A Networked Radar Task Scheduling Method Based on Adaptive Scheduling

Abstract

This invention discloses a network radar task scheduling method based on adaptive scheduling, mainly addressing the problem that existing network radar task scheduling methods generate a large amount of communication data while improving the time resource utilization of the network radar system. The implementation scheme is as follows: 1) Establish a network radar task scheduling constraint model; 2) Construct a task scheduling capacity calculation function; 3) Establish a network radar task scheduling optimization problem based on the task scheduling model and the task scheduling capacity function; 4) Calculate the comprehensive priority of tasks according to the optimization problem and sort the tasks; 5) Use an adaptive task scheduling algorithm to schedule the sorted tasks, obtaining the solution to the task scheduling optimization problem. The network radar task scheduling method of this invention reduces the number of times tasks for the same target are cross-scheduled on different radars, reduces the pressure on communication transmission, improves the real-time performance of task scheduling, and can be used in coherent cooperative network radar systems.

Description

Technical Field

[0001] This invention belongs to the field of radar technology, and specifically relates to a network radar task scheduling method that can be used in coherent cooperative network radar systems. Background Technology

[0002] With the increasing performance bottlenecks of single-station radars and their growing difficulties in responding to new threats, radar detection systems are shifting from single-radar detection to collaborative detection using networked radars. Compared to monostatic radars, networked radars can collect target scattering information from multiple dimensions, including spatial, frequency, and polarization dimensions, and have the potential to expand the functionality and performance of radar systems. To fully utilize the performance of networked radar systems and achieve a rational allocation of system resources among multiple radar tasks, designing an effective task scheduling algorithm is a crucial issue. In monostatic radar task scheduling algorithms, adaptive scheduling is a commonly used algorithm. To improve the success rate and time utilization of task scheduling, the concept of a time window is introduced, making task scheduling more flexible and effective. However, compared to monostatic radar task scheduling, networked radar task scheduling requires not only determining the actual execution time of the task but also the node where the task will be executed.

[0003] Patent application CN114912246A discloses a "task scheduling method for phased array networked radar". The task scheduling utility is composed of task priority and execution time. First, considering the characteristics of phased array networked radar, the maximization problem of phased array networked radar is formalized. Then, a heuristic algorithm for maximizing task scheduling utility in phased array networked radar is proposed, simultaneously considering task priority and expected execution time to solve the task scheduling problem in phased array networked radar.

[0004] In their paper "Beam Camping Scheduling for Distributed Radar Networking Systems Based on Pulse Interleaving," published in the 2022 Journal of Radars, Cheng Ting, Heng Siyu, Li Zhongzhu, and others proposed a real-time beam camping scheduling algorithm based on pulse interleaving. This algorithm introduces a time pointer vector, assigning the beam camping task with the highest overall priority at the time indicated by the time pointer vector to the radar node with the lowest interleaving time utilization, effectively reducing the time gaps introduced during scheduling. To determine whether the assigned beam camping task can be successfully scheduled and executed on the corresponding radar node, online pulse interleaving analysis is employed. This analysis introduces a time slot occupancy matrix and an energy consumption matrix to characterize the time and energy resource usage of each radar node, simplifying the interleaving analysis process and enabling interleaving between beam camping tasks with different pulse repetition periods and numbers.

[0005] The aforementioned network radar task scheduling method achieves better scheduling performance than independent scheduling of each radar node by uniformly allocating tasks and performing cross-scheduling, where the requesting node and execution node of a task are different. Cross-scheduling introduces additional communication data to the entire system. However, since cross-scheduling trades increased communication data for higher time utilization, this method only considers improving system time and resource utilization during task scheduling, neglecting the large amount of additional communication data generated during task scheduling. This places a huge transmission burden on the communication system and cannot guarantee the real-time performance of system task scheduling. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology and propose a network radar task scheduling method based on adaptive scheduling, so as to reduce the additional communication data generated during task scheduling, reduce the data transmission pressure on the communication system, and improve the real-time performance of system task scheduling.

