Multi-attribute task scheduling method with timeliness guarantee and application thereof

Abstract

The invention relates to a multi-attribute task scheduling method with timeliness guarantee and application thereof. The multi-attribute task scheduling method is applied to pervasive edge computing. A distributed computing power sensing mechanism is adopted, an IPv6 expansion header is designed, computing power sensing is achieved in a broadcast mode, and task equipment is generated to detect surrounding resources and load conditions; according to the method, battlefield tasks are subjected to multi-attribute tasks according to actual conditions, an unloading decision is made by comprehensively considering factors such as calculation, storage, transmission and load conditions by using multi-attribute decisions according to different task types, and a TOP3 pre-inquiry process is added. According to the multi-attribute task scheduling method with timeliness guarantee and reference thereof, resources in a battlefield environment are fully utilized, the load condition in a battlefield scene is optimized, stable task unloading is guaranteed, the deterministic time delay requirement of a time delay sensitive task is met, and the task completion rate is increased.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a multi-attribute task scheduling method with timeliness guarantee and application thereof. Background Art

[0002] The popularity of mobile Internet and the rapid development of Internet of Things technology have led to an explosive growth in mobile communication traffic. More and more mobile applications are trying to complete more complex logical computing functions on the terminal. In order to meet the needs of various businesses for low latency, high reliability, and privacy protection, edge computing has emerged. In some scenarios where there is a lack of infrastructure (such as battlefields in the wild), due to environmental limitations, there is no centralized management and the terminal device can only rely on its own resources and perception capabilities for computing and storage. The concept of pervasive edge computing has emerged. Pervasive edge computing refers to a type of edge computing that relies only on edge devices with perception, storage, and communication capabilities to achieve point-to-point offloading without the need for centralized management.

[0003] In a field battlefield environment lacking infrastructure, the resources of various devices are unevenly distributed. For example, the wearable devices (smart watches, helmets, etc.) on soldiers have very limited resources, while various vehicles (such as tanks, infantry fighting vehicles, etc.) have more powerful resources and processing capabilities. If the resources of terminal devices are shared, and resource-rich devices assist resource-poor devices to complete tasks, the efficiency of task completion can be improved and the problem of some devices having difficulty completing computing tasks can be solved.

[0004] The essence of edge computing is to offload tasks from terminal devices to edge servers or remote clouds. Pervasive edge computing relies solely on edge devices with sensing, storage, and communication capabilities to achieve point-to-point offloading without the need for centralized management, and is more suitable for scenarios where servers are lacking. Existing technologies only consider the deadline in task division, and only consider the minimum uplink and download time in offloading decisions. The types of tasks in actual battlefield environments are numerous and complex, and it is necessary to consider task attributes and resource types in many aspects to make offloading decisions. In terms of computing power perception, due to the scarcity of resources in field scenarios, high environmental dynamics, and diverse task types, it is necessary to make full use of the resources in the battlefield environment, optimize the load conditions in the battlefield scenario, and ensure the stable progress of task offloading. However, current research cannot solve this problem well. Summary of the invention

[0005] The purpose of the present invention is to provide a multi-attribute task scheduling method with timeliness guarantee, which fully utilizes the resources in the battlefield environment, optimizes the load situation in the battlefield scene, ensures the stable progress of task offloading, meets the deterministic delay requirements of delay-sensitive tasks, and improves the task completion rate.

[0006] In order to solve the above technical problems, the present invention provides a multi-attribute task scheduling method with timeliness guarantee, comprising:

[0007] Conduct integrated measurement of computing power and develop a distributed computing power perception mechanism;

[0008] The mobile terminal generates tasks, classifies the tasks according to their attributes based on the actual battlefield situation, and designs the IPv6 header to encapsulate the task attributes and computing resource attributes;

[0009] Analyze the delay during the unloading process of the task;

[0010] After the device generates a task, it makes an offloading decision based on the task type and perception data;

[0011] According to the uninstallation decision, an uninstallation process is performed, and a pre-inquiry is performed before the uninstallation process to ensure the timeliness of the task.

[0012] In a preferred embodiment of the present invention, the fusion measurement of computing power includes:

[0013] Computing and network resources are packaged into computing power and integrated into metrics. The metrics of the integrated metrics include:

[0014]

[0015] Among them, comp is a computing metric, CPU Cycle is a CPU cycle, FLOPS is the number of floating-point operations per second, CPI is the number of cycles per instruction, RT is the response time, Stor is a storage metric, Memory is the storage capacity, IOPS is the number of input and output per second, Throughout is the throughput, Del is a transmission metric, Bandwidth is the transmission bandwidth, Jitter is the jitter parameter, Delay is the delay parameter, and Capacity is the capacity parameter;

[0016] The distributed computing power perception mechanism includes:

[0017] The central node abstracts its own computing power information and location information, and notifies the computing power information and location information to the surrounding nodes within the communication range at a set time interval to realize the computing power perception process;

[0018] Each node maintains a computing power information table, and each node has a topological map of the surrounding computing power:

[0019] Hashrate Matrix

[0020] Among them, ID represents the unique identification ID of the node, C, S, and D represent the computing, storage, and transmission power information of the node, loc represents the location information of the node, and w represents the estimated waiting time in the scheduling queue of the node.

