CN112835637B - Task unloading method for vehicle user mobile edge calculation - Google Patents
Task unloading method for vehicle user mobile edge calculation Download PDFInfo
- Publication number
- CN112835637B CN112835637B CN202110102608.5A CN202110102608A CN112835637B CN 112835637 B CN112835637 B CN 112835637B CN 202110102608 A CN202110102608 A CN 202110102608A CN 112835637 B CN112835637 B CN 112835637B
- Authority
- CN
- China
- Prior art keywords
- task
- user
- unloading
- server
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000004364 calculation method Methods 0.000 title claims description 66
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 86
- 238000005265 energy consumption Methods 0.000 claims abstract description 84
- 238000002922 simulated annealing Methods 0.000 claims abstract description 19
- 230000005540 biological transmission Effects 0.000 claims description 21
- 230000000694 effects Effects 0.000 claims description 15
- 238000004891 communication Methods 0.000 claims description 14
- 238000005516 engineering process Methods 0.000 claims description 14
- 238000000137 annealing Methods 0.000 claims description 12
- 230000007246 mechanism Effects 0.000 claims description 12
- 230000009467 reduction Effects 0.000 claims description 12
- 230000007704 transition Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000001816 cooling Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 230000007423 decrease Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- 238000011068 loading method Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000000903 blocking effect Effects 0.000 claims description 3
- 229920003087 methylethyl cellulose Polymers 0.000 claims 1
- 238000002474 experimental method Methods 0.000 abstract description 6
- 230000006872 improvement Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 4
- 230000010365 information processing Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
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
- G06F9/44—Arrangements for executing specific programs
- G06F9/445—Program loading or initiating
- G06F9/44594—Unloading
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- 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]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
-
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- Mathematical Physics (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
A task unloading method for vehicle user mobile edge computing belongs to the field of Internet of things and mainly considers that at a crossroad with a plurality of roadside devices RSUs, each roadside device is provided with an MEC server and can provide computing unloading services for mobile users, mobile devices located around an MEC server group can unload tasks through any one server, the task computing time and energy consumption of vehicle users are comprehensively considered through user unloading effectiveness, task unloading decisions of the users are optimized through a simulated annealing algorithm, a scheme with the highest task unloading effectiveness in the current environment is obtained, and experiments prove that the method effectively reduces energy consumption while computing task time is reduced.
Description
Technical Field
The invention belongs to the field of Internet of things, and particularly relates to a task unloading method for vehicle user mobile edge computing.
Background
With the continuous progress of the technology, the rapid development of intelligent equipment and wireless communication technology brings possibility to the interconnection of everything and changes to all aspects of life and work of people. Meanwhile, modern traffic systems are continuously enriched due to the great popularization of automobiles, the mode of mass living and traveling is improved, automobiles are developed towards a more intelligent direction, intelligent vehicles have intelligent environment sensing capability, safety assessment is conducted on the running state of the vehicles, and more suitable running routes are found. The emergence of intelligent automobiles is a corollary to the continual evolution of intelligent technology.
However, the increasing number of automobiles also puts a great pressure on the modern society, such as low traffic efficiency, frequent traffic accidents, urban road congestion and aggravation of environmental pollution. Therefore, the improvement of driving safety, the reduction of accident rate, the improvement of road traffic jam and the improvement of Transportation efficiency become social concerns and efforts, and the construction of Intelligent Transportation Systems (ITS) has become a trend.
Considering the requirement for real-time performance of information processing, if all information processing is put on the cloud platform to be executed, it is obvious that the requirements of low latency and high quality of service (QoS) cannot be satisfied. The mobile edge computing technology is that the mobile equipment is a development platform which is close to the network edge of an object or a source and integrates network, computing, storing and applying core capability, and edge computing service is provided nearby. The appearance of the edge computing technology expands cloud computing functions and services to the edge of a network, and due to the fact that the edge computing technology is close to equipment, the real-time performance of information processing is guaranteed, and the performance of data transmission is improved. By adopting the MEC technology, the computing task of the cloud platform can be executed on the edge server by deploying light-weight infrastructure such as the edge server, so that the task execution pressure of the cloud platform is reduced, and higher-quality service is brought to users. When the unloading is performed, if all users unload their tasks, the computing capacity of the MEC server is also greatly stressed, and therefore, the task unloading decision becomes a critical issue for realizing efficient computing unloading.
Disclosure of Invention
The invention aims to optimize the task unloading decision of a user by using a simulated annealing algorithm and obtain a scheme with the highest task unloading utility in the current environment. The invention considers that the intersection with a plurality of roadside equipment RSUs is mainly considered, each roadside equipment is provided with an MEC server, calculation unloading services can be provided for mobile users, the mobile equipment positioned around an MEC server group can unload tasks through any one server, the task calculation time and energy consumption of vehicle users are comprehensively considered through user unloading effectiveness, and a simulated annealing algorithm is used for optimizing the task unloading decision of the users to obtain a scheme with the highest task unloading effectiveness in the current environment.
The invention relates to a task unloading method for vehicle user mobile edge calculation, which mainly comprises the following key steps:
1, basic principle:
1.1, a scene model;
1.2, calculating tasks by a user;
1.3, executing the task locally;
1.4, unloading the task to an edge server;
and 2, unloading utility of the user task:
3, description of algorithm:
3.1, simulating relevant contents of an annealing mechanism;
3.2, initial unloading strategy;
3.3, state transition mechanism;
3.4, an edge calculation task unloading algorithm based on a simulated annealing mechanism;
further, in step 1.1, the scene is set at an intersection with dense vehicles, a plurality of roadside devices RSUs are arranged at the intersection, and the intersection is provided with the MEC server, and the server group is regarded as a server group. Surrounding vehicles (vehicles) are mobile users that can communicate with any one of the server farms and offload task resources to the servers. Therefore, the user and server groups are represented as follows: v ═ M ∈ M, and RSU and MEC are in one-to-one correspondence, and therefore M ∈ M can be used instead.
