is a flow chart of the present invention;

fig. 2 is a scene diagram according to an embodiment of the present invention.

Detailed Description

The following describes embodiments of the present invention in further detail with reference to the accompanying drawings.

The invention provides a low-delay service space-based computing method, which provides routing and computing strategies according to the conditions of flow load, computing load and the like, selects a path in real time, can adapt to network flow change and node computing load change, has strong self-adaptability, realizes a low-delay processing method for information acquisition, information extraction and information transmission by fully utilizing space-based computing and communication resources, and solves the problem of high-efficiency processing and transmission of low-delay service of a low-orbit satellite computing platform. The method can improve the use efficiency of satellite communication and computing resources, reduce the service transmission processing time delay and meet the service quality requirement of low-time-delay service.

As shown in fig. 1, a method for computing space-based services with low latency includes the following steps:

11) task decomposition

Provided with satellites S_sThe task to be transmitted is a task T (D, R), the D is input data to be processed by the task T, and the R is output data after processing by the task T. And decomposing the T into parallel subtasks to jointly complete the task T. And when all the subtasks of the T are completed, the task T is completed.

12) The network state acquisition is required for realizing path acquisition and computing node selection as follows

Each satellite of the low-orbit network has 4 intersatellite links of 2 front and back in the orbit and 2 left and right between the orbits, and all the satellites S in the low-orbit network_iPeriodically reporting the network load of the satellite to the high orbit satellite, wherein the reported content comprises the available bandwidth B of the four inter-satellite links_i＝{B_i,1,B_i,2,B_i,3,B_i,4Mbps, current self-available computing power C_iMflops。

The high orbit satellites periodically broadcast low orbit satellite network states.

13) The computing node and the transmission path selection method are as follows

According to the network state obtained by 2) and the sub-tasks and the sub-task attributes decomposed by 1), firstly sorting the sub-tasks, then arranging the tasks on the candidate transmission paths in sequence, and mapping the sub-tasks to the corresponding satellite nodes;

14) transmission and computation

And 3) sending the data of the corresponding to-be-processed tasks of the subtasks to corresponding satellite nodes according to the arrangement result of the node 3), loading a processing program by the satellite nodes according to the calculation requirements of the subtasks, processing the received subtask data and sending the processing result to the target node. And setting the processing result as a high-priority service, and sending the high-priority service through reserved bandwidth.

15) Subtask aggregation

And after each subtask is correctly executed, feeding back to the task source node, and executing the failed task. And redistributing the transmission calculation process by the task source node until the calculation result is correct, and converging the subtasks at the target node after all subtasks are correctly executed to complete the transmission calculation of the whole task.

2. The sorting method described in step 1, 13) is as follows:

let the j-th subtask have a ranking factor of λ_j，

Take lambda_j＝size(D_j)/size(D_max)-c_{task_j}/c_maxWherein D is_jFor data to be processed for task j, size (D)_j) For data size, size (D)_max) The size of the input data to be processed corresponding to the subtask with the largest input data size, c_{task_j}The amount of computation required to process task j, c_maxThe calculation amount of the subtask with the largest required calculation amount is obtained.

Each subtask is ordered by an ordering factor from large to small.

3. The method for acquiring the candidate transmission path node set comprises the following steps:

abstracting low-orbit satellite network into 2-dimensional mesh network, and slave task source node S_sTo the target node S_dAll nodes contained in the path set with the minimum hop count form a candidate transmission path node set, and the total number of the nodes in the set is L.