[0021] Specifically, the device notifies the surrounding nodes of its own computing power attributes at specified time intervals. Each device owns and maintains a computing power resource topology map to maintain awareness of surrounding resources.

[0022] Each node maintains a computing power information table to update and record the computing power information of the current surrounding environment; when the node receives the computing power information from the central node, it records it.

[0023] In a preferred embodiment of the present invention, the mobile terminal generates tasks, and classifies the tasks according to attributes according to the actual battlefield situation, including the following steps: classifying the services that may actually be generated, the types of which include: soldier information, environmental monitoring, collaborative combat, reconnaissance target identification, and intrusion detection;

[0024] Or / and, set the ipv6 header, encapsulate the task attributes and computing resource attributes, including the following steps: After the task is classified, different decision functions are formulated for different businesses to select the unloading node.

[0025] Specifically, considering that different types of services have different preferences for computing, storage, transmission and other attributes, in order to evenly load the network and improve network resource utilization and task completion rate.

[0026] In a preferred embodiment of the present invention, the specific fields of the extended header in the ipv6 header include: TS, TkT, tID, sID, C, S, D, pos_info, T ddl ; Among them, the TKT field stores the task type; the TS field is used to distinguish the message type and status; when the TS value is 0, it means that the message is a computing power information message, which is used by the central node to notify its own resource information. When the TS value is not 0, the message is a task message, which is used for information exchange between the task node and the central node when unloading tasks. Different TS values ​​represent different task states;

[0027] The tID field stores the unique ID of the task, the sID field stores the TN node ID that generates the task, and the TS field stores the task status. ddl The field stores the maximum waiting time of the task; the pos_info field stores the location information of the node; the C field stores the computing resources required to complete the task, and the S field stores the size of the task required to complete the task.

[0028] In a preferred embodiment of the present invention, after the device generates a task, an unloading decision is made according to the task type and the perception data, including:

[0029] The weight values ​​of different task types based on different attributes are given through the analytic hierarchy process (AHP);

[0030] Construct a judgment matrix A based on relative importance;

[0031]

[0032] According to the judgment matrix A, the weights of each attribute are obtained by arithmetic mean method:

[0033]

[0034] Use the TOPSIS method to judge the advantages and disadvantages of the available solutions and finally come up with the unloading solution;

[0035] The node calculates and extracts the computing power, storage, transmission, queue waiting time and distance from itself of the surrounding nodes based on the collected computing power information, and obtains the original data matrix:

[0036]

[0037] Normalize the matrix:

[0038]

[0039] Get the normalized matrix:

[0040]

[0041] The optimal solution Z is obtained by standardizing the matrix + and the worst solution Z - ,in It is composed of the maximum value of each column element in Z, It is composed of the minimum value of each column element in Z;

[0042] According to Z + , Z - And ω obtained by formula (4) i Calculate the distance between each evaluation object and them and Then calculate how close the solution is to the optimal solution:

[0043]

[0044]

[0045] Where 0≤C i≤1,C i →1 indicates that the evaluation object is better; according to C i Sort and give results; through comprehensive evaluation method and hierarchical analysis method, each task gets targeted decision according to the task attributes and the surrounding resource conditions at that time.

[0046] Specifically, the TOPSIS method, or Technique for Order Preference by Similarity to Ideal Solution, is a commonly used comprehensive evaluation method that can make full use of the information of the original data, and the results can accurately reflect the gap between the evaluation schemes. AHP, or Analytic Hierarchy Process, is a decision analysis method that classifies and layers complex problems to reduce the complexity of decision making.

[0047] In a preferred embodiment of the present invention, the delay in the unloading process of the task is analyzed, wherein the task delay analysis includes: calculation delay, communication delay, queue waiting delay;

[0048] The computational delay includes: at a certain moment, the task node i generates task t and offloads it to the central node j for execution. The computational time required to complete the task on j can be expressed as:

[0049]

[0050] Among them, S t is the data size of the task, C t Refers to the computing resources required per MB to complete the task, C j is the computing resource allocated by the central node j to task t, in GHz;

[0051] or / and,

[0052] The communication delay includes: when the task node encounters a task that needs to be unloaded, it needs to select a suitable central node to unload the task to; accordingly, after the CN node calculates the task, it sends the result back to the TN node, and the proposed task is transmitted in TNi and CNj; in this process, the communication delay in the node consists of two parts, the sending delay and propagation delay X(i,j);

[0053] Propagation delay refers to the time it takes for an electromagnetic signal to propagate a distance D in a propagation medium. They are expressed as:

[0054]

[0055]

[0056] Among them, B tis the bandwidth, R is the propagation speed of electromagnetic waves in the current medium; D(i,j) is the physical distance between node i and node j;

[0057] or / and,

[0058] The queue waiting delay includes: the estimated processing time of the new task and queue delay It is expressed as:

[0059]

[0060]

[0061] Where m is the number of tasks in the queue when task t enters the pending queue of CN j, and α is the proportional coefficient of the read and write time required for a unit of data volume;

[0062] For the tth task generated by TN i, its actual service delay is:

[0063]

[0064] Where l≥4, the specific value of l is related to D t The number of times the request is forwarded is related to is the offloading function. When task t is finally offloaded to CN j, When offloading to other CN nodes,