Step 1.2 the user calculation tasks are as follows: let the Task (Task) be denoted by T, and let T be any user v who has a computing Task to perform at any timevAnd this task cannot be broken down into smaller subtasks. Each task has two parameters, namely the Size (Size) of the processing task and the workload, which are necessary to execute the taskThe required size of the calculation (Cycles), and can therefore be expressed as Tv=<ds[bits],dc[cycles]>。
The locally executed tasks in step 1.3 are as follows: each computing task may be executed locally (Local) or off-loaded to the edge server, and although off-loading the task to the edge server may reduce the consumption of Local computing, it may correspondingly increase the time consumption and partial energy consumption of the upload of the corresponding task. Unifying the types of all vehicle users, the ability of the users to perform calculations locally being the same, lv[cycles/s],dcvFor workloads, user v executes computing task T locally, incorporating the computing resources required for each taskvThe time required is as shown in equation (1):
when the user performs the calculation task locally, the energy consumption power of the user v for performing the calculation isThe required time isThe energy consumption of the computing task when executed locally can be obtained as follows:
the task is unloaded to the edge server in step 1.4 as follows: if the task is uploaded to the server, the time consumption of the task is increased, the time can be divided into transmission time and time for execution on the server, and meanwhile, the execution of the calculation task also generates energy consumption.
User v may transmit data in any sub-band of any server m within communication range, which may be denoted asSince the user may choose to perform the computing task locally or on the server, assumeIn time, it means that the user is unloaded to the edge server in the computing task, so the constraint condition can be obtained:
in addition, in the use of the orthogonal frequency division multiplexing technology, certain noise and interference exist, which can affect the signal reception and cause the reduction of the communication quality, and can be represented by a signal-to-interference-plus-noise ratio, a signal gain matrix from a user to a server in a range represented by H, and a signal gain matrix from the user to the server represented by HTo represent the output power matrix of the user byRepresenting the option to offload the computing task to all users on the edge server, it can therefore be computed that:
dividing the bandwidth B into N identical sub-bands, each having a size of W ═ B/N [ Hz ], and combining with shannon's formula, the transmission rate of the information can be obtained as follows:
Cv,m=W log2(1+sv,m) (5)
wherein s isv,mFor all sub-bands on the edge server s, i.e.:
it can be known from equation (5) that the transmission of information is not only related to the bandwidth of the channel, but also affected by the signal-to-noise ratio. In addition, combining the size of the calculation task obtained before, the time for the user v to transmit the task can be obtained as follows:
when a service receives a task uploaded by a user through a sub-band, the task execution work can be carried out at an edge server side, and the calculation execution capacity of a single server is rm[cycles/s]Because the same server can process a plurality of tasks at the same time, the constraint condition can be obtained:
v is a user who offloads the computation task to the edge server, that is, the sum of the computation execution capacity of each task allocated to the server m cannot exceed the total computation capacity of the server, and by combining the computation task model, the execution time of the task at the server side can be obtained as follows:
in summary, when user v chooses to offload a computing task to an edge server for execution, the total time consumed is:
the user v unloads the task to the edge server for execution, the energy consumption of the user is mainly generated during the task transmission, and the energy consumption generated during the transmission of the user can be obtained as follows:
further, in step 2.1, in a given environment, the user experience of the computing task mainly comes from the execution completion time of the task and the energy consumption generated during the execution, and the objective optimization function is synthesized, and the single user task unloading effect QvComprises the following steps:
Qva higher value of (c) indicates better unloading effectiveness. Wherein λ is1And λ2Represents the bias weight of the user to the time consumption and energy consumption in the calculation task, and satisfies { lambda }1+λ2=1|λ1∈[0,1],λ2∈[0,1]}. In the case of a better demand for time consumption, λ can be increased1Taking value of (A), otherwise increasing λ2The value of (a).
The new task unloading method based on the mobile edge computing technology is that the execution time of the task and the energy consumption of the task execution are improved by optimizing the execution decision of the computing task selected by each user, and the maximum user unloading utility meeting the conditions is obtained, namely:
Qmax=max(∑v∈VQv) (13)
after expanding equation (13), the following equation can be made:
on the premise of balancing and considering unloading time consumption and energy consumption, consider lambda1=λ2As c is 0.5, formula (14) can be converted to:
whereinIs a fixed value. And optimizing the optimal solution of the user unloading utility function by combining a simulated annealing algorithm.
Further, the simulated annealing algorithm in step 3.1 is a heuristic algorithm and is also a greedy algorithm, and the difference is that in the searching process, due to the introduction of a random factor, when a feasible solution is iteratively updated, an element worse than a current value is received at a certain probability, so that the simulated annealing algorithm can jump out a local optimal solution, and thus a global optimal solution is obtained.
Initial temperature T0Simulating the temperature at which the annealing algorithm starts, considering the number of user vehicles as the initial temperature, i.e. T0=V。
The temperature reduction coefficient α is an exponential decrease method according to the literature, and therefore T (n +1) ═ α T (n) is included, and α is a temperature reduction coefficient and has a value of 0.8 to 0.99.
The termination temperature, if there is no state that can be updated in case of several iterations, or a set termination temperature is reached, the annealing is considered to be complete.
And (3) assuming that the current state is f (n), and the state is changed into f (n +1) at the next moment, the probability p that the system is changed from f (n) to f (n +1) is as follows:
where Δ f ═ f (n) -f (n + 1).
The higher the temperature, i.e. the optimization is started, the greater the probability of cooling down.
