Heterogeneous task scheduling method and device for rotating and sweeping ultra-wideband satellites

Abstract

In the heterogeneous task scheduling method, device and storage medium for rotating and swinging ultra-wideband satellites proposed in this application, the directed acyclic graph DAG of the task is obtained, and candidate decoding is generated according to the DAG, and the initial time and initial energy consumption required for scheduling the task using the candidate decoding are calculated, and then a random number r is generated. If r is less than the probability threshold p, the candidate decoding is decomposed, and then the decomposed candidate decoding is reconstructed according to the optimal candidate solution judgment rule to obtain the optimal candidate decoding. If the time and energy consumption required for scheduling the task using the optimal candidate decoding reach the threshold, the optimal candidate decoding is output. Among them, the candidate decoding represents the scheduling and resource allocation plan of the task. In this application, the task scheduling is optimized with discrete multi-objectives, which avoids the coding conversion overhead in the continuous algorithm, reduces the complexity of the algorithm, and improves the convergence of solving multi-objective problems.

Description

Technical Field

[0001] The present application relates to the field of satellite-borne high-performance computing technology, and in particular to a heterogeneous task scheduling method, device and storage medium for a rotating, swing-scanning ultra-wideband satellite. Background Art

[0002] With the rapid development of earth observation technology, satellites in orbit with different widths and resolutions in high, medium and low orbits together constitute an earth observation network. Among them, ultra-wide-width remote sensing satellites adopt a new ultra-wide coverage imaging system with vertical track scanning and along-track splicing. The satellite orbit altitude is 500km, the sub-satellite spatial resolution is 1 meter, and the imaging width is 3000km, which can achieve full coverage of the earth's surface in a single day. However, the amount of remote sensing image data per unit time acquired by ultra-wide-width satellites is as high as 420Gbps, which far exceeds the data transmission bandwidth and storage space of ultra-wide-width satellites. Therefore, ultra-wide-width satellites need to integrate heterogeneous resources to perform heterogeneous multi-core / multi-task scheduling and processing of complex computing tasks.

[0003] However, when dealing with heterogeneous multi-core task scheduling, the relevant methods only take the time required to process the task as a single optimization target, without considering the energy consumption required for task processing. In addition, heterogeneous multi-core / multi-task scheduling is a discrete problem. In the relevant technology, the continuous optimization algorithm is converted into a discrete code to solve the discrete problem, which destroys the original solution performance of the algorithm and increases the complexity of the algorithm. Summary of the invention

[0004] The present application provides a heterogeneous task scheduling method, device and storage medium for a rotating and sweeping ultra-wideband satellite, so as to at least solve the technical problems of not considering the energy consumption required for task processing and the high algorithm complexity in the related technology.

[0005] The first aspect of the present application provides a heterogeneous task scheduling method for a rotating swing-scan ultra-wideband satellite, the method comprising:

[0006] S1. Obtain a directed acyclic graph (DAG) of the task, where the DAG includes the execution relationship of the task;

[0007] S2. Generate candidate decodings according to the DAG, and calculate the initial time and initial energy consumption required for scheduling the task using the candidate decodings, wherein the candidate decodings represent the scheduling and resource allocation scheme of the task;

[0008] S3, generate a random number r, if r is less than the probability threshold p, continue to execute step S4, where 0<r<1, 0<p<1;

[0009] S4, decomposing the candidate decoding;

[0010] S5, reconstructing the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code;

[0011] S6. If the time and energy consumption required for scheduling the task using the optimal candidate decoding both reach a threshold, outputting the optimal candidate decoding.

[0012] The second embodiment of the present application proposes a heterogeneous task scheduling device for a rotating swing-sweep ultra-wideband satellite, including:

[0013] An acquisition module, used to acquire a directed acyclic graph (DAG) of a task, wherein the DAG includes an execution relationship of the task;

[0014] A generation module generates candidate decodings according to the DAG, and calculates the initial time and initial energy consumption required for scheduling the task using the candidate decodings, wherein the candidate decodings represent a scheduling and resource allocation scheme for the task;

[0015] A judgment module is used to generate a random number r. If r is less than a probability threshold p, then continue to execute step S4, where 0<r<1, 0<p<1;

[0016] A decomposition module, used for decomposing the candidate decoding;

[0017] A processing module, used for reconstructing the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code;

[0018] The output module is used to output the optimal candidate decoding if the time and energy consumption required for scheduling the task using the optimal candidate decoding reach a threshold.