[0065] In a preferred embodiment of the present invention, an uninstallation process is performed according to the uninstallation decision, and a pre-inquiry is performed before the uninstallation process to ensure the timeliness of the task, including:

[0066] According to the offloading decision result, the TOP-3 solutions pm = {x 1 ,x 2 ,x 3}, the plan records the three most suitable CN nodes for unloading; in order to ensure the timeliness of the task, a pre-query method is used to query node x 1 Send inquiry message, sID 1 Estimated waiting time for the task after receiving the inquiry message and estimated processing time If the task can be completed within the deadline, a confirmation message is returned, otherwise the task is forwarded to node x. 2 , x 2 After receiving the inquiry message, determine whether it can be completed in the same way; and so on;

[0067] When all three nodes cannot complete the task, node i will not receive a response from CN and decides to abandon the task after exceeding the task deadline event;

[0068] When the three nodes can complete the task, if i receives any confirmation message, it will immediately unload the task and start the final unloading process.

[0069] In a preferred embodiment of the present invention, a multi-attribute task scheduling method and device with timeliness guarantee includes:

[0070] The fusion measurement module is used to perform fusion measurement on computing power and formulate a distributed computing power perception mechanism;

[0071] The task generation module is used to generate tasks for mobile terminals, classify tasks according to their attributes based on the actual battlefield situation, and design the IPv6 header to encapsulate the task attributes and computing resource attributes;

[0072] The delay analysis module is used to analyze the delay in the unloading process of the task;

[0073] The offloading decision module is used to make offloading decisions based on the task type and perception data after the device generates a task;

[0074] The pre-inquiry uninstallation module is used to perform the uninstallation process according to the uninstallation decision, and to perform a pre-inquiry before the uninstallation process to ensure the timeliness of the task.

[0075] In a preferred embodiment of the present invention, a device for scheduling multi-attribute tasks with timeliness guarantee includes:

[0076] Memory;

[0077] processor;

[0078] as well as

[0079] Computer programs;

[0080] The computer program is stored in the memory and is configured to be executed by the processor to implement the multi-attribute task scheduling method with timeliness guarantee.

[0081] In a preferred embodiment of the present invention, a storage medium for multi-attribute task scheduling with timeliness guarantee is provided, characterized in that a computer program is stored thereon, and when the computer program is executed by a processor, the multi-attribute task scheduling method with timeliness guarantee is implemented.

[0082] In a preferred embodiment of the present invention, the nodes in the scene without network infrastructure are defined as two categories, one is the task nodes TN (Task Nodes), which are numerous and have high mobility, but lack computing resources and are basically unable to process the tasks generated by themselves, but can complete the forwarding of tasks; for example, the smart wearable devices of soldiers. The other is the center nodes CN (Centre Nodes), which are rich in computing resources, but the number is not less than that of the task nodes TN, and are used to assist the task nodes TN in completing tasks. Such as tanks, infantry fighting vehicles and other vehicles. The center nodes CN and task nodes TN are represented by the sets K = {1,2,3,…,k} and N = {1,2,3,…,n} respectively. Tasks are abstractly represented as octets in the system: D t ={tID,sID,TS,C t ,S t ,T ddl ,TkT}, where tID is the unique ID of the task, sID is the ID of the TN node that generates the task, TS represents the state of the task, C t ,S t are the computing resources required to complete the task and the size of the task, T ddl Represents the maximum waiting time for the task. In fact, the TN node that generates the task will wait slightly longer than T ddl The source node waits for the response from the CN node for a certain period of time, but a timeout of too long will cause the source node to abandon the task. TkT represents the type of the task.

[0083] The present invention solves the defects existing in the background technology, and the beneficial effects of the present invention are as follows:

[0084] A multi-attribute task scheduling method with timeliness guarantee and its application; fully utilize the resources in the battlefield environment, optimize the load situation in the battlefield scene, ensure the stable progress of task offloading, meet the deterministic delay requirements of delay-sensitive tasks, and improve the task completion rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0086] Figure 1 A framework diagram of a task scheduling method provided for a preferred embodiment of the present invention;

[0087] Figure 2A network environment diagram of ubiquitous edge computing provided by a preferred embodiment of the present invention;

[0088] Figure 3 A design diagram of the IPv6 packet extension part provided for the preferred embodiment of the present invention;

[0089] Figure 4 A task offloading flow chart of a multi-attribute task scheduling method with timeliness guarantee provided in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0090] The core of the present invention is to provide a multi-attribute task scheduling method with timeliness guarantee, which can meet the deterministic delay requirements of delay-sensitive tasks and improve the task completion rate in the scenario.

[0091] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0092] Embodiment 1

[0093] A multi-attribute task scheduling method with timeliness guarantee, comprising:

[0094] Conduct integrated measurement of computing power and develop a distributed computing power perception mechanism.

