CN115037751A - Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method - Google Patents

Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method Download PDF

Info

Publication number
CN115037751A
CN115037751A CN202210744842.2A CN202210744842A CN115037751A CN 115037751 A CN115037751 A CN 115037751A CN 202210744842 A CN202210744842 A CN 202210744842A CN 115037751 A CN115037751 A CN 115037751A
Authority
CN
China
Prior art keywords
vehicle
unloading
task
vehicles
unmanned aerial
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202210744842.2A
Other languages
Chinese (zh)
Other versions
CN115037751B (en
Inventor
宋晓勤
王书墨
宋铁成
彭昱捷
杨雨露
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute Of Southeast University
Original Assignee
Shenzhen Institute Of Southeast University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute Of Southeast University filed Critical Shenzhen Institute Of Southeast University
Priority to CN202210744842.2A priority Critical patent/CN115037751B/en
Publication of CN115037751A publication Critical patent/CN115037751A/en
Application granted granted Critical
Publication of CN115037751B publication Critical patent/CN115037751B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/104—Peer-to-peer [P2P] networks
    • H04L67/1074—Peer-to-peer [P2P] networks for supporting data block transmission mechanisms
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • H04W4/46—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for vehicle-to-vehicle communication [V2V]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides an unmanned aerial vehicle assisted heterogeneous Internet of vehicles task migration and resource allocation method, aiming at a scene of cooperative computing unloading of a mobile edge server and an unmanned aerial vehicle, firstly, a decision of whether a vehicle unloads is obtained through a potential game, namely, the vehicle decides local computing or unloads to an MEC server or an unmanned aerial vehicle for computing. For the vehicles deciding the task unloading, a distributed resource allocation method is adopted, a base station is not required to centrally schedule channel state information, each vehicle deciding the task unloading is regarded as an intelligent agent, deep reinforcement learning is trained through DDQN, and each vehicle deciding the task unloading selects an unloading node and emission power based on local observation state information. The algorithm can minimize the system time delay under the limit of the maximum transmitting power, and a good balance is obtained between the complexity and the performance.