[0019] The computer storage medium proposed in the third aspect of the present application, wherein the computer storage medium stores computer-schedulable instructions; after the computer-schedulable instructions are scheduled by the processor, the method described in the first aspect above can be implemented.

[0020] The technical solution provided by the embodiments of the present application brings at least the following beneficial effects:

[0021] In the heterogeneous task scheduling method, device and storage medium for rotating swing ultra-wideband satellites proposed in this application, the directed acyclic graph DAG of the task is obtained, and candidate decoding is generated according to the DAG, and the initial time and initial energy consumption required for scheduling the task using the candidate decoding are calculated, and then a random number r is generated. If r is less than the probability threshold p, the candidate decoding is decomposed, and then the decomposed candidate decoding is reconstructed according to the optimal candidate solution judgment rule to obtain the optimal candidate decoding. If the time and energy consumption required for scheduling the task using the optimal candidate decoding reach the threshold, the optimal candidate decoding is output. Among them, the candidate decoding represents the scheduling and resource allocation scheme of the task. Therefore, in the method provided by this application, when selecting the optimal task scheduling method, not only the time required for scheduling the task is considered, but also the energy consumption required for scheduling the task is considered, and the task scheduling is optimized with discrete multi-objectives, so that the satellite can achieve efficient processing of the task while consuming lower energy consumption. At the same time, this application avoids the coding conversion overhead in the continuous algorithm and reduces the complexity of the algorithm by adopting multi-level coding as the discretization operation operator.

[0022] In addition, the method proposed in this application is based on Pareto to determine the optimal candidate solution, which improves the convergence of solving multi-objective problems.

[0023] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A schematic flow chart of a heterogeneous task scheduling method for a rotating swing-scan ultra-wideband satellite provided according to an embodiment of the present application;

[0026] Figure 2 A schematic diagram of the structure of a DAG of a task provided according to an embodiment of the present application;

[0027] Figure 3 A schematic diagram of the structure of hierarchical coding provided according to an embodiment of the present application;

[0028] Figure 4 A flowchart of a method for generating candidate decodings based on a DAG according to an embodiment of the present application is shown;

[0029] Figure 5 A schematic diagram of a process of mutating a candidate decoding according to an embodiment of the present application;

[0030] Figure 6 A schematic diagram of a process of decomposing candidate decoding according to an embodiment of the present application;

[0031] Figure 7 A flowchart of a method for reconstructing decomposed candidate solution codes according to an optimal candidate solution judgment rule proposed according to an embodiment of the present application;

[0032] Figure 8 The present invention is a flow chart of a heterogeneous task scheduling device for a rotating and sweeping ultra-wideband satellite according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0034] The following describes a heterogeneous task scheduling method and device for a rotating and sweeping ultra-wideband satellite according to an embodiment of the present application with reference to the accompanying drawings.

[0035] Embodiment 1

[0036] Figure 1 The flowchart of the heterogeneous task scheduling method for a rotating swing-scan ultra-wideband satellite provided according to an embodiment of the present application is as follows: Figure 1 As shown, this may include:

[0037] Step S1: Obtain the DAG (Directed Acyclic Graph) of the task.

[0038] Among them, in this application, DAG includes the execution relationship of tasks.

[0039] Furthermore, in the present application, the levels corresponding to the tasks can be divided according to the execution relationship between the tasks in the DAG.

[0040] For example, in this application, Figure 2 This is a diagram of the DAG structure of the tasks provided for this application. Figure 2As shown, it is assumed that the task to be completed is the geometric correction task in on-orbit high-precision positioning, and the task is divided into Task 0, Task 1, Task 2, Task 3, and Task 4. Specifically, Task 0 is to obtain the image data without geometric correction, Task 1 is to establish the same local plane coordinate system, Task 2 is to perform forward calculation for each scanned frame, Task 3 is to project each frame to the plane coordinate system and stitch the whole scene, and Task 4 is to output the corrected image after back-calculation using the RPC (RPC, Rational Polynomial Coefficients) model. And, according to the execution relationship between Task 0-Task 4, it can be concluded that Figure 2 The DAG of the task is shown.