[0095] Specifically, the integrated measurement of computing power includes:

[0096] Computing and network resources are packaged into computing power and integrated into metrics. The metrics of the integrated metrics include:

[0097]

[0098] Among them, comp is a computing metric, CPU Cycle is a CPU cycle, FLOPS is the number of floating-point operations per second, CPI is the number of cycles per instruction, RT is the response time, Stor is a storage metric, Memory is the storage capacity, IOPS is the number of input and output per second, Throughout is the throughput, Del is a transmission metric, Bandwidth is the transmission bandwidth, Jitter is the jitter parameter, Delay is the delay parameter, and Capacity is the capacity parameter;

[0099] The distributed computing power perception mechanism includes:

[0100] The central node abstracts its own computing power information and location information, and notifies the computing power information and location information to the surrounding nodes within the communication range at a set time interval to realize the computing power perception process;

[0101] Each node maintains a computing power information table, and each node has a topological map of the surrounding computing power:

[0102]

[0103] Specifically, the device notifies the surrounding nodes of its own computing power attributes at specified time intervals. Each device owns and maintains a computing power resource topology map to maintain awareness of surrounding resources.

[0104] Each node maintains a computing power information table to update and record the computing power information of the current surrounding environment; when the node receives the computing power information from the central node, it records it.

[0105] The mobile terminal generates tasks, classifies the tasks according to their attributes based on the actual battlefield situation, and designs the IPv6 header to encapsulate the task attributes and computing resource attributes.

[0106] Specifically, the mobile terminal generates tasks, and classifies the tasks according to attributes based on the actual battlefield situation, including the following steps: Taking into account that the types of tasks generated by each device are different, the actual possible services are classified into types including: soldier information, environmental monitoring, collaborative operations, reconnaissance target identification, and intrusion detection.

[0107] Designing the IPv6 header to encapsulate the task attributes and computing resource attributes includes the following steps:

[0108] After task classification, different decision functions are formulated to select the offloading nodes for different businesses.

[0109] Specifically, considering that different types of services have different preferences for computing, storage, transmission and other attributes, in order to make the network load uniform and improve network resource utilization and task completion rate. The specific fields of the extended header in the ipv6 header include: TS, TkT, tID, sID, C, S, D, pos_info, T ddl; Among them, the TKT field stores the task type; the TS field is used to distinguish the message type and status; when the TS value is 0, it means that the message is a computing power information message, which is a message used by the central node to notify its own resource information. When the TS value is not 0, the message is a task message, which is used for information exchange between the task node and the central node when unloading tasks. Different TS values ​​represent different task states; the tID field stores the unique identification ID of the task, the sID field stores the TN node ID that generates the task, and the TS field stores the task status. ddl The field stores the maximum waiting time for the task.

[0110] Analyze the delay during the offloading process in the task.

[0111] Specifically, the delay in the task unloading process is analyzed, where the task delay analysis includes: calculation delay, communication delay, and queue waiting delay;

[0112] Calculation delay includes:

[0113] At a certain moment, it is planned that task node i generates task t and offloads it to central node j for execution. The computing time required to complete the task on j can be expressed as:

[0114]

[0115] Among them, S t is the data size of the task, C t Refers to the computing resources required per MB to complete the task, C j It is the computing resource allocated by the central node j to task t, in GHz.

[0116] Communication delays include:

[0117] When a task node encounters a task that needs to be offloaded, it needs to select a suitable central node to offload the task to. Accordingly, after the CN node calculates the task, it sends the result back to the TN node, and the task is planned to be transmitted in TN i and CN j. In this process, the communication delay in the node consists of two parts: the sending delay and the and propagation delay X(i,j);

[0118] Propagation delay refers to the time it takes for an electromagnetic signal to propagate a distance D in a propagation medium. They are expressed as:

[0119]

[0120]

[0121] Among them, B tis the bandwidth, R is the propagation speed of electromagnetic waves in the current medium; D(i,j) is the physical distance between node i and node j.

[0122] The queue waiting delay includes:

[0123] Estimated processing time for new tasks and queue delay It is expressed as:

[0124]

[0125]

[0126] Where m is the number of tasks in the queue when task t enters the pending queue of CN j, and α is the proportional coefficient of the read and write time required for a unit of data volume;

[0127] For the tth task generated by TN i, its actual service delay is:

[0128]

[0129] Where l≥4, the specific value of l is related to D t The number of times the request is forwarded is related to is the offloading function. When task t is finally offloaded to CN j, When offloading to other CN nodes,

[0130] After the device generates a task, it makes an offloading decision based on the task type and perception data.

[0131] Specifically, after the device generates a task, it makes an offloading decision based on the task type and perception data, including:

[0132] The weight values ​​of different task types based on different attributes are given through the analytic hierarchy process (AHP);

[0133] Construct a judgment matrix A based on relative importance;

[0134]

[0135] According to the judgment matrix A, the weights of each attribute are obtained by arithmetic mean method:

[0136]

[0137] Use the TOPSIS method to judge the advantages and disadvantages of the available solutions and finally come up with the unloading solution;

[0138] The node calculates and extracts the computing power, storage, transmission, queue waiting time and distance from itself of the surrounding nodes based on the collected computing power information, and obtains the original data matrix:

[0139]

[0140] Normalize the matrix:

[0141]

[0142] Get the normalized matrix:

[0143]

[0144] The optimal solution Z is obtained by standardizing the matrix + and the worst solution Z - ,in It is composed of the maximum value of each column element in Z, It is composed of the minimum value of each column of elements in Z.

