Method for task planning of space information network based resource interchange

Abstract

Disclosed is a method for task planning of a space information network based on resource interchange. The method includes: initializing basic parameters of the space information network; dividing a planning horizon into K time slots of equal length, and constructing a resource time-varying graph for the space information network; sampling a feasible resource combination space of each task, and obtaining a candidate resource combination set comprised of the resource combinations with independence greater than or equal to a threshold n; calculating a conflict relation between resource combinations, and constructing a resource combination conflict graph; obtaining a maximum independent set of the resource combination conflict graph to obtain a global planning result; and searching a neighborhood of the global planning result, and completing a local adjustment of a task planning scheme through the resource interchange, to complete the task planning based on characteristics of the resource interchange.

Description

CROSS-REFERENCE TO RELATED APPLICATION

The present disclosure claims priority to Chinese Patent Application No. 202011079368.3 filed on Oct. 10, 2020, which is hereby incorporated by reference in its entirety.

TECHNICAL FIELD

The present disclosure belongs to the field of space information technologies, and relates to a method for task planning of a space information network based on resource interchange.

BACKGROUND

The space information network is a network system that uses space platforms such as satellites as carriers to obtain, transmit and process space information in real time. In view of its significant advantages such as wide coverage, flexible networking, and freedom from geographical environment restrictions, the space information network has played an important role in the military and civilian fields in recent years. However, due to a wide variety of tasks and resources in the space information network, and characteristics such as dynamic randomness and non-uniformity of the temporal and spatial distribution of the tasks and resources in the space information network, the phenomenon of shortages of certain resources in some time periods and space of the network occurs from time to time, which causes some tasks in the network to be unable to be processed in time, thereby greatly limiting the performance of the space information network. In the face of ever-increasing task demands, it is important to reasonably schedule limited resources to solve a great task demand timely.

By using inherent properties such as resource mobility in the space information network and emerging technologies such as a wireless software-defined payload, it is possible to replace or transform scarce resources with idle resources, thereby providing conditions for improving resource utilization efficiency and ensuring effective completion of tasks. Currently, some researches have been carried out on the resource mobility in the space information network such as resource interchange and resource aggregation, and management methods for network resources are provided based on a characteristic of the resource interchange of the space information network.

However, existing methods also have certain shortcomings. First, these methods only qualitatively use the characteristic of the resource interchange of the space information network, and do not quantitatively give conditions for the interchange of resource combinations. Second, the discontinuity of network resources and task requirements leads to a complex combination optimization problem in task planning of the space information network. Although the resource interchange greatly increases matching freedom between resource combinations and tasks, the resource interchange also greatly increases a range of feasible solution space, thereby significantly increasing the computational complexity of obtaining the optimal solution. Existing work mostly considers a problem of simple scenarios with a limited number of satellites and limited types of resources, or designing heuristic algorithms to split the original combinational optimization problem into several sub-problems from a dimension of time or a task execution stage, while ignores a correlation between resources.

SUMMARY

The present disclosure aims to overcome the above defects in the related art, and provides a method for task planning of a space information network based on resource interchange, capable of reducing the computational the complexity of the task planning while increasing a task completion rate.

In order to achieve the above objects, the method for the task planning of the space information network based on resource interchange provided by the present disclosure includes the following steps of:

• • 1) initializing basic parameters of the space information network;
  • 2) dividing a planning horizon into K time slots of equal length, where a length of each time slot is ti, and constructing a resource time-varying graph G_K(V, A) for the space information network, where V and A respectively represent a set of vertices and a set of resource arcs in the resource time-varying graph;
  • 3) sampling the feasible resource combination space of each task om_i∈OM, and obtaining a candidate resource combination set, denoted by P_i, comprised of the resource combinations with independence greater than or equal to a threshold n;
  • 4) calculating the conflict relation among resource combinations, and constructing a resource combination conflict graph RCG(P_C, E_C), where P_C and E_C respectively represent a set of vertices and a set of edges of the resource combination conflict graph;
  • 5) obtaining the maximum independent set of the resource combination conflict graph, and completing large-scale global task planning based on a candidate resource combination set, to obtain a global planning result; and
  • 6) searching the neighborhood of the global planning result, and completing a local adjustment of a task planning scheme through the resource interchange, to complete the task planning based on characteristics of the resource interchange.

The basic parameters of the space information network include a set of Earth observation satellites OS, a set of relay satellites RS, a set of ground stations GS, a data processing center dc, a set of observation targets OB, a set of tasks OM, and a planning horizon T.