Source node S_sTo the target node S_dMinimum jumpThe set of all nodes in the number path range is S ═ S_iAnd the low-orbit satellite network can be abstracted to a 2-dimensional mesh network, and in the 2-dimensional mesh network topology, the inter-orbit links are marked in the x-axis direction and the intra-orbit links are marked in the y-axis direction. For a certain communication, the data communication source node is S_s(x_s,y_s) The destination node is S_d(x_d,y_d) If the constellation is a constellation without a reverse seam, the task source node S_sTarget node S_dThe difference between the x-direction communication logic distance and the y-direction communication logic distance is as follows:

if the constellation is a constellation with reverse gaps, the task source node S_sTarget node S_dThe difference between the x-direction communication logic distance and the y-direction communication logic distance is as follows:

wherein N is the number of orbital planes, and M is the number of satellites on one orbital plane;

then the slave S_sTo S_dAt most share

The hop distance of the path with the least number of hops is delta x + delta y; the set of nodes on these paths is the candidate transmission path node set, and the path set with the least number of hops forms the candidate transmission path set.

If the constellation is a constellation without a reverse seam, the path in the x-axis direction in the path with the least hop number is as follows:

(a) if | x_d-x_s|≤N-|x_d-x_sIf the path along the x-axis is:

when x is_s>x_dWhen x_s→x_s-1→x_s-2→…→x_d；

When x is_s≤x_dWhen x_s→x_s+1→x_s+2→…→x_d；

(b) If | x_d-x_s|>N-|x_d-x_sIf the path along the x-axis is:

when x is_s≤x_dWhen x_s→x_s-1→x_s-2→…→1→N→…→x_d；

When x is_s>x_dWhen x_s→x_s+1→x_s+2→…→N→1→…→x_d。

Otherwise if the constellation with the reverse seam is:

when x is_s>x_dWhen x_s→x_s-1→x_s-2→…→x_d；

When x is_s≤x_dWhen x_s→x_s+1→x_s+2→…→x_d。

The constellation without reverse slits and the constellation with reverse slits have the same path in the Y-axis direction, and are both

(a) If y_d-y_s|≤M-|y_d-y_sIf the y-axis direction is:

when y is_s>y_dWhen y is_s→y_s-1→y_s-2→…→y_d；

When y is_s≤y_dWhen y is_s→y_s+1→y_s+2→…→y_d；

(b) If y_d-y_s|>M-|y_d-y_sIf the y-axis direction is:

when y is_s≤y_dWhen y is_s→y_s-1→y_s-2→…→1→M→…→y_d；

When y is_s>y_dWhen y is_s→y_s+1→y_s+2→…→M→1→…→y_d。

4. For each subtask after sorting:

for candidate transmission pathsEach satellite node S in the node set_k(k 1., L), k 1., L, k 1, S₁Is a source node S_sWhen k is L, S_LIs a target node S_dThe following processes are performed:

is a node S_kThe size of the computing resource allocated to the jth sub-task,

is a node S_kThe size of available computing resources;

for all slave source nodes S_sTo node S_kThe maximum available bandwidth in the set of candidate transmission paths of (1), the maximum available bandwidth being determined by the available bandwidth of the inter-satellite link in step 12),

for the slave source node S_sTo node S_kAllocating communication bandwidth of the jth sub-task on the candidate transmission path,

to place the jth sub-task in the node S_kReduced latency for execution, switch being the source node S_sTo node S_kTotal delay, size (D) of the satellite node data exchange on the transmission path_j) The size of input data to be processed for the jth subtask, Task being the set of all subtasks, c_{task_j}The amount of computation required to process the jth sub-task;

if it is not

The jth subtask is scheduled to the node S_kAnd updating the system state as follows:

task- { subtask j }

For all nodes S_sTo node S_kNode S on the maximum bandwidth path_i：

Otherwise, the jth subtask is arranged to the source node to be executed, and the system state is updated as follows:

task- { subtask j }

And repeating the subtask processing process until all subtasks are arranged.

An embodiment scenario is shown in fig. 2, which is a part of a constellation, where the entire constellation includes 24 orbital planes, and each orbital plane includes 24 satellites. Wherein the satellite51611 generates data and the processed data is transmitted to its subordinate nodes via the satellite 51712.