Description

Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method
Technical Field
The invention relates to an internet of vehicles technology, in particular to an unmanned aerial vehicle-assisted task migration and resource allocation method for the internet of vehicles, and more particularly to an unmanned aerial vehicle-assisted heterogeneous internet of vehicles task migration and resource allocation method.
Background
With the development of the internet of vehicles, various vehicle applications such as route planning, autopilot, and infotainment applications have emerged. The applications can ensure the safety of travel and can also provide entertainment interconnection in the journey. However, most of these applications are delay sensitive, resource intensive, computationally complex, and energy demanding. Currently, many vehicles still have limited storage capacity and insufficient computing resources to meet the stringent delay constraints of these applications. Mobile Edge Computing (MEC) can provide low-latency computing services for vehicles by deploying computing and storage resources at the edge of the network, and in a typical scenario of an internet of vehicles, MEC servers are deployed on roadside units to provide computing services for vehicles. In wisdom road construction, unmanned aerial vehicle is used for road patrol, bridge inspection and road damage inspection. When the unmanned aerial vehicle patrols in a certain area, the powerful computing power of the unmanned aerial vehicle can also be used as an MEC server.
Currently, unmanned aerial vehicle-assisted mobile edge computing is still in a starting stage, and only a few studies have conducted detailed studies on this field. In addition, the existing research mainly optimizes a calculation unloading strategy, but does not fully consider the problem of cooperation and communication resource allocation between heterogeneous MEC servers under a time-varying channel.
Therefore, the invention provides an unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method, which aims at the scene of cooperative computing and unloading of a mobile edge server and an unmanned aerial vehicle, takes minimization of system time delay as an optimization target of task migration and resource allocation, and achieves good balance between complexity and performance.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems in the prior art, the unmanned aerial vehicle assisted heterogeneous Internet of vehicles task migration and resource allocation method is provided, and the unmanned aerial vehicle can provide computing resources for vehicles. The method adopts a mixed frequency spectrum access technology to carry out transmission, and realizes the minimization of system time delay.
The technical scheme is as follows: aiming at the scene of cooperative calculation unloading of the mobile edge server and the unmanned aerial vehicle, the purpose of minimizing system delay is achieved by reasonably and efficiently calculating unloading decisions and resource allocation. In order to reduce system time delay and improve the frequency spectrum utilization rate, a hybrid frequency spectrum access technology is adopted for transmission, a vehicle unloads tasks to an MEC server on a roadside unit for calculation through a vehicle to roadside facility (V2I) link, the tasks are unloaded to an unmanned aerial vehicle for calculation through a vehicle to vehicle (V2V) link, and the V2I link and the V2V link are accessed to different slices through a 5G slice technology and do not interfere with each other. Firstly, the decision of whether the vehicle is unloaded is obtained through potential game, namely the vehicle decides local calculation or is unloaded to an MEC server or unmanned aerial vehicle calculation. And for the vehicles deciding the task unloading, a distributed resource allocation method is adopted, the base station is not required to centrally schedule the channel state information, each vehicle deciding the task unloading is regarded as an intelligent agent, and the transmitting power is selected based on the local observation state information. By establishing the Deep reinforcement Learning model, the Deep reinforcement Learning model is optimized by using a Deep Double Q-Learning algorithm (DDQN). And obtaining the unloading node and the transmitting power of each vehicle for deciding the task unloading according to the optimized DDQN model. The invention is realized by the following technical scheme: an unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method comprises the following steps:
(1) deploying a Mobile Edge Computing (MEC) server at a Road Side Unit (RSU), deploying an unmanned aerial vehicle in the system to provide computing service for a vehicle, and locally processing computing tasks of the vehicle and unloading the computing tasks to the unmanned aerial vehicle or the MEC server;
(2) establishing a communication model and a calculation model comprising N vehicles and M unmanned aerial vehicles, and further establishing a joint calculation migration and resource allocation model;
(3) each vehicle acquires the positions of the unmanned aerial vehicle and the MEC, the occupation condition of computing resources and task information;
(4) obtaining a decision whether each vehicle is unloaded or not based on the potential game, and establishing a deep reinforcement learning model for the vehicle determining task unloading with the goal of reducing system time delay according to the obtained vehicle unloading decision;
(5) training a deep reinforcement learning model based on the DDQN;
(6) in the execution stage, the vehicle n with the calculation task judges whether the task is unloaded or not through potential game, and determines the unloaded vehicle n 0 Obtaining a current state from local observations
Figure BSA0000276505790000021
Obtaining unloading nodes and transmitting power of the vehicle by using the trained deep reinforcement learning model;
further, the step (2) comprises the following specific steps:
(2a) the system comprises N vehicles, M unmanned aerial vehicles, a road side unit for deploying an MEC server, and a set for vehicles
Figure BSA0000276505790000022
Representation, collection
Figure BSA0000276505790000023
Representing unmanned vehicles, the task of vehicle n may be represented as
Figure BSA0000276505790000024
c n Representing the number of CPU cycles, s, required for vehicle n to complete a task n Indicating the size of the amount of task data unloaded by vehicle n,
Figure BSA0000276505790000025
representing the maximum tolerable delay of the vehicle n for the task execution. Considering each time slot, the vehicle generates a task, and the unloading decision of the vehicle task is used
Figure BSA0000276505790000026
Is shown as a n 0 denotes that vehicle n performs the calculation task locally, a n 1 denotes the off-load of vehicle n to MEC server computation over V2I link, a n 2 means vehicle n offloads mission drone calculations over the V2V link. V2V communication and V2I communication do not interfere with each other using 5G slicing technique,
by collections
Figure BSA0000276505790000027
Indicating the task computation location, where loc, uav [ m ]]And MEC respectively represents that a calculation task is executed locally, the task is unloaded to the mth unmanned aerial vehicle for calculation, and the task is unloaded to the MEC server for calculation. For task-unloading positions
Figure BSA0000276505790000028
A sign, wherein
Figure BSA0000276505790000029
The calculation task representing vehicle n is performed at location z,
Figure BSA00002765057900000210
the calculation task representing vehicle n is not performed at location z.