[0145] According to Z + , Z - And ω obtained by formula (4) i Calculate the distance between each evaluation object and them and Then calculate how close the solution is to the optimal solution:

[0146]

[0147]

[0148] Where 0≤C i ≤1,C i →1 indicates that the evaluation object is better; according to C i Sort and give the results; through the AHP and TOPSIS methods, each task gets a targeted decision based on the task attributes and the surrounding resource conditions at the time.

[0149] According to the uninstallation decision, the uninstallation process is performed, and a pre-query is performed before the uninstallation process to ensure the timeliness of the task. Specifically, it includes: according to the uninstallation decision result, the TOP-3 solutions pm={x 1 ,x 2 ,x 3}, the plan records the three most suitable CN nodes for unloading; in order to ensure the timeliness of the task, a pre-query method is used to query node x 1 Send inquiry message, sID 1 Estimated waiting time for the task after receiving the inquiry message and estimated processing time If the task can be completed within the deadline, a confirmation message is returned, otherwise the task is forwarded to node x. 2 , x 2 After receiving the inquiry message, determine whether it can be completed in the same way; and so on;

[0150] When all three nodes cannot complete the task, node i will not receive a response from CN and decides to abandon the task after exceeding the task deadline event;

[0151] When the three nodes can complete the task, if i receives any confirmation message, it will immediately unload the task and start the final unloading process.

[0152] Embodiment 2

[0153] Based on the first embodiment, a multi-attribute task scheduling method and device with timeliness guarantee includes:

[0154] The fusion measurement module is used to perform fusion measurement on computing power and formulate a distributed computing power perception mechanism;

[0155] The task generation module is used to generate tasks for mobile terminals, classify tasks according to their attributes based on the actual battlefield situation, and design the IPv6 header to encapsulate the task attributes and computing resource attributes;

[0156] The delay analysis module is used to analyze the delay in the unloading process of the task;

[0157] The offloading decision module is used to make offloading decisions based on the task type and perception data after the device generates a task;

[0158] The pre-inquiry uninstallation module is used to perform the uninstallation process according to the uninstallation decision, and to perform a pre-inquiry before the uninstallation process to ensure the timeliness of the task.

[0159] Embodiment 3

[0160] Based on the first embodiment, a device for scheduling multi-attribute tasks with timeliness guarantee includes:

[0161] Memory;

[0162] processor;

[0163] as well as

[0164] Computer programs;

[0165] The computer program is stored in the memory and is configured to be executed by the processor to implement the multi-attribute task scheduling method with timeliness guarantee.

[0166] Embodiment 4

[0167] Based on the first embodiment, a storage medium for multi-attribute task scheduling with timeliness guarantee is provided, characterized in that a computer program is stored thereon, and when the computer program is executed by a processor, the multi-attribute task scheduling method with timeliness guarantee is implemented.

[0168] Embodiment 5

[0169] Please refer to Figure 1 , Figure 1 A flowchart of a multi-attribute task scheduling method with timeliness guarantee provided by an embodiment of the present invention; the method may include:

[0170] Step s110: CN generates a computing power perception information message and notifies it to surrounding nodes. After receiving the message, the node stores the information in its own resource topology map and saves the surrounding computing power information. The attributes of the task are abstracted as follows: {tID, sID, TS, C t ,S t ,T ddl ,TkT}. Among them, tID is the unique ID of the task, sID is the ID of the TN node that generates the task, TS represents the state of the task, C t ,S t are the computing resources required to complete the task and the size of the task, T ddl Represents the maximum waiting time for the task. In fact, the TN node that generates the task will wait slightly longer than T ddl The source node waits for the response of the CN node for a certain period of time, but a timeout of too long will cause the source node to abandon the task. TkT represents the type of the task. The design of the extended header field of the IPv6 packet can be found in the introduction of the relevant technology, which will not be repeated here.

[0171] Step s120: classify the task types in the scene according to various computing power attributes.

[0172] Soldier information: soldier equipment loss and quantity to maintain normal operation monitoring, soldier vital signs information, etc.

[0173] Environmental monitoring: monitor the surrounding environment and terrain.

[0174] Collaborative combat: Transmit combat assistance data between soldiers (such as sniper and infantry collaboration).

[0175] Reconnaissance target identification: Upload pictures, videos and other intelligence containing target information for judgment.

[0176] Intrusion detection: When an enemy invades, it is necessary to quickly locate, identify and alarm the intrusion target.

[0177] Step s130: After the task node generates the task, it uses the scheduling algorithm to obtain the TOP3 optimal strategy as the basis for unloading according to the task type and the locally stored resource topology map.

[0178] Santy's 1-9 scale method is used to give the relative importance. For example, for the task type of environmental monitoring, its task data will involve the transmission of pictures and videos, and is not sensitive to time, so this type of task considers the storage space attribute to be the most important, while computing resources and distance attributes are relatively unimportant. We give its subjective evaluation degree as {3,9,7,5,1}. For the task type of reconnaissance target recognition, the task data will also involve large data such as pictures, but it has high requirements for latency, and will use recognition algorithms, which have high requirements for computing resources. Therefore, the subjective evaluation degree given to it is {9,7,5,3,5}.