The low-delay service space-based computing method comprises the following steps:

11) task decomposition

The processing task T to be transmitted of the satellite51611 is in the remote sensing imageAnd target identification, wherein the input data D is remote sensing data of 2000 x 10000 x 8 bits. Decompose T into 2 subtasks { Ta₁，Ta₂And finishing the task T together. The data amount of each subtask is 2000 × 5000 × 8 bit-80 Mbit, and the calculation amount required for processing each subtask is 100 gflips. And when all the subtasks of the T are completed, the task T is completed.

12) The network state acquisition is required for realizing path acquisition and computing node selection as follows

All satellites S in the low-earth-orbit network_iPeriodically reporting the network load of the satellite to the high orbit satellite, wherein the reported content comprises the available bandwidth B of the four inter-satellite links_i＝{B_i,1,B_i,2,B_i,3,B_i,4Mbps, current self-available computing power C_iMflops。

The high orbit satellites periodically broadcast low orbit satellite network states.

The current computing power and communication power of each satellite are shown in tables 1 and 2.

TABLE 1 satellite node computing power C

Satellite identification	Current available computing power
		Satellite51611	6Gflops
Satellite51612	10Gflops
		Satellite51613	5Gflops
Satellite51710	4Gflops
		Satellite51711	11Gflops
Satellite51712	7Gflops

TABLE 2 available Bandwidth B between satellites

13) The computing node and the transmission path selection method are as follows

According to the network state obtained by 2) and the sub-tasks and the sub-task requirements decomposed by 1), firstly sorting the sub-tasks, then arranging the tasks on the candidate transmission paths according to the sorting, and mapping the sub-tasks to the corresponding satellite nodes;

14) transmission and computation

And 3) sending the data of the corresponding to-be-processed tasks of the subtasks to corresponding satellite nodes according to the arrangement result of the node 3), loading a processing program by the satellite nodes according to the calculation requirements of the subtasks, processing the received subtask data and sending the processing result to the target node. And setting the processing result as a high-priority service, and sending the high-priority service through reserved bandwidth.

15) Subtask aggregation

And after each subtask is correctly executed, feeding back to the task source node, and executing the failed task. And redistributing the transmission calculation process by the task source node until the calculation result is correct, and converging the subtasks at the target node after all subtasks are correctly executed to complete the transmission calculation of the whole task.

2. The results of the sorting method described in step 1) 13) are as follows:

in this example λ₁＝λ₂I.e. ordered as { Ta₁，Ta₂}。

3. According to the method of the invention, the set of candidate transmission path nodes is { Satellite51611, Satellite51612, Satellite51613, Satellite51710, Satellite51711, Satellite51712},

the candidate transmission paths are:

in this example satellite51611 to satellite51712

Shortest path, hop count is 3:

Satellite51611->Satellite51612->Satellite51613->Satellite51712,

Satellite51611->Satellite51612->Satellite51711->Satellite51712

Satellite51611->Satellite51710->Satellite51711->Satellite51712

4. the arrangement method according to step 1 of the invention results in the following:

considering that the propagation delay is of the order of ms and the calculation and transmission delays are of the order of seconds, the propagation delay is ignored in this embodiment.

Satellite51611 assigns a subtask Ta1 to each node, and the calculated Δ t is shown in Table 3 as

TABLE 3 task 1 latency

Satellite	Δt
		Satellite51612	-9.52380952380953
Satellite51613	11.3725490196078
		Satellite51710	22.8205128205128
Satellite51711	-8.48484848484848
		Satellite51712	2.51082251082251

Because of the fact that

The subtask Ta1 is scheduled to node Satellite51612 for calculation. The update information is:

TABLE 4 satellite node computing power

TABLE 5 available Bandwidth and propagation delay between satellites

Satellite51611 allocates subtasks Ta2 to each node, and the calculated Δ t is

TABLE 6 task 2 latency

Satellite	Δt
		Satellite51613	11.3725490196078
Satellite51710	22.8205128205128
		Satellite51711	-8.48484848484848
Satellite51712	2.51082251082251

Because of the fact that

So the subtask Ta2 is assigned to the Satellite 51711.