(2b) The signal-to-interference-and-noise ratio (SINR) at which vehicle n offloads the mission to drone m is expressed as:
Figure BSA00002765057900000211
the transfer rate at which vehicle n offloads the task to drone m is expressed as:
Figure BSA00002765057900000212
wherein the content of the first and second substances,
Figure BSA00002765057900000213
transmission bandwidth, P, representing the vehicle offloading of tasks to the drone n Representing the transmitted power, σ, of the vehicle n 2 Representing work of noiseRate, h n，uav[m] Representing the channel gain from vehicle n to drone m,
Figure BSA0000276505790000031
indicating interference of vehicles other than vehicle n in offloading tasks to drone m' on vehicle n
Figure BSA0000276505790000032
Wherein
Figure BSA0000276505790000033
a n′ When 2, J (a) n′ 2) 1, otherwise J (a) n′ ＝2)＝0，P n′ Representing the transmitted power, h, of the vehicle n n′，uav[m] Representing the channel gain from vehicle n' to drone m;
(2c) likewise, the signal to interference plus noise ratio (SINR) at which vehicle n offloads the task to the MEC server is expressed as:
Figure BSA0000276505790000034
the transmission rate at which vehicle n offloads tasks to the MEC server is expressed as:
Figure BSA0000276505790000035
wherein the content of the first and second substances,
Figure BSA0000276505790000036
transmission bandwidth, P, representing the vehicle offloading of tasks to the MEC server n Representing the transmitted power, σ, of the vehicle n 2 Representing the noise power, h n，mec Representing the channel gain of vehicle n to the MEC server,
Figure BSA0000276505790000037
indicating that vehicles other than vehicle n offload tasks to MEC server for vehicle nInterference
Figure BSA0000276505790000038
Wherein
Figure BSA0000276505790000039
a n′ When 1, J (a) n′ 1, otherwise J (a) n′ ＝1)＝0，P n′ Representing the transmitted power, h, of the vehicle n n′，mec Representing the channel gain of vehicle n' to the MEC server;
(2d) establishing a calculation model, a n 0 means that vehicle n performs the calculation task locally,
Figure BSA00002765057900000310
representing the local computing power of vehicle n, the local computing latency is then:
Figure BSA00002765057900000311
a n 1 represents the calculation of unloading the task to the MEC server by the vehicle n through the V2R link, and the uploading time delay of the vehicle n to the MEC server is:
Figure BSA00002765057900000312
the computation time delay for vehicle n to upload a task to the MEC server is:
Figure BSA00002765057900000313
Figure BSA00002765057900000314
computing power allocated to the task of vehicle n for the MEC server.
a n 2 denotes a vehiclen is calculated by unloading the task to the unmanned aerial vehicle through the V2R link, and the uploading time delay of the task uploaded to the unmanned aerial vehicle m by the vehicle n is
Figure BSA0000276505790000041
The calculation time delay of the vehicle n uploading the task to the unmanned aerial vehicle m is
Figure BSA0000276505790000042
Figure BSA0000276505790000043
The computing power of the task allocated to vehicle n for drone m ignores the issue delay, so the delay of vehicle n offloading the task to the MEC server is
Figure BSA0000276505790000044
The time delay for vehicle n to offload a task to drone m is:
Figure BSA0000276505790000045
the time delay for local computation, task offloading to the MEC server, and task offloading to the drone may be expressed as:
Figure BSA0000276505790000046
(2e) in summary, the following objective functions and constraints can be established:
Figure BSA0000276505790000047
wherein, the constraint conditions C1 and C2 are shown in the tableThe tasks can be executed only locally, and are unloaded to the MEC server computation or the unmanned plane computation, each computation task can only select one computation mode, the constraint condition C3 indicates the local computation capability range of the vehicle n,
Figure BSA0000276505790000051
is the vehicle n local maximum computing power, constraints C4 and C5 mean that the computing power allocated to the vehicle by the MEC server and drone is not negative, constraints C6 and C7 indicate that the computing power allocated to the vehicle by the MEC server and drone cannot exceed their maximum computing power, F mec Is the maximum computing power of the MEC server, F uav[m] Is the maximum computing power of drone m; constraints C8 and C9 indicate that the task for vehicle n is offloaded to the MEC server or drone to perform calculations that satisfy its maximum latency constraint; the C10 constraint indicates that vehicle n transmit power is non-negative and satisfies its maximum transmit power constraint;
further, the step (4) comprises the following specific steps:
(4a) obtaining a decision whether each vehicle is unloaded or not based on the potential game, modeling the unloading decision of the task vehicle into the potential game, and expressing the potential game as
Figure BSA0000276505790000052
Wherein
Figure BSA0000276505790000053
Is a collection of vehicles, a n For the unloading decision of vehicle n, u n As a cost function for vehicle n.
In the game model, each vehicle is a resource competitor, so that N vehicles compete for limited resources in a network, and each vehicle can choose to unload calculation or execute task calculation locally, wherein a n E 0, 1 is the unloading decision for vehicle n,
Figure BSA0000276505790000054
set of unloading decisions representing all vehicles, a n 0 means that vehicle n performs the calculation task locally, a n 1 indicates that vehicle n is off-loading the missionTo the MEC server or drone. When the unloading decision of the vehicle n is a n When, its cost function is expressed as u n (a n ，a -n ) Wherein a is -n Representing the set of unloading decisions for all vehicles except vehicle n. Each vehicle may wish to minimize its own cost by finding the optimal unloading decision, i.e.
Figure BSA0000276505790000055
The potential game converges on Nash equilibrium, that is, the unloading decision is found through the optimal response iteration
Figure BSA0000276505790000056
The absence of all vehicles to change the current unloading decision can minimize its own cost.
(4b) Based on offloading decisions
Figure BSA0000276505790000057
By collections
Figure BSA0000276505790000058
Offloading decisions in a vehicle
Figure BSA0000276505790000059
Vehicle, N 0 To represent
Figure BSA00002765057900000510
The number of vehicles, defining states s as observation information and low-dimensional fingerprint information related to transmitting power and unloading nodes, including vehicles n 0 To unmanned aerial vehicle
Figure BSA00002765057900000511
Channel state information of
Figure BSA00002765057900000512
Vehicle n 0 Channel state information to MEC
Figure BSA00002765057900000513
Vehicle n 0 To unmanned aerial vehicle
Figure BSA00002765057900000514
Received of
Figure BSA00002765057900000515
Vehicle n 0 Received interference to MEC
Figure BSA00002765057900000516
Vehicle n 0 Task information of
Figure BSA00002765057900000517
Training turns e and the random exploration variable ε in the ε -greedy algorithm, i.e.
Figure BSA00002765057900000518
Will be provided with
Figure BSA00002765057900000519
Vehicles being treated as agents, each time based on current state
Figure BSA00002765057900000520
Selecting an unloading node and transmitting power;
(4c) defining each unloading-deciding vehicle n 0 Is selected offload node and transmit power, denoted as
Figure BSA00002765057900000521
Figure BSA0000276505790000061
For vehicles n 0 The selected one of the task off-load nodes,
Figure BSA0000276505790000062
for vehicles n 0 Discrete transmit power levels;
(4d) statorDefining a reward function r, the objective of offloading being an offloading decision