[0041] Step S2: Generate candidate decodings according to the DAG, and calculate the initial time and initial energy consumption required for scheduling tasks using the candidate decodings.

[0042] In this application, candidate decoding is used to represent the scheduling and resource allocation scheme of the task. The candidate decoding may be a hierarchical coding. Specifically, Figure 3 This is a schematic diagram of the hierarchical coding structure proposed in this application. Figure 3 As shown, the hierarchical coding includes three layers, the first layer is the execution sequence of tasks, the second layer is the processor sequence composed of processor numbers, the processor sequence corresponds one-to-one with the scheduling sequence of tasks in the first layer, and the third layer is the DVFS (Dynamic Voltage and Frequency Scaling) value of the processor. Each DVFS value corresponds to the voltage / frequency in the processor number of the second layer to realize the dynamic change of the processor voltage / frequency. In this application, multi-level coding is used as a discretization operation operator to avoid the coding conversion overhead in the continuous algorithm.

[0043] For example, in this application, it is assumed that 0, 1, 2, and 3 are used to represent the DVFS value. The larger the value, the higher the voltage / frequency and the higher the performance of the processor. Therefore, the DVFS technology can be used to reasonably adjust the voltage / frequency of the processor according to the task size and save energy.

[0044] refer to Figure 3 As shown, task 2 in the candidate decoding is processed by the processor with processor number 0 at a voltage / frequency of DVFS 1.

[0045] And, the method for generating candidate decoding according to DAG in the present application may include the following steps:

[0046] Step S21, dividing the execution level of the task according to the pre-order and post-order constraints of the DAG;

[0047] Step S22, randomly sorting the tasks in each level to obtain a task sequence in each level;

[0048] Step S23: sequentially combine the task sequences in each level to obtain the first layer code in the candidate decodes, and then randomly generate the second layer code and the third layer code in the candidate decodes.

[0049] Specifically, Figure 4 This is a flowchart of the method for generating candidate decoding based on DAG proposed in this application. Figure 4 As shown, the successor of Task 0 is Task 1 and Task 2; the predecessor of Task 1 is Task 0, and the successor is Task 3 and Task 4, so Task 0 is task level one, and so on, Task 1 and Task 2 are task level two, Task 3 is task level three, and Task 4 is task level four.

[0050] Among them, in this application, by randomly sorting the tasks in task level two, a task sequence of 2, 1 in task level two can be obtained.

[0051] Further, in the present application, according to the level of the task, the task sequence in each level is combined from low to high to obtain the candidate decoding. Figure 4 The first layer code is shown. And, in the present application, the second layer code and the third layer code in the candidate solution code can be randomly generated.

[0052] Also, in the present application, after candidate decodings are generated according to the DAG, it is necessary to calculate the initial time and initial energy consumption required for scheduling tasks using the scheduling and resource allocation scheme of the candidate decodings.

[0053] Step S3: Generate a random number r. If r is less than the probability threshold p, continue scheduling step S4.

[0054] In this application, 0<r<1, 0<p<1.

[0055] And, in the present application, if r is greater than or equal to p, the candidate solution is mutated, and step S4 is continued using the candidate solution obtained after the mutation.

[0056] Among them, in the present application, the method for mutating the candidate solution may include: randomly selecting a position in the first layer encoding of the candidate solution, and randomly inserting the task at the position into the remaining positions between the preceding task and the succeeding task of the position.

[0057] Specifically, Figure 5 Schematic diagram of the process of mutating candidate decoding proposed in this application. Figure 5As shown in the figure, a mutation operation is performed on task 2, whose predecessor and successor tasks are task 0 and task 3 respectively. Then there is only one position where task 2 can be inserted. The candidate decoding codes after mutation are as follows Figure 5 shown.

[0058] Step S4: Decompose the candidate decoding codes.

[0059] Among them, in this application, the method for decomposing candidate decoding includes the following steps:

[0060] Step S41, obtaining task parameter d.

[0061] In this application, d represents the reserved part of the candidate decoding, d is less than the number of tasks, and d is a positive integer.

[0062] Step S42: randomly select d consecutive positions in the candidate decoded codes, and clear the second layer codes and the third layer codes of the remaining positions.