The method for obtaining the initial temperature by adopting the initial unloading strategy in the step 3.2 comprises the following steps: and under the condition that the unloading constraint condition is met, selecting a mode meeting the maximum unloading effect according to a formula (12) to execute the task, so as to obtain an initial unloading strategy, namely an initial temperature.
The state transition method of step 3.3 is as follows: the state transition is that under the condition of meeting the set constraint, a user selects a set of calculation task execution modes, and changes the calculation task execution modes of the user through random probability, namely whether to execute unloading and selects an unloaded sub-band to establish a new unloading strategy.
Step 3.4 the edge calculation task unloading algorithm based on the simulated annealing mechanism is executed as follows:
step 1: obtaining the number of vehicle users in the communication range according to the communication range of the edge server to obtain the initial temperature;
step 2: and (3) constructing an initial strategy for user unloading through the algorithm 1 according to the formula (12) and combining with user unloading constraints, and calculating the initial system unloading utility according to the formula (15).
And 3, step 3: if the set termination temperature is not reached, a new unloading strategy is obtained through the algorithm 2 within the specified iteration frequency range, and then the new system unloading effectiveness is calculated according to the formula (15);
and 4, step 4: comparing the values of the unloading utility of the system at the current time and the last time, if the value is greater than the unloading utility of the system at the last time, accepting the unloading strategy at the current time, updating the current solution of the function, otherwise, obtaining the cooling probability p according to a formula (16), generating a random number which is uniformly distributed, if p is greater than the random number, accepting the unloading strategy, otherwise, abandoning the unloading strategy at the current time;
and 5: adjusting the current temperature according to the temperature reduction coefficient, and repeatedly executing the steps (2) - (4)
Step 6: and when the end temperature is reached or the specified iteration times are reached, the annealing is ended, and the obtained user unloading effect is the optimal solution of the function.
The invention has the advantages and positive effects that:
a task unloading method for vehicle user mobile edge computing mainly studies how to realize efficient computing unloading key problem. The method mainly considers that at a crossroad with a plurality of roadside equipment RSUs, each roadside equipment is provided with an MEC server, calculation unloading services can be provided for mobile users, mobile equipment located around an MEC server group can unload tasks through any one server, task calculation time and energy consumption of vehicle users are comprehensively considered through user unloading effectiveness, task unloading decisions of the users are optimized through a simulated annealing algorithm, a scheme with the highest task unloading effectiveness under the current environment is obtained, experiments prove that the method effectively reduces energy consumption while reducing calculation task time, and has certain practical value.
Drawings
FIG. 1 is a scene model;
FIG. 2 is a diagram comparing OFDM and OFDMA modes of operation;
FIG. 3 is a graph of task computation times under different time consumption preferences;
FIG. 4 is a graph of task energy consumption under different time consumption preferences;
FIG. 5 is a graph of energy consumption versus number of MEC servers;
FIG. 6 is a graph of time consumption versus number of MEC servers;
FIG. 7 is a graph comparing the user task offload utility for different MEC server numbers;
FIG. 8 is a graph of time consumption versus vehicle flow;
FIG. 9 is a graph comparing energy consumption at different vehicle flows;
FIG. 10 is a graph comparing user offload effectiveness at different traffic flows;
FIG. 11 is a traffic flow statistical plot over a time period of a day;
FIG. 12 is a comparison graph of task computation times over a length of time of day;
FIG. 13 is a graph comparing user energy consumption over a length of time of day;
FIG. 14 is a graph of user offload utility versus time of day length;
FIG. 15 is a flowchart of a task offloading method for vehicle user mobile edge computing in accordance with the present invention.
Detailed Description
Example 1
The MATLAB platform is used in the experiment to test the set algorithm TOSA. Through comparing the three aspects of calculation tasks of different quantities of servers and different sizes in different time periods, and comparing the three aspects of calculation tasks in other task unloading methods, experiments prove that the TOSA algorithm has certain improvement in time consumption and energy consumption. The task unloading method for vehicle user moving edge calculation in the present embodiment refers to fig. 15, and mainly includes the following key steps:
1, basic principle:
1.1, a scene model;
1.2, calculating tasks by a user;
1.3, executing the task locally;
1.4, unloading the task to an edge server;
and 2, unloading utility of the user task:
3, description of algorithm:
3.1, simulating relevant contents of an annealing mechanism;
3.2, initial unloading strategy;
3.3, state transition mechanism;
3.4, an edge calculation task unloading algorithm based on a simulated annealing mechanism;
in the step 1.1, scenes are arranged at a crossroad with dense vehicles, a plurality of roadside devices RSUs are arranged at the crossroad, and an MEC server is arranged at the crossroad and is regarded as a server group. Surrounding vehicles (vehicles) are mobile users that can communicate with any one of the server farms and offload task resources to the servers. Therefore, the user and server groups are represented as follows: v ═ 1,2, 3.. times, V), M ═ 1,2, 3.. times, n, and RSU and MEC are in one-to-one correspondence, and therefore M ∈ M can be used instead.
Step 1.2 the user calculation tasks are as follows: let the Task (Task) be denoted by T, assuming that any one user v, at any one time, has one userThe computing task to be performed is TvAnd this task cannot be broken down into smaller subtasks. Each task corresponds to two necessary parameters, namely the Size of the processing task (Size) and the work load, i.e. the Size of the computations (Cycles) required to perform the task, and can therefore be denoted Tv=<ds[btts],dc[cycles]>。
The locally executed tasks in step 1.3 are as follows: each computing task may be executed locally (Local) or off-loaded to the edge server, and although off-loading the task to the edge server may reduce the consumption of Local computing, it may correspondingly increase the time consumption and partial energy consumption of the upload of the corresponding task. Unifying the types of all vehicle users, the ability of the users to perform calculations locally being the same, lv[cycles/s],dcvFor workloads, user v executes computing task T locally, incorporating the computing resources required for each taskvThe time required is as shown in equation (1):
when the user performs the calculation task locally, the energy consumption power of the user v for performing the calculation isThe energy consumption of the computing task when executed locally can be obtained as follows:
the task is unloaded to the edge server in step 1.4 as follows: if the task is uploaded to the server, the time consumption of the task is increased, the time can be divided into transmission time and time for execution on the server, and meanwhile, the execution of the calculation task also generates energy consumption.