Figure BSA0000276505790000063
The vehicle selects an offload node and transmit power that minimizes all offload decisions under the constraint of satisfying maximum transmit power
Figure BSA0000276505790000064
The task processing latency of the vehicle, so the reward function can be expressed as:
Figure BSA0000276505790000065
where b is a fixed value used to adjust the value of the reward function,
(4e) establishing a deep reinforcement learning model on the basis of Q learning according to the established state, action and reward function, and determining n vehicles to be unloaded 0 Establishing corresponding evaluation function
Figure BSA0000276505790000066
Indicating vehicle n deciding to unload 0 Slave status
Figure BSA0000276505790000067
Performing an action
Figure BSA0000276505790000068
The resulting discount reward, the Q-value update function is:
Figure BSA0000276505790000069
wherein r is t For the instant reward function, gamma is the discount factor,
Figure BSA00002765057900000610
for deciding the number of unloaded vehicles n 0 Acquiring observation information and low-dimensional fingerprint information related to transmitting power and an unloading node at the moment t,
Figure BSA00002765057900000611
indicating vehicle n deciding to unload 0 Is carried out at time t
Figure BSA00002765057900000612
In the latter state of the system, the system is,
Figure BSA00002765057900000613
is an action
Figure BSA00002765057900000614
The formed motion space.
Further, the step (5) comprises the following specific steps:
(5a) starting the environment simulator to initialize the predicted network parameters of each agent
Figure BSA00002765057900000615
And target network parameters
Figure BSA00002765057900000616
(5b) Initializing the training round number P;
(5c) updating the vehicle position and the unmanned aerial vehicle position, acquiring the occupation condition of unmanned aerial vehicle and MEC calculation resources, task information and the like, and initializing a time step t in the P round;
(5d) asynchronously operating the prediction network for each agent based on the input state
Figure BSA00002765057900000617
Output motion
Figure BSA00002765057900000618
And obtain an instant prize r t While going to the next state
Figure BSA00002765057900000619
Thereby obtaining training data
Figure BSA00002765057900000620
(5e) To train data
Figure BSA00002765057900000621
Storing the experience data into respective experience playback pools;
(5f) each agent randomly samples N from the experience replay pool k Training data
Figure BSA00002765057900000622
Forming a data set D, inputting into a prediction network
Figure BSA00002765057900000623
(5g) Each agent calculates Loss value Loss (n) through the prediction network and the target network 0 ) Updating the agent prediction network by back propagation of the neural network using a small batch gradient descent strategy
Figure BSA00002765057900000624
The parameters of (1);
(5h) training times reach the target network updating interval, and network parameters are predicted according to the training times
Figure BSA0000276505790000071
Updating target network parameters
Figure BSA0000276505790000072
(5i) Judging whether t is less than K, wherein K is the total time step in the p round, if so, t is t +1, and entering the step (5c), otherwise, entering the step (5 j);
(5j) judging whether p is less than I or not, setting a threshold value for the training round number by I, if so, setting p to be p +1, and entering the step (5c), otherwise, finishing the optimization to obtain an optimized deep reinforcement learning model;
further, the step (6) comprises the following specific steps:
(6a) calculating resource occupation according to the position of the unmanned aerial vehicle and the MECAnd obtaining unloading decisions of the vehicles by the task information through potential game, and calculating the unloading decisions of the vehicles n which are not locally calculated for each unloading decision 0 Obtaining the state information of the moment
Figure BSA0000276505790000073
(6b) Each unloading decision not being a locally calculated vehicle n 0 Inputting state information by using a trained deep reinforcement learning model
Figure BSA0000276505790000074
(6c) Outputting the optimal action strategy, namely each unloading decision is not the vehicle n calculated locally 0 Offload node to obtain optimal vehicle selection
Figure BSA0000276505790000075
And transmit power
Figure BSA0000276505790000076
Has the advantages that: the invention provides an unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method, which adopts a mixed frequency spectrum access technology to transmit aiming at a scene of cooperative calculation unloading of a mobile edge server and an unmanned aerial vehicle, accesses different slices to a V2V link and a V2I based on a 5G slicing technology without mutual interference, obtains a decision of whether a vehicle is locally calculated through potential game, optimizes unloading nodes and transmitting power of the unloaded vehicle by adopting deep double-Q learning, and realizes task calculation by minimizing system delay. .
In conclusion, under the scene of collaborative computing and unloading of the mobile edge server and the unmanned aerial vehicle, the unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method provided by the invention is superior in minimizing system time delay.
Drawings
Fig. 1 is a flowchart of a method for task migration and resource allocation in heterogeneous internet of vehicles assisted by an unmanned aerial vehicle according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a system for task migration and resource allocation in heterogeneous internet of vehicles assisted by an unmanned aerial vehicle according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a DDQN algorithm framework provided by an embodiment of the present invention;
Detailed Description
The core idea of the invention is that: aiming at the scene of collaborative computing unloading of the mobile edge server and the unmanned aerial vehicle, a mixed spectrum access technology is adopted for transmission, a V2V link and a V2I are accessed into different slices based on a 5G slicing technology and are not interfered with each other, a decision of whether the vehicle is unloaded or not is obtained through potential game, the unloaded vehicle is regarded as an intelligent body, a deep reinforcement learning model is established, and the deep reinforcement learning model is optimized through deep double-Q learning. And obtaining the optimal unloading node and transmitting power for determining the unloading vehicle according to the optimized deep reinforcement learning model, thereby achieving the purpose of minimizing the system time delay.
The present invention is described in further detail below.
Step (1), a mobile edge computing server (MEC server) is deployed at a Road Side Unit (RSU), an unmanned aerial vehicle is deployed in the system to provide computing service for a vehicle, and computing tasks of the vehicle can be processed locally and unloaded to the unmanned aerial vehicle or the MEC server;
step (2), a communication model and a calculation model including N vehicles and M unmanned aerial vehicles are established, and then a joint calculation migration and resource allocation model is established, specifically:
(2a) the system comprises N vehicles, M unmanned aerial vehicles, a road side unit for deploying an MEC server, and a set for vehicles
Figure BSA0000276505790000081
Representation, collection
Figure BSA0000276505790000082
Denotes unmanned plane, vehicle nCan be expressed as
Figure BSA0000276505790000083
c n Representing the number of CPU cycles, s, required for vehicle n to complete a task n Indicating the size of the amount of task data unloaded by vehicle n,
