CN-121985375-A - Vehicle-mounted computing task processing method, device and equipment based on mobile edge computing

CN121985375ACN 121985375 ACN121985375 ACN 121985375ACN-121985375-A

Abstract

The application discloses a vehicle-mounted computing task processing method based on mobile edge computing, which comprises the steps of responding to detection of a target computing task to be processed, determining the current node position of an edge computing node, determining link channel gain for a target communication link between a vehicle and the edge computing node based on the current vehicle speed, the current vehicle position, the current environment type and the current node position, determining a state observation vector based on the current vehicle speed, the current vehicle position and the link channel gain, and inputting the state observation vector into a generating type intelligent agent of the vehicle so as to output a task processing strategy of the target computing task through the generating type intelligent agent, wherein the task processing strategy is used for processing the target computing task, determining task time delay, task energy consumption and task completion degree of the target computing task, and determining a reward signal for optimizing the generating type intelligent agent. The application jointly optimizes the task unloading and transmission energy efficiency.

Inventors

HU BINTAO
XU SHUGONG
ZHANG WENZHANG
ZHAO JIKANG
WANG JINGCHEN

Assignees

西交利物浦大学

Dates

Publication Date: 20260505
Application Date: 20260205

Claims (10)

1. A method for processing a vehicle-mounted computing task based on mobile edge computing, characterized by being executed by a vehicle-mounted terminal, the method comprising: determining a current node position of an edge computing node in response to detecting a target computing task to be processed; Determining a link channel gain for a target communication link between the host vehicle and the edge computing node based on a current host vehicle speed, a current host vehicle position, a current environment type, and the current node position; Determining a state observation vector based on the current vehicle speed, the current vehicle position and the link channel gain, and inputting the state observation vector into a generating type intelligent agent deployed on the vehicle to output a task processing strategy of the target computing task through the generating type intelligent agent, wherein the task processing strategy comprises a task unloading proportion and a power control factor; and processing the target computing task based on the task unloading proportion and the power control factor, and determining task time delay, task energy consumption and task completion degree of the target computing task, wherein the task time delay, the task energy consumption and the task completion degree are used for determining a reward signal to optimize the generated intelligent agent.
2. The method of claim 1, wherein the processing the target computing task based on the task offload ratio and the power control factor and determining a task latency, a task energy consumption, and a task completion of the target computing task comprises: processing the target computing task based on the task unloading proportion to obtain a sub-computing task, wherein the sub-computing task comprises a local sub-task executed by a self-vehicle and/or an edge sub-task executed by the edge computing node; controlling the basic transmitting power of the own vehicle based on the power control factor to obtain the task transmitting power matched with the edge subtask; The task transmitting power is adopted to respectively transmit the edge subtasks to the corresponding edge computing nodes, so that the edge computing nodes process the edge subtasks and determine edge time delay and edge energy consumption of the edge subtasks; Processing the local subtasks through a local computing unit deployed in the own vehicle, and determining the local time delay and the local energy consumption of the local subtasks; Determining task time delay of the target computing task based on the edge time delay and the local time delay, and determining task energy consumption of the target computing task based on the edge energy consumption and the local energy consumption; and determining the task completion degree of the target computing task according to the task time delay and the maximum time delay threshold of the target computing task.
3. The method of claim 2, wherein said determining a local latency and a local energy consumption of the local subtask comprises: Determining the task size of the local subtask, the calculation amount required by unit data and the calculation capacity of the local calculation unit; and determining the local time delay and the local energy consumption required by a local computing unit deployed in the own vehicle to process the local subtasks based on the task size, the calculated amount required by the unit data and the computing capacity.
4. The method of claim 2, wherein the determining the edge latency and edge energy consumption of the edge subtask comprises: determining a transmission rate of the target communication link based on a bandwidth share of the edge computing node and a communication link quality of the target communication link, wherein the bandwidth share is dynamically allocated for the edge computing node based on an associated link number, an associated link quality, an associated link transmit power, and a transmission file size of the associated link; determining a transmission delay of the edge subtask based on a task size of the edge subtask and a transmission rate of the target communication link; Determining the transmission energy consumption of the edge subtask based on the transmission delay of the edge subtask and the task transmitting power corresponding to the edge subtask; Determining a target computing power allocated to the edge subtask based on the total computing resource amount of the edge computing node, the maximum delay threshold of the edge subtask and the computing amount required by unit data; determining the calculation time delay of the edge subtask based on the calculated amount required by the unit data of the edge subtask and the target calculation force; Determining the computing energy consumption of the edge subtasks based on the computing time delay and the target computing force; And determining the edge time delay of the edge subtask based on the transmission time delay and the calculation time delay, and determining the edge energy consumption of the edge subtask based on the transmission energy consumption and the calculation energy consumption.
5. The method of claim 2, wherein after determining the task latency, task energy consumption, and task completion of the target computing task, the method further comprises: Calculating a reward signal corresponding to the target calculation task based on the task time delay, the task energy consumption and the task completion degree of the target calculation task; storing state observation vectors, task processing strategies and rewarding signals corresponding to the target computing task as historical interaction data; And calculating a strategy gradient through the historical interaction data of the generating intelligent agent, and updating strategy network parameters based on the strategy gradient.
6. The method of claim 1, wherein the determining a link channel gain for a target communication link between a host vehicle and the edge computing node based on a current host vehicle speed, a current host vehicle location, a current environment type, and the current node location comprises: determining an uplink angle and a three-dimensional spatial distance between a host vehicle and the edge computing node based on the current host vehicle position and the current node position; determining a line-of-sight probability based on the uplink angle and the current environment type, and determining a path loss of the target communication link based on the line-of-sight probability, the three-dimensional spatial distance, and the current environment type; a link channel gain is determined for a target communication link between the host vehicle and the edge computing node based on the current host vehicle speed and a path loss of the target communication link.
7. The method of claim 1, wherein the edge computing node is an unmanned aerial vehicle with a mobile edge computing server, and wherein the unmanned aerial vehicle is deployed in a surrounding hover within a predetermined airspace around a ground base station.
8. An on-vehicle computing task processing device based on mobile edge computing, which is characterized by being configured in an on-vehicle terminal, the device comprising: the node position determining module is used for determining the current node position of the edge computing node in response to the detection of the target computing task to be processed; The channel gain determining module is used for determining a link channel gain for a target communication link between the own vehicle and the edge computing node based on the current own vehicle speed, the current own vehicle position, the current environment type and the current node position; The processing strategy determining module is used for determining a state observation vector based on the current vehicle speed, the current vehicle position and the link channel gain, and inputting the state observation vector into a generating type intelligent agent deployed on a vehicle so as to output a task processing strategy of the target computing task through the generating type intelligent agent, wherein the task processing strategy comprises a task unloading proportion and a power control factor; The evaluation index determining module is used for processing the target computing task based on the task unloading proportion and the power control factor and determining task time delay, task energy consumption and task completion degree of the target computing task, wherein the task time delay, the task energy consumption and the task completion degree are used for determining a reward signal to optimize the generated intelligent agent.
9. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when executed by a processor, implements the mobile edge computing-based on-board computing task processing method as claimed in any one of claims 1 to 7.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the mobile edge computing based on-vehicle computing task processing method as claimed in any one of claims 1-7 when executing the computer program.