[0063] Specifically, Figure 6 This is a flow chart of the decomposition of candidate decoding proposed in this application. Figure 6 As shown, assuming that the acquired task parameter d is 2, two consecutive positions in the dotted box in the candidate solution are randomly selected, and the second layer code and the third layer code of the remaining positions are cleared.

[0064] Step S5: reconstruct the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code.

[0065] Among them, in the present application, the decomposed candidate solution codes are reconstructed according to the optimal candidate solution judgment rule, and the prevention of obtaining the optimal candidate solution code may include the following steps:

[0066] Step S51: combine all processor numbers and DVFS values ​​to obtain a reconstruction candidate solution set, wherein each solution in the reconstruction candidate solution set includes a processor code and a corresponding DVFS value;

[0067] Step S52, calculating the time and energy consumption required for scheduling tasks at the reconstructed position for each solution in the reconstruction candidate set, wherein the unreconstructed position still uses the candidate solution encoding before clearing;

[0068] Step S53, placing the solutions whose time and energy consumption required for the scheduling task are both less than the initial time and the initial energy consumption into the Pareto solution set at the reconstruction position;

[0069] Step S54: randomly select a solution in the Pareto solution set as the second layer code and the third layer code at the reconstruction position.

[0070] Specifically, in this application, Figure 7 This is a flow chart of the method proposed in this application for reconstructing the decomposed candidate solution encoding according to the optimal candidate solution judgment rule. Figure 7 As shown, the solutions in the reconstruction candidate solution set are {(0,0), (0,1), (0,2), (0,3), (1,0), (1,1),

[0071] (1, 2), (1, 3), (2, 0), (2, 1), (2, 2), (2, 3)}. Among them, when reconstructing task 0, each solution in the reconstructed candidate set is released from the second layer coding and the third layer coding of task 0 in turn, and the time and energy consumption required for scheduling tasks using the candidate decoding constructed by each solution are calculated. At this time, task 3 and task 4 still use the candidate decoding before clearing. Assuming that the time and energy consumption required for scheduling tasks using the candidate decoding reconstructed by (1, 2) and (2, 1) in the reconstructed candidate solution set are both less than the initial time and initial energy consumption, the Pareto solution set corresponding to task 0 is {(1, 2), (2, 1)}, then a solution in the Pareto solution set corresponding to task 0 can be randomly selected as the second layer coding and the third layer coding on task 0. And, in the present application, steps S51 to S54 are repeated to complete the reconstruction of tasks 3 and 4.

[0072] Step S6: If the time and energy consumption required to schedule the task using the optimal candidate decoding reach the threshold, the optimal candidate decoding is output.

[0073] In this application, the time and energy consumption required for scheduling tasks correspond to the time threshold and energy consumption threshold, respectively. If the time and energy consumption required for scheduling tasks using the optimal candidate decoding reach the time threshold and energy consumption threshold, respectively, the optimal candidate decoding is output. Otherwise, steps S3 to S5 are re-executed until the time and energy consumption required for scheduling tasks using the optimal candidate decoding reach the threshold.

[0074] The heterogeneous task scheduling method for rotating and swinging ultra-wideband satellites proposed in this application obtains the directed acyclic graph DAG of the task, generates candidate decoding according to the DAG, and calculates the initial time and initial energy consumption required for scheduling the task using the candidate decoding, and then generates a random number r. If r is less than the probability threshold p, the candidate decoding is decomposed, and then the decomposed candidate decoding is reconstructed according to the optimal candidate solution judgment rule to obtain the optimal candidate decoding. If the time and energy consumption required for scheduling the task using the optimal candidate decoding reach the threshold, the optimal candidate decoding is output. Among them, the candidate decoding represents the scheduling and resource allocation scheme of the task. Therefore, in the method provided by this application, when selecting the optimal task scheduling method, not only the time required for scheduling the task is considered, but also the energy consumption required for scheduling the task is considered, and the task scheduling is optimized with discrete multi-objectives, so that the satellite can achieve efficient processing of the task while consuming lower energy consumption. At the same time, this application avoids the coding conversion overhead in the continuous algorithm by adopting multi-level coding as the discretization operation operator, and reduces the complexity of the algorithm.

[0075] In addition, the method proposed in this application is based on Pareto to determine the optimal candidate solution, which improves the convergence of solving multi-objective problems.