As shown in fig. 2, in the Orthogonal Frequency Division Multiple Access (OFDMA) technique, as an evolution of OFDM, after a channel is subcarrier-converted by OFDM, transmission data is loaded on a part of subcarriers, so that a user can select a subchannel with a better channel condition for data transmission, and it can be ensured that each subcarrier is used by a user with a better corresponding channel, thereby obtaining diversity gain on frequency.
Assuming that the bandwidth B is divided into N identical sub-bands, each having a size of W ═ B/N [ Hz ], in the physical layer of the communication network for car networking, in combination with OFDM technology, 7 sub-channels of 10MHz are divided for transmission, so it can be seen that each server is divided by 7 sub-channels.
User v may transmit data in any sub-band of any server m within communication range, which may be denoted asSince the user may choose to perform the computing task locally or on the server, assumeIn time, it means that the user is unloaded to the edge server in the computing task, so the constraint condition can be obtained:
in addition, in the use of the orthogonal frequency division multiplexing technology, certain noise and interference exist, which can affect the signal reception and cause the reduction of the communication quality, and can be represented by a signal-to-interference-plus-noise ratio, a signal gain matrix from a user to a server in a range represented by H, and a signal gain matrix from the user to the server represented by HTo represent the output power matrix of the user byRepresenting the option to offload the computing task to all users on the edge server, it can therefore be computed that:
combining with a shannon formula, the obtained information transmission rate is as follows:
Cv,m=W log2(1+sv,m) (5)
wherein s isv,mFor all sub-bands on the edge server s, i.e.:
it can be known from equation (5) that the transmission of information is not only related to the bandwidth of the channel, but also affected by the signal-to-noise ratio. In addition, combining the size of the calculation task obtained before, the time for the user v to transmit the task can be obtained as follows:
when a service receives a task uploaded by a user through a sub-band, the task execution work can be carried out at an edge server end, and the calculation execution capacity of a single server is rm[cycles/s]Since the same server can process multiple tasks simultaneously, the constraint condition can be obtained:
v is a user who offloads the computation task to the edge server, that is, the sum of the computation execution capacity of each task allocated to the server m cannot exceed the total computation capacity of the server, and by combining the computation task model, the execution time of the task at the server side can be obtained as follows:
in summary, when user v chooses to offload a computing task to an edge server for execution, the total time consumed is:
the user v unloads the task to the edge server for execution, the energy consumption of the user is mainly generated during the task transmission, and the energy consumption generated during the transmission of the user can be obtained as follows:
further, in step 2.1, in a given environment, the user experience of the computing task mainly comes from the execution completion time of the task and the energy consumption generated during the execution, and the objective optimization function, namely the unloading effect of the single user task, is synthesized as follows:
Qva higher value of (b) indicates better unloading effectiveness. Wherein λ is1And λ2Represents the bias weight of the user to the time consumption and energy in the calculation task, and satisfies { lambda1+λ2=1|λ1∈[0,1],λ2∈[0,1]}. In the case of a better demand for time consumption, λ can be increased1Taking value of (A), otherwise increasing λ2The value of (a).
The new task unloading method based on the mobile edge computing technology is that the execution time of the task and the energy consumption of the task execution are improved by optimizing the execution decision of the computing task selected by each user, and the maximum user unloading utility meeting the conditions is obtained, namely:
Qmax=max(∑v∈VQv) (13)
after expanding equation (13), the following equation can be made:
on the premise of balancing and considering unloading time consumption and energy consumption, consider lambda1=λ2C is 0.5, so equation (14) can be converted to:
whereinIs a fixed value. And optimizing the optimal solution of the user unloading utility function by combining a simulated annealing algorithm.
Further, the simulated annealing algorithm in step 3.1 is a heuristic algorithm and is also a greedy algorithm, and the difference is that in the searching process, due to the introduction of a random factor, when a feasible solution is iteratively updated, an element worse than a current value is received at a certain probability, so that the simulated annealing algorithm can jump out a local optimal solution, and thus a global optimal solution is obtained.
Initial temperature T0Simulating the temperature at which the annealing algorithm starts, considering the number of user vehicles as the initial temperature, i.e. T0=V。
The temperature reduction coefficient α is an exponential decrease method according to the literature, and therefore T (n +1) ═ α T (n) is included, and α is a temperature reduction coefficient and has a value of 0.8 to 0.99.
And (4) a termination temperature, wherein if no state can be updated in the case of a plurality of iterations or the set termination temperature is reached, the annealing is considered to be finished.
And (3) assuming that the current state is f (n), and the state is changed into f (n +1) at the next moment, the probability p that the system is changed from f (n) to f (n +1) is as follows:
where Δ f ═ f (n) -f (n + 1).
The higher the temperature, i.e. the optimization is started, the greater the probability of cooling down.
The method for obtaining the initial temperature by adopting the initial unloading strategy in the step 3.2 comprises the following steps: and under the condition that the unloading constraint condition is met, selecting a mode meeting the maximum unloading effect according to a formula (12) to execute the task, so as to obtain an initial unloading strategy, namely an initial temperature.
Step 3.3 the state transition method is as follows: the state transition is that under the condition of meeting the set constraint, a user selects a set of calculation task execution modes, and changes the calculation task execution modes of the user through random probability, namely whether to execute unloading and selects an unloaded sub-band to establish a new unloading strategy.
