CN-121979652-A - Micro-exploration computation-intensive dynamic allocation method and imaging system

CN121979652ACN 121979652 ACN121979652 ACN 121979652ACN-121979652-A

Abstract

The application relates to the technical field of micro-exploration calculation, in particular to a micro-exploration calculation intensive dynamic allocation method and an imaging system, which are used for decomposing an instant imaging task in an edge server into a plurality of subtasks, dividing the subtasks into the subtasks of serial tasks and the subtasks of parallel tasks according to the dependency relationship among the subtasks, determining an optimal execution sequence according to the dependency degree of the subtasks, evaluating the real-time resource state of each sensor node, determining a sensor node set suitable for bearing tasks, combining the transmission delay and the transmission energy consumption required by the subtasks from the sensor node set suitable for bearing tasks, minimizing the total transmission delay and the total transmission energy consumption, solving an optimal task allocation scheme, unloading the subtasks of the serial tasks to the sensor nodes according to the optimal execution sequence and unloading the subtasks of the parallel tasks without the dependency relationship to the plurality of sensor nodes for simultaneous processing according to the optimal task allocation scheme.

Inventors

TIAN RUYUN
WANG HONGYU
ZHANG YUXING
CHEN YUYANG
XU JIAHUI
WEI ZHONG

Assignees

南京信息工程大学

Dates

Publication Date: 20260505
Application Date: 20251209

Claims (10)

1. A micro-prospecting computationally intensive dynamic allocation method, comprising: Decomposing the instant imaging task in the edge server into a plurality of subtasks; dividing the subtasks into subtasks of serial tasks and subtasks of parallel tasks according to the dependency relationship between the subtasks; Determining an optimal execution sequence according to the dependency degree of the subtasks; evaluating the real-time resource state of each sensor node to determine a sensor node set suitable for bearing tasks; combining transmission delay and transmission energy consumption required by subtasks from a sensor node set suitable for bearing tasks, minimizing total transmission delay and total transmission energy consumption, and solving an optimal task allocation scheme; And unloading the subtasks of the serial tasks to the sensor nodes according to the optimal execution sequence and the optimal task allocation scheme, and unloading the subtasks of the parallel tasks without the dependency relationship to a plurality of sensor nodes for simultaneous processing.
2. The micro-prospecting computation intensive dynamic allocation method according to claim 1, wherein determining the optimal execution order according to the degree of dependency of the subtasks comprises: Calculating the degree of subtasks; selecting a subtask with the incidence degree of 0 as an initial task; during the sorting process, each subtask After execution, all subsequent subtasks are updated If the subtask is updated If the degree of entry of (1) becomes 0, the subtask is executed Adding into a queue; And repeating the operation until the degree of penetration of all subtasks is 0.
3. The micro-motion exploration computation-intensive dynamic allocation method according to claim 1, wherein minimizing total transmission delay and total transmission energy consumption, finding an optimal task allocation scheme comprises: Setting a single target cost function by adopting total transmission delay and total transmission energy consumption; Unloading the task with unique constraint, computing resource constraint, link bandwidth constraint, energy constraint and deadline constraint as constraint conditions; And minimizing the single objective cost function under the constraint condition to obtain an optimal decision variable, wherein the task allocation scheme corresponding to the optimal decision variable is an optimal task allocation scheme.
4. A micro-prospecting computation intensive dynamic allocation method according to claim 3, wherein said single objective cost function is: . Wherein the method comprises the steps of As a weight factor for the total transmission delay, The weight factor for the total transmission energy consumption is as follows , Is the transmission delay used for normalization under the current network conditions and the transmission delay used for normalization under the current network conditions, Is the transmission energy consumption reference value used for normalization under the current network conditions.
5. The micro-motion exploration computation intensive dynamic allocation method according to claim 3, wherein said task offloading uniqueness constraint is that each subtask can only be scheduled to one sensor node at the same time; The computational resource constraint is that the computational complexity of the subtasks cannot exceed the self capacity of the sensor node; The link bandwidth constraint is that the total link transmission rate must not exceed the wireless transmission bandwidth; The energy constraint is that the transmission energy consumption of the subtasks cannot exceed the residual energy of the sensor nodes; the deadline constraint is that the subtasks for setting the deadline need to be completed before the deadline.
6. The micro-prospecting computation intensive dynamic allocation method according to claim 1, further comprising triggering a dynamic adjustment task offload path policy to adjust an offload path for a sub-task when a trigger condition is satisfied, wherein the trigger condition comprises: the residual energy of the sensor node is smaller than an energy triggering threshold; The real-time transmission delay is greater than a transmission delay trigger threshold; the average transmission delay is greater than the target transmission delay and the average transmission energy consumption is greater than the target transmission energy consumption.
7. The micro-survey computationally intensive dynamic allocation method according to claim 6, wherein triggering the dynamic adjustment task offload path policy adjusts the offload paths of the subtasks, comprising: Constructing a historical task path database for storing and continuously updating historical task unloading path data; after each new subtask is unloaded, updating the historical average transmission delay and the historical average transmission energy consumption of the new subtask by adopting weighted average; Normalizing the updated historical average transmission delay and the updated historical average transmission energy consumption; scoring each unloading path according to the normalization result; And selecting the unloading path with the lowest score as a task unloading path to update the unloading path of the subtask.
8. A micro-prospecting computation intensive dynamic allocation method according to claim 3, wherein the total transmission delay is the sum of the transmission delays of all serial task groups and all parallel task groups; calculating the unloading transmission delay, calculating the sum of the transmission delay and the queuing transmission delay when the subtasks need to be unloaded to the sensor node for processing, and obtaining the transmission delay of the subtasks; Adding the transmission delays of the subtasks and all the precursor subtasks to obtain the transmission delay of the serial task; The transmission delay of the parallel task group comprises the steps of calculating the transmission delay of the subtasks, and taking the transmission delay of the largest subtask in the parallel task group as the transmission delay of the parallel task group.
9. A micro-prospecting computation-intensive dynamic allocation method according to claim 3, wherein the calculation of the total transmitted energy consumption comprises: calculating the sum of unloading transmission energy consumption and calculating transmission energy consumption when the subtask needs to be unloaded to a sensor node for processing, and obtaining the transmission energy consumption of the subtask; and summing the transmission energy consumption of all the subtasks to obtain the total transmission energy consumption.
10. A micro-prospecting computationally intensive imaging system for performing the dynamic allocation method according to any one of claims 1 to 9, comprising an edge server for decomposing an instant imaging task into a plurality of subtasks; dividing the subtasks into subtasks of serial tasks and subtasks of parallel tasks according to the dependency relationship between the subtasks; Determining an optimal execution sequence according to the dependency degree of the subtasks; evaluating the real-time resource state of each sensor node to determine a sensor node set suitable for bearing tasks; combining transmission delay and transmission energy consumption required by subtasks from a sensor node set suitable for bearing tasks, minimizing total transmission delay and total transmission energy consumption, and solving an optimal task allocation scheme; According to the optimal execution sequence and the optimal task allocation scheme, the subtasks of the serial tasks are unloaded to the sensor nodes, and the subtasks of the parallel tasks without dependency relationship are unloaded to a plurality of sensor nodes for simultaneous processing; And receiving a processing result returned by the sensor node, and imaging according to the processing result.