[0076] Embodiment 2

[0077] Further, Figure 8 The structure diagram of a heterogeneous task scheduling device for a rotating swing-scan ultra-wideband satellite provided according to an embodiment of the present application is as follows: Figure 8 As shown, this may include:

[0078] An acquisition module 801 is used to acquire a directed acyclic graph (DAG) of a task, where the DAG includes an execution relationship of the task;

[0079] A generation module 802 generates candidate decodings according to the DAG, and calculates the initial time and initial energy consumption required for scheduling tasks using the candidate decodings, wherein the candidate decodings represent a scheduling and resource allocation scheme for the task;

[0080] The judgment module 803 is used to generate a random number r. If r is less than the probability threshold p, the step S4 is continued, where 0<r<1, 0<p<1;

[0081] A decomposition module 804, used for decomposing candidate decoding;

[0082] A processing module 805 is used to reconstruct the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code;

[0083] The output module 806 is configured to output the optimal candidate decoding if the time and energy consumption required for scheduling the task using the optimal candidate decoding both reach a threshold.

[0084] Among them, in this application, the above device also includes:

[0085] The mutation module 807 is used to mutate the candidate solution if r is greater than or equal to p, and continue to execute the above step S4 using the candidate solution obtained after the mutation.

[0086] Furthermore, in the present application, the above-mentioned output module is also used to re-execute the above-mentioned steps S3 to S5 if the time and energy consumption required for scheduling the task using the optimal candidate decoding do not reach the threshold, until the time and energy consumption required for scheduling the task using the optimal candidate decoding both reach the threshold.

[0087] In order to implement the above embodiments, the present disclosure also proposes a computer storage medium.

[0088] The computer storage medium provided in the embodiment of the present disclosure stores computer executable instructions; after the computer executable instructions are executed by the processor, the following can be achieved: Figure 1 The method shown.

[0089] In the heterogeneous task scheduling method, device and storage medium for rotating swing ultra-wideband satellites proposed in this application, the directed acyclic graph DAG of the task is obtained, and candidate decoding is generated according to the DAG, and the initial time and initial energy consumption required for scheduling the task using the candidate decoding are calculated, and then a random number r is generated. If r is less than the probability threshold p, the candidate decoding is decomposed, and then the decomposed candidate decoding is reconstructed according to the optimal candidate solution judgment rule to obtain the optimal candidate decoding. If the time and energy consumption required for scheduling the task using the optimal candidate decoding reach the threshold, the optimal candidate decoding is output. Among them, the candidate decoding represents the scheduling and resource allocation scheme of the task. Therefore, in the method provided by this application, when selecting the optimal task scheduling method, not only the time required for scheduling the task is considered, but also the energy consumption required for scheduling the task is considered, and the task scheduling is optimized with discrete multi-objectives, so that the satellite can achieve efficient processing of the task while consuming lower energy consumption. At the same time, this application avoids the coding conversion overhead in the continuous algorithm and reduces the complexity of the algorithm by adopting multi-level coding as the discretization operation operator.

[0090] In addition, the method proposed in this application is based on Pareto to determine the optimal candidate solution, which improves the convergence of solving multi-objective problems.

[0091] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0092] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code that includes one or more schedulable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be scheduled in the order shown or discussed, including scheduling functions in a substantially simultaneous manner or in reverse order based on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0093] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in the field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A heterogeneous task scheduling method for rotating and sweeping ultra-wideband satellites, characterized in that: The method comprises: S1. Obtain a directed acyclic graph (DAG) of the task, where the DAG includes the execution relationship of the task; S2. Generate candidate decodings according to the DAG, and calculate the initial time and initial energy consumption required for scheduling the task using the candidate decodings, wherein the candidate decodings represent the scheduling and resource allocation scheme of the task; S3. Generate random numbers r ,like r If it is less than the probability threshold p, then continue to execute step S4, where 0<r<1, 0<p<1; S4, decomposing the candidate decoding; S5, reconstructing the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code; S6. If the time and energy consumption required for scheduling the task using the optimal candidate decoding reach a threshold, outputting the optimal candidate decoding; The candidate decoding is a hierarchical coding, the first level of the hierarchical coding is the execution sequence of the task, the second level of the hierarchical coding is the processor sequence composed of processor numbers, the processor sequence corresponds one-to-one with the execution sequence of the task, and the third level of the hierarchical coding is the dynamic voltage frequency adjustment technology DVFS value of the processor; The reconstructing the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code comprises the following steps: Combining all processor numbers and DVFS values ​​to obtain a reconstruction candidate solution set, wherein each solution in the reconstruction candidate solution set includes a processor code and a corresponding DVFS value; Calculate the time and energy consumption required for scheduling the task at the reconstructed position for each solution in the reconstruction candidate set, wherein the unreconstructed position is still encoded using the candidate solution before clearing; The solution where the time and energy consumption required for scheduling the task are both less than the initial time and the initial energy consumption is placed in the Pareto solution set at the reconstruction position; A solution is randomly selected from the Pareto solution set as the second layer code and the third layer code at the reconstruction position.