Step 3.4 the edge calculation task unloading algorithm based on the simulated annealing mechanism is executed as follows:
step 1: obtaining the number of vehicle users in the communication range according to the communication range of the edge server to obtain the initial temperature;
step 2: and (3) constructing an initial strategy for user unloading through the algorithm 1 according to the formula (12) and combining with user unloading constraints, and calculating the initial system unloading utility according to the formula (15).
And step 3: if the set termination temperature is not reached, a new unloading strategy is obtained through the algorithm 2 within the specified iteration frequency range, and then the new system unloading effectiveness is calculated according to the formula (15);
and 4, step 4: comparing the values of the unloading utility of the system at the current time and the last time, if the value is greater than the unloading utility of the system at the last time, accepting the unloading strategy at the current time, updating the current solution of the function, otherwise, obtaining the cooling probability p according to a formula (16), generating a random number which is uniformly distributed, if p is greater than the random number, accepting the unloading strategy, otherwise, abandoning the unloading strategy at the current time;
and 5: adjusting the current temperature according to the temperature reduction coefficient, and repeatedly executing the steps (2) - (4)
Step 6: and when the end temperature is reached or the specified iteration times are reached, the annealing is ended, and the obtained user unloading effect is the optimal solution of the function.
The MATLAB platform is used in the experiment to test the set algorithm TOSA. Through comparing the three aspects of calculation tasks of different time periods (namely different traffic flow environments), different numbers of servers and different sizes, and comparing the three aspects of calculation tasks with other task unloading methods, experiments prove that the TOSA algorithm has certain improvement in time consumption and energy consumption.
The simulation will consider three performance indicators, which are:
1. different traffic flow environments. By comparing the TOSA algorithm with other task unloading methods under different traffic flow environments, the improvement of the algorithm in terms of time consumption and energy consumption is observed.
2. A different number of servers. By using the TOSA algorithm in comparison with other task offloading methods at different numbers of servers, the improvement of the algorithm in terms of time consumption and energy consumption was observed.
3. Different sizes of computational tasks. By comparing the TOSA algorithm with other task unloading methods under different sizes of computing tasks, the improvement of the algorithm in terms of time consumption and energy consumption is observed.
The results of the simulation experiments for this example are as follows:
1) user energy efficiency consumption conditions under different time consumption preference values
As can be seen from fig. 3, as the user preference for time consumption increases, the average time consumed by each user in performing the calculation task decreases, and the larger the traffic flow, the more significant the decrease in calculation time. In a high traffic scenario, the average per-user task computation time drops by 72% when the user's preference for time reaches a maximum.
Fig. 4 is a schematic diagram of energy consumption of users under different time consumption preference values, and it can be seen that as the preference of the users for time increases, the energy consumption of task execution of the users increases, and the energy consumption of the users increases more obviously in an environment with a large traffic flow. This is because, in a low traffic flow scenario, most users choose to offload computing tasks to the MEC server for execution, and thus, the energy consumption for executing tasks is not large, but only the energy consumption for transmission. On the other hand, when the time consumption preference value of the user is increased, the TOSA pays more attention to the time of task calculation, that is, the weight of energy consumption is reduced, so that a scheme with smaller time consumption is selected in the decision of task unloading, for example, the task is locally executed, which results in the energy consumption of task execution being increased. In the case of a large traffic volume, the MEC server has a large load, so in order to reduce the time consumption, more tasks are executed locally, and thus the energy consumption is increased more obviously.
2) Energy consumption situation of MEC server number users executing calculation task under different traffic flow environments
Fig. 5 is a schematic diagram of energy consumption of computing tasks executed by users with different numbers of MEC servers in two traffic flow environments, and it can be seen that, because the SMSEF is mainly optimized from the energy consumption of computing tasks of users, the users can choose to offload tasks to the MEC servers as much as possible, and thus the energy consumption of the users themselves is not large. However, the TOSA algorithm and the LSO algorithm provided herein take both the time consumption and the energy consumption of the calculation task into consideration, so the energy consumption is relatively high, and in an environment with a large traffic flow, the energy consumption of users of the TOSA, the LSO and the SMSEF algorithms is increased significantly. On the other hand, as the number of MEC servers increases, the three algorithms all decrease in energy consumption. When the MEC server number reaches the maximum, the TOSA algorithm reduces the energy consumption by 45% compared with the LSO algorithm.
3) Time consumption situation under different traffic flow environments
Fig. 6 is a comparison graph of task computation time in two traffic flow environments, and it can be seen from the graph that as the number of MEC servers increases, more computation tasks can be offloaded to the MEC servers for execution, and thus the time consumption is reduced. The TOSA algorithm proposed herein reduces the task computation time by 24% as the number of MEC servers increases, which is about 55% lower than the SMSEF algorithm and about 33% lower than the LSO algorithm.
4) User task offload utility under different MEC server numbers
The computing task unloading utility of the three algorithms is shown in figure 7 by combining two aspects of time consumption and energy consumption. As can be seen from the figure, when the MEC number is small, the task offloading effect is much lower than that of the TOSA algorithm and the LSO algorithm because the SMSEF algorithm only considers the energy consumption in the low traffic flow environment, and the task offloading effect is less than 0 in the high traffic flow environment, that is, the task offloading is not suitable for the task offloading, according to the formula (12) to the formula (15), because the SMSEF algorithm only considers the energy consumption but neglects the time consumption. With the increase of the number of MEC servers, the task unloading utility of the three algorithms is improved, wherein the TOSA algorithm provided by the invention comprehensively considers two aspects of time consumption and energy consumption, so that the task unloading utility is superior to the other two algorithms in performance, and the performance is more obvious in the environment with larger traffic flow. When the number of servers reaches a maximum, the task offload utility is 40% higher than the LSO algorithm and 29% higher than the SMSEF algorithm.