[0007] To achieve the above objectives, the technical solution adopted by the present invention includes the following steps:

[0008] (1) Establish a network radar task scheduling constraint model:

[0009] Establish a system containing M radars I = {i1,i2,…,i...} m ,…,i M}, N tasks R = {r1, r2, ..., r n ,…,r N The task scheduling start time is t0, the task scheduling interval is SI, and the execution queue of M radars is Q = {q1, q2, ..., q}. m ,…,q M}, and a network radar task scheduling constraint model in which any two radar deployments satisfy coherence constraints, wherein:

[0010] i m Let q represent the m-th radar. m This represents the execution queue of the m-th radar, where m = 1, 2, ..., M, and M ≥ 2.

[0011] r n Let n represent the nth task, where n = 1, 2, ..., N, and N ≥ M.

[0012] r n ={P n N frn N ton ,t bn ,t en ,t sn ,t dwn ,t wn ,TFbn ,TF en}, P n N frn N ton t bn t en t sn t dwn t wn TF bn TF en Representing task r respectively n The priority of the work mode, the node from which the task request comes, the node to which the task is assigned, the expected start time, the task deadline, the actual execution time, the task dwell time, the time window width, the start time of the available time period, and the end time of the available time period.

[0013] (2) Construct the independent variable as the working mode priority p n and length of stay t dwn Task scheduling capacity calculation function f(p) n ,t dwn ):

[0014]

[0015] Where t and k represent the working mode priority p, respectively. n Duration of stay t dwn The proportion of the task scheduling capacity, t,k≥0,t+k=1;

[0016] (3) Establishing a network radar task scheduling optimization problem:

[0017] By setting constraints, we establish a network radar task scheduling optimization problem with the total scheduling capacity F of the network radar tasks as the objective function:

[0018]

[0019]

[0020]

[0021] Among them, constraints <1> Represents task r n Actual execution time t sn Within the time interval [t0, t0+SI], and satisfying the time window constraint t en -t wn <t sn <t en +t wn Constraints <2> This indicates that the length of the available time period is not less than that of task r. nThe dwell time; U(x) represents the step function, x = t0 + SI - t sn -t dwn ;

[0022] (4) Calculate and sort the overall task priority:

[0023] (4a) Based on the priority of work method P n Expected start time t bn Task deadline t en The overall priority of the tasks is jointly determined, and the task r is calculated using a hybrid priority calculation method. n Overall task priority PRI n ;

[0024] (4b) Set the task set R = {r1, r2, ..., r...} n ,…,r N}Based on overall task priority PRI n Arrange the tasks in descending order to obtain a new task set R' = {r'1, r'2, ..., r' n ,…,r' N};

[0025] (5) Use an adaptive task scheduling algorithm for task scheduling:

[0026] (5a) Determine if the new task set R' is empty:

[0027] If so, return to the task in the execution queue Q, and task scheduling ends;

[0028] Otherwise, select the task r with the highest overall priority from set R'. n ', and initialize the radar node set φ to empty, and execute step (5b);

[0029] (5b) Use an adaptive task scheduling algorithm to determine each radar i m Can task r be executed? n ':

[0030] If so, add the radar number m to the radar node set φ;

[0031] Otherwise, the radar node set φ remains unchanged;

[0032] (5c) Determine if set φ is empty:

[0033] If so, return to step (5a).

[0034] Otherwise, select a radar from set φ to perform task r. n ', Execute step (5d);

[0035] (5d) Task r n 'Add the selected radar to its execution queue and use an adaptive task scheduling algorithm to determine task r.' n 'Actual execution time t on the selected radar' sn The available time period of the selected radar [TF] bn ,TF en Updated to [TF] bn ,t sn ]、[t sn +t dwn ,TF en Return to step (5a).