Specifically, step 4) includes:

• • 4a) adding each candidate resource combination in set P_se=Å_1≤i≤|OM|P_i as a vertex in the resource combination conflict graph; and
  • 4b) for any pair of vertices p_i,k and p_j,l in the resource combination conflict graph, adding an edge between the vertices p_i,k and p_j,l if resource combinations p_i,k and p_j,l have a conflict relation or satisfy i=j and k≠l, where p_i,k represents the k-th candidate resource combination of the task om_i.

Specifically, step 5) includes:

• • 5a) initializing output set P_su=∅;
  • 5b) sorting all vertices in P_C based on their degree in a descending order, and selecting a vertex p₀ with the largest degree;
  • 5c) adding the vertex p₀ with the largest degree to output set P_su,
  • 5d) deleting vertex p₀ and all the vertices adjacent to p₀ from P_C, and deleting all the edges associated with the deleted vertices from E_C; and
  • 5e) proceeding to step 5b) when P_C≠∅; otherwise, outputting the set P_su, and obtaining, based on P_su, a set of planned tasks and a resource allocation scheme of the planned tasks.

Specifically, step 6) includes:

• • 6a) obtaining the resource time-varying graph G_K(V, A) and its subgraph G_K′(V′, A′) by removing the resource arcs represented by the resources occupied by the planned tasks, and obtaining the set of tasks that is successfully planned, denoted by OM_y, and the set of tasks that has not been planned, denoted by OM_w;
  • 6b) calculating resource shortage degrees of all tasks in OM_w and sorting the resource shortage degrees from high to low;
  • 6c) initializing a counter n=1;
  • 6d) finding an idle resource for the n-th task om_n in OM_w through the resource interchange;
  • 6e) increasing the counter n by 1; and
  • 6f) proceeding to step 6d) when n is smaller than or equal to |OM_w|; otherwise, outputting task set OM_y.

Specifically, step 6d) includes:

• • 6d1) finding a feasible resource combination with the least resources occupied by a planned task for the task om_n, which is denoted by Pd_n, and using Pz_n to represent the set of resources in Pd_n and occupied by the planned tasks;
  • 6d2) traversing each resource arc a∈Pz_n; for the task that occupies the resource arc a, denoted by fom(a), re-allocating a new resource combination for fom(a) through resource interchange, so that the resource arc a is no longer occupied; and if the resource interchange succeeds, updating G_K′(V, A′), and deleting a resource a from the set of resources Pz_n; otherwise, skipping to step 6e); and
  • 6d3) finding a path from ob_n^┌tsn/τ┐ to dc^└ten/τ┘ in G_K′(V′, A′) as the new resource combination allocated to task om_n, denoted by p_n, updating G_K′(V′, A′), and adding the task om_n to the set of tasks OM_y that is successfully planned.

The present disclosure has the following beneficial effects.

During specific operations, the method for the task planning of the space information network based on resource interchange of the present disclosure, based on a mechanism of space information resource interchange, converts non-scarce resources into scarce resources based on the scarcity of task resources, such that the scarce resources can be released, thereby improving a resource utilization rate and enhancing task execution efficiency of the space information network. In addition, on a basis of the resource time-varying graph, a task planning problem in the space information network is transformed into a flow problem in graph theory. According to a relation between resource combinations and a degree of independence of the resource combinations, a task planning process is divided into two stages: large-scale global optimization and a small-scale local adjustment, such that an integer linear programming problem is simplified, and a feasible resource combination space of each task is sampled to obtain a set of independent resource combinations. The set of independent resource combinations is used as an optimization space for global task planning to reduce the complexity of global optimization.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a schematic diagram illustrating a scenario used by the present disclosure;

FIG. 2 is an overall flowchart of the present disclosure;

FIG. 3 is a resource time-varying diagram in the present disclosure;

FIG. 4 is a flowchart illustrating a sampling algorithm of a resource combination;

FIG. 5 is a resource combination conflict diagram in the present disclosure;

FIG. 6 is a flowchart illustrating an algorithm for completing a large-scale global task planning based on a conflict relation between candidate resource combinations; and

FIG. 7 is a flowchart illustrating an algorithm for completing a local adjustment of a task planning scheme based on resource interchange.

DESCRIPTION OF EMBODIMENTS

The present disclosure will be further described in detail below in combination with the accompanying drawings.

With reference to FIG. 1 and FIG. 2, a method for task planning of a space information network based on resource interchange according to the present disclosure includes the following steps.

