CN111240821B - Collaborative cloud computing migration method based on Internet of vehicles application security grading - Google Patents

Collaborative cloud computing migration method based on Internet of vehicles application security grading Download PDF

Info

Publication number
CN111240821B
CN111240821B CN202010034600.5A CN202010034600A CN111240821B CN 111240821 B CN111240821 B CN 111240821B CN 202010034600 A CN202010034600 A CN 202010034600A CN 111240821 B CN111240821 B CN 111240821B
Authority
CN
China
Prior art keywords
task request
task
cloud computing
vehicle
computing
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.)
Active
Application number
CN202010034600.5A
Other languages
Chinese (zh)
Other versions
CN111240821A (en
Inventor
胡斌杰
段海波
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.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
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 South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN202010034600.5A priority Critical patent/CN111240821B/en
Publication of CN111240821A publication Critical patent/CN111240821A/en
Application granted granted Critical
Publication of CN111240821B publication Critical patent/CN111240821B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5061—Partitioning or combining of resources
    • G06F9/5072—Grid computing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/0231—Traffic management, e.g. flow control or congestion control based on communication conditions
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/08—Load balancing or load distribution
    • 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]
    • H04W4/44—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for communication between vehicles and infrastructures, e.g. vehicle-to-cloud [V2C] or vehicle-to-home [V2H]
    • 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]
    • H04W4/48—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for in-vehicle communication
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/50—Indexing scheme relating to G06F9/50
    • G06F2209/502—Proximity

Abstract

The invention discloses a collaborative cloud computing migration method based on Internet of vehicles application security grading, which comprises the following steps: acquiring a base station and relevant data information of task requests generated by vehicles within the range of the base station, and determining the safety factor of each task request, wherein the coefficient is the quantification of the influence of the task requests on the safety of the vehicles; deploying a task request meeting certain requirements on a vehicle-mounted terminal for execution; the method comprises the steps of coordinating edge computing and cloud computing, dividing task requests which cannot be deployed at a vehicle-mounted terminal into two types, and updating feasible solution domains of each type; and combining the improved discrete artificial bee colony algorithm with a greedy selection strategy to carry out optimization solution, thereby obtaining a migration decision and resource allocation scheme for ensuring the safety priority of the vehicle. The method has the advantages that the influence of the task request on the safety of the vehicle is graded, the complementary advantages of the vehicle-mounted terminal, the edge computing and the cloud computing are fully utilized, and the safety of the vehicle is greatly improved.