[0179] The weight values ​​of different task types based on different attributes are given through the analytic hierarchy process (AHP).

[0180] The TOPSIS method is used to judge the advantages and disadvantages of the available solutions and finally the unloading solution pm = {x 1 ,x 2 ,x 3}.

[0181] Step s140: TN makes a preliminary inquiry before formal uninstallation according to the policy pm. First, 1 Send inquiry message, sID 1 Estimated waiting time for the task after receiving the inquiry message and estimated processing time If the task can be completed within the deadline, a confirmation message is returned, otherwise the task is forwarded to node x. 2 , x 2 After receiving the inquiry message, it determines whether it can be completed in the same way. If all three nodes cannot complete the task, node i will not receive a response from CN and decide to abandon the task after exceeding the task deadline. On the contrary, if TN i receives any confirmation message, it will immediately unload the task and start the final unloading process.

[0182] Step s150: After receiving the confirmation message, the mobile terminal starts to unload, sends the task package with the real task data to the destination node for unloading, and the unloading is completed after receiving the result.

[0183] Embodiment 6

[0184] The following is an introduction to the ubiquitous edge computing environment provided by the present invention. Please refer to Figure 2, Figure 2 A network environment diagram of ubiquitous edge computing provided by an embodiment of the present invention.

[0185] Consider a field battlefield environment without network infrastructure, which is composed of many heterogeneous mobile devices with uneven resources and short-distance communication. We define the nodes in the scene into two categories. One is task nodes TN (Task Nodes), such as soldiers' smart wearable devices. There are many such devices with high mobility, but they lack computing resources and basically cannot process the tasks generated by themselves, but they can complete the forwarding of tasks. The other is center nodes CN (Centre Nodes). These nodes have rich computing resources but are not as many in number (such as tanks, infantry fighting vehicles and other vehicles). They are used to assist TN in completing tasks. CN and TN are represented by the sets K = {1,2,3,…,k} and N = {1,2,3,…,n} respectively. For example, there are 3 center nodes and 8 task nodes in a specific time period. Many nodes in the scene generate tasks and offload them to different CNs for calculation. For TN-2, the possible reason for choosing node CN-2 instead of CN-1 is that CN-2 has more computing resources and better communication channels. For situations like task-1, TN-1 initially chooses to send a task request to CN-2. Limited by the communication distance, it can only choose one TN as a relay node to forward the task request packet to CN-2. At this time, since CN-2 is overloaded with tasks, the estimated execution time is too long and it cannot meet the requirement of task-1 to be completed within the specified time, so CN-2 forwards its task request to CN-3. Unfortunately, CN-3 also needs to wait for a long time, so the task-1 request is forwarded to CN-1 again. CN-1 can meet the task requirements of task-1, so CN-1 finally calculates task-1 and returns the result to TN-1.

[0186] Embodiment 7

[0187] The field design of the IPv6 packet provided by the present invention is introduced below. Please refer to Figure 3 , Figure 3 The invention provides an IPv6 packet extension part design.

[0188] The extension header is placed between the IPv6 header and the upper layer protocol data unit, and the three together constitute the IPv6 packet. In the extension header section, we edited the fields to define the properties of the task.

[0189] The TKT field stores the task type, and the TS field is used to distinguish the message type and status. If the TS value is 0, it means that the message is a computing power information message, which is used by the central node to announce its own resource information. Otherwise, the message is a task message, which is used for information exchange between the task node and the central node when unloading tasks. Different TS values ​​represent different task states (request status, return result status, etc.). For example, suppose a node in the network receives message 1, and the fields of message 1 are {0, 0, 0, 5, 3000, 4000, 300, (197, 81), 0}. The TS field is 0, which means that the message is a computing power information message from node 5. 3000, 4000, and 300 respectively represent the computing power resource information of node 5, and 197 and 81 refer to the horizontal and vertical coordinates of the position of node 5. In another case, a node in the network receives message 2 with fields of {1,3,271,18,300,400,0,0,0,120}. The TS field of the message is 1, indicating that the message is a task message and the task is in a request state. It is task No. 271 from node No. 18. 300 and 400 represent the computing resources and storage space required to complete the task. 120 indicates that the deadline for the task is 120s.

[0190] Embodiment 8

[0191] The following is an introduction to the multi-attribute task scheduling method with timeliness guarantee provided by the present invention. Please refer to Figure 4 , Figure 4 A flowchart of a multi-attribute task scheduling method with timeliness guarantee provided by an embodiment of the present invention. The task offloading process mainly includes the following steps:

[0192] (1) Computing power perception. The central node will broadcast its current remaining resources and location information to all nodes within the surrounding communication range at a certain time interval int. At this time, the data packet only carries the header related to the computing power information, and the message content is empty. The header will contain the node identifier (sID), the node's computing power information (C, S, D), and the node's location information (pos_info). The message type TS is set to 0, indicating that the current message is a computing power information message. When the node receives the data packet, it counts the data in the message into the computing power information table.