At step 1), basic parameters of the space information network are initialized.

The basic parameters of the space information network include a set of Earth observation satellites OS={os₁, os₂, . . . }, a set of relay satellites RS={rs₁, rs₂, . . . }, a set of ground stations GS={gs₁, gs₂, . . . }, a data processing center dc, a set of observation targets OB={ob₁, ob₂, . . . }, a set of tasks OM={om₁, om₂, . . . } and a planning horizon T.

At step 2), a planning horizon T is divided into K time slots of equal length, where a length of each time slot is τ, and a resource time-varying graph G_K(V, A) for the space information network is constructed, where V and A respectively represent the set of vertices and the set of resource arcs in the resource time-varying graph. The resource time-varying graph G_K(V, A) for the space information network is a K-layered directed graph.

The set of vertices V in the resource time-varying graph contains a set of duplicates of all nodes in the space information network in different time slots. That is, V=V_OB∪V_IM∪V_PS∪V_ST∪V_RS∪V_GS∪V_DC, where V_OB, V_IM, V_PS, V_ST, V_RS, V_GS and V_DC respectively correspond to a set of duplicate vertices in each time slot of observation targets, observation satellite imagers, processors, transceivers and storage units, relay satellites, ground stations and data processing center.

The set of resource arcs A in the resource time-varying graph contains different types of resources in a network. That is, A=A_O∪A_L∪A_S∪A_C, where A_O, A_L, A_S, and A_C respectively represent a set of observation arcs, a set of transmission arcs, a set of storage arcs, and a set of calculation arcs.

At step 3), the feasible resource combination space of each task om_i∈OM is sampled, and a set of resource combinations P_i with independence greater than or equal to a threshold n is determined as a candidate resource combination set.

Specifically, step 3) includes the following steps.

At step 3a), obtaining a set of resources in which the maximum degree of independence of each task om_i is not smaller than n is modeled as an optimization problem:

$P1\text{:}\max \sum _{p\in P_{a,i}}{x}_p.$

That is, the number of elements in the candidate resource combination set is maximized, where P_a,i represents a set of all feasible resource combinations of the task om_i, and x_p represents a Boolean variable.

At step 3a1), constraint C1 is added to restrict that the same resource can only be used by one resource combination in n time slots. That is, independence between any two resource combinations with x_p=1 is at least n:

$C1\text{:}Z\left(p_1,p_2\right)⩾n,\forall p_1,p_2\in \left\{p❘x_p=1,p\in P_{a,i}\right\};$

At step 3a2), constraint C2 is added. x_p being equal to 1 means that a resource combination p is within the set of candidate resources; otherwise, the resource combination p is not within the set of candidate resources:

$C2\text{:}{x}_p\in \left\{0,1\right\}.$

Referring to FIG. 4, a resource combination sampling algorithm is designed based on the optimization problem P1 in 3a):

At step 3b), P_i=∅ is initialized.

At step 3c), each resource arc a in a time slot from ┌ts_i/τ┐ to └te_i/τ┘ in the resource time-varying graph is traversed. If a is a storage arc, a weight w(a) of the resource arc a is set to 0; otherwise, w(a) is set to the number of resource arcs whose degree of independence from a is smaller than n.

At step 3d) a shortest path from a source node ob_i^┌tsi/τ┐ to a destination node dc^└tei/τ┘ on the resource time-varying graph is obtained and recorded as p_s. Let P_i=P_i∪{p_s}.

At step 3e), each resource arc a in p_s is traversed. If a is an observation arc, a calculation arc or a transmission arc, all resource arcs whose degree of independence from a is smaller than n are deleted from G_K(V, A).

At step 3f), if there is still a path between ob_i^┌tsi/τ┐ and dc^└tei/τ┘ in the resource time-varying graph, skip to step 3c); otherwise, output a candidate resource combination P_i.

At step 4), a conflict relation between resource combinations is calculated, and a resource combination conflict graph RCG (P_C, E_C) is constructed, where P_C and E_C respectively represent a set of vertices and a set of edges of the resource combination conflict graph.

Specifically, step 4) includes the following steps.

At step 4a), p_i,k is set as representing the k-th candidate resource combination of task om_i. For resource combinations p_i,k and p_j,l, if one of the following conditions is met:

(i) resources (v_m1^r, v_n1^r)∈p_i,k and (v_m2^t, v_n2^t)∈p_j,l existing, and v_m1^r=v_m2^t v_m1^r∈V_ST being satisfied; and

(ii) resources (v_m1^r, v_n1^r)∈p_i,k and (v_m2^t, v_n2^t)∈p_j,l existing, and v_n1^r=v_n2^t v_n1^r∈V_IM∪V_RS∪V_GS being satisfied;

The resource combinations p_i,k and p_j,l are defined to conflict with each other.