Description

Collaborative cloud computing migration method based on Internet of vehicles application security grading
Technical Field
The invention relates to the technical field of Internet of vehicles and cloud computing, in particular to a collaborative cloud computing migration method based on Internet of vehicles application security grading.
Background
In recent years, with the dramatic increase in the number of automobiles, road bearing capacity has come to near saturation in many large cities, and traffic safety, efficiency, and environmental issues have become more prominent. The internet of vehicles also gradually enters the visual field of people, and the internet of vehicles technology means that vehicles interact with other vehicles (V2V), roadside devices (V2I), pedestrians (V2P) and the like through information communication technology, so that the intelligence of the vehicles is improved. In particular, the rapid development of applications such as collision avoidance, driving assistance, and automatic parking have made these types of applications have a large impact on the safety of the vehicle, so there is an urgent need to prioritize the successful implementation of these applications. However, the computing resources of the vehicle-mounted terminal are generally limited, and many tasks with low delay constraints and computational intensive performance cannot be executed at the local terminal.
The cloud computing has been developed rapidly in recent years because of its advantages such as high reliability, low cost, high scalability, etc., and users send task request data to a cloud computing center for processing, and then download computing results to clients. The cloud computing server occupies a mainstream market for a long time due to abundant computing resources, but is often far away from users, and a high time delay is generated in data transmission, so that execution of some applications with strict time delay constraints is greatly limited, and especially some car networking applications related to vehicle safety cannot be successfully executed in time. Subsequently ETSI proposed Mobile Edge Computing (Mobile Edge Computing), which later evolved into Multi-Access Edge Computing (Multi-Access Edge Computing) as demand varied. The edge computing mainly means that a server with certain computing resources and storage capacity is deployed near a base station, namely, at a position which is one-hop away from a user, and compared with the traditional cloud computing, the edge computing greatly saves transmission delay. However, since the computing resources of the edge cloud are generally limited, only a limited number of requests can be processed, and the edge cloud cannot completely meet the requirements of task requests in the car networking environment in the face of more and more car networking task requests.
During the research and the proposal of the invention, at least: in order to ensure the safety of the vehicle, the safety application and the non-safety application in the internet of vehicles are treated distinctively, and the cloud computing and the edge computing are combined to perform cooperative auxiliary processing on the task request, so that the defects among the task request and the task request can be greatly overcome, and finally, the application which has the larger influence on the safety of the vehicle is successfully executed with higher priority.
Disclosure of Invention
In view of the above, the invention provides a collaborative cloud computing migration method based on vehicle networking application security classification, which aims to solve the problems that no consideration is given to vehicle security when cloud computing migration decision is made in the existing vehicle networking, and the existing research mainly only considers task execution of an edge computing auxiliary vehicle and neglects advantage complementation of edge computing and cloud computing, so that quantitative classification is realized according to the importance degree of vehicle networking task requests on vehicle security, complementary advantages of cloud computing and edge computing are fully utilized, priority migration of safe task requests is realized, and as many task requests as possible are successfully executed while vehicle security is guaranteed.
The technical purpose of the invention is realized by the following technical scheme:
a collaborative cloud computing migration method based on Internet of vehicles application security grading is characterized in that an edge computing server is deployed near a single base station and connected with a cloud computing center, the base station adopts an OFDMA access mode, an uncertain number of vehicles exist in the coverage range of the base station, task requests generated by the vehicles can be selected to be executed at a vehicle-mounted terminal, an edge computing end or a cloud computing end, and the collaborative cloud computing migration method comprises the following steps:
s1, acquiring the information of the current base station and the relevant data of each task request, and determining the safety factor of each task request: assuming that m task requests are generated in total, the data information St of the corresponding base station, the task request set T and the data information T of any task request i are obtainediRespectively expressed as follows:
St＝[N，a，F，spkb，nmax]
T＝{0，1，…，m-1}
Ti＝[Di，Ci，τi，Si]
wherein N, a, F, spkb, NmaxRespectively representing the total number of sub-channels available for the current base station, the bandwidth of a single sub-channel, the available computing resources of an edge computing terminal, the transmission time delay from the transmission unit data amount of the base station to a cloud computing terminal, the maximum number of the assignable sub-channels of each task request, and Di、Ci、τi、SiRespectively representing the data input quantity of the task request i, the execution cycle number corresponding to the unit data quantity, the time delay constraint and the safety factor; wherein a safety factor SiThe influence of the task request i on the safety of the vehicle is quantified when the safety factor S is reachedi≥SthThen, the task request belongs to a secure task request, wherein SthIs a safe threshold;
s2, sequentially judging whether the task requests in the task request set T can be executed at the vehicle-mounted terminal, wherein the principle that the task request i belongs to T and can be executed at the vehicle-mounted terminal is as follows:
Figure GDA0003487709630000031
wherein
Figure GDA0003487709630000032
Deploying the time delay executed at the vehicle-mounted terminal for the task request i,
Figure GDA0003487709630000033
initializing a task request set L (phi) executed by the vehicle-mounted terminal for a corresponding vehicle-mounted terminal computing resource, and if a task request i meets the conditions, then L (L + i) and T (T \ i), wherein L (L + i) represents adding an element i to the set L, and T (T \ i) represents removing the element i from the set T;
the migration decision of any task request i in T obtained in step S2 is xiE { -1, 0, 1}, which respectively represent that the sub-channel can be migrated to an edge computing end, abandoned to be executed and migrated to a cloud computing end, and the distribution number of the sub-channels is ni∈{0，1，…，nmax}，nmaxThe number of the sub-channels which can be allocated at most is shown in each task request, and the calculation resource allocation is fi∈[0，F](ii) a And when the mobile terminal is migrated to the edge computing terminal, the time delay of downloading the computing result is ignored, when the mobile terminal is migrated to the cloud computing terminal, the computing time delay of the cloud computing terminal and the time delay of downloading the computing result are ignored, and the time delays of the two migration modes