Figure BSA0000276505790000084
representing the maximum tolerable delay of the vehicle n to the task execution. Considering each time slot, the vehicle generates a task, and the unloading decision of the vehicle task is used
Figure BSA0000276505790000085
Is shown as a n 0 denotes that vehicle n performs the calculation task locally, a n 1 denotes the offloading of tasks by vehicle n to MEC server computation over V2I link, a n 2 means vehicle n offloads mission drone calculations over the V2V link. V2V communication and V2I communication do not interfere with each other using 5G slicing technique,
by collections
Figure BSA0000276505790000086
Indicating the task computation location, where loc, uav [ m ]]And MEC respectively represents that a calculation task is executed locally, the task is unloaded to the mth unmanned aerial vehicle for calculation, and the task is unloaded to the MEC server for calculation. For task-unloading positions
Figure BSA0000276505790000087
A sign, wherein
Figure BSA0000276505790000088
The calculation task representing vehicle n is performed at location z,
Figure BSA0000276505790000089
the calculation task representing vehicle n is not performed at location z.
(2b) The signal-to-interference-and-noise ratio (SINR) at which vehicle n offloads the mission to drone m is expressed as:
Figure BSA00002765057900000810
the transfer rate at which vehicle n offloads the task to drone m is expressed as:
Figure BSA00002765057900000811
wherein the content of the first and second substances,
Figure BSA00002765057900000812
transmission bandwidth, P, representing the vehicle offloading of tasks to the drone n Representing the transmitted power, σ, of the vehicle n 2 Represents the noise power, h n，uav[m] Representing the channel gain from vehicle n to drone m,
Figure BSA00002765057900000813
indicating interference of vehicles other than vehicle n in offloading tasks to drone m' on vehicle n
Figure BSA00002765057900000814
Wherein
Figure BSA00002765057900000815
a n′ When 2, J (a) n′ 2) 1, otherwise J (a) n′ ＝2)＝0，P n′ Representing the transmitted power, h, of the vehicle n n′，uav[m] Representing the channel gain from vehicle n' to drone m;
(2c) likewise, the signal to interference plus noise ratio (SINR) at which vehicle n offloads the task to the MEC server is expressed as:
Figure BSA0000276505790000091
the transmission rate at which vehicle n offloads tasks to the MEC server is expressed as:
Figure BSA0000276505790000092
wherein the content of the first and second substances,
Figure BSA0000276505790000093
transmission bandwidth, P, representing the vehicle offloading tasks to the MEC server n Representing the transmitted power, σ, of the vehicle n 2 Representing the noise power, h n，mec Representing the channel gain of the vehicle n to the MEC server,
Figure BSA0000276505790000094
representing disturbances to vehicle n by vehicles other than vehicle n offloading tasks to MEC server
Figure BSA0000276505790000095
Wherein
Figure BSA0000276505790000096
a n′ When 1, J (a) n′ 1, otherwise J (a) n′ ＝1)＝0，P n′ Representing the transmitted power, h, of the vehicle n n′，mec Representing the channel gain of vehicle n' to the MEC server;
(2d) establishing a calculation model, a n 0 means that vehicle n performs the calculation task locally,
Figure BSA0000276505790000097
representing the local computing power of vehicle n, the local computing latency is:
Figure BSA0000276505790000098
a n 1 represents the calculation of unloading the task to the MEC server by the vehicle n through the V2R link, and the uploading time delay of the vehicle n to the MEC server is:
Figure BSA0000276505790000099
the computation time delay for vehicle n to upload a task to the MEC server is:
Figure BSA00002765057900000910
Figure BSA00002765057900000911
computing power allocated to the task of vehicle n for the MEC server.
a n 2 represents that the vehicle n unloads the task through the V2R link, and the uploading time delay of the vehicle n uploading the task to the unmanned aerial vehicle m is
Figure BSA00002765057900000912
The calculation time delay of the vehicle n uploading the task to the unmanned aerial vehicle m is
Figure BSA00002765057900000913
Figure BSA00002765057900000914
The computing power of the task allocated to vehicle n for unmanned aerial vehicle m ignores the issue delay, so the delay generated when vehicle n unloads the task to the MEC server is
Figure BSA0000276505790000101
The time delay for the vehicle n to unload the task to the drone m is:
Figure BSA0000276505790000102
the latency of local computation, task offloading to the MEC server, and task offloading to the drone may be expressed as:
Figure BSA0000276505790000103
(2e) in summary, the following objective functions and constraints can be established:
Figure BSA0000276505790000104
wherein, the constraint conditions C1 and C2 indicate that the tasks can be executed locally, are unloaded to an MEC server for calculation or are unloaded to an unmanned aerial vehicle for calculation, each calculation task can only select one calculation mode, the constraint condition C3 indicates the local calculation capability range of the vehicle n,
Figure BSA0000276505790000105
is the vehicle n local maximum computing power, constraints C4 and C5 mean that the computing power allocated to the vehicle by the MEC server and drone is not negative, constraints C6 and C7 indicate that the computing power allocated to the vehicle by the MEC server and drone cannot exceed their maximum computing power, F mec Is the maximum computing power of the MEC server, F uav[m] Is the maximum computing power of drone m; constraints C8 and C9 indicate that the task of vehicle n is offloaded to the MEC server or the drone performs the calculation to meet its maximum latency constraint; the C10 constraint indicates that vehicle n transmit power is non-negative and satisfies its maximum transmit power constraint;
step (3), each vehicle acquires the positions of the unmanned aerial vehicle and the MEC, the occupation condition of computing resources and task information;
and (4) obtaining a decision whether each vehicle is unloaded or not based on the potential game, and establishing a deep reinforcement learning model for the vehicle determining task unloading with the goal of reducing system time delay according to the obtained vehicle unloading decision, wherein the method comprises the following specific steps:
(4a) obtaining each vehicle based on potential gameAnd (4) making a decision of unloading, namely modeling the unloading decision of the task vehicle into a potential game, which is expressed as
Figure BSA0000276505790000111
Wherein
Figure BSA0000276505790000112
Is a collection of vehicles, a n For the unloading decision of vehicle n, u n Cost function n for vehicle n
In the game model, each vehicle is a resource competitor, so that N vehicles compete for limited resources in a network, each vehicle can choose to unload calculation or execute task calculation locally, wherein a n E 0, 1 is the unloading decision for vehicle n,
Figure BSA0000276505790000113
set of unloading decisions representing all vehicles, a n 0 means that vehicle n performs the calculation task locally, a n 1 means that vehicle n offloads the task to the MEC server or drone for computation. When the unloading decision of the vehicle n is a n When, its cost function is expressed as u n (a n ，a -n ) Wherein a is -n Representing the set of unloading decisions for all vehicles except vehicle n. Each vehicle may wish to minimize its own cost by finding an optimal unloading decision, i.e.
Figure BSA0000276505790000114
The potential game converges on Nash equilibrium, that is, the unloading decision is found through the optimal response iteration