The total time delay of the task execution obtained by the invention is max {100/5+80/21, 100/5.5+50/12} -, 23.810 seconds.

The above embodiments illustrate that the present invention can effectively reduce the service delay compared with the conventional independent processing method of computing communication.

Those skilled in the art will appreciate that the invention may be practiced without these specific details.

Claims (7)

1. A low-delay service space-based computing method is characterized by comprising the following steps:

1) satellite S with task source node_sThe task to be transmitted and processed is a task T (D, R), wherein D is input data to be processed by the task T, and R is output data processed by the task T; decomposing the T into a plurality of parallel subtasks, and completing the task T when each subtask of the T is completed;

2) each satellite in the low-orbit network periodically reports the network state of the satellite to the high-orbit satellite, and the high-orbit satellite periodically broadcasts the network state of the low-orbit satellite;

each satellite of the low-orbit network has 4 inter-satellite links of 2 front and back in the orbit and 2 left and right between the orbits, and the ith satellite S_iThe reported content comprises an available bandwidth set B of four intersatellite links_iMbps and currently available computing resources

Mflops；B_i＝{B_i,1,B_i,2,B_i,3,B_i,4In which B is_i,1For the ith satellite S_iAvailable bandwidth of the first inter-satellite link, B_i,2For the ith satellite S_iAvailable bandwidth of the second inter-satellite link, B_i,3For the ith satellite S_iAvailable bandwidth of the third inter-satellite link, B_i,4For the ith satellite S_iThe available bandwidth of the fourth inter-satellite link;

3) determining a candidate transmission path node set according to the low earth orbit satellite network state, sequencing subtasks according to the subtasks and subtask attributes decomposed in the step 1), then arranging the subtasks on the candidate transmission path node set according to the sequencing, and mapping the subtasks to corresponding satellite nodes;

4) according to the arrangement result, sending input data to be processed of the subtasks to corresponding satellite nodes, loading a processing program by the satellite nodes according to the calculation requirement of the subtasks, processing the received subtask data and sending the processing result to a target node, wherein the processing result is set as a high-priority service and is sent through a reserved bandwidth;

5) and feeding back each subtask to the task source node after the subtask is correctly executed, and for the task which fails to be executed, redistributing the transmission calculation process by the task source node until the calculation result is correct, and converging the subtasks at the target node after all subtasks are correctly executed to complete the transmission calculation of the whole task.

2. The method for computing space-based services with low latency as claimed in claim 1, wherein in the step 3), the subtasks are sorted according to the following method:

let the j-th subtask have a ranking factor of λ_j，

Take lambda_j＝size(D_j)/size(D_max)-c_{task_j}/c_maxWherein D is_jInput data to be processed for the jth subtask, size (D)_j) Size of input data to be processed for jth sub-task, D_maxFor the input data to be processed corresponding to the subtask with the largest size of the input data to be processed in the subtask set, c_{task_j}The amount of computation required to process the jth sub-task, c_maxIs the calculated amount, lambda, of the subtask in the subtask set with the largest required calculated amount_jA task ordering factor;

each subtask is ordered by an ordering factor from large to small.

3. The method according to claim 1, wherein in step 3), the method for determining the candidate transmission path node set is as follows:

abstracting low-orbit satellite network into 2-dimensional mesh network, and slave task source node S_sTo the target node S_dAll nodes contained in the path set with the minimum hop count form a candidate transmission path node set, and the total number of the nodes in the set is L.