Figure GDA0003487709630000034
And
Figure GDA0003487709630000035
the expressions are respectively as follows:
Figure GDA0003487709630000036
Figure GDA0003487709630000037
wherein
Figure GDA0003487709630000041
RiThe spkb respectively represents a data uploading rate when a single sub-channel is allocated to the task request i, a transmission time delay from a base station to a cloud computing end, and Ri、di、α、|hi|、σ2And respectively representing the transmitting power of the vehicle-mounted terminal corresponding to the task request i, the distance between the corresponding vehicle-mounted terminal and the base station, a path attenuation factor, a Rayleigh fading factor and Gaussian white noise.
S3, classifying the task requests in the set T into the set T one by one according to whether the task requests are migrated to the cloud computing end to be executed and can meet the time delay constraintmecOr TbothFinally, determining a feasible solution domain of each task request in the set T;
wherein T ismecThe task request in (1) cannot be selectively migrated to the cloud computing end for execution, and TbothThe task requests in the method can be selectively migrated to an edge computing end and a cloud computing end for execution, and any classified task request i meets the condition that the classified task request i belongs to (i belongs to T ^ i belongs to T)mec)∨(i∈T∧i∈Tboth)；
S31, classifying all task requests in the task request set T into the set TmecOr TbothIn, assume an initial Tmec＝φ，TbothFor any task request i e T, the corresponding classification criteria are:
Figure GDA0003487709630000042
wherein
Figure GDA0003487709630000043
Refers to the assignment of n to task request imaxTime delay generated by execution of each subchannel and migration to a cloud computing end is judged, if the time delay still does not meet time delay constraint, the task request can be successfully executed only by migration to edge computing, and the task request is classified into a set TmecOtherwise, classify the task request into the set TbothPerforming the following steps;
s32, the pair satisfies (i belongs to T ^ i belongs to Tmec)∨(i∈T∧i∈Tboth) The principle of all task requests i to update the feasible solution domain is as follows:
Figure GDA0003487709630000044
Figure GDA0003487709630000045
wherein
Figure GDA0003487709630000046
Wherein
Figure GDA0003487709630000047
Representing any task request i ∈ TbothMigrating to a cloud computing end to execute the number of sub-channels R required for correspondingiIndicating the data upload rate at which task request i is assigned a single subchannel,ceil (x) denotes rounding up x; at this time, the number of sub-channels allocated to the task request i
Figure GDA0003487709630000051
When the selection is moved to the cloud computing end for execution, when
Figure GDA0003487709630000052
Considering the execution of the edge computing end or abandoning the execution; the range of the feasible solution is actually reduced through the updating of the feasible solution domain, so that the complexity of subsequent algorithm processing can be reduced;
s4, migration decision and resource allocation solution are carried out on the task requests in the task request set T through the combination of the improved discrete artificial bee colony algorithm and the greedy selection strategy:
s41, the fitness function in the improved discrete artificial bee colony algorithm of the invention is as follows:
Figure GDA0003487709630000053
wherein
Figure GDA0003487709630000054
fit is the fitness value, fit (i) is the reward value for successful execution of task request i, and fsafe(Si) Is a reward value for the successful execution of a secure task request i, which is positively correlated with a safety factor funsafe(Si) Is a reward value for the successful execution of the non-secure task request i, which is independent of the safety factor; in order to improve the safety of the vehicle, the reward value required by the safe task request is required to be higher than that required by the non-safe task, and the reward value corresponding to the safe task request with higher safety coefficient is higher;
s42, distributing the number n of sub-channels by the variable to be optimized in the improved discrete artificial bee colony algorithmiAssuming there are k task requests to be migrated for decision, the pth honey source n in the algorithmpExpressed as:
Figure GDA0003487709630000055
wherein
Figure GDA0003487709630000056
Represents the sub-channel distribution number corresponding to the ith task request in the pth honey source, and i belongs to [0, 1, …, k-1 ∈]；
The improved discrete artificial bee colony algorithm comprises three stages of operation of hiring bees, following bees and scouting bees, and comprises the following specific steps of:
s421, adopting a neighborhood searching mode based on the combination of DE/best/1 strategy and self-adaptive dynamic coefficient in the bee hiring stage, and assuming that the optimal honey source is n at the momentbestThen, the neighborhood search mode is expressed as:
Figure GDA0003487709630000061
Figure GDA0003487709630000062
wherein
Figure GDA0003487709630000063
The number of sub-channels allocated corresponding to the ith task request in the current best honey source, g (iter) is a dynamic coefficient, beta is a convergence factor, gthThe dynamic coefficient threshold is used, as shown in the expression, neighborhood search of hiring bees is carried out towards the direction of the optimal solution, and the search amplitude is dynamically reduced along with the increase of the number of search iterations, so that the search capability of the algorithm can be improved in the initial search stage, and the dynamic coefficient is unchanged in the later search stage, so that the convergence speed can be accelerated while the search capability is ensured;
s422, in the bee following stage, a neighborhood searching mode based on the combination of a DE/random/1 strategy and a self-adaptive dynamic coefficient is adopted, and the neighborhood searching mode is expressed as follows:
Figure GDA0003487709630000064
the neighborhood searching mode following the bee stage can better reduce the possibility that the bee-hiring stage is trapped in a local optimal solution;
s423, if the fitness value of a certain honey source is not improved after continuous iteration for a certain number of times, initializing the honey source by the detection bee, wherein the detection bee stage uses an optimal/worst solution combined honey source initialization mode, and the optimal honey source is assumed to be nbestThe worst honey source is nworstThen the initialization is expressed as:
Figure GDA0003487709630000065
the expression can be used for knowing that the initialized honey source is always moved towards the direction of the optimal solution when the detection bees are initialized, so that the optimal initial honey source can be obtained more easily;
s424, obtained by operating at each stage
Figure GDA0003487709630000071
May not be within its reasonable value range and so need to be derived for each stage
Figure GDA0003487709630000072