Figure BSA0000276505790000115
The absence of all vehicles changing the current unloading decision minimizes its own cost.
(4b) Based on offload decisions
Figure BSA0000276505790000116
By collections
Figure BSA0000276505790000117
Decision making for unloading in vehicle
Figure BSA0000276505790000118
Vehicle, N 0 To represent
Figure BSA0000276505790000119
The number of vehicles, defining states s as observation information and low-dimensional fingerprint information related to transmitting power and unloading nodes, including vehicles n 0 To unmanned aerial vehicle
Figure BSA00002765057900001110
Channel state information of
Figure BSA00002765057900001111
Vehicle n 0 Channel state information to MEC
Figure BSA00002765057900001112
Vehicle n 0 To unmanned aerial vehicle
Figure BSA00002765057900001113
Received of
Figure BSA00002765057900001114
Vehicle n 0 Received interference to MEC
Figure BSA00002765057900001115
Vehicle n 0 Task information of
Figure BSA00002765057900001116
Training round numbers e and random exploratory variables ε in the ε -greedy algorithm
Figure BSA00002765057900001117
Will be provided with
Figure BSA00002765057900001118
Vehicle-to-vehicle agent, each time based on current state
Figure BSA00002765057900001119
Selecting an unloading node and transmitting power;
(4c) defining each unloading-deciding vehicle n 0 Is selected offload node and transmit power, denoted as
Figure BSA00002765057900001120
Figure BSA00002765057900001121
For vehicles n 0 The selected task of the off-load node,
Figure BSA00002765057900001122
for vehicles n 0 Discrete transmit power levels;
(4d) defining a reward function r, the objective of offloading being an offloading decision
Figure BSA00002765057900001123
The vehicle selects an offload node and transmit power that minimizes all offload decisions under the constraint of satisfying maximum transmit power
Figure BSA00002765057900001124
The mission handling delay of the vehicle, so the reward function can be expressed as:
Figure BSA00002765057900001125
where b is a fixed value used to adjust the value of the reward function,
(4e) according to the established state, action and reward function, a deep reinforcement learning model is established on the basis of Q learning, and each vehicle determines n unloaded vehicles 0 Establishing corresponding evaluation function
Figure BSA0000276505790000121
Indicating vehicle n deciding to unload 0 Slave status
Figure BSA0000276505790000122
Performing an action
Figure BSA0000276505790000123
The resulting discount reward, the Q-value update function is:
Figure BSA0000276505790000124
wherein r is t For the instant reward function, gamma is the discount factor,
Figure BSA0000276505790000125
for deciding the number of unloaded vehicles n 0 Acquiring observation information and low-dimensional fingerprint information related to transmitting power and an unloading node at the moment t,
Figure BSA0000276505790000126
indicating vehicle n deciding to unload 0 Is carried out at time t
Figure BSA0000276505790000127
In the latter state of the system, the system is,
Figure BSA0000276505790000128
is an action
Figure BSA0000276505790000129
The formed motion space.
Step (5), training a deep reinforcement learning model based on the DDQN, and specifically comprising the following steps:
(5a) starting the environment simulator to initialize the predicted network parameters of each agent
Figure BSA00002765057900001210
And target network parameters
Figure BSA00002765057900001211
(5b) Initializing training round number P;
(5c) updating the vehicle position and the unmanned aerial vehicle position, acquiring the occupation condition of unmanned aerial vehicle and MEC calculation resources, task information and the like, and initializing a time step t in the P round;
(5d) asynchronously operating the prediction network for each agent based on the input state
Figure BSA00002765057900001212
Output motion
Figure BSA00002765057900001213
And obtains the real-time reward rt and simultaneously turns to the next state
Figure BSA00002765057900001214
Thereby obtaining training data
Figure BSA00002765057900001215
(5e) To train data
Figure BSA00002765057900001216
Storing the experience data into respective experience playback pools;
(5f) each agent randomly samples N from the experience replay pool k Training data
Figure BSA00002765057900001217
Forming a data set D, inputting into a prediction network
Figure BSA00002765057900001218
(5g) Each agent calculates Loss value Loss (n) through the prediction network and the target network 0 ) Updating the agent prediction network by back propagation of the neural network using a small batch gradient descent strategy
Figure BSA00002765057900001219
The parameters of (1);
(5h) training times reach the target network updating interval, and network parameters are predicted according to the training times
Figure BSA00002765057900001220
Updating target network parameters
Figure BSA00002765057900001221
(5i) Judging whether t is less than K, wherein K is the total time step in the p round, if so, t is t +1, and entering the step (5c), otherwise, entering the step (5 j);
(5j) judging whether p is less than I or not, setting a threshold value for the training round number by I, if so, setting p to be p +1, and entering the step (5c), otherwise, finishing the optimization to obtain an optimized deep reinforcement learning model;
step (6), in the execution stage, judging whether the task is unloaded or not by the vehicle n with the calculation task through potential game, and determining the unloaded vehicle n 0 Obtaining a current state from local observations
Figure BSA0000276505790000131
The method for obtaining the unloading node and the transmitting power of the vehicle by using the trained deep reinforcement learning model specifically comprises the following steps:
(6a) obtaining unloading decisions of vehicles through potential game according to the position of the unmanned aerial vehicle, the occupation condition of MEC computing resources and task information, and carrying out local computation on each unloading decision 0 Obtaining the state information of the moment
Figure BSA0000276505790000132
(6b) Each unloading decision not being a locally calculated vehicle n 0 Inputting state information by using a trained deep reinforcement learning model
Figure BSA0000276505790000133
(6c) Output the optimal action strategy, i.e. eachVehicle n with vehicle unloading decisions not calculated locally 0 Offload node to obtain optimal vehicle selection
Figure BSA0000276505790000134
And transmit power
Figure BSA0000276505790000135
In fig. 1, a flow chart of an unmanned aerial vehicle-assisted heterogeneous internet of vehicles task migration and resource allocation method is described, and first, whether a task is unloaded is judged by a vehicle through a potential game, and the unloaded vehicle is determined to select an unloading node and transmission power based on a DDQN trained deep reinforcement learning model.
In fig. 2, a system model of drone-assisted heterogeneous internet of vehicles task migration and resource allocation is described, and it can be seen that the MEC server and the drone can provide computing services for the vehicle.
In fig. 3, an algorithmic framework of a DDQN is depicted, the DDQN comprising two networks, a predicted network and a target network, respectively.
According to the description of the invention, it should be obvious to those skilled in the art that the invention provides the unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method, which can effectively reduce the system delay and achieve good balance between complexity and performance.
Details not described in the present application are well within the skill of those in the art.