[0036] Compared with the prior art, the present invention has the following advantages:

[0037] This invention establishes a task scheduling model based on coherence among radars, enabling the scheduling of the same task on any radar. Furthermore, by adding a mapping relationship from the requesting radar to the executing radar within the task scheduling model, an optimization problem for maximizing the task scheduling capacity of networked radars is established. Additionally, by employing a selection principle that first determines whether each radar can execute the task, then selects the radar to execute the task, prioritizing the requesting radar, and finally selecting the radar with the smallest sequence number, the number of times the same target task is scheduled on different radars is reduced.

[0038] Compared with existing technologies, this invention improves time utilization while reducing the amount of communication data generated during task scheduling, reducing the pressure on communication transmission, and improving the real-time performance of task scheduling. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the implementation of the present invention;

[0040] Figure 2 This is a simulation comparison chart of the time utilization rate of the present invention and the prior art;

[0041] Figure 3 This is a simulation comparison chart of the scheduling success rates of the present invention and existing technologies;

[0042] Figure 4 This is a simulation comparison chart of the total scheduling capacity of the present invention and existing technologies;

[0043] Figure 5 This is a simulation comparison chart of time utilization between the present invention and existing technologies under the same amount of communication data. Detailed Implementation

[0044] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0045] Reference Figure 1 The implementation steps for this example are as follows:

[0046] Step 1: Establish a network radar task scheduling constraint model.

[0047] Establish a network radar task scheduling constraint model comprising M radars (I), N tasks, task scheduling start time t0, task scheduling interval SI, and execution queue Q for the M radars, wherein the deployment of any two radars within the network satisfies coherence constraints, where:

[0048] I = {i1, i2, ..., i m ,…,i M}, R={r1,r2,…,r n ,,r N}, Q={q1,q2,…,q m ,,q M},

[0049] i m Let q represent the m-th radar. m This represents the execution queue of the m-th radar, where m = 1, 2, ..., M, and M ≥ 2.

[0050] r n Let n represent the nth task, where n = 1, 2, ..., N, and N ≥ M.

[0051] r n ={P n N frn N ton ,t bn ,t en ,t sn ,t dwn ,t wn ,TF bn ,TF en},

[0052] P n Represents task r n Work method priority, N frn Indicates the request task r n Radar serial number, N ton Indicates the execution of task r n Radar serial number, t bn Represents task r n Expected start time, t en Represents task r n Deadline, t sn Represents task r n Actual execution time, t dwn Represents task r nDuration of stay, t wn Represents task r n Time window width, TF bn Indicates the start time of the available time period, TF en Indicates the end time of the available time period;

[0053] The coherence between any two radars is a prerequisite for the execution of a mission targeting the same target on any radar. The deployment constraints for radar coherence are as follows:

[0054]

[0055] Where d represents the distance between different radars, L represents the distance from the target to the radar, λ represents the wavelength of the transmitted signal, and D represents the target size.

[0056] Step 2: Construct a task scheduling capacity calculation function.

[0057] Based on the priority principle and time utilization principle in task scheduling, namely, the higher the priority of the work mode, the larger the task scheduling capacity, and the higher the time utilization rate, the larger the task scheduling capacity, a task scheduling capacity calculation function f(p) of the following form is constructed. n ,t dwn ):

[0058]

[0059] Where t represents the priority of the working mode p n The proportion of task scheduling capacity, k represents the dwell time t. dwn The proportion of the task scheduling capacity, t,k≥0,t+k=1.

[0060] Step 3: Establish the network radar task scheduling optimization problem.

[0061] Under the above task scheduling model, in order to improve time utilization while reducing the amount of communication data generated during task scheduling of the networked radar system, and to obtain the task scheduling result with the maximum total task scheduling capacity, it is necessary to establish a networked radar task scheduling optimization problem, which is implemented as follows:

[0062] 3.1) Based on the task scheduling capacity calculation function in step 2, establish the objective function to maximize the total task scheduling capacity:

[0063]

[0064]

[0065] Where U(x) represents the step function, x = t0 + SI - t sn -t dwn;

[0066] 3.2) Set the first constraint condition according to the task scheduling time window constraint principle:

[0067] max(t0,t en -t wn )≤t sn ≤min(t en +t wn ,t0+SI-t dwn )

[0068] This constraint condition indicates that task r n Actual execution time t sn Within the time interval [t0, t0+SI], and task r n The actual execution time is within the time window.