2. The method according to claim 1, characterized in that Generating candidate decoding according to the DAG comprises the following steps: Dividing the execution levels of the tasks according to the pre-order and post-order constraints of the DAG; Randomly sort the tasks within each level to obtain the task sequence within each level; The task sequences in each level are combined in turn to obtain the first layer of codes in the candidate decodes, and then the second layer of codes and the third layer of codes in the candidate decodes are randomly generated.

3. The method according to claim 1, characterized in that like r If the probability is greater than or equal to the probability threshold, the candidate solution is mutated, and step S4 is continued using the candidate solution obtained after the mutation.

4. The method according to claim 1, characterized in that: The mutating the candidate solution comprises randomly selecting a position in the first layer code of the candidate solution, and randomly inserting the task at the position into the remaining positions between the preceding task and the succeeding task of the position.

5. The method according to claim 1, characterized in that Decomposing the candidate decoding comprises the following steps: Obtaining a task parameter d, wherein d represents a reserved portion of the candidate decoding, d is less than the task quantity value, and d is a positive integer; Randomly select d consecutive positions in the candidate solution codes, and clear the second layer codes and the third layer codes of the remaining positions.

6. The method according to claim 1, characterized in that The method further comprises: If the time and energy consumption required for scheduling the task using the optimal candidate decoding do not reach the threshold, steps S3 to S5 are executed again.

7. A heterogeneous task scheduling device for rotating and sweeping ultra-wideband satellites, characterized in that: The device comprises: An acquisition module, used to acquire a directed acyclic graph (DAG) of a task, wherein the DAG includes an execution relationship of the task; A generation module generates candidate decodings according to the DAG, and calculates the initial time and initial energy consumption required for scheduling the task using the candidate decodings, wherein the candidate decodings represent a scheduling and resource allocation scheme for the task; Judgment module, used to generate random numbers r ,like r If it is less than the probability threshold p, then continue to execute step S4, where 0<r<1, 0<p<1; A decomposition module, used for decomposing the candidate decoding; A processing module, used for reconstructing the decomposed candidate solution codes according to the optimal candidate solution judgment rule to obtain the optimal candidate solution code; An output module, configured to output the optimal candidate decoding if the time and energy consumption required for scheduling the task using the optimal candidate decoding reach a threshold; The candidate decoding is a hierarchical coding, the first level of the hierarchical coding is the execution sequence of the task, the second level of the hierarchical coding is the processor sequence composed of processor numbers, the processor sequence corresponds one-to-one with the execution sequence of the task, and the third level of the hierarchical coding is the dynamic voltage frequency adjustment technology DVFS value of the processor; The processing module is specifically used for: Combining all processor numbers and DVFS values ​​to obtain a reconstruction candidate solution set, wherein each solution in the reconstruction candidate solution set includes a processor code and a corresponding DVFS value; Calculate the time and energy consumption required for scheduling the task at the reconstructed position for each solution in the reconstruction candidate set, wherein the unreconstructed position is still encoded using the candidate solution before clearing; The solution where the time and energy consumption required for scheduling the task are both less than the initial time and the initial energy consumption is placed in the Pareto solution set at the reconstruction position; A solution is randomly selected from the Pareto solution set as the second layer code and the third layer code at the reconstruction position.

8. A computer storage medium, wherein: The computer storage medium stores computer executable instructions; after the computer executable instructions are executed by the processor, any method described in claims 1-6 can be implemented.

Images