Description

Micro-exploration computation-intensive dynamic allocation method and imaging system Technical Field The application relates to the technical field of micro-motion exploration calculation, in particular to a micro-motion exploration calculation intensive dynamic allocation method and an imaging system. Background Micro-exploration has become a leading field and a key direction of seismology research and application by utilizing subsurface noise data to perform shear wave velocity structural imaging. The traditional micro-exploration method adopts sensor nodes to collect noise data, and then the data are transmitted back to a central server for centralized processing and imaging. The cluster architecture and the centralized management mode not only increase the resource cost and the operation complexity, but also cause serious transmission delay of imaging tasks due to the problems of long exploration and construction period, complex large-scale data recovery work and the like, and can not acquire underground structure information in real time. The lack of an instant imaging method has become a key bottleneck restricting the development of micro-exploration methods in the field of geophysical exploration. The edge computing technology meets the low transmission delay requirement required by instant imaging by deploying lightweight edge servers near sensor nodes to provide highly flexible and responsive computing services. The physical proximity enables the edge server to rapidly respond to imaging task requests, and effectively solves the problem of imaging transmission delay inherent in the traditional micro-prospecting method. In addition, the adoption of the edge computing technology for immediate imaging can avoid complicated data recovery work and expensive cluster architecture construction. The edge calculation is applied to the micro-exploration system, and aims to construct a micro-exploration edge network with low transmission delay and high reliability. However, the cost limitations and limited computational power of edge servers in micro-prospecting edge networks are major challenges. The task allocation method ensures reasonable scheduling of resources in the micro exploration edge network by allocating the calculation tasks to the sensor nodes for processing. The method not only improves the resource utilization rate of the sensor node, but also reduces the operation pressure of the edge server. Therefore, optimizing task allocation strategies in the micro-exploration edge network is important, so that the transmission delay of instant imaging can be remarkably reduced, and the overall reliability of the system can be improved. Existing research typically employs smart search algorithms, meta-heuristic algorithms, or artificial intelligence to optimize task allocation. Although the intelligent search algorithm can find the optimal task allocation scheme through iteration, the dilemma of a local optimal solution is easily trapped. Therefore, in practical application, the effect of the intelligent search algorithm has fluctuation, and the system performance is difficult to ensure. However, the artificial intelligence method requires a great deal of model training, which not only increases the calculation cost, but also is limited by hardware conditions. Furthermore, the point-in-time imaging task contains numerous dependency-aware subtasks, but prior studies mostly neglect the impact of this dependency on task allocation. Similarly, most methods focus on optimization strategies in static environments, and there are few studies on dynamic scenes. In the micro-exploration edge network task, sensor nodes are limited by environmental interference and residual energy, the working condition is severely constrained, and the requirements of the traditional static task allocation method are difficult to meet. Disclosure of Invention The application aims to overcome the current situation that the sensor node is limited by environmental interference and residual energy and the running condition of the sensor node has a plurality of constraints, and provides a micro-exploration computation-intensive dynamic allocation method and an imaging system. According to an embodiment of the first aspect of the application, a micro-exploration computation-intensive dynamic allocation method comprises the following steps: Decomposing the instant imaging task in the edge server into a plurality of subtasks; dividing the subtasks into subtasks of serial tasks and subtasks of parallel tasks according to the dependency relationship between the subtasks; Determining an optimal execution sequence according to the dependency degree of the subtasks; evaluating the real-time resource state of each sensor node to determine a sensor node set suitable for bearing tasks; combining transmission delay and transmission energy consumption required by subtasks from a sensor node set suitable for bearing tasks, minimizing total transmissio