Description

Vehicle-mounted computing task processing method, device and equipment based on mobile edge computing Technical Field The application relates to the field of intelligent traffic systems, in particular to a vehicle-mounted computing task processing method and device based on mobile edge computing, a readable medium and electronic equipment. Background In recent years, with the development of sixth generation mobile communication technology, the capability of internet of vehicles in terms of high speed, low time delay and large-scale data processing is continuously improved by means of artificial intelligence technology. The advent of diversified intelligent applications such as automatic driving, real-time navigation and the like has put forward higher requirements on vehicle-mounted computing resources. However, limited by volume and power consumption, the computational power of on-board units is difficult to meet the increasingly complex task processing demands. Mobile edge computing provides an effective solution for the internet of vehicles by sinking computing resources to the network edge. The prior research provides a task unloading algorithm based on a dichotomy, which can distribute an overloaded task to an idle vehicle-mounted unit, and a heuristic algorithm and a deep reinforcement learning method are also used for optimizing resource allocation. However, as the vehicle edge computing scene becomes more complex, the existing method is difficult to realize the collaborative optimization of task offloading and global resource allocation when dealing with a high-dimensional state space, and the overall performance of the system under a dynamic environment is limited. Disclosure of Invention The application provides a vehicle-mounted computing task processing method, a device, a medium and electronic equipment based on mobile edge computing, which can realize the task unloading and transmission energy efficiency combined optimization of a vehicle in a dynamic environment, and achieve the purposes of reducing task time delay, optimizing system energy efficiency and improving task completion degree. According to a first aspect of the present application, there is provided a vehicle-mounted computing task processing method based on mobile edge computing, executed by a vehicle-mounted terminal, the method comprising: determining a current node position of an edge computing node in response to detecting a target computing task to be processed; Determining a link channel gain for a target communication link between the host vehicle and the edge computing node based on a current host vehicle speed, a current host vehicle position, a current environment type, and the current node position; Determining a state observation vector based on the current vehicle speed, the current vehicle position and the link channel gain, and inputting the state observation vector into a generating type intelligent agent deployed on the vehicle to output a task processing strategy of the target computing task through the generating type intelligent agent, wherein the task processing strategy comprises a task unloading proportion and a power control factor; and processing the target computing task based on the task unloading proportion and the power control factor, and determining task time delay, task energy consumption and task completion degree of the target computing task, wherein the task time delay, the task energy consumption and the task completion degree are used for determining a reward signal to optimize the generated intelligent agent. According to a second aspect of the present application, there is provided an in-vehicle computing task processing device based on mobile edge computing, configured in an in-vehicle terminal, the device comprising: the node position determining module is used for determining the current node position of the edge computing node in response to the detection of the target computing task to be processed; The channel gain determining module is used for determining a link channel gain for a target communication link between the own vehicle and the edge computing node based on the current own vehicle speed, the current own vehicle position, the current environment type and the current node position; The processing strategy determining module is used for determining a state observation vector based on the current vehicle speed, the current vehicle position and the link channel gain, and inputting the state observation vector into a generating type intelligent agent deployed on a vehicle so as to output a task processing strategy of the target computing task through the generating type intelligent agent, wherein the task processing strategy comprises a task unloading proportion and a power control factor; The evaluation index determining module is used for processing the target computing task based on the task unloading proportion and the power control factor and determining task time delay, task