Carrying out normalization processing on legal value domains and determining correspondingly required computing resources fi(ii) a All task requests in the honey source obtained through each stage finally meet the time delay constraint, and partial or all task requests need to be selected according to a certain rule for migration decision;
s43, the greedy selection strategy is to process the new honey source obtained in each stage in the steps S421-S423; the invention adopts a greedy selection strategy for realizing the priority selection strategy of the task requests according to pro (i) decreasing for all the task requests in each honey source under the condition of satisfying the available resource constraint condition, wherein:
Figure GDA0003487709630000073
wherein
Figure GDA0003487709630000074
pro(i)、fit(i)、
Figure GDA0003487709630000075
Respectively representing the average reward value and the successfully executed reward value of each subchannel occupied by the task request i in the effective honeypot p, the actual effective subchannel distribution number distributed by the task request i in the effective honeypot p, and fsafe(Si) Is a reward value, f, for successful execution of a secure task request iunsafe(Si) Is a reward value for successful execution of the non-secure task request i; the greedy selection strategy is mainly adopted because no matter the task request is migrated to the edge computing end or the cloud computing end, a certain number of sub-channels need to be allocated to the task request for data transmission, and at the moment, it is reasonable to perform greedy selection on the task request for migration according to the descending order of pro (i).
Compared with the prior art, the invention has the following advantages and effects:
(1) according to the importance degree of the influence of different types of task requests on the safety of the vehicle, a safety factor is quantitatively given to each task request for the first time, and the safety of the vehicle is considered conveniently and seriously.
(2) According to the advantages of edge computing over cloud computing, whether the time delay can be met when each task to be migrated is requested to be migrated to the cloud computing end or not is further classified into a set TmecOr TbothIn the method, the selectable migration decision and the resource allocation value range of the task request in the two sets are re-determined, the range of feasible solutions is narrowed, and the rate of obtaining the optimal result can be accelerated.
(3) Available resources of the vehicle-mounted terminal, the edge computing terminal and the cloud computing terminal are fully utilized, the total amount of resources which can be selected and used by the task request is increased, and the possibility of successful execution of the task is greatly improved.
(4) By combining the improved discrete artificial bee colony algorithm with the greedy selection strategy, the approximate optimal solution can be quickly solved, and the solving complexity is low.
(5) The solving method provided by the invention has the potential of carrying out reasonable dynamic adjustment on the deployment of the computing resource amount of the edge computing terminal according to the arrival rule of task requests in different time periods in different daily areas when the number of the available sub-channels is constant, thereby better meeting the vehicle requirements.
Drawings
FIG. 1 is a flow chart of a collaborative cloud computing migration method based on vehicle networking application security as disclosed herein;
FIG. 2 is a diagram of a scenario in an urban environment in an embodiment of the invention;
fig. 3 is a diagram illustrating a relationship between the total number of successfully migrated task requests and the number of iterations, which is obtained by solving a task request that cannot be executed locally by using an improved discrete artificial bee colony algorithm and a greedy selection strategy in combination, when the total number of available subchannels and the edge computing resources are constant in the embodiment of the present invention;
fig. 4 is a diagram of a relationship between fitness values and iteration times obtained by solving a task request that cannot be executed locally by using an improved discrete artificial bee colony algorithm and a greedy selection strategy in combination when the total available sub-channel number and edge computing resources are fixed in the embodiment of the present invention;
fig. 5 is a relationship diagram of the adaptive value and the available computing resource of the edge computing end when the total number of available sub-channels, the total number of task requests that cannot be executed at the in-vehicle terminal, and the number of iterations are uniform in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
In the present embodiment, as shown in fig. 2, there is a base station St at the edge of the urban road, one edge computing server is deployed near each base station, and a cloud computing center is deployed at a long distance; meanwhile, a certain number of vehicles exist in the coverage radius of the base station, each vehicle can generate task requests such as automatic parking, collision avoidance and video call at variable time, and the corresponding task requests can be selected to be assisted by using a local vehicle-mounted terminal, an edge computing terminal or a cloud computing terminal or be abandoned for execution due to limited resources.
In combination with the migration method flow in fig. 1, the embodiment discloses a collaborative cloud computing migration method based on internet of vehicles application security classification, which includes the following steps:
s1, acquiring the information of the current base station and the relevant data of each task request, and determining the safety factor of each task request: the method is characterized in that an edge computing server is deployed near a single base station and connected with a cloud computing center, the base station adopts an OFDMA access mode, a plurality of vehicles are arranged in the coverage range of the base station, task requests generated by the vehicles can be selected to be executed at a vehicle-mounted terminal, an edge computing end or a cloud computing end, and if m task requests coexist, relevant data information St, a task request set T and relevant data information T of any task request i of the corresponding base station are obtainediRespectively as follows:
St＝[N，a，F，spkb，nmax]
T＝{0，1，…，m-1}
Ti＝[Di，Ci，τi，Si]
wherein N, a, F, spkb, NmaxRespectively representing the total number of sub-channels available for the current base station, the bandwidth of a single sub-channel, the available computing resources of an edge computing terminal, the transmission time delay from the transmission unit data amount of the base station to a cloud computing terminal, the maximum number of the assignable sub-channels of each task request, and Di、Ci、τi、SiRespectively representing the data input quantity and unit of the task request iThe number of execution cycles (i.e., the computation density), the delay constraint and the safety factor corresponding to the data volume; wherein a safety factor SiThe influence of the task request i on the safety of the vehicle is quantified when the safety factor S is reachedi≥SthWhen the task request belongs to the safe task request, wherein SthIs a safety threshold.
Arbitrary task request
Figure GDA00034877096300001010
Performing time delay at vehicle-mounted terminal
Figure GDA0003487709630000101
Migration to edge compute end execution delay
Figure GDA0003487709630000102
And migration to cloud computing end execution time delay