At step 4b), the resource combination conflict graph RCG (P_C, E_C) as illustrated in FIG. 5 is constructed based on the conflict relationship between the resource combinations at step 4a), where P_C and E_C respectively represent the set of vertices and the set of edges of the resource combination conflict graph. Specifically, step 4b) includes the following steps.

At step 4b1), the candidate resource combination in a set p_se=∪_1≤i≤|OMP_i is added as a vertex in the resource combination conflict graph.

At step 4b2), for any pair of vertices (p_i,k and p_j,l) in the resource combination conflict graph, if resource combinations p_i,k and p_j,l conflict with each other, or i=j and k≠l, an edge is added between vertices p_i,k and p_j,l.

At step 5), a maximum independent set of the resource combination conflict graph is obtained, and large-scale global task planning based on a candidate resource combination is completed, to obtain a global planning result.

Specifically, referring to FIG. 6, step 5) includes the following steps.

At step 5a), output set P_su=∅ is initialized.

At step 5b), all vertices in P_C are sorted based on their degree in a descending order, and a vertex p₀ with the largest degree is selected.

At step 5c), the vertex p₀ with the largest degree is added to output set P_su.

At step 5d), the vertex p₀ and all the vertices adjacent to the vertex p₀ are deleted from P_C, and all the edges associated with the deleted vertices are deleted from E_C.

At step 5e), if P_C≠∅, the method proceeds to step 5b); otherwise, the set P_su is outputted, and a set of planned tasks and a resource allocation scheme of the planned tasks are obtained based on P_su.

At step 6), a neighborhood of the global planning result is searched, and a local adjustment of a task planning scheme is completed through the resource interchange, to complete the task planning based on characteristics of the resource interchange.

Specifically, step 6) includes the following steps.

At step 6a), the resource time-varying graph G_K(V, A) and its subgraph G_K′(V′, A′) that has removed the resource arcs represented by the resources occupied by planned tasks are obtained, and the set of tasks that are successfully planned and the set of tasks that have not been planned, which are respectively denoted by OM_y and OM_w, are obtained.

At step 6b), resource shortage degrees of all tasks in OM_w are calculated, and the resource shortage degrees are sorted from high to low.

The minimum number of resources occupied by all feasible task combinations of the task om_i is defined as the resource shortage degree

${\eta }_i=\underset{p_{i,k}\in P_i}{\min }{\eta }_{i,k}$ of the task, where η_i,k represents the number of resources occupied by the planned task for the k-th resource combination of om_i. A specific process of obtaining the resource shortage degree is as follows.

At step 6b1), weights are assigned to the observation arcs, storage arcs, and calculation arcs on the resource time-varying graph G_K(V, A). If a resource corresponding to the arc is already occupied by the planned task, the weight of the arc is assigned to 1; otherwise, the weight is assigned to 0.

At step 6b2), a path between ob_i^┌tsi/τ┐ and dc^└tei/τ┘ is obtained by using a shortest path algorithm. A path length is the resource shortage degree η_i of the task.

On a basis of the resource shortage degree obtained in (6b), the algorithm as illustrated in FIG. 7 for completing the local adjustment of the task planning scheme based on the resource interchange is designed.

At step 6c), a counter n=1 is initialized.

At step 6d), an idle resource for the n-th task om_n in OM_w is found through the resource interchange.

At step 6e), the counter n is increased by 1.

At step 6f), if n is smaller than or equal to |OM_w|, the method proceeds to step 6d); otherwise, task set OM_y is outputted.

Specifically, step 6d) includes the following steps.

At step 6d1), a feasible resource combination Pd_n with the least resources occupied by planned tasks for the task om_n is found, and Pz_n denotes the set of resources occupied by the planned tasks and in Pd_n.

At step 6d2), each resource arc a∈Pz_n is traversed. For a task fom(a) that occupies the resource arc a, a new resource combination for the task fom(a) is re-allocated through the resource interchange, so that the resource arc a is no longer occupied. If the resource interchange succeeds, G_K′(V′, A′) is updated, and a resource a is deleted from the set of resources Pz_n; otherwise, skip to step 6e).