5) Unloading decision comparison of three methods under different traffic flows
FIGS. 8, 9 and 10 show the unloading decision comparison of three algorithms for two size calculation tasks under different traffic flow conditions, and the preference values of the user in time consumption and energy consumption are considered to be the same, namely lambda1=λ 21/2. As can be seen from fig. 8, in the case that the number of MEC servers is fixed, as the traffic flow and the size of the calculation task increase, the three algorithms both increase in task calculation time, and as the calculation task increases, the task calculation time of the SMSEF algorithm is significantly increased. This is because SMSEF offloads a large number of tasks to be executed on the MEC server, resulting inThe time consumption of the computing task is exacerbated. When the maximum traffic flow is reached, the time consumption of the TOSA algorithm is 7.8% lower than that of the LSO algorithm and 60% lower than that of the SMSEF algorithm.
6) Influence of traffic flow on energy consumption of users
Figure 9 illustrates the effect of traffic flow on energy consumption by a user. The energy consumption of the user is increased along with the increase of the traffic flow, and under the condition of uniformly considering the time consumption and the energy consumption preference value, the TOSA algorithm selects the calculation task of the user to be executed locally, so that the energy consumption of the user is increased. But at maximum traffic flow, the energy consumption of the TOSA algorithm is 21% lower than the LSO algorithm.
7) Impact of traffic flow on user offload utility
With the integration of task computation time and user energy consumption, FIG. 10 illustrates the impact of traffic flow on user offload utility. In a simulation experiment, the user unloading effectiveness of the three algorithms is increased along with the change of the traffic flow environment, but the user unloading effectiveness is slightly lower than that of a small-sized computing task when a large computing task is faced, which also indicates that the MEC server is mainly suitable for processing tasks with small computing amount.
8) Energy consumption situation of three algorithms under different preference values
FIG. 12 is a schematic diagram of the user's energy consumption for the three algorithms over a length of time of day under different preference values. It can be seen that the TOSA and LSO algorithms are more energy consuming during high traffic periods, since when the traffic is large, part of the calculation tasks are directly performed locally, thus increasing the energy consumption, when the decision is biased towards energy consumption (λ |)10.3) the energy consumption drops, the average user energy consumption of the TOSA algorithm is about 42% lower than the LSO algorithm.
FIG. 11 is a statistical graph of traffic flow during a time of day, and FIGS. 12, 13 and 14 are graphs obtained by selecting lambda in accordance with the statistical graph of traffic flow during the time of day in combination with time consumption and energy consumption preference values for vehicle users1=0.3,λ20.7 (more towards energy consumption) and λ1=0.7,λ2Three algorithms compared at 0.3 (more time consuming) within one dayTask computation time, task energy consumption and user offload utility comparison map.
And integrating task calculation time in one day and user energy consumption. Figure 14 shows a comparison of user offload utility for three algorithms. It can be seen that the TOSA algorithm herein outperforms the LSO algorithm in both biased mission computation time and biased energy consumption situations. The user unloading utility of the SMSEF algorithm is unstable, when the traffic flow is small, the task is unloaded to the MEC server to be executed, the too much blocking cannot be caused, but when the traffic flow is increased, the task blocking is serious, so the user unloading utility is lower than 0 and is not suitable for unloading, the user unloading utility is 45% higher than the average user unloading utility of the LSO algorithm under the condition of time consumption preference, and the user unloading utility is 27% higher than the LSO algorithm under the condition of energy consumption preference.
Claims (4)
1. A task unloading method for vehicle user mobile edge calculation is characterized by mainly comprising the following steps:
1, basic principle:
1.1, a scene model;
1.2, calculating a task by a user;
1.3, executing the task locally;
1.4, unloading the task to an edge server;
and 2, unloading utility of the user task:
2.1, unloading utility of a single user task;
3, description of algorithm:
3.1, simulating relevant contents of an annealing mechanism;
3.2, initial unloading strategy;
3.3, state transition mechanism;
3.4, an edge calculation task unloading algorithm based on a simulated annealing mechanism;
in step 1.1, the scene is set at an intersection with dense vehicles, a plurality of roadside devices RSU are arranged at the intersection, and MEC servers are configured, and the intersection is regarded as a server group, the surrounding vehicles (vehicles) are mobile users, the users communicate with any one of the server group and unload task resources to the servers, and the users and the server group are represented as follows: v ═ M ∈ M, since RSUs and MECs are in one-to-one correspondence, all of them can be replaced by M ∈ M;
step 1.2 the user calculation tasks are as follows: let the Task (Task) be denoted by T, and let T be any user v who has a computing Task to perform at any timevAnd this task cannot be decomposed into smaller subtasks any more, each corresponding to two necessary parameters, namely the Size of the processing task (Size) and the workload that needs to be done, i.e. the Size of the calculations (Cycles) needed to execute the task, hence denoted Tv=<ds[bits],dc[cycles]>;
The locally executed tasks in step 1.3 are as follows: each calculation task is executed or unloaded to the edge server locally (Local), although unloading the task to the edge server can reduce the consumption of Local calculation, correspondingly, the time consumption and partial uploading energy consumption are increased when the corresponding task is uploaded, the types of all vehicle users are unified, and the capacity of the users for executing the calculation locally is the same, namely lv[cycles/s],dcvFor the workload, in combination with the computing resources required for each task, it is obtained that user v, if it executes computing task T locallyvThe time required is as shown in equation (1):
when the user performs the calculation task locally, the energy consumption power of the user v for performing the calculation isThe required time isObtaining computing tasks to execute locallyThe energy consumption in running is as follows:
the task is unloaded to the edge server in step 1.4 as follows: if the task is uploaded to the server, the time consumption of the task is increased, the time is mainly divided into transmission time and time for execution on the server, and meanwhile, the execution of the calculation task also generates energy consumption;