Figure GDA0003487709630000103
The calculation of (c) is as follows:
Figure GDA0003487709630000104
Figure GDA0003487709630000105
Figure GDA0003487709630000106
wherein
Figure GDA0003487709630000107
Wherein
Figure GDA0003487709630000108
Ri、ni、fiRespectively representing vehicle-mounted terminal computing resources corresponding to task requests iThe uploading rate when distributing single sub-channel, the number of distributed sub-channels and the computing resources of the distributed edge computing end; at the same time Pi、diAnd h and sigma respectively represent the transmitting power of the vehicle-mounted terminal corresponding to the task request i, the distance between the vehicle-mounted terminal and the current base station, a Rayleigh fading factor and Gaussian white noise.
S2, deploying the task request meeting the requirements in the task request set T on the vehicle-mounted terminal to execute:
initializing the task request set L as phi executed in the vehicle-mounted terminal if the task request set L is phi
Figure GDA0003487709630000109
If the value of L is L + i, and if the value of T is T \ i, the execution is selected at the vehicle-mounted terminal, otherwise, a subsequent migration decision is made; wherein L ═ L + i denotes the addition of element i to set L, and T ═ T \ i denotes the removal of element i from set T;
s3, according to whether each task request is migrated to a cloud computing end to be executed and meets the time delay constraint or not, aggregating
Figure GDA00034877096300001011
In which all task requests are classified into a set TmecOr TbothThen, the feasible solution domain of each task request is re-determined, and the specific process includes:
(1) because task request data need to pass through a core network from a cloud computing end, part of task requests can not be processed by means of cloud computing, and for any i e T, the classification standard is as follows:
Figure GDA0003487709630000111
wherein
Figure GDA0003487709630000112
Wherein
Figure GDA0003487709630000113
Refers to the assignment of n to task request imaxWhen it is a subchannel, itIf the time delay still does not meet the time delay constraint, the task request can be successfully executed only by migrating to the edge computing, so that the task request is classified into a set TmecOtherwise, classify the task request into the set TbothPerforming the following steps;
(2) for the condition that (i belongs to T ^ i belongs to Tmec)∨(i∈T∧i∈Tboth) The principle of all task requests i to update the feasible solution domain is as follows:
Figure GDA0003487709630000114
Figure GDA0003487709630000115
wherein
Figure GDA0003487709630000116
ceil (x) denotes rounding up x;
so far, a feasible solution domain corresponding to a migration decision and resource allocation is determined for all task requests in the set T;
s4, performing migration decision and resource allocation solution on the task requests in the set T by combining an improved discrete artificial bee colony algorithm and a greedy selection strategy, wherein the specific process comprises the following steps:
(1) determining a fitness function:
assuming that there are k task requests in the set T, the fitness function of the optimization problem in the present invention is as follows:
Figure GDA0003487709630000117
wherein
Figure GDA0003487709630000118
In the above expression, fit is a fitness value, and fit (i) is the successful execution of task request iThe prize value of the line, and fsafe(Si) Is a reward value, f, for successful execution of a secure task request iunsafe(Si) Is a reward value for successful execution of the non-secure task request i;
in this example
Figure GDA0003487709630000121
(2) Determining optimization variables of the algorithm and honey sources:
variable to be optimized in improved discrete artificial bee colony algorithm is used for distributing number n to sub-channelsiAssuming there are k task requests to be migrated for decision, the pth honey source n in the algorithmpExpressed as:
Figure GDA0003487709630000122
wherein
Figure GDA0003487709630000123
Represents the sub-channel distribution number corresponding to the ith task request in the pth honey source, and i belongs to [0, 1, …, k-1 ∈]；
In this embodiment, for convenience of simulation, two different honey sources are set for the pth honey source, which are respectively:
Figure GDA0003487709630000124
Figure GDA0003487709630000125
wherein n isp(op) is the honey source to be optimized processed at each stage in the modified discrete artificial bee colony algorithm, and n isp(valid) is the number of n obtained each timep(op) is updated according to a certain rule and is actually used for calculating fiEffective honey source of fit (i);
(3) determining
Figure GDA0003487709630000126
And
Figure GDA0003487709630000127
feasible solution domain and implementation of
Figure GDA0003487709630000128
Initialization:
in the invention, the resources primarily allocated to all task requests can meet the corresponding time delay constraint, and then the greedy selection strategy is used for selecting the task requests in the available resource constraint, so that the task requests are selected in the available resource constraint
Figure GDA0003487709630000129
And
Figure GDA00034877096300001210
the feasible solution domain of (1) must be able to satisfy the delay constraint of any task request i e T, and the specific process is as follows:
(3.1) solving the lower limit:
to pair
Figure GDA00034877096300001211
All of them may choose to migrate to the edge compute side, so it must satisfy
Figure GDA00034877096300001212
Thereby obtaining
Figure GDA00034877096300001213
And
Figure GDA00034877096300001214
the lower limit values are all:
Figure GDA00034877096300001215
wherein R isiIndicating a corresponding transmission rate when a subchannel is allocated to the task request i, ceil (x) indicating that x is rounded up;
(3.2) solving the upper limit value
Definition of
Figure GDA0003487709630000131
Has an upper limit value of
Figure GDA0003487709630000132
At the same time, the user can select the desired position,
Figure GDA0003487709630000133
has an upper limit value of
Figure GDA0003487709630000134
To pair
Figure GDA0003487709630000135
Since it is only possible to select the edge compute side for migration, the corresponding
Figure GDA0003487709630000136
And
Figure GDA0003487709630000137
all upper limit values of (are n)maxI.e. by
Figure GDA0003487709630000138
Figure GDA0003487709630000139
To pair
Figure GDA00034877096300001310
The method can be selectively migrated to an edge computing end and a cloud computing end, and when the number of the distributed sub-channels reaches the number
Figure GDA00034877096300001311
The method selects to migrate to the cloud computing end for execution, and if the execution is finished, the method selects to migrate to the cloud computing end for execution
Figure GDA00034877096300001312
Will choose to migrate to the edge compute side for execution and always satisfy according to the previous definition
Figure GDA00034877096300001313
So as to obtain corresponding
Figure GDA00034877096300001314
Upper limit value of (2):
Figure GDA00034877096300001315
at the same time due to
Figure GDA00034877096300001316
The method is a honey source object actually processed by an improved discrete artificial bee colony algorithm, and in the method, in order to realize that the probability of task request i migrating to an edge computing end is equal to that of task request i migrating to a cloud computing end, the probability of task request i migrating to the edge computing end is equal to that of task request i migrating to the cloud computing end, so that the task request i is migrated to the cloud computing end