At step 6d3), a path p_n from ob_n^┌tsn/τ┐ and dc^└ten/τ┘ is found in G_K′(V′, A′) as the new resource combination allocated to task om_n, G_K′(V′, A′) is updated, and task om_n is added to set OM_y.

The above description is only a specific example of the present disclosure. Obviously, it is possible for those skilled in the art to make various modifications and changes in form and details without departing from the principle and structure of the present disclosure after understanding the content and principle of the present disclosure. However, these modifications and changes based on the concept of the present disclosure are still within the protection scope of the attached claims of the present disclosure.

Claims

1. A method for task planning of a space information network based on resource interchange, comprising: step 1: initializing basic parameters of the space information network;step 2: dividing a planning horizon into K time slots of equal length, where a length of each time slot is ti, and constructing a resource time-varying graph GK(V, A) for the space information network, where V and A respectively represent a set of vertices and a set of resource arcs in the resource time-varying graph;step 3: sampling a feasible resource combination space of each task omi∈OM, and obtaining a candidate resource combination set, denoted by Pi, comprised of the resource combinations with independence greater than or equal to a threshold n;step 4: calculating a conflict relation between resource combinations, and constructing a resource combination conflict graph RCG(PC, EC), where PC and EC respectively represent a set of vertices and a set of edges of the resource combination conflict graph;step 5: obtaining a maximum independent set of the resource combination conflict graph, and completing large-scale global task planning based on a candidate resource combination, to obtain a global planning result; andstep 6: searching a neighborhood of the global planning result, and completing a local adjustment of a task planning scheme through the resource interchange, to complete the task planning based on characteristics of the resource interchange.

2. The method according to claim 1, wherein the basic parameters of the space information network comprise a set of Earth observation satellites OS, a set of relay satellites RS, a set of ground stations GS, a data processing center dc, a set of observation targets OB, a set of tasks OM, and a planning horizon T.

3. The method according to claim 1, wherein the step 4 comprises: step 4a: adding each candidate resource combination in set pse=∪1≤i≤|OM|Pi as a vertex in the resource combination conflict graph; andstep 4b: for any pair of vertices pi,k and pj,l in the resource combination conflict graph, adding an edge between vertices pi,k and pj,l when resource combinations pi,k and pj,l have a conflict relation or satisfy i=j and k≠l, where pi,k represents a k-th candidate resource combination of the task omi.

4. The method according to claim 1, wherein the step 5 comprises: step 5a: initializing an output set Psu=∅;step 5b: sorting all vertices in PC based on respective degrees in a descending order, and selecting a vertex p0 with the largest degree;step 5c: adding the vertex p0 with the largest degree to the output set Psu;step 5d: deleting vertex p0 and all the vertices adjacent to p0 from PC, and deleting all the edges associated with the deleted vertices from EC; andstep 5e: proceeding to the step 5b if PC≠∅; otherwise, outputting the set Psu, and obtaining, based on Psu, a set of planned tasks and a resource allocation scheme of the planned tasks.

5. The method according to claim 1, wherein the step 6 comprises: step 6a: obtaining the resource time-varying graph GK(V, A) and a subgraph GK′(V′, A′) that has removed resource arcs represented by resources occupied by planned tasks, and obtaining a set of tasks OMy that is successfully planned and a set of tasks OMw that has not been planned;step 6b: calculating resource shortage degrees of all tasks in OMw, and sorting the resource shortage degrees from high to low;step 6c: initializing a counter n=1;step 6d: finding an idle resource combination for the n-th task omn in OMw through the resource interchange;step 6e: increasing the counter n by 1; andstep 6f: proceeding to the step 6d when n is smaller than or equal to |OMw|; otherwise, outputting the set of tasks OMy.

6. The method according to claim 5, wherein the step 6d comprises: step 6d1: finding a feasible resource combination Pdn with the least resources occupied by planned tasks for task omn, and setting Pzn as representing a set of resources occupied by the planned tasks and in Pdn;step 6d2: traversing each resource arc a∈Pzn; for the task that occupies resource arc a, denoted by fom(a), re-allocating a new resource combination for fom(a) through resource interchange, so that the resource arc a is no longer occupied; and if the resource interchange succeeds, updating GK′(V, A′), and deleting resource a from Pzn; otherwise, skipping to the step 6e; andstep 6d3: finding a path pn from obn┌tsn/τ┐ and dc└ten/τ┘ in GK′(V′, A′) as the new resource combination allocated to task omn, updating GK′(V′, A′), and adding task omn to the set OMy.