user v transmits data in any sub-band of any server m within communication range, denoted as Since the user chooses to perform a computing task locally or on a server, assumeWhen, it means that the user is off-loaded to the edge server in the computing task, therefore we derive the constraint:
in addition, in the use of the ofdm technology, there is a certain amount of noise and interference, which affect the reception of signals and cause a reduction in communication quality, represented by the signal-to-interference-plus-noise ratio, represented by H, the signal gain matrix from the user to the server within the range, and represented by HTo represent the output power matrix of the user byRepresenting the option to offload the computing task to all users on the edge server, therefore compute:
dividing the bandwidth B into N identical sub-bands, each having a size of W ═ B/N [ Hz ], and obtaining, by combining with a shannon formula, a transmission rate of information as:
Cv,m=W log2(1+sv,m) (5)
wherein s isv,mFor all sub-bands on the edge server s, i.e.:
it is known from formula (5) that the transmission of information is not only related to the bandwidth of the channel, but also affected by the snr, and in addition, the time of the transmission task of user v is obtained by combining the size of the calculation task obtained before:
when a service receives a task uploaded by a user through a sub-band, the task execution work can be carried out at an edge server end, and the calculation execution capacity of a single server is rm[cycles/s]Because the same server can process a plurality of tasks simultaneously, the constraint condition is obtained:
v is a user for unloading the computing task to the edge server, namely the sum of the computing execution capacity of each task allocated to the server m cannot exceed the total computing capacity of the server, and the execution time of the task at the server end is obtained by combining a computing task model as follows:
in summary, when user v chooses to offload a computing task to an edge server for execution, the total time consumed is:
the user unloads the task to the edge server for execution, the energy consumption of the user is mainly generated during the task transmission, and the energy consumption generated by the user v during the transmission is obtained as follows:
in the given environment in step 2.1, the user experience of the computing task comes from the execution completion time of the task and the energy consumption generated during the execution, the two are combined to optimize the objective function, and the unloading effect of the single user task is as follows:
Qva higher value of (A) indicates better unloading effect, wherein λ1And λ2Represents the bias weight of the user to the time consumption and energy in the calculation task, and satisfies { lambda1+λ2=1|λ1∈[0,1],λ2∈[0,1]In case of better demand on time consumption, λ can be increased1Taking value of (A), otherwise increasing λ2Taking the value of (A);
on the other hand, if too many computing tasks are offloaded to the serverOn-execution, which increases the pressure on the MEC server, causes a blocking of tasks on the server, and also increases energy consumption, in which case the user task off-loading utility is lower than when performing computing tasks locally, when Q isvWhen the task is less than or equal to 0, the task is far less effective than the task executed locally on the MEC server, so the computing task should not be unloaded to the MEC server for execution;
therefore, the new task offloading method based on the mobile edge computing technology improves the task execution time and the energy consumption of task execution by optimizing the execution decision of the computing task selected by each user, and obtains the maximum user offloading utility satisfying the conditions, that is:
Qmax=max(∑v∈VQv) (13)
in conjunction with the previous specification, the following constraints are obtained:
(1) the vehicle user can select the task execution mode, and the task execution mode is locally executed or transmitted to the edge server;
(2) the same edge server processes the computing tasks of a plurality of users;
(3) selecting any sub-frequency band of a certain edge server in the range to transmit data;
(4) if the user selects to unload the computing task to a certain server, but the current server has too much load to cause the unloading effect of the user to be too low, the task unloading is not executed;
after expanding equation (13), the following equation is made:
on the premise of balancing unloading time consumption and energy consumption, the lambda is considered1=λ2C is 0.5, thus equation (14) translates to:
whereinOptimizing the optimal solution of the user unloading utility function for a fixed value by combining a simulated annealing algorithm;
the simulated annealing algorithm in the step 3.1 is a heuristic algorithm and is also a greedy algorithm, and the difference is that in the searching process, because random factors are introduced, when a feasible solution is updated iteratively, an element which is worse than a current value is received at a certain probability, so that the simulated annealing algorithm jumps out of a local optimal solution, and a global optimal solution is obtained;
initial temperature T0Simulating the temperature at which the annealing algorithm starts, considering the number of user vehicles as the initial temperature, i.e. T0=V;
The temperature reduction coefficient α is an exponential decrease method according to literature, and therefore T (n +1) ═ α T (n) exists, and α is a temperature reduction coefficient and has a value of 0.8 to 0.99;
a termination temperature, if there is no updated state under several iterations, or a set termination temperature is reached, the annealing is considered complete;
and (3) assuming that the current state is f (n), and the state is changed into f (n +1) at the next moment, the probability p that the system is changed from f (n) to f (n +1) is as follows:
wherein Δ f ═ f (n) -f (n + 1);
the higher the temperature, i.e. the optimization is started, the greater the probability of cooling down.
2. The task offloading method towards vehicle user mobile edge computing as recited in claim 1 wherein in step 3.2, an initial offloading strategy is used to obtain the initial temperature by: and under the condition that the unloading constraint condition is met, selecting a mode meeting the maximum unloading effect per se according to a formula (12) to execute the task, thereby obtaining an initial unloading strategy, namely the initial temperature.
3. Task off-loading method for vehicle user mobile edge computing oriented according to claim 1, characterized in that step 3.3 said state transition method is as follows: the state transition is that under the condition of meeting the set constraint, a user selects a set of calculation task execution modes, and changes the calculation task execution modes of the user through random probability, namely whether to execute the unloading and selects the unloaded sub-band to establish a new unloading strategy.
4. The task offloading method for vehicle user mobile edge computing as recited in claim 1, wherein the step 3.4 of the edge computing task offloading algorithm based on simulated annealing mechanism is performed as follows:
step 1: obtaining the number of vehicle users in the communication range according to the communication range of the edge server, and obtaining the initial temperature;
step 2: constructing an initial strategy of user unloading by combining user unloading constraint according to a formula (12), and calculating initial system unloading utility according to a formula (15);
and step 3: if the set termination temperature is not reached, a new unloading strategy is obtained within the specified iteration frequency range, and then the new unloading effect of the system is calculated according to a formula (15);
and 4, step 4: comparing the values of the unloading utility of the system at the current time and the last time, if the value is greater than the unloading utility of the system at the last time, accepting the unloading strategy at the current time, updating the current solution of the function, otherwise, obtaining the cooling probability p according to a formula (16), and generating a random number which is uniformly distributed, if the p is greater than the random number, accepting the unloading strategy, otherwise, abandoning the unloading strategy at the current time;
and 5: adjusting the current temperature according to the cooling coefficient, and repeatedly executing the steps (2) - (4);
step 6: and when the end temperature is reached or the specified iteration times are reached, the annealing is ended, and the obtained user unloading effect is the optimal solution of the function.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110102608.5A CN112835637B (en) | 2021-01-26 | 2021-01-26 | Task unloading method for vehicle user mobile edge calculation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110102608.5A CN112835637B (en) | 2021-01-26 | 2021-01-26 | Task unloading method for vehicle user mobile edge calculation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112835637A CN112835637A (en) | 2021-05-25 |
CN112835637B true CN112835637B (en) | 2022-05-17 |
Family
ID=75931617
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110102608.5A Expired - Fee Related CN112835637B (en) | 2021-01-26 | 2021-01-26 | Task unloading method for vehicle user mobile edge calculation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112835637B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113364860B (en) * | 2021-06-03 | 2022-06-24 | 吉林大学 | Method and system for joint calculation resource allocation and unloading decision in MEC |
CN113542357B (en) * | 2021-06-15 | 2022-05-31 | 长沙理工大学 | Electric vehicle auxiliary mobile edge calculation unloading method with minimized energy consumption cost |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108920279A (en) * | 2018-07-13 | 2018-11-30 | 哈尔滨工业大学 | A kind of mobile edge calculations task discharging method under multi-user scene |
CN109814951A (en) * | 2019-01-22 | 2019-05-28 | 南京邮电大学 | The combined optimization method of task unloading and resource allocation in mobile edge calculations network |
CN112256349A (en) * | 2020-10-26 | 2021-01-22 | 重庆邮电大学 | SSA-SA algorithm-based mobile edge computing task unloading method |
-
2021
- 2021-01-26 CN CN202110102608.5A patent/CN112835637B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108920279A (en) * | 2018-07-13 | 2018-11-30 | 哈尔滨工业大学 | A kind of mobile edge calculations task discharging method under multi-user scene |
CN109814951A (en) * | 2019-01-22 | 2019-05-28 | 南京邮电大学 | The combined optimization method of task unloading and resource allocation in mobile edge calculations network |
CN112256349A (en) * | 2020-10-26 | 2021-01-22 | 重庆邮电大学 | SSA-SA algorithm-based mobile edge computing task unloading method |
Also Published As
Publication number | Publication date |
---|---|
CN112835637A (en) | 2021-05-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110035410B (en) | Method for joint resource allocation and computational offloading in software-defined vehicle-mounted edge network | |
CN111836283B (en) | Internet of vehicles resource allocation method based on MEC multi-server | |
CN110312231B (en) | Content caching decision and resource allocation optimization method based on MEC in Internet of vehicles | |
CN109413615B (en) | Energy latency tradeoff for MEC-based energy-aware offloading in the Internet of vehicles | |
CN111918245B (en) | Multi-agent-based vehicle speed perception calculation task unloading and resource allocation method | |
CN111010684B (en) | Internet of vehicles resource allocation method based on MEC cache service | |
CN112835637B (en) | Task unloading method for vehicle user mobile edge calculation | |
Huang et al. | Vehicle speed aware computing task offloading and resource allocation based on multi-agent reinforcement learning in a vehicular edge computing network | |
CN111132074B (en) | Multi-access edge computing unloading and frame time slot resource allocation method in Internet of vehicles environment | |
CN112188627B (en) | Dynamic resource allocation strategy based on state prediction | |
CN115297171B (en) | Edge computing and unloading method and system for hierarchical decision of cellular Internet of vehicles | |
CN112698940B (en) | Vehicle auxiliary edge calculation task distribution system for vehicle-road cooperation | |
CN115134242B (en) | Vehicle-mounted computing task unloading method based on deep reinforcement learning strategy | |
CN113641417B (en) | Vehicle security task unloading method based on branch-and-bound method | |
CN113286329A (en) | Communication and computing resource joint optimization method based on mobile edge computing | |
CN116566838A (en) | Internet of vehicles task unloading and content caching method with cooperative blockchain and edge calculation | |
CN114826454B (en) | Intelligent resource management method in MEC-assisted Internet of vehicles communication system | |
CN114928611A (en) | Internet of vehicles energy-saving calculation unloading optimization method based on IEEE802.11p protocol | |
Wang et al. | Research on V2I/V2V hybrid multi-hop edge computing offloading algorithm in IoV environment | |
CN114374949A (en) | Power control mechanism based on information freshness optimization in Internet of vehicles | |
CN113709249A (en) | Safe balanced unloading method and system for driving assisting service | |
CN111311091B (en) | Expressway task detection and scheduling method and system based on vehicle-mounted cloud and unmanned aerial vehicle | |
CN117156493A (en) | QoE-based vehicle networking task unloading and resource allocation method | |
CN114928893B (en) | Architecture based on intelligent reflecting surface and task unloading method | |
Ma et al. | Edge computing and UAV swarm cooperative task offloading in vehicular networks |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220517 |