4. The method as claimed in claim 3, wherein the task source node S is configured to perform the calculation_sTo the target node S_dSet of candidate transmission path nodes is S ═ S_iAbstracting a low-orbit satellite network into a 2-dimensional mesh network, and in the 2-dimensional mesh network topology, marking links between orbits as an x-axis direction and links in the orbits as a y-axis direction; for a certain communication, a task source node S is set_sThe coordinate is (x)_s,y_s) The target node S_dThe coordinate is (x)_d,y_d)；

If the constellation is a constellation without a reverse seam, the task source node S_sTarget node S_dX and y directions therebetweenThe communication logic distance difference is:

if the constellation is a constellation with reverse gaps, the task source node S_sTarget node S_dThe difference between the x-direction communication logic distance and the y-direction communication logic distance is as follows:

wherein N is the number of orbital planes, and M is the number of satellites on one orbital plane;

then the slave S_sTo S_dAt most share

The hop distance of the path with the least number of hops is delta x + delta y; the set of nodes on these paths is the candidate transmission path node set, and the path set with the least number of hops forms the candidate transmission path set.

5. The method according to claim 4, wherein if the constellation without reverse seam is used, the path in the x-axis direction in the path with the least number of hops is:

(a) if | x_d-x_s|≤N-|x_d-x_sIf the path along the x-axis is:

when x is_s>x_dWhen x_s→x_s-1→x_s-2→…→x_d；

When x is_s≤x_dWhen x_s→x_s+1→x_s+2→…→x_d；

(b) If | x_d-x_s|>N-|x_d-x_sIf the path along the x-axis is:

when x is_s≤x_dWhen x_s→x_s-1→x_s-2→…→1→N→…→x_d；

When x is_s>x_dWhen x_s→x_s+1→x_s+2→…→N→1→…→x_d。

Otherwise if the constellation with the reverse seam is:

when x is_s>x_dWhen x_s→x_s-1→x_s-2→…→x_d；

When x is_s≤x_dWhen x_s→x_s+1→x_s+2→…→x_d。

6. The method according to claim 5, wherein the constellation without reverse slit and the constellation with reverse slit have the same path in the Y-axis direction, and both have the same path

(a) If y_d-y_s|≤M-|y_d-y_sIf the y-axis direction is:

when y is_s>y_dWhen y is_s→y_s-1→y_s-2→…→y_d；

When y is_s≤y_dWhen y is_s→y_s+1→y_s+2→…→y_d；

(b) If y_d-y_s|>M-|y_d-y_sIf the y-axis direction is:

when y is_s≤y_dWhen y is_s→y_s-1→y_s-2→…→1→M→…→y_d；

When y is_s>y_dWhen y is_s→y_s+1→y_s+2→…→M→1→…→y_d。

7. The method for computing space-based services with low latency as claimed in claim 6, wherein in the step 3), the method for arranging the subtasks is as follows:

for each subtask after sorting:

for each satellite node S in the set of candidate transmission path nodes_k1, L, k, S₁Is a source node S_sWhen k is L, S_LIs a target node S_dThe following processes are performed:

is a node S_kThe size of the computing resource allocated to the jth sub-task,

is a node S_kThe size of available computing resources;

for all slave source nodes S_sTo node S_kThe maximum available bandwidth in the set of candidate transmission paths of step 2), the maximum available bandwidth being determined by the available bandwidth of the inter-satellite link in step 2),

for the slave source node S_sTo node S_kAllocating communication bandwidth of the jth sub-task on the candidate transmission path,

to place the jth sub-task in the node S_kReduced latency for execution, switch being the source node S_sTo node S_kTotal delay, size (D) of the satellite node data exchange on the transmission path_j) The size of input data to be processed for the jth subtask, Task being the set of all subtasks, c_{task_j}The amount of computation required to process the jth sub-task;

if it is not

The jth subtask is scheduled to the node S_kAnd updating the system state as follows:

task- { subtask j }

For all nodes S_sTo node S_kNode S on the maximum bandwidth path_i：

Otherwise, the jth subtask is arranged to the source node to be executed, and the system state is updated as follows:

task- { subtask j }

Images