[0193] (2) Offloading decision. When task node i generates a task, the MADM algorithm needs to be used to make an offloading decision. The algorithm flow is shown in the figure. The input of the algorithm is task D tThe task type TKT and the computing power matrix R are first used as the input of the AHP algorithm to calculate the weight ω of each attribute for the task type, and then the TOPSIS algorithm is used to judge the pros and cons of the solution based on the weight ω and R. The output is the task D at the current moment in the current environment t The TOP-3 decision solutions pm = {x 1 ,x 2 ,x 3}.

[0194] (3) Scheduling query. After task node i obtains the decision solution for the task, in order to ensure timeliness, it first sends a query to the destination node sID of the optimal solution before scheduling. 1 Send a data packet to inquire. The header of this data packet will contain tID, sID, and the computing power C required to complete the task. t , the data size of the task S t and the task deadline T ddl , and in order to distinguish the task status, the TS field is set to 1. 1 After receiving the inquiry packet, it calculates its own estimated completion time. If it can be completed in time, it returns a confirmation packet, which contains the node's sID and the TS field is set to 2. If it cannot be completed in time, it forwards the inquiry packet to the sID 2 And complete the judgment process. Similarly, sID 3 Follow the same steps. If the final sID 3 If it cannot be completed, the task will be discarded.

[0195] (4) Complete the offloading. When the task node receives the confirmation message from the central node, it sends a data packet containing the offloading task content to the node. The message header at this time contains tID and sID. After the central node receives the task, it adds it to the task queue and waits for processing. After processing, it returns the calculation result to the source node.

[0196] In the several embodiments provided in the present invention, it should be understood that the network environment of ubiquitous edge computing and the complex tasks of directed acyclic graph (DAG) modeling involved are merely schematic, and they only represent one form of the model, which can be expressed in many ways.

[0197] In summary, the multi-attribute task scheduling method with timeliness guarantee provided by the present invention, in the scenario of ubiquitous edge computing, first designs a distributed computing power perception mechanism to ensure that the task generation equipment detects the surrounding resources and load conditions. On this basis, we perform multi-attribute tasks on battlefield tasks according to actual conditions, and formulate unloading strategies for different task types by comprehensively considering factors such as calculation, storage, transmission and load conditions. The present invention can meet the deterministic delay requirements of delay-sensitive tasks and has a high task completion rate.

[0198] The technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a mobile terminal (which can be a mobile phone, or a tablet computer, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0199] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0200] The network model, tasks, IPv6 extension headers and task offloading process of the multi-attribute task scheduling method with timeliness guarantee provided by the present invention are introduced in detail above. The present invention uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A multi-attribute task scheduling method with timeliness guarantee, characterized by: Conduct integrated measurement of computing power and develop a distributed computing power perception mechanism; The mobile terminal generates tasks, classifies the tasks according to their attributes based on the actual battlefield situation, and designs the IPv6 header to encapsulate the task attributes and computing resource attributes; Analyze the delay during the unloading process of the task; After the device generates a task, it makes an offloading decision based on the task type and perception data; According to the uninstallation decision, an uninstallation process is performed, and a pre-inquiry is performed before the uninstallation process to ensure the timeliness of the task.

2. The multi-attribute task scheduling method with timeliness guarantee as claimed in claim 1, characterized in that: The integrated measurement of computing power includes: Computing and network resources are encapsulated into computing power and integrated and measured. The indicators of the integrated measurement include: Among them, comp is a computing metric, CPU Cycle is a CPU cycle, FLOPS is the number of floating-point operations per second, CPI is the number of cycles per instruction, RT is the response time, Stor is a storage metric, Memory is the storage capacity, IOPS is the number of input and output per second, Throughout is the throughput, Del is a transmission metric, Bandwidth is the transmission bandwidth, Jitter is the jitter parameter, Delay is the delay parameter, and Capacity is the capacity parameter; The distributed computing power perception mechanism includes: The central node abstracts its own computing power information and location information, and notifies the computing power information and location information to the surrounding nodes within the communication range at a set time interval to realize the computing power perception process; Each node maintains a computing power information table, and each node has a topological map of the surrounding computing power: Hashrate Matrix Where ID represents the unique identification ID of the node, C, S, and D represent the computing, storage, and transmission power information of the node, loc represents the location information of the node, and w represents the estimated waiting time in the scheduling queue of the node.

3. The multi-attribute task scheduling method with timeliness guarantee as claimed in claim 1, characterized in that: The mobile terminal generates tasks and classifies the tasks according to their attributes based on the actual battlefield situation, including the following steps: Classify the actual services that may be generated, including soldier information, environmental monitoring, collaborative operations, reconnaissance target identification, and intrusion detection; Or / and, set the IPv6 header to encapsulate the task attributes and computing resource attributes, including the following steps: After task classification, different decision functions are formulated to select offloading nodes for different businesses.

4. The multi-attribute task scheduling method with timeliness guarantee as claimed in claim 3, characterized in that: The specific fields of the extended header in the IPv6 header include: TS, TkT, tID, sID, C, S, D, pos_info, T ddl ; Among them, the TKT field stores the task type; the TS field is used to distinguish the message type and status; when the TS value is 0, it means that the message is a computing power information message, which is used by the central node to notify its own resource information. When the TS value is not 0, the message is a task message, which is used for information exchange between the task node and the central node when unloading tasks. Different TS values ​​represent different task states; The tID field stores the unique ID of the task, the sID field stores the TN node ID that generates the task, and the TS field stores the task status. ddl The field stores the maximum waiting time of the task; the pos_info field stores the location information of the node; the C field stores the computing resources required to complete the task, and the S field stores the size of the task required to complete the task.