Figure GDA00034877096300001317
The upper limit value of (A) is required to satisfy
Figure GDA00034877096300001318
Correspond to obtain
Figure GDA00034877096300001319
In view of the above, it can be seen that
Figure GDA00034877096300001320
Figure GDA00034877096300001321
So for i ∈ T, the corresponding
Figure GDA00034877096300001322
Initialization:
Figure GDA00034877096300001323
(4) bee hiring stage:
adopting a neighborhood searching mode based on the combination of DE/best/1 strategy and adaptive dynamic coefficient in the bee employment stage, and assuming that the optimal honey source is n at the momentbest(op), then to the honey source npThe neighborhood search operation performed on the subchannel allocation number of any task request i in (op) is as follows:
Figure GDA0003487709630000141
Figure GDA0003487709630000142
wherein
Figure GDA0003487709630000143
The number of sub-channels allocated for the ith task request in the current best honey source, g (iter) is a dynamic coefficient, beta is a convergence factor, gthIs a dynamic coefficient threshold;
(5) following the bee stage:
in the following bee stage, a neighborhood searching mode based on the combination of DE/random/1 strategy and self-adaptive dynamic coefficient is adopted, and then the honey source n is subjected topThe neighborhood search operation for any task request i in (op) is represented as:
Figure GDA0003487709630000144
a random neighborhood searching mode is adopted in the bee following stage, so that the bee hiring stage can be better prevented from entering a local optimal solution, and the probability of obtaining the optimal solution is improved;
(6) a bee investigation stage:
in the optimization process, if a honey source is still unchanged after a certain number of iterations, the honey source is re-detected in the bee investigation stageInitialization, used here is a principle based on the best/worst solution, assuming that the current best and worst honey sources are n, respectivelybest(op) and nworst(op), then initializing a specific expression of the honey source as follows:
Figure GDA0003487709630000145
the expression can be used for knowing that the initialized honey source is always moved towards the direction of the optimal solution when the detection bees are initialized, so that the optimal initial honey source can be obtained more easily;
(7) the bee source to be optimized obtained after processing for each stage of the processes (4) to (6) may exceed the feasible solution domain, so that the bee source to be optimized needs to be normalized:
as the three stages of processing of the hiring bee, the following bee and the reconnaissance bee are carried out, the processing of each stage is obtained
Figure GDA0003487709630000146
May be out of its reasonable range, and so further normalization is required, as follows:
Figure GDA0003487709630000151
the new honey source generated after each stage operation changes, and the required pair corresponds to
Figure GDA0003487709630000152
Is/are as follows
Figure GDA0003487709630000153
The updating is carried out, and the updating rule is as follows:
Figure GDA0003487709630000154
the correspondence can be calculated to obtain fi、fit(i)；
(8) After each stage is processed and the obtained honey sources are normalized and updated as shown in (7), a greedy selection strategy is adopted to sequentially select the effective honey sources n from pro (i) in a descending orderp(valid) selecting a task request for migration, wherein:
Figure GDA0003487709630000155
as shown above, pro (i) indicates that the task request i in the effective honey source p occupies the average reward value of each sub-channel, and task requests are selected one by one according to the decreasing sequence of pro (i) by using a greedy selection strategy to perform migration under the condition of meeting resource constraints, so that task requests with larger reward values of single sub-channels are migrated preferentially, and a better migration decision can be finally obtained;
and (5) performing iterative optimization processing on the processes (4) - (8), and stopping optimization until the maximum iteration number iter _ max is reached to obtain a final migration decision and resource allocation scheme.
The collaborative cloud computing migration method based on the car networking application security classification provided by the invention is simulated in a Spyder platform by using Python language, a simulation parameter setting table is shown in table 1, a simulation result is a relation graph of the total number of task requests successfully migrated and iteration times in fig. 3, a relation graph of fitness value and iteration times in fig. 4, and a relation graph of the fitness value and available computing resources of an edge computing end in fig. 5.
TABLE 1 simulation parameter setup table
Simulation parameters Value of
Base station coverage radius 200m
Total number of task requests 150
Task request input data volume Di 800-1200kb
Number of cycles C of unit data of task requesti 800-1500cycle/bit
Single task delay constraint range Ti 0.5-1.5s
Available bandwidth 20MHz
Single subchannel bandwidth 15KHz
Number of available subchannels 1365
Edge compute side available compute resource F 100Gcycle/s
Unit data transmission delay spkb from base station to cloud computing end 6*10^(-4)s/kb
Convergence factor beta 0.94
Dynamic coefficient threshold gth 0.4
Maximum number of iterations 300
Fig. 3 and 4 show that when the number of available subchannels and the edge computing end computing resources are constant, the total number of successfully migrated task requests and the fitness value change with the number of iterations, and it can be seen from the figures that the fitness value obtained by the cooperative cloud computing migration method and the total number of successfully migrated task requests both increase with the number of iterations, and finally reach a stable value and do not change any more, i.e., convergence is achieved, and an optimal solution is achieved around the 120 th iteration; compared with a migration method only using edge computing, the cooperative cloud computing migration method can successfully migrate more task requests and obtain a higher fitness value, and can better assist the vehicle in processing the task requests.
Fig. 5 is a diagram showing a relationship between a fitness value and an available computing resource of an edge computing side when the number of available sub-channels is constant, and it can be seen from the diagram that the fitness value gradually increases with the increase of the available computing resource of the edge computing side, but the increase becomes slower and slower, mainly because no matter whether a task request is migrated to the edge computing side or the cloud computing side, a certain number of sub-channels need to be allocated for data transmission, at this time, the number of available sub-channels becomes a main constraint factor, which is also consistent with logic; meanwhile, the simulation result shows that the migration method has the potential of dynamically adjusting the available computing resources of the edge computing terminal according to the arrival rule of the task requests in different areas at different moments when the number of the available sub-channels is fixed, so that the safety of the vehicle can be better improved.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (6)

1. A collaborative cloud computing migration method based on Internet of vehicles application security grading is characterized in that an application scene is that an edge computing server is deployed near a single base station and is connected with a cloud computing center, the base station adopts an OFDMA access mode, a certain number of vehicles are arranged in the coverage range of the base station, and task requests generated by the vehicles are selected to be executed at a vehicle-mounted terminal, an edge computing end or a cloud computing end, and the collaborative cloud computing migration method comprises the following steps:
s1, acquiring the information of the current base station and the relevant data of each task request, and determining the safety factor of each task request: assuming that m task requests are generated in total, the data information St of the corresponding base station, the task request set T and the data information T of any task request i are obtainediRespectively expressed as follows:
St＝[N，a，F，spkb，nmax]
T＝{0，1，...，m-1}
Ti＝[Di，Ci，τi，Si]
wherein N, a, F, spkb, NmaxRespectively representing the total number of sub-channels available for the current base station, the bandwidth of a single sub-channel, the available computing resources of an edge computing terminal, the transmission time delay from the transmission unit data amount of the base station to a cloud computing terminal, the maximum number of the assignable sub-channels of each task request, and Di、Ci、τi、SiRespectively representing the data input quantity of the task request i, the execution cycle number corresponding to the unit data quantity, the time delay constraint and the safety factor; wherein a safety factor SiThe influence of the task request i on the safety of the vehicle is quantified when the safety factor S is reachedi≥SthThen, the task request belongs to a secure task request, wherein SthIs a safe threshold;
s2, deploying the task request meeting certain requirements on the vehicle-mounted terminal to execute: judging whether the task request in the task request set T can be deployed in the vehicle-mounted terminal for execution, wherein the principle that the task request i belongs to T and is selected to be executed in the vehicle-mounted terminal is as follows:
Figure FDA0003487709620000021
wherein
Figure FDA0003487709620000022
Execution delay, f, when deploying a task request i at a vehicle-mounted terminali locInitializing a task request set L ═ phi executed by the vehicle-mounted terminal for a corresponding vehicle-mounted terminal computing resource, and if the condition is met, then L ═ L + i and T ═ T \ i, wherein L ═ L + i represents that an element i is added to the set L, and T ═ T \ i represents that the element i is removed from the set T;
s3, classifying the task requests in the set T into the set T according to whether the time delay constraint can be met when each task request is migrated to the cloud computing end for executionmecOr TbothFinally, re-determining the feasible solution domain of each task request in the set T;
wherein T ismecThe task request in (1) cannot be selectively migrated to the cloud computing end for execution, and TbothThe task request in the method can be selectively migrated to an edge computing end and a cloud computing end for execution, and any classified task request i meets the condition that (i belongs to T ^ i belongs to T ∈ T ∈ ^ i ^ T ∈ ^ Tmec)∨(i∈T∧i∈Tboth)；
S4, performing migration decision and resource allocation solution on the task requests in the task request set T by combining an improved discrete artificial bee colony algorithm and a greedy selection strategy; the fitness function in the modified discrete artificial bee colony algorithm in the step S4 is as follows:
Figure FDA0003487709620000023
wherein
Figure FDA0003487709620000024
fit is a fitness value, and fit (i) is a reward value for successful execution of task request iAnd f issafe(Si) Is a reward value, f, for successful execution of a secure task requestunsafe(Si) Is a reward value for the non-secure task request to execute successfully.
2. The collaborative cloud computing migration method based on Internet of vehicles application security classification as claimed in claim 1, wherein in step S2, the migration decision of any task request i in the task request set T is xiE { -1, 0, 1}, which respectively represent that the sub-channel can be migrated to an edge computing end, abandoned to be executed and migrated to a cloud computing end, and the distribution number of the sub-channels is ni∈{0，1，...，nmaxIn which n ismaxRequesting the assignable maximum number of sub-channels for each task, calculating the resource allocation as fi∈[0，F](ii) a And when the mobile terminal is migrated to the edge computing terminal, the time delay of downloading the computing result is ignored, when the mobile terminal is migrated to the cloud computing terminal, the computing time delay of the cloud computing terminal and the time delay of downloading the computing result are ignored, and the time delays of the two migration modes
Figure FDA0003487709620000031
And
Figure FDA0003487709620000032
the expressions are respectively as follows:
Figure FDA0003487709620000033
Figure FDA0003487709620000034
wherein
Figure FDA0003487709620000035
RiAnd spkb respectively represents the data uploading rate when a single sub-channel is allocated to the task request i, and the unit data volume transmitted by the base station to the cloud computingTransmission delay of the terminal, Pi、di、α、|hi|、σ2And respectively representing the transmitting power of the vehicle-mounted terminal corresponding to the task request i, the distance between the corresponding vehicle-mounted terminal and the base station, a path attenuation factor, a Rayleigh fading factor and Gaussian white noise.
3. The collaborative cloud computing migration method based on Internet of vehicles application security classification as claimed in claim 1, wherein in step S3, all task requests in task request set T are further classified into set TmecOr TbothIn, assume an initial Tmec＝φ，TbothFor any task request i e T, the corresponding classification criteria are:
Figure FDA0003487709620000036
wherein
Figure FDA0003487709620000037
Indicating assignment of n to task request imaxExecution delay, R, when a subchannel migrates to a cloud computing endiIndicating the data upload rate when a single subchannel is allocated to task request i.
4. The collaborative cloud computing migration method based on Internet of vehicles application safety grading according to claim 1, characterized in that in step S3, the requirement of satisfying (i e T ∈ >mec)∨(i∈T∧i∈Tboth) All task requests i to update the feasible solution domain, and the updating principle is as follows:
Figure FDA0003487709620000041
Figure FDA0003487709620000042
wherein
Figure FDA0003487709620000043
Figure FDA0003487709620000044
The number of sub-channels, R, required for migration of the task request i to the cloud computing end to execute is representediIndicating that task request i is assigned a data upload rate of a single subchannel, ceil (x) indicating that x is rounded up.
5. The collaborative cloud computing migration method based on Internet of vehicles application security ranking of claim 1, wherein the variable to be optimized in the modified discrete artificial bee colony algorithm of step S4 is the sub-channel allocation number niAssuming there are k task requests to be migrated for decision, the pth honey source n in the algorithmpExpressed as:
Figure FDA0003487709620000045
the improved discrete artificial bee colony algorithm comprises three stages of operation of hiring bees, following bees and scout bees, wherein a neighborhood search mode based on a DE/best/1 strategy and self-adaptive dynamic coefficients is adopted in the hiring bee stage, a neighborhood search mode based on a DE/random/1 strategy is adopted in the following bee stage, and an optimal/worst solution combined honey source initialization mode is adopted in the scout bee stage; and obtained for each stage operation
Figure FDA0003487709620000047
Normalization processing is required to be carried out, so that the feasible solution domain can be kept, correspondingly required computing resources are determined, finally, a new honey source is obtained at each stage, and the task requests in the new honey source meet the time delay constraint;
wherein
Figure FDA0003487709620000046
The number of sub-channel allocations corresponding to the ith task request in the pth honey source is represented, and i belongs to [0, 1]。
6. The collaborative cloud computing migration method based on Internet of vehicles application security classification as claimed in claim 1, wherein the greedy selection strategy in step S4 is to process new honey sources obtained at each stage in the improved discrete artificial bee colony algorithm, and is to implement a priority selection strategy for task requests according to a greedy selection strategy that pro (i) is decreased for all task requests in each honey source under the condition that available resource constraints are satisfied, and finally, the selected task request is to be migrated, and then, a migration decision is determined, wherein:
Figure FDA0003487709620000051
wherein
Figure FDA0003487709620000052
pro(i)、fit(i)、
Figure FDA0003487709620000053
Respectively representing the average reward value and successfully executed reward value of each subchannel occupied by the effective task request i in the honey source p, the actual effective subchannel distribution number distributed by the effective task request i in the honey source p, and fsafe(Si) Is a reward value, f, for successful execution of a secure task requestunsafe(Si) Is a reward value for the non-secure task request to execute successfully.
CN202010034600.5A 2020-01-14 2020-01-14 Collaborative cloud computing migration method based on Internet of vehicles application security grading Active CN111240821B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010034600.5A CN111240821B (en) 2020-01-14 2020-01-14 Collaborative cloud computing migration method based on Internet of vehicles application security grading

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010034600.5A CN111240821B (en) 2020-01-14 2020-01-14 Collaborative cloud computing migration method based on Internet of vehicles application security grading

Publications (2)

Publication Number Publication Date
CN111240821A CN111240821A (en) 2020-06-05
CN111240821B true CN111240821B (en) 2022-04-22

Family

ID=70874668

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010034600.5A Active CN111240821B (en) 2020-01-14 2020-01-14 Collaborative cloud computing migration method based on Internet of vehicles application security grading

Country Status (1)

Country Link
CN (1) CN111240821B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111901424B (en) * 2020-07-28 2021-09-07 苏州大学 Cloud edge cooperative network resource smooth migration and reconstruction method and system
CN112039992B (en) * 2020-09-01 2022-10-28 平安资产管理有限责任公司 Model management method and system based on cloud computing architecture
CN112052149B (en) * 2020-09-06 2022-02-22 厦门理工学院 Big data information acquisition system and use method
CN112469001B (en) * 2020-11-17 2022-07-26 国网浙江省电力有限公司信息通信分公司 Application migration method and device, electronic equipment and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109302709A (en) * 2018-09-14 2019-02-01 重庆邮电大学 The unloading of car networking task and resource allocation policy towards mobile edge calculations
CN109600419A (en) * 2018-11-12 2019-04-09 南京信息工程大学 The computation migration method of car networking application is supported under a kind of mobile edge calculations environment
CN109656703A (en) * 2018-12-19 2019-04-19 重庆邮电大学 A kind of mobile edge calculations auxiliary vehicle task discharging method
CN109684075A (en) * 2018-11-28 2019-04-26 深圳供电局有限公司 A method of calculating task unloading is carried out based on edge calculations and cloud computing collaboration

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180089460A1 (en) * 2016-09-23 2018-03-29 Ford Global Technologies, Llc Patient-centered vehicle emergency application and web interface that provides emergency personnel with vehicle data and personalized occupant health information

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109302709A (en) * 2018-09-14 2019-02-01 重庆邮电大学 The unloading of car networking task and resource allocation policy towards mobile edge calculations
CN109600419A (en) * 2018-11-12 2019-04-09 南京信息工程大学 The computation migration method of car networking application is supported under a kind of mobile edge calculations environment
CN109684075A (en) * 2018-11-28 2019-04-26 深圳供电局有限公司 A method of calculating task unloading is carried out based on edge calculations and cloud computing collaboration
CN109656703A (en) * 2018-12-19 2019-04-19 重庆邮电大学 A kind of mobile edge calculations auxiliary vehicle task discharging method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Exploring Computation Offloading in Vehicular Clouds;Alisson Barbosa de Souza et all;《2019 IEEE 8th International Conference on Cloud Networking (CloudNet)》;20191106;第1-4页 *
Mobile-Edge Computing for Vehicular Networks: A Promising Network Paradigm with Predictive Off-Loading;Ke Zhang et al;《 IEEE Vehicular Technology Magazine》;20170424;第12卷(第2期);第36-44页 *
车联网环境下多接入边缘计算与云计算协同迁移策略的研究;段海波;《中国优秀硕士学位论文全文数据工程科技II辑》;20210228(第2期);第C034-986页 *

Also Published As

Publication number Publication date
CN111240821A (en) 2020-06-05

Similar Documents

Publication Publication Date Title
CN111240821B (en) Collaborative cloud computing migration method based on Internet of vehicles application security grading
CN111132077B (en) Multi-access edge computing task unloading method based on D2D in Internet of vehicles environment
CN111918245B (en) Multi-agent-based vehicle speed perception calculation task unloading and resource allocation method
CN111711666B (en) Internet of vehicles cloud computing resource optimization method based on reinforcement learning
CN113543074B (en) Joint computing migration and resource allocation method based on vehicle-road cloud cooperation
CN111132074B (en) Multi-access edge computing unloading and frame time slot resource allocation method in Internet of vehicles environment
CN113687875B (en) Method and device for unloading vehicle tasks in Internet of vehicles
CN113645273B (en) Internet of vehicles task unloading method based on service priority
CN115297171B (en) Edge computing and unloading method and system for hierarchical decision of cellular Internet of vehicles
CN112512121A (en) Radio frequency spectrum dynamic allocation method and device based on reinforcement learning algorithm
CN110418143B (en) SVC video transmission method in Internet of vehicles
He et al. An offloading scheduling strategy with minimized power overhead for internet of vehicles based on mobile edge computing
CN112888021B (en) Task unloading method for avoiding interruption in Internet of vehicles
CN112437449B (en) Joint resource allocation method
CN117221951A (en) Task unloading method based on deep reinforcement learning in vehicle-mounted edge environment
CN113535390A (en) Method, system, device and medium for distributing multi-access edge computing node resources
Laroui et al. Virtual mobile edge computing based on IoT devices resources in smart cities
CN115052262A (en) Potential game-based vehicle networking computing unloading and power optimization method
CN112906745B (en) Integrity intelligent network training method based on edge cooperation
CN114513814A (en) Edge network computing resource dynamic optimization method based on unmanned aerial vehicle auxiliary node
CN114138466A (en) Task cooperative processing method and device for intelligent highway and storage medium
CN111741050A (en) Data distribution method and system based on roadside unit
CN115373856B (en) Unloading task allocation method for intelligent vehicle in end edge cloud network framework
CN114390624B (en) Non-uniform vehicle task unloading method for dealing with cell switching problem
Zeng et al. Research on Task Offloading and Typical Application Based on Deep Reinforcement Learning and Device-Edge-Cloud Collaboration

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