[0069] 3.3) Set a second constraint based on the principle of length of stay:

[0070] TF bn ≤t sn <t sn +t dwi ≤TF en

[0071] This constraint states that the length of the available time period is not less than the task r. n The length of stay;

[0072] 3.4) Using the above objective function and constraints, the following network radar task scheduling optimization problem is formed:

[0073]

[0074] Step 4: Calculate and sort the overall task priority.

[0075] To solve the above optimization problem using an adaptive task scheduling algorithm and obtain the task scheduling result with the maximum scheduling capacity, it is necessary to first calculate the overall priority of N tasks and then sort the N tasks according to their overall priority. The specific implementation is as follows:

[0076] (4.1) Based on the priority P of the work method n Expected start time t bn Task deadline t en The task r is computed using a hybrid priority calculation method. n Overall task priority PRI n :

[0077] (4.1.1) Set the task set R = {r1, r2, ..., r...} n ,…,rN Task r in} n According to the priority of work method P n Expected start time t bn Task deadline t en Sort the results to obtain task r. n Sort sequence number q n l n d n ;

[0078] (4.1.2) Calculate the overall priority PRI of the tasks based on their sorted sequence numbers. n :

[0079] PRI n =aq n +bl n +cd n

[0080] Where 'a' represents the priority of the working mode P. n The weight in the overall task priority, b represents the expected start time t. bn The weight in the overall task priority, 'c' represents the task deadline 't'. en The proportions of each task in the overall priority are a,b,c≥0,a+b+c=1;

[0081] (4.2) Set the task set R = {r1, r2, ..., r...} n ,…,r N}Based on overall task priority PRI n Arrange the tasks in descending order to obtain a new task set R' = {r'1, r'2, ..., r' n ,…,r' N}

[0082] Step 5: Use an adaptive task scheduling algorithm to schedule tasks.

[0083] To obtain the result of the network radar task scheduling optimization problem in step (3), an adaptive task scheduling algorithm based on task comprehensive priority is needed to schedule the N tasks sorted in step (4). The specific implementation is as follows:

[0084] (5.1) Determine if the new task set R' is empty:

[0085] If so, return to the task in the execution queue Q, and task scheduling ends;

[0086] Otherwise, select the task r' with the highest overall priority from the new task set R'. n And initialize the radar node set φ to empty, and execute step (5.2);

[0087] (5.2) Use an adaptive task scheduling algorithm to sequentially determine each radar i m Can task r' be executed? n :

[0088] (5.2.1) Determine task r' n Whether it can be executed at the expected time, i.e., the time period [t] bn ,t bn +t dwn Is it available?

[0089] If so, then task r' n In radar i m The upper-level scheduling execution adds the radar number m to the radar node set φ;

[0090] Otherwise, proceed to step (5.2.2);

[0091] (5.2.2) Determine task r' n Available time period [TF] bn ,TF en The following constraints must be met:

[0092] TF bn ≤t sn <t sn +t dwn ≤TF en

[0093] Where max(t0,t) en -t wn )≤t sn ≤min(t en +t wn ,t0+SI-t dwn );

[0094] (5.2.3) Check if there is an available time period [TF] that meets the conditions of step (5.2.2). bn ,TF en ]:

[0095] If so, then determine task r' n In radar i m The system can schedule and execute the operation, and add the radar number m to the radar node set φ, and execute step (5.4);

[0096] Otherwise, task r' n In radar i m If the above cannot be scheduled for execution, the radar node set φ remains unchanged, and step (5.3) is executed;

[0097] (5.3) Determine if set φ is empty:

[0098] If so, return to step (5.1).

[0099] Otherwise, proceed to step (5.4);

[0100] (5.4) Determine the requesting task r' n Does the radar belong to set φ?