5. The multi-attribute task scheduling method with timeliness guarantee as claimed in claim 1, characterized in that: After the device generates a task, it makes an offloading decision based on the task type and perception data, including: The weight values ​​of different task types based on different attributes are given through the hierarchical analysis method; the judgment matrix A is constructed according to the relative importance; According to the judgment matrix A, the weights of each attribute are obtained by arithmetic mean method: Use the good-bad solution distance method to judge the advantages and disadvantages of the available solutions and finally come up with an unloading solution; The node calculates the original data matrix based on the collected computing power information: x ij are the main elements of the original data matrix, representing the node's computing power, storage, transmission, queue waiting time and distance information from itself. Normalize the matrix to get the standardized matrix: The optimal solution Z is obtained by standardizing the matrix + and the worst solution Z - ,in It is composed of the maximum value of each column element in Z, It is composed of the minimum value of each column element in Z; According to Z + , Z - And ω obtained by formula (4) i Calculate the distance between each evaluation object and them and Then calculate how close the solution is to the optimal solution: Where 0≤C i ≤1,C i →1 indicates that the evaluation object is better; according to the closeness of the solution to the optimal solution, C i Sort and give results; through comprehensive evaluation method and hierarchical analysis method, each task gets targeted decision according to the task attributes and the surrounding resource conditions at that time.

6. The multi-attribute task scheduling method with timeliness guarantee as claimed in claim 1, characterized in that: Analyze the delay in the task unloading process, where the task delay analysis includes: calculation delay, communication delay, and queue waiting delay; The computational delay includes: at a certain moment, the task node i generates task t and offloads it to the central node j for execution. The computational time required to complete the task on j can be expressed as: Among them, S t is the data size of the task, C t Refers to the computing resources required per MB to complete the task, C j is the computing resource allocated by the central node j to task t, in GHz; or / and, The communication delay includes: when the task node encounters a task that needs to be unloaded, it needs to select a suitable central node to unload the task to; accordingly, after the CN node calculates the task, it sends the result back to the TN node, and the proposed task is transmitted in TNi and CNj; in this process, the communication delay in the node consists of two parts, the sending delay and propagation delay X(i,j); Propagation delay refers to the time it takes for an electromagnetic signal to propagate a distance D in a propagation medium. They are expressed as: Among them, B t is the bandwidth, R is the propagation speed of electromagnetic waves in the current medium; D(i,j) is the physical distance between node i and node j; or / and, The queue waiting delay includes: the estimated processing time of the new task and queue delay It is expressed as: Among them, m is the number of tasks in the queue when task t enters the waiting queue of CNj, and α is the proportional coefficient of the reading and writing time required for unit data volume; For the tth task generated by TNi, its actual service delay is: Where l≥4, the specific value of l is related to D t The number of times the request is forwarded is related to is the unloading function. When task t is finally unloaded to CNj, When offloading to other CN nodes, 7. The multi-attribute task scheduling method with timeliness guarantee as claimed in claim 1, characterized in that: According to the uninstallation decision, the uninstallation process is carried out, and a preliminary inquiry is conducted before the uninstallation process to ensure the timeliness of the task, including: According to the offloading decision result, the TOP-3 solution pm={x1,x2,x3} is recorded for task t. The solution records the three CN nodes that are most suitable for offloading. In order to ensure the timeliness of the task, a pre-query method is used to send a query message to node x1. After receiving the query message, sID1 estimates the waiting time of the task. and estimated processing time If the task can be completed within the deadline, a confirmation message is returned, otherwise the task is forwarded to node x2. After receiving the query message, x2 determines whether it can be completed in the same way; and so on; When all three nodes cannot complete the task, node i will not receive a response from CN and decides to abandon the task after exceeding the task deadline event; When the three nodes can complete the task, if i receives any confirmation message, it will immediately unload the task and start the final unloading process.

8. A multi-attribute task scheduling device with timeliness guarantee, characterized in that: include: The fusion measurement module is used to perform fusion measurement on computing power and formulate a distributed computing power perception mechanism; The task generation module is used to generate tasks for mobile terminals, classify tasks according to their attributes based on the actual battlefield situation, and design the IPv6 header to encapsulate the task attributes and computing resource attributes; The delay analysis module is used to analyze the delay in the unloading process of the task; The offloading decision module is used to make offloading decisions based on the task type and perception data after the device generates a task; The pre-inquiry uninstallation module is used to perform the uninstallation process according to the uninstallation decision, and to perform a pre-inquiry before the uninstallation process to ensure the timeliness of the task.

9. A multi-attribute task scheduling device with timeliness guarantee, characterized in that: include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and is configured to be executed by the processor to implement the multi-attribute task scheduling method with timeliness guarantee as described in any one of claims 1 to 7.

10. A storage medium for multi-attribute task scheduling with timeliness guarantee, characterized in that: A computer program is stored thereon, and the computer program is executed by a processor to implement the multi-attribute task scheduling method with timeliness guarantee as described in any one of claims 1 to 7.

Images