Claims (1)

1. An unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method is characterized by comprising the following steps:
(1) deploying a Mobile Edge Computing (MEC) server at a Road Side Unit (RSU), deploying an unmanned aerial vehicle in the system to provide computing service for a vehicle, and locally processing computing tasks of the vehicle and unloading the computing tasks to the unmanned aerial vehicle or the MEC server;
(2) establishing a communication model and a calculation model comprising N vehicles and M unmanned aerial vehicles, and further establishing a joint calculation migration and resource allocation model;
(3) each vehicle acquires the positions of the unmanned aerial vehicle and the MEC, the occupation condition of computing resources and task information;
(4) obtaining a decision whether each vehicle is unloaded or not based on the potential game, and establishing a deep reinforcement learning model for the vehicle determining task unloading with the goal of reducing system time delay according to the obtained vehicle unloading decision;
(5) training a deep reinforcement learning model based on the DDQN;
(6) in the execution stage, the vehicle n with the calculation task judges whether the task is unloaded through potential game, and determines the unloaded vehicle n 0 Obtaining a current state from local observations
Figure FSA0000276505780000011
Obtaining unloading nodes and transmitting power of the vehicle by using the trained deep reinforcement learning model;
further, the step (4) comprises the following specific steps:
(4a) obtaining a decision whether each vehicle is unloaded or not based on the potential game, modeling the unloading decision of the task vehicle into the potential game, and expressing the potential game as
Figure FSA0000276505780000012
Wherein
Figure FSA0000276505780000013
Is a collection of vehicles, a n For the unloading decision of vehicle n, u n As a cost function of the vehicles N, each vehicle is a resource competitor in the game model, so that limited resources exist in a competition network of N vehicles, each vehicle can choose to unload calculation or locally execute task calculation, wherein a n E 0, 1 is the unloading decision for vehicle n,
Figure FSA0000276505780000014
set of unloading decisions representing all vehicles, a n 0 means that vehicle n performs the calculation task locally, a n 1 means that vehicle n offloads task to MEC server or drone for computationWhen the unloading decision of the vehicle n is a n When, its cost function is expressed as u n (a n ，a -n ) Wherein a is -n Representing a set of unloading decisions for all vehicles except vehicle n, each of which may wish to minimize its own cost by finding the optimal unloading decision, i.e. the vehicle n
Figure FSA0000276505780000015
Wherein
Figure FSA0000276505780000016
The time delay of the task is calculated locally for the vehicle n,
Figure FSA0000276505780000017
for the time delay for vehicle n to offload tasks to the MEC server,
Figure FSA0000276505780000018
the potential game converges to nash equilibrium for the time delay of unloading the task to the unmanned plane calculation for vehicle n, i.e. the unloading decision is found through the optimal response iteration
Figure FSA0000276505780000019
All vehicles do not change the current unloading decision so that the cost per se can be minimized;
(4b) based on offload decisions
Figure FSA00002765057800000110
By collections
Figure FSA00002765057800000111
Decision making for unloading in vehicle
Figure FSA00002765057800000112
Vehicle, N 0 To represent
Figure FSA00002765057800000113
Number of vehicles, decision to unload from vehicle
Figure FSA00002765057800000114
The vehicle is regarded as an intelligent agent, and the state s is defined as observation information and low-dimensional fingerprint information related to the transmitting power and the unloading node, including the vehicle n 0 To unmanned aerial vehicle
Figure FSA00002765057800000115
Channel state information of
Figure FSA00002765057800000116
Vehicle n 0 Channel state information to MEC
Figure FSA00002765057800000117
Vehicle n 0 To unmanned aerial vehicle
Figure FSA00002765057800000118
Received of
Figure FSA00002765057800000119
Vehicle n 0 Received interference to MEC
Figure FSA0000276505780000021
Vehicle n 0 Task information of
Figure FSA0000276505780000022
Training round numbers e and random exploratory variables ε in the ε -greedy algorithm
Figure FSA0000276505780000023
Will be provided with
Figure FSA0000276505780000024
The vehicle is regarded as an intelligent agent, and n is taken every time 0 Based on the current state
Figure FSA0000276505780000025
Selecting an unloading node and transmitting power;
(4c) defining each unloading-deciding vehicle n 0 Is selected offload node and transmit power, denoted as
Figure FSA0000276505780000026
Figure FSA0000276505780000027
For vehicles n 0 The selected task of the off-load node,
Figure FSA0000276505780000028
for vehicles n 0 A set of selectable task off-load nodes,
Figure FSA0000276505780000029
for vehicles n 0 Discrete transmit power levels;
(4d) defining a reward function r, the objective of offloading being an offloading decision
Figure FSA00002765057800000210
The vehicle selects an offload node and transmit power that minimizes all offload decisions under the constraint of satisfying maximum transmit power
Figure FSA00002765057800000211
The mission handling delay of the vehicle, so the reward function can be expressed as:
Figure FSA00002765057800000212
where b is a fixed value used to adjust the value of the reward function,
Figure FSA00002765057800000213
indicating a vehicle n 0 At position z 0 The execution is carried out in such a way that,
Figure FSA00002765057800000214
indicating a vehicle n 0 Is not at position z 0 The execution is carried out in such a way that,
Figure FSA00002765057800000215
indicating a vehicle n 0 At position z 0 The time delay of execution;
(4e) according to the established state, action and reward function, a deep reinforcement learning model is established on the basis of Q learning, and each vehicle determines n unloaded vehicles 0 Establishing corresponding evaluation function
Figure FSA00002765057800000216
Indicating vehicle n deciding to unload 0 Slave status
Figure FSA00002765057800000217
Performing an action
Figure FSA00002765057800000218
The resulting discount reward, the Q-value update function is:
Figure FSA00002765057800000219
wherein r is t For the instant reward function, gamma is the discount factor,
Figure FSA00002765057800000220
for deciding the number of unloaded vehicles n 0 Acquiring observation information and low-dimensional fingerprint information related to transmitting power and an unloading node at the moment t,
Figure FSA00002765057800000221
indicating vehicle n deciding to unload 0 Is carried out at time t
Figure FSA00002765057800000222
In the latter state of the system, the system is,
Figure FSA00002765057800000223
is an action
Figure FSA00002765057800000224
The formed motion space.
CN202210744842.2A 2022-06-28 2022-06-28 Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method Active CN115037751B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210744842.2A CN115037751B (en) 2022-06-28 2022-06-28 Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210744842.2A CN115037751B (en) 2022-06-28 2022-06-28 Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method