[0101] If so, then select request task r' n The radar performs this task;

[0102] Otherwise, select the radar with the smallest index in set φ to perform the task;

[0103] (5.5) Transfer task r' n Add the radar selected in step (5.4) to its execution queue, and use the adaptive task scheduling algorithm to determine task r'. n Actual execution time t on the selected radar sn The specific implementation is as follows:

[0104] (5.5.1) Determine the time period [t] bn ,t bn +t dwn Is it available?

[0105] If so, then task r' n Actual execution time t sn =t en ;

[0106] Otherwise, proceed to step (5.5.2);

[0107] (5.5.2) Determine whether the following constraints are satisfied and TF bn Minimum available time period [TF] bn ,TF en ]:

[0108] TF bn ≤t sn <t sn +t dwn ≤TF en

[0109] Where max(t0,t) en -t wn )≤t sn ≤min(t en +t wn ,t0+SI-t dwn );

[0110] (5.5.3) The time period [TF] obtained in step (5.5.2) bn,TF en ] start time TF bn As task r n 'Actual execution time t sn ;

[0111] (5.6) Select the available time period of the radar [TF] bn ,TF en Updated to [TF] bn ,t sn ]、[t sn +t dwn ,TF en Return to step (5.1).

[0112] The effects of the present invention will be further described below with reference to simulation experiments:

[0113] 1. Simulation Experiment Conditions

[0114] The simulation software environment is an Intel(R) Core(TM) i7-11800H CPU@2.30GHz (16 CPUs) and Matlab R2019a running Windows 11 Home Chinese Edition 64-bit operating system.

[0115] The simulation uses three radars, each with an equal number of request tasks. There are six types of request tasks, and the three radars randomly generate events of each type.

[0116] In this simulation, the overall task priority is given directly. In actual radar scheduling, task priority is calculated from target information. It is also assumed that the amount of communication data resulting from cross-scheduling of each task is the same. In the task capacity function, the weights for task priority and dwell time are m = 0.6 and k = 0.4.

[0117] The specific parameter settings for the task are shown in Table 1.

[0118] Table 1

[0119] Confirm Task 2 20 6 Lost the mission 5 30 5 Precision tracking 4 20 4 Normal tracking 2 50 3 High-priority search 4 30 2 Low priority search 6 50 1

[0120] 2. Simulation Experiment Content

[0121] Simulation Experiment 1: Under the simulation conditions described above, the task scheduling method of this invention, the existing task scheduling method for phased array networked radars, and the beam dwell scheduling method for distributed radar network systems based on pulse interleaving were used to schedule six types of tasks. The time resource utilization rate of the networked radar system was calculated, and their respective time utilization rate curves were plotted and compared with the time utilization rate curves when each radar independently scheduled tasks. The results are as follows: Figure 2 As shown.

[0122] from Figure 2It is evident that the time utilization rate of all three network radar task scheduling methods is higher than that of individual radar task scheduling. As the total number of tasks increases, the time utilization rates of the three task scheduling methods become similar.

[0123] Simulation Experiment 2: Under the above simulation conditions, the task scheduling method of this invention, the existing task scheduling method for phased array networked radars, and the beam dwell scheduling method for distributed radar network systems based on pulse interleaving were used to schedule six types of tasks. The success rate of task scheduling was calculated, and the respective task scheduling success rate curves were plotted and compared with the task scheduling success rate curves when each radar independently schedules tasks. The results are as follows: Figure 3 As shown.

[0124] from Figure 3 It is evident that when the total number of tasks is small, the task scheduling success rate of the three network radar task scheduling methods is higher than that of the task scheduling success rate of each radar independently scheduling tasks. When the total number of tasks is large, the task scheduling success rate of the present invention is higher than that of the prior art.

[0125] Simulation Experiment 3: Under the above simulation conditions, the task scheduling method of this invention, the existing task scheduling method for phased array networked radars, and the beam dwell scheduling method for distributed radar network systems based on pulse interleaving were used to schedule six types of tasks. The total task scheduling capacity was calculated, and the total task scheduling capacity curves for each were plotted and compared with the total task scheduling capacity curves when each radar independently schedules tasks. The results are as follows: Figure 4 As shown.

[0126] from Figure 4 It is evident that the total task scheduling capacity of the three network radar task scheduling methods is higher than that of the task scheduling capacity of each radar independently scheduling tasks, but the total task scheduling capacity of the present invention is the highest.

[0127] Simulation Experiment 4: Under the above simulation conditions, the task scheduling method of this invention, the existing task scheduling method for phased array networked radar, and the beam dwell scheduling method for distributed radar network systems based on pulse interleaving were used to schedule six types of tasks. The ratio of the time utilization increment to the number of task cross-scheduling times for each method was calculated.

[0128]

[0129] Based on the calculation results, plot the curves showing the ratio of the time utilization increment to the number of task cross-scheduling operations for each method, such as... Figure 5 As shown.

[0130] from Figure 5It is evident that among the three network radar task scheduling methods, the ratio of the time utilization increment to the number of task cross-scheduling times of the present invention is the largest, indicating that when the time resource utilization of the network radar system is the same, the task scheduling method of the present invention generates the least amount of communication data during task scheduling.

[0131] The simulation experiments above show that, compared with the prior art, the present invention has higher time utilization, task scheduling success rate and total task scheduling capacity. While improving time utilization, it reduces the number of times the same target task is cross-scheduled on different radars, reduces the amount of communication data generated during task scheduling, and reduces the communication transmission pressure of the networked radar system. Under the condition of unchanged communication transmission rate, it reduces the time for transmitting task scheduling information and improves the real-time performance of task scheduling.

Claims

1. A network radar task scheduling method based on adaptive scheduling, characterized in that, Includes the following steps: (1) Establish a network radar task scheduling constraint model: Establish a system containing M radars I = {i1,i2,…,i...} m ,…,i M }, N tasks R = {r1, r2, ..., r n ,…,r N The task scheduling start time is t0, the task scheduling interval is SI, and the execution queue of M radars is Q = {q1, q2, ..., q}. m ,…,q M }, and a network radar task scheduling constraint model in which any two radar deployments satisfy coherence constraints, wherein: i m Let q represent the m-th radar. m This represents the execution queue of the m-th radar, where m = 1, 2, ..., M, and M ≥ 2. r n Let n represent the nth task, where n = 1, 2, ..., N, and N ≥ M. r n ={P n N frn N ton ,t bn ,t en ,t sn ,t dwn ,t wn ,TF bn ,TF en }, P n N frn N ton t bn t en t sn t dwn t wn TF bn TF en Representing task r respectively n The priority of the work mode, the node from which the task request comes, the node to which the task is assigned, the expected start time, the task deadline, the actual execution time, the task dwell time, the time window width, the start time of the available time period, and the end time of the available time period. (2) Construct the independent variable as the working mode priority p n and length of stay t dwn Task scheduling capacity calculation function f(p) n ,t dwn ): Where t and k represent the working mode priority p, respectively. n Duration of stay t dwn The proportion of the task scheduling capacity, t,k≥0,t+k=1; (3) Establishing a network radar task scheduling optimization problem: By setting constraints, we establish a network radar task scheduling optimization problem with the total scheduling capacity F of the network radar tasks as the objective function: Among them, constraints <1> Indicates task r n Actual execution time t sn Within the time interval [t0, t0+SI], and satisfying the time window constraint t en -t wn <t sn <t en +t wn Constraints <2> This indicates that the length of the available time period is not less than that of task r. n The dwell time; U(x) represents the step function, x = t0 + SI - t sn -t dwn ; (4) Calculate and sort the overall task priority: (4a) Based on the priority of work method P n Expected start time t bn Task deadline t en The overall priority of the tasks is determined jointly, and the task r is calculated using a hybrid priority calculation method. n Overall Task Priority (PRI) n ; (4b) Set the task set R = {r1, r2, ..., r...} n ,…,r N }Based on overall task priority PRI n Arrange the tasks in descending order to obtain a new task set R′={r1′,r′2,…,r′ n ,…,r′ N }; (5) Use an adaptive task scheduling algorithm for task scheduling: (5a) Determine if the new task set R′ is empty: If so, return to the task in the execution queue Q, and task scheduling ends; Otherwise, select the task r′ with the highest overall priority for the current task from set R′. n And initialize the radar node set φ to empty, and execute step (5b); (5b) Use an adaptive task scheduling algorithm to determine each radar i m Can task r′ be executed? n : If so, add the radar number m to the radar node set φ; Otherwise, the radar node set φ remains unchanged; (5c) Determine if set φ is empty: If so, return to step (5a). Otherwise, select a radar from set φ to perform task r′. n Execute step (5d); (5d) Task r′ n Add the selected radar to its execution queue and use an adaptive task scheduling algorithm to determine task r′. n Actual execution time t on the selected radar sn The available time period of the selected radar [TF] bn ,TF en Updated to [TF] bn ,t sn ]、[t sn +t dwn ,TF en Return to step (5a).

2. The method according to claim 1, characterized in that, The coherence constraint conditions that any two radar stations in step (1) must satisfy are expressed as follows: Where d represents the distance between different radars, L represents the distance from the target to the radar, λ represents the wavelength of the transmitted signal, and D represents the target size.

3. The method according to claim 1, characterized in that, In step (4a), the combined task priority PRI is calculated using the hybrid priority calculation method. n The implementation steps are as follows: (4a1) Let the task set R = {r1,r2,…,r} n ,…,r N Task r in} n According to the priority of work method P n Expected start time t bn Task deadline t en Sort the results to obtain task r. n Sort sequence number q n l n d n ; (4a2) Calculate the overall task priority (PRI) based on the sorted task numbers. n : PRI n =aq n +bl n +cd n Where a, b, and c represent the working mode priority P, respectively. n Expected start time t bn Task deadline t en The weight of each task in the overall priority is a,b,c≥0,a+b+c=1.

4. The method according to claim 1, characterized in that, In step (5b), an adaptive task scheduling algorithm is used to determine each radar i m Can task r′ be executed? n The implementation steps are as follows: (5b1) Determine task r′ n Whether it can be executed at the expected time, i.e., the time period [t] bn ,t bn +t dwn Is it available? If so, then it is classified as task r′. n In radar i m The upper-level scheduling execution; Otherwise, proceed to step (5b2); (5b2) Determine task r′ n Available time period [TF] bn ,TF en The constraints TF must be satisfied. bn ≤t sn <t sn +t dwn ≤TF en Task r′ n Actual execution time t sn The constraint condition that is satisfied is max(t0,t). en -t wn )≤t sn ≤min(t en +t wn ,t0+SI-t dwn ); (5b3) Check if there is an available time period that meets the conditions of step (5b2) [TF] bn ,TF en ]: If so, then it is classified as task r′. n In radar i m The upper-level scheduling execution; Otherwise, classify it as task r′ n In radar i m It cannot be scheduled for execution.

5. The method according to claim 1, characterized in that, In step (5c), a radar is selected from the set φ to perform task r′. n It is by judging the request task r′ n The selection is based on whether the radar belongs to set φ. If request task r′ n If the radar belongs to set φ, then the requested task r′ is selected. n Radar; Otherwise, select the radar with the smallest index in set φ.

6. The method according to claim 1, characterized in that, In step (5d), an adaptive task scheduling algorithm is used to determine task r′. n Actual execution time t on the selected radar sn The implementation steps are as follows: (5d1) Determine the time period [t] bn ,t bn +t dwn Is it available? If so, then task r′ n Actual execution time t sn =t en ; Otherwise, proceed to step (5d2); (5d2) Determine whether the following constraints are satisfied and TF bn Minimum available time period [TF] bn ,TF en ]: TF bn ≤t sn <t sn +t dwn ≤TF en max(t0,t en -t wn )≤t sn ≤min(t en +t wn ,t0+SI-t dwn ); (5d3) The time period [TF] obtained in step (5d2) bn ,TF en ] start time TF bn As task r′ n Actual execution time t sn .

Images