Publications (2)

Publication Number Publication Date
CN115037751A true CN115037751A (en) 2022-09-09
CN115037751B CN115037751B (en) 2023-05-05

Family

ID=83127119

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210744842.2A Active CN115037751B (en) 2022-06-28 2022-06-28 Unmanned aerial vehicle-assisted heterogeneous Internet of vehicles task migration and resource allocation method

Country Status (1)

Country Link
CN (1) CN115037751B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200186964A1 (en) * 2018-12-07 2020-06-11 T-Mobile Usa, Inc. Uav supported vehicle-to-vehicle communication
CN111713126A (en) * 2017-11-15 2020-09-25 联想(北京)有限公司 Exchange of UL interference detection related information
CN112543115A (en) * 2020-11-13 2021-03-23 北京科技大学 Unmanned aerial vehicle edge computing network resource allocation method and device based on block chain
CN113543074A (en) * 2021-06-15 2021-10-22 南京航空航天大学 Joint computing migration and resource allocation method based on vehicle-road cloud cooperation
CN114584951A (en) * 2022-03-08 2022-06-03 南京航空航天大学 Combined computing unloading and resource allocation method based on multi-agent DDQN
CN114626298A (en) * 2022-03-14 2022-06-14 北京邮电大学 State updating method for efficient caching and task unloading in unmanned aerial vehicle-assisted Internet of vehicles
CN114650567A (en) * 2022-03-17 2022-06-21 江苏科技大学 Unmanned aerial vehicle-assisted V2I network task unloading method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111713126A (en) * 2017-11-15 2020-09-25 联想(北京)有限公司 Exchange of UL interference detection related information
US20200186964A1 (en) * 2018-12-07 2020-06-11 T-Mobile Usa, Inc. Uav supported vehicle-to-vehicle communication
CN112543115A (en) * 2020-11-13 2021-03-23 北京科技大学 Unmanned aerial vehicle edge computing network resource allocation method and device based on block chain
CN113543074A (en) * 2021-06-15 2021-10-22 南京航空航天大学 Joint computing migration and resource allocation method based on vehicle-road cloud cooperation
CN114584951A (en) * 2022-03-08 2022-06-03 南京航空航天大学 Combined computing unloading and resource allocation method based on multi-agent DDQN
CN114626298A (en) * 2022-03-14 2022-06-14 北京邮电大学 State updating method for efficient caching and task unloading in unmanned aerial vehicle-assisted Internet of vehicles
CN114650567A (en) * 2022-03-17 2022-06-21 江苏科技大学 Unmanned aerial vehicle-assisted V2I network task unloading method

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
. MESSOUS; HICHEM SEDJELMACI; NOUREDDIN HOUARI; S. SENOUCI: "Computation offloading game for an UAV network in mobile edge computing", 2017 IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS (ICC) *
ATA KHALILI; SHEYDA ZARANDI; MEHDI RASTI: "Joint Resource Allocation and Offloading Decision in Mobile Edge Computing", IEEE COMMUNICATIONS LETTERS *
YANLING REN; ZHIBIN XIE; ZHENFENG DING; XIYUAN SUN;JIE XIA; YUBO TIAN: "Computation offloading game in multiple unmanned aerial vehicle-enabled mobile edge computing networks", IET COMMUNICATION *
宋晓勤; 谈雅竹; 董莉; 王健康; 胡静; 宋铁成: "基于拟牛顿内点法的认知车联网能效优先资源分配算法", 东南大学学报（自然科学版） *
董超; 沈赟; 屈毓锛: "基于无人机的边缘智能计算研究综述", 智能科学与技术学报 *

Also Published As

Publication number Publication date
CN115037751B (en) 2023-05-05

Similar Documents

Publication Publication Date Title
CN111414252B (en) Task unloading method based on deep reinforcement learning
CN113543074B (en) Joint computing migration and resource allocation method based on vehicle-road cloud cooperation
CN113254188B (en) Scheduling optimization method and device, electronic equipment and storage medium
CN113032904B (en) Model construction method, task allocation method, device, equipment and medium
Callegaro et al. Optimal edge computing for infrastructure-assisted UAV systems
CN113395654A (en) Method for task unloading and resource allocation of multiple unmanned aerial vehicles of edge computing system
CN112929849B (en) Reliable vehicle-mounted edge calculation unloading method based on reinforcement learning
CN115640131A (en) Unmanned aerial vehicle auxiliary computing migration method based on depth certainty strategy gradient
CN114630397B (en) Unmanned aerial vehicle access selection method based on time slot division
CN111132074A (en) Multi-access edge computing unloading and frame time slot resource allocation method in Internet of vehicles environment
CN115134242B (en) Vehicle-mounted computing task unloading method based on deep reinforcement learning strategy
CN114205353B (en) Calculation unloading method based on hybrid action space reinforcement learning algorithm
CN113645273B (en) Internet of vehicles task unloading method based on service priority
CN115174481B (en) Method for assisting edge vehicle networking service and content hybrid caching by unmanned aerial vehicle
WO2022242468A1 (en) Task offloading method and apparatus, scheduling optimization method and apparatus, electronic device, and storage medium
CN114650567A (en) Unmanned aerial vehicle-assisted V2I network task unloading method
CN116916386A (en) Large model auxiliary edge task unloading method considering user competition and load
CN117580105B (en) Unmanned aerial vehicle task unloading optimization method for power grid inspection
CN116030623A (en) Collaborative path planning and scheduling method based on blockchain in cognitive Internet of vehicles scene
CN114626298A (en) State updating method for efficient caching and task unloading in unmanned aerial vehicle-assisted Internet of vehicles
CN115052262A (en) Potential game-based vehicle networking computing unloading and power optimization method
CN117528649A (en) Method for establishing end-edge cloud system architecture, task unloading and resource allocation optimization method and end-edge cloud system architecture
CN117580063A (en) Multi-dimensional resource collaborative management method in vehicle-to-vehicle network
CN116634401A (en) Task unloading method for maximizing satisfaction of vehicle-mounted user under edge calculation
CN115967430A (en) Cost-optimal air-ground network task unloading method based on deep reinforcement learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant