CN109376374B - Multi-user computing migration method based on multi-radio frequency communication - Google Patents
Multi-user computing migration method based on multi-radio frequency communication Download PDFInfo
- Publication number
- CN109376374B CN109376374B CN201811017232.2A CN201811017232A CN109376374B CN 109376374 B CN109376374 B CN 109376374B CN 201811017232 A CN201811017232 A CN 201811017232A CN 109376374 B CN109376374 B CN 109376374B
- Authority
- CN
- China
- Prior art keywords
- user
- radio frequency
- sub
- migration
- task
- 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
Links
Images
Classifications
-
- 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
-
- 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
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention belongs to the field of mobile cloud computing, and particularly relates to a multi-user computing migration method based on multi-radio frequency communication, which comprises the following steps of: the controller utilizes a multi-radio frequency multi-user computation migration modeling method based on parameter analysis to model a multi-radio frequency multi-user computation migration problem according to the collected overall user information and the system target to obtain an initial problem model theta; the controller relaxes and converts the model by utilizing a relaxation and linearization method based on McCormick envelope according to the characteristics of the initial problem model theta obtained in the last step to obtain a relaxed problem model theta * (ii) a The controller utilizes a branch limit method to solve the problem model theta obtained in the last step after relaxation * The solution is performed so that the total beneficial users are the most and the user migration cost is reduced. The method provided by the invention considers the problems of the calculation time delay and the calculation energy of the user at the same time, can meet the diversity of the user requirements, increases the data transmission rate of the user, and greatly improves the utilization efficiency of multiple radio frequencies.
Description
Technical Field
The invention belongs to the field of mobile cloud computing, and particularly relates to a multi-user computing migration method based on multi-radio frequency communication.
Background
With the popularization of smart phones, more and more mobile phone applications appear, such as: speech recognition, face recognition, augmented reality, etc. These applications consume a lot of resources, including the computing resources (CPU) of the mobile phone and the battery power of the mobile phone, and the small size of the mobile phone determines that the computing resources and the battery power of the mobile phone are limited. Therefore, the mobile phone resource shortage and the application resource demand are too large, and the relationship between the two presents a great challenge to the development of a future mobile phone platform. Computational migration is an emerging technology that addresses the above-mentioned problems. The computing task of the mobile terminal is migrated to the cloud facility with sufficient resources for computing by using the wireless network, namely, the mobile cloud computing is adopted, so that the problem of insufficient resources of the mobile terminal is solved.
The main problem of computing migration is how to improve the device performance of the user and to save the energy consumption of the user device. Much work has previously been done to investigate the issue of computing migration saving power consumption and latency. The first category mainly studies the problem of computation migration under the condition of single user and multiple radio frequencies. It optimizes the communication resources when one user uploads tasks using multiple radio frequencies to minimize the task execution energy consumption, but does not take into account the case of simultaneous transmission of multiple users (Mahmoodi, s.em, k.p.subalkshmi, and video player. "Cloud streaming for multi-radio enabled mobile devices." Communications (ICC), "IEEE International Conference on.ieee, 2015.). The second type studies the computational migration problem under multi-user single-channel conditions. Which minimizes the sum of the transmit power of all users by optimizing the channel resources and the transmit power of the users. However, it only considers the case where a plurality of users share one channel for data transmission, and does not consider the case where a plurality of frequency bands are used for data transmission (barbarosa, s. Sardelitti, and p.d. lorenza. "Joint allocation of calculation and communication resources in multi-user mobile closed calculation." 395.6 (2013): 26-30.). The third category mainly studies the problem of computation migration under multi-user multi-channel conditions. Which minimizes the sum of the overall user task performance costs by optimizing the multi-channel communication resources. However, it only considers dividing a frequency band into multiple channels for transmission by different users, and does not consider that multiple frequency band resources can be utilized simultaneously in multi-radio communication (d.huang, p.wang, and d.niyato, "a dynamic streaming algorithm for mobile computing." IEEE Transactions on Wireless Communications, vol.11, no.6, pp.1991-1995, 2012.). Furthermore, existing patents on computational migration, by search, do not deal with multi-user multi-radio scenarios.
In summary, the following problems mainly exist in the prior art:
(1) In the multi-radio frequency communication, only a single user is considered, and the problem of resource competition among users is not considered.
(2) Only single-channel communication and multi-channel communication are considered in multi-user communication, and communication by using multiple radio frequencies is not considered. Transmission resources of a plurality of frequency bands are not fully utilized, and a plurality of channels cannot be simultaneously utilized for transmission.
Disclosure of Invention
The invention provides a multi-user computing migration method based on multi-radio frequency communication. The method has the main idea that modeling is carried out on multi-user multi-radio frequency calculation migration; then, carrying out conversion relaxation on the model by using an Mccormick envelope method; and finally, solving the problem by using the branch limit to obtain an optimal resource allocation scheme.
Before resource optimization, the following operations need to be completed. Firstly, a user acquires information such as equipment residual capacity, computing capacity, radio frequency number, each radio frequency transmission power and the like; then, the user acquires the task attribute to be calculated. Calculating the data size of a task and calculating the required CPU period information; finally, the user sets own preference information for calculating delay and energy and sends all the user-related information to the controller. The controller is generally deployed at a server end and used for controlling the operation of the whole process, and the main tasks include collecting user data, modeling a multi-user calculation migration problem based on multi-radio frequency communication, relaxing and optimizing the model, solving the model and the like.
A multi-user calculation migration method based on multi-radio frequency communication comprises the following steps:
(1.1) the controller utilizes a multi-radio frequency multi-user computation migration modeling method based on parameter analysis to model a multi-radio frequency multi-user computation migration problem according to the collected overall user information and the system target to obtain an initial problem model theta;
(1.2) the controller relaxes and converts the model by using a relaxation and linearization method based on McCormick envelope according to the characteristics of the initial problem model theta obtained in the last step to obtain a relaxed problem model theta * ；
(1.3) the controller utilizes a branch limit method to carry out processing on the relaxed problem model theta obtained in the last step * Solving for the most users that benefit overall and user migrationThe shift cost is reduced.
The step (1.1) comprises the following steps:
(2.1) respectively calculating that the user i completes the task J locally i Required time T i (m) And the energy consumed E i (m) ：
Wherein task J i @(g i ,D i )，g i Represents the completion of task J i Number of CPU cycles required, D i Represents the completion of task J i Amount of data required to upload, f i (m) Representing the device computing power (number of CPU cycles), p, of the user i i Representing the energy consumed by the equipment of the user i per CPU cycle, U is a user set, N is the number of users,
is a radio frequency set, K is the number of radio frequencies;
(2.2) calculating the local calculation cost of the user i
Wherein the content of the first and second substances,
respectively representing preference weights of user i for computing delay and computing energy consumption, wherein
And->
(2.3) if user i chooses to count in the cloudIf yes, calculating the transmission rate of the user i on the jth radio frequency
The invention adopts a wireless competition channel transmission model to transmit data;
wherein the content of the first and second substances,
represents the proportion of user i that performs task transmission with radio frequency j, based on the comparison result>
(2.4) respectively calculating the data uploading time of the user i on the radio frequency j
And the energy consumed by uploading data by the user i through the radio frequency j>
Wherein D is i Indicating user i completed task J i Amount of data to be uploaded, P i j The transmission power of radio frequency j on the mobile device representing user i;
(2.5) calculation task J i Calculating the required time T at the cloud i,cloud And energy consumed by the equipment in idle state
Wherein, f i (c) Representing the computing power, P, allocated by user i in the cloud i (IDLE) Represents the power of the device of user i in the idle state;
(2.6) calculating task Transmission time T i (mtc) And sum of energy consumed by users at each radio frequency
Wherein the content of the first and second substances,
uploading the energy consumed by data for a user i through a radio frequency j;
(2.7) calculating the calculation cost of the user i in the cloud
Wherein the content of the first and second substances,
energy consumed in an idle state of the equipment;
(2.8) since the user benefits only if the cost calculated by the user in the cloud is less than the cost calculated locally, the migration is performed, so the following constraints are obtained:
(2.9) due to
And &>
Represents whether the user has migrated, so there are the following constraints:
(2.10) the number of persons who will benefit totally
As a target, the constraint of the target is ^ greater or greater>
Satisfy a constraint by solving>
The maximum beneficial number of people is obtained.
The step (1.2) comprises:
(3.1) mixing
Conversion to the following form to remove the non-linear max function:
(3.2) introduction of variables
Will->
Equivalent replacement is by ^ or ^>
The following four conditions need to be satisfied: />
(1) If it is not
Then there should be +>
(2) If it is not
Then there should be->
(3) If it is not
Then there should be->
(4) If it is not
Then there should be->
Two constraints are as follows:
(3.3) to be restricted
Equivalent substitutions are made with the following constraints:
(3.4) introduction of New variables
Because it is->
There are therefore the following constraints:
①
②
③
④
(3.5) simplifying the constraint, and obtaining the following constraint after replacement:
(3.6) due to
And according to the physical meaning: />
Thus is used->
Replacement->
Namely, it is
Get the target function as->
The convex optimization problem as described above is constrained.
The step (1.3) comprises:
(4.1) setting an upper bound UB 0 = + ∞, lower bound LB = - ∞, objective function value obj =0, wait queue wlist = Φ, iteration count variable k =0;
(4.2) set the relaxor problem SP (ω) kth ,β kth ) Where kth represents the kth sub-problem and the constraint β = {0,1} is relaxed to 0 ≦ β ≦ 1, adding the relaxed sub-problem to the queue, wlist
(4.3) judging whether the queue is empty, if not, executing the following steps;
(4.4) selecting a sub-problem from the queue, removing the queue and solving the sub-problem;
(4.5) if the sub-problem is unsolved, pruning the branch;
(4.6) if the sub-problem optimal solution is obtained and all
All integers, set LB = obj, and remove all UBs in the queue k A node < LB;
(4.7) if the sub-problem optimal solution is obtained
Not all integers, pair problem SP (omega) k ,β k ) Adding constraints
Get the sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k = obj; then, the sub-problem SP (ω) k ,β k ) Adding a restraint->
To sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k ＝obj；
(4.8) jumping to the step (4.3) until an optimal solution is obtained.
The invention has the beneficial effects that:
(1) The method provided by the invention simultaneously considers the problems of the calculation time delay and the calculation energy of the user, and can meet the diversity of the user requirements.
(2) The method provided by the invention simultaneously utilizes a plurality of radio frequencies to carry out data transmission, the radio frequencies do not interfere with each other, and the data transmission rate of a user is greatly improved.
(3) The resource allocation method provided by the invention optimizes the task calculation place of the user, the radio frequency set utilized by the user and the data volume transmitted by each radio frequency, so that the user benefits the most overall, and the utilization efficiency of multiple radio frequencies is greatly improved.
(4) The resource allocation method provided by the invention considers the calculation cost of the user in the local and cloud simultaneously, and effectively reduces the calculation cost of the total user.
Drawings
FIG. 1 is a flow chart of a multi-user computation migration method based on multi-radio frequency communication;
FIG. 2 is a flow chart of user problem modeling in a multi-user computational migration method based on multi-radio frequency communication;
FIG. 3 is a flow chart of problem transformation by using newly-built variables and McCormick evenlayer relaxation in a multi-user computation migration method based on multi-radio frequency communication;
FIG. 4 is a flow chart of a solution of a multi-user computation migration method based on multi-radio frequency communication by using a branch and bound method;
FIG. 5 is a diagram of an embodiment of the present invention;
Detailed Description
The invention is further described below with reference to the accompanying drawings.
The invention belongs to the field of mobile cloud computing, and particularly relates to a multi-user computing migration method based on multi-radio frequency communication.
With the popularization of smart phones, more and more mobile phone applications appear, such as: speech recognition, face recognition, augmented reality, etc. These applications consume a lot of resources, including the computing resources (CPU) of the mobile phone and the battery power of the mobile phone, and the small size of the mobile phone determines that the computing resources and the battery power of the mobile phone are limited. Therefore, the mobile phone resource shortage and the application resource demand are too large, and the relationship between the mobile phone resource shortage and the application resource demand provides a great challenge for the development of a future mobile phone platform. Computational migration is an emerging technology that addresses the above-mentioned problems. The computing task of the mobile terminal is migrated to the cloud facility with sufficient resources for computing by using the wireless network, namely, the mobile cloud computing is adopted, so that the problem of insufficient resources of the mobile terminal is solved.
The main problem of computing migration is how to improve the device performance of the user and to save the energy consumption of the user device. Much work has previously been done to investigate the issue of computing migration saving power consumption and latency. The first category mainly studies the problem of computation migration under the condition of single user and multiple radio frequencies. It optimizes the communication resources when one user uploads tasks using multiple radio frequencies to minimize the task execution energy consumption, but does not take into account the case of simultaneous transmission of multiple users (Mahmoodi, s.em, k.p.subalkshmi, and video player. "Cloud streaming for multi-radio enabled mobile devices." Communications (ICC), "IEEE International Conference on.ieee, 2015.). The second type studies the computational migration problem under multi-user single channel conditions. Which minimizes the sum of the transmit power of all users by optimizing the channel resources and the transmit power of the users. However, it only considers the case that a plurality of users share one channel for data transmission, and does not consider the case that a plurality of frequency bands are used for data transmission (barbarosas, s., s.sardelitti, and p.d. lorenzo. "Joint allocation of transmission and communication resources in multi-user mobile closed transmission." 395.6 (2013): 26-30.). The third category mainly studies the problem of computation migration under multi-user multi-channel conditions. Which minimizes the sum of the overall user task performance costs by optimizing the multi-channel communication resources. However, it only considers dividing a frequency band into multiple channels for transmission by different users, and does not consider that multiple frequency band resources can be utilized simultaneously in multi-radio communication (d.huang, p.wang, and d.niyato, "a dynamic streaming algorithm for mobile computing." IEEE Transactions on Wireless Communications, vol.11, no.6, pp.1991-1995, 2012.). Furthermore, the existing patents on computing migration do not relate to multi-user multi-radio scenarios, as retrieved.
In summary, the following problems mainly exist in the current research work:
only single user is considered in multi-radio frequency communication, and the problem of resource competition among users is not considered
Only single-channel communication and multi-channel communication are considered in multi-user communication, and communication by using multiple radio frequencies is not considered. Transmission resources of a plurality of frequency bands are not fully utilized, and a plurality of channels cannot be simultaneously utilized for transmission.
According to the above analysis, the deficiency of the current research work is mainly due to the following key factors that need to be additionally considered during the migration of multi-radio frequency calculation:
(1) How the user selects the appropriate set of radio frequencies to communicate may make it most beneficial to the user.
(2) The user utilizes the ratio of different radio frequency transmission tasks.
(3) How multiple users perform resource allocation.
A user is said to be a beneficiary user if the overhead (computational delay and power consumption) incurred by the computational migration is less than that performed locally. The concept of beneficiary users is very important in computing migration. First, from a user perspective, it ensures that the user is rational, as the user will only migrate if the migration cost is less than the local execution cost; second, from an operator perspective, more benefitting users means higher utilization of cloud servers and more revenue for computing services.
Therefore, the invention researches the calculation migration problem under multi-user multi-radio frequency, each user can simultaneously utilize a plurality of frequency bands for transmission, the multi-user calculation migration problem based on multi-radio frequency conditions is provided, the resource allocation is optimized, the total beneficial users can be maximized, and the total calculation cost is obviously reduced. In other words, the invention optimizes the problem by comprehensively considering the following elements: whether each user performs task migration or not, (2) a migration user selects a proper radio frequency set for migration, (3) the proportion of each radio frequency migration data used by the user, and (4) how to distribute the users who benefit the most overall.
The invention provides a multi-user computing migration method based on multi-radio frequency communication. The method has the main idea that modeling is carried out on multi-user multi-radio frequency calculation migration; then, carrying out conversion relaxation on the model by using an Mccormick envelope method; and finally, solving the problem by using the branch limit to obtain an optimal resource allocation scheme.
Before resource optimization, the following operations need to be completed. Firstly, a user acquires information such as residual electric quantity, calculation capacity, radio frequency number, transmission power of each radio frequency and the like of equipment; then, the user acquires the task attribute to be calculated. Calculating the data size of a task and calculating the required CPU period information; finally, the user sets the preference information of the user for calculating delay and energy and sends the relevant information of all the users to the controller. The controller is generally deployed at a server end and used for controlling the operation of the whole process, and the main tasks include collecting user data, modeling a multi-user calculation migration problem based on multi-radio frequency communication, relaxing and optimizing the model, solving the model and the like.
The multi-user calculation migration method based on multi-radio frequency communication mainly comprises the following specific steps:
(1) The controller utilizes a multi-radio frequency multi-user computation migration modeling method based on parameter analysis to model a multi-radio frequency multi-user computation migration problem according to the collected information of all users and the target of the system, and an initial problem model theta is obtained.
(2) The controller relaxes and converts the model by utilizing a relaxation and linearization method based on McCormick envelope according to the characteristics of the initial problem model theta obtained in the last step to obtain a relaxed problem model theta * 。
The controller utilizes a branch limit method to solve the problem model theta obtained in the last step after relaxation * The solution is performed so that the overall benefit is the most users and the user migration cost is reduced.
The multi-radio frequency multi-user computation migration modeling method based on parameter analysis in the step (1) specifically further comprises the following steps:
(1.1) respectively calculating that the user i completes the task J locally i Time required
And the energy expended is->
Wherein task J i @(g i ,D i )，g i Represents the completion of task J i Number of CPU cycles required, D i Represents the completion of task J i Amount of data required to upload, f i (m) Representing the device computing power (number of CPU cycles), p, of the user i i Representing the energy consumed by the device of user i per CPU cycle. The user set is U, the number of users is N, and the radio frequency set is U
The number of radio frequencies is K.
(1.2) calculating the local calculation cost of the user i
Wherein
Respectively representing preference weights of user i for computing delay and computing energy consumption, wherein
And->
(1.3) if the user i selects to calculate in the cloud, calculating the transmission rate of the user i on the jth radio frequency
The invention adopts a wireless competition channel transmission model to transmit data.
Wherein
Represents the proportion of user i that utilizes radio frequency j for task transmission, and
(1.4) respectively calculating the data uploading time of the user i on the radio frequency j
And the energy consumed by uploading data by the user i through the radio frequency j>
Wherein D i Indicating user i completed task J i Amount of data to be uploaded, P i j Representing the transmission power of radio frequency j on the mobile device of user i.
(1.5) calculation task J i Calculating the required time T at the cloud i,cloud And the energy consumed by the equipment in idle state
Wherein f is i (c) Representing the computing power, P, allocated by user i in the cloud i (IDLE) Representing the power of the device of user i in an idle state (i.e., when no data is being transmitted).
(1.6) calculating task Transmission time T i (mtc) And sum of energy consumed by users at each radio frequency
(1.7) calculating the calculation cost of the user i in the cloud
(1.8) since the user benefits only when the cost calculated in the cloud is less than the cost calculated in the local, the migration can be performed, so that the following constraints can be obtained:
(1.9) due to
And &>
Represents whether the user has migrated, so there are the following constraints:
(1.10) the number of persons who will benefit totally
As a target, the constraint of the target is ^ greater or greater>
Satisfy a constraint by solving>
The maximum beneficial number of people is obtained.
The relaxation and linearization method based on McCormick envelope in the step (2) specifically further comprises the following steps:
(2.1) mixing
Conversion to the following form to remove the non-linear max function:
(2.2) introduction of variables
To be +>
Equivalent replacement is by ^ or ^>
The following four conditions need to be satisfied:
(1) if it is not
Then there should be +>
(2) If it is not
Then there should be->
/>
(3) If it is not
Then there should be->
(4) If it is not
Then there should be->
Two constraints are as follows:
(2.3) to constrain
Equivalent substitutions are made with the following constraints:
(2.4) introduction of New variables
Because +>
There are therefore the following constraints:
①
②
③
④
(2.5) simplifying the constraint, and obtaining the following constraint after replacement:
(2.6) due to
And according to the physical meaning: />
Thus is used->
Replacement->
Namely that
Get an objective function of>
The convex optimization problem as described above is constrained.
The branch and limit method in step (3) further includes the following steps:
(3.1) setting an upper bound UB 0 = + ∞, lower bound LB = - ∞, objective function value obj =0, wait queue wlist = Φ, iteration count variable k =0.
(3.2) set the relaxor problem SP (ω) kth ,β kth ) Where kth represents the kth sub-problem and the constraint β = {0,1} is relaxed to 0 ≦ β ≦ 1. The slack problem is added to the queue, wlist.
(3.3) when the queue is not empty, performing the steps of:
and (3.4) selecting a sub-problem from the queue, removing the queue and solving the sub-problem.
(3.5) if the subproblem is unsolved, pruning the branch.
(3.6) if the sub-problem optimal solution is obtained and is complete
All integers, set LB = obj, and remove all queuesUB k < LB's node.
(3.7) if the optimal solution of the subproblem is obtained
Not all integers, pair sub-problem SP (ω) k ,β k ) Adding a restraint->
Get the sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k = obj. Then, the sub-problem SP (ω) k ,β k ) Adding a restraint->
To sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k ＝obj。
And (3.8) jumping to the step (3.3) until an optimal solution is obtained.
The patent provides a multi-user computing migration method based on multi-radio frequency communication, which comprises the following steps:
(1) The method provided by the invention considers the problems of the calculation time delay and the calculation energy of the user at the same time, and can meet the diversity of the user requirements.
(2) The method provided by the invention simultaneously utilizes a plurality of radio frequencies to transmit data, the radio frequencies do not interfere with each other, and the data transmission rate of a user is greatly improved.
(3) The resource allocation method provided by the invention optimizes the task calculation place of the user, the radio frequency set utilized by the user and the data volume transmitted by each radio frequency, so that the user benefits the most overall, and the utilization efficiency of multiple radio frequencies is greatly improved.
(4) The resource allocation method provided by the invention considers the calculation cost of the user in the local and cloud simultaneously, and effectively reduces the calculation cost of the total user.
FIG. 1 is a flow chart of a multi-user computing migration method based on multi-radio frequency communication.
FIG. 2 is a flow chart of user problem modeling in a multi-user computational migration method based on multi-radio frequency communication.
Fig. 3 is a flow chart of problem transformation by using new variables and McCormick envelope relaxation in a multi-user computation migration method based on multi-radio frequency communication.
FIG. 4 is a flow chart of a multi-user computation migration method based on multi-radio frequency communication and a solving method using a branch and bound method.
In this embodiment, as shown in fig. 5, there are a total of two users, user1 and User2, and there are two radio frequencies on the mobile device of each User. Each user has a computing task J i ＝(g i ,D i )，g i Represents the completion of task J i Number of CPU cycles required, D i Represents the completion of task J i The amount of data that needs to be uploaded. From the figure, task J of User1 can be seen 1 = (1GHZ, 5000KB), user2 task J 2 = (1.2GHZ, 5100KB). Both User1 and User2 can upload data through AP 1 and AP 2. Before starting the calculation, the user performs the following operation. Firstly, respectively acquiring the Computing Power (CPU), the radio frequency number (both 2) and the transmission power of radio frequency of equipment by using 1 and using 2; secondly, the User1 and the User2 respectively obtain the attributes of the calculation tasks. Wherein the calculation amount required by the calculation task of User1 is 1GHZ, the data amount required to be uploaded is 5000KB, the calculation amount required by the calculation task of User2 is 1.2GHZ, and the data amount required to be uploaded is 5100KB; then, user1 and User2 set their own weights for calculating delay and energy preference, respectively
And &>
Finally, user1 and User2 send the above information to the controller.
Then, the multi-user calculation migration method based on multi-radio frequency communication mainly comprises the following specific steps:
(1) The controller calculates the cost calculated locally according to the information uploaded by the User1 and the User2
(2) Controller adding constraints
And restrict>
The constraints represent that the local cost of the user is less than or equal to the cost calculated by the user in the cloud and the migration decision equation of the user respectively. />
Represents whether the user has migrated. At this time, it is>
When the representative user makes a calculation locally>
The user performs calculation in the cloud. At this time, an initial model of the multi-user computational migration problem based on multi-radio frequency communication is obtained.
(3) Controller introduces new variables using the proposed method
Represents whether user1 performs a data transmission using radio frequency 1, when->
It means that User1 does not use radio frequency 1 for transmission, when +>
Representing User1 transmitting data using radio frequency 1. The other variables are treated similarly.
(4) The controller adds constraints:
wherein the first two constraints are
Replacement by means of>
Sufficient requirements of (a); the latter four constraints are the constraints that occur when performing McCormick envelope relaxation. Then, will restrict >>
The following constraints translate:
(6) Controller setting upper bound UB 0 = + ∞, lower bound LB = - ∞, objective function value obj =0, wait queue wlist = Φ, iteration count variable k =0, the problem is solved using the following branch-and-bound method.
(7) Set the relaxor problem SP (ω) kth ,β kth ) Where kth represents the 1 st sub-problem and the constraint β = {0,1} is relaxed to 0 ≦ β ≦ 1. The slack problem is added to the queue wlan.
(8) When the wlan is not empty, the following steps are performed:
(9) A subproblem is selected from the wlan, removed from the queue and solved by a common convex optimization solver (matlab, cvx).
(10) If the sub-problem is not solved, the branch is pruned.
(11) If the sub-problem optimal solution is obtained and all
All integers, set LB = obj, and remove all UBs in the queue k < LB's node.
(12) If the optimal solution of the sub-problem is obtained
Not all integers, pair sub-problem SP (ω) k ,β k ) Adding a restraint->
Get the sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k = obj. Then, the sub-problem SP (ω) k ,β k ) Addition of constraints>
To sub-problem SP (ω) k+1 ,β k+1 ) Set k = k +1, ub k ＝obj。
(13) And jumping to the step 11 until an optimal solution is obtained.
Finally, the patent proposes a multi-user computing migration method based on multi-radio frequency communication, which can achieve the following beneficial effects:
(1) The method provided by the invention simultaneously considers the problems of the calculation time delay and the calculation energy of the user, and can meet the diversity of the user requirements.
(2) The method provided by the invention simultaneously utilizes a plurality of radio frequencies to transmit data, the radio frequencies do not interfere with each other, and the data transmission rate of a user is greatly improved.
(3) The resource allocation method provided by the invention optimizes the task calculation place of the user, the radio frequency set utilized by the user and the data volume transmitted by each radio frequency, so that the user benefits the most overall, and the utilization efficiency of multiple radio frequencies is greatly improved.
(4) The resource allocation method provided by the invention considers the calculation cost of the user in the local and cloud simultaneously, and effectively reduces the calculation cost of the total user.
Claims (1)
1. A multi-user computing migration method based on multi-radio frequency communication is characterized by comprising the following steps:
(1.1) the controller utilizes a multi-radio frequency multi-user computation migration modeling method based on parameter analysis to model a multi-radio frequency multi-user computation migration problem according to the collected information of all users and the target of the system, and an initial problem model theta is obtained;
(1.2) the controller relaxes and converts the model by using a relaxation and linearization method based on McCormick envelope according to the characteristics of the initial problem model theta obtained in the last step to obtain a relaxed problem model theta * ；
(1.3) the controller utilizes a branch limit method to carry out processing on the relaxed problem model theta obtained in the last step * Solving is carried out, so that the total beneficial users are the most, and the user migration cost is reduced;
the step (1.1) comprises:
(2.1) respectively calculating that the user i completes the task J locally i Required time T i (m) And the energy consumed
Wherein task J i @(g i ,D i )，g i Represents the completion of task J i Number of CPU cycles required, D i Represents the completion of task J i Amount of data required to upload, f i (m) Representing the device computing power (number of CPU cycles), p, of the user i i Representing the energy consumed by the equipment of the user i per CPU cycle, U being the user set, N being the userThe number of the first and second groups is,
is a radio frequency set, and K is the number of radio frequencies;
(2.2) calculating the calculation cost of the user i in the local
Wherein the content of the first and second substances,
represents a preference weighting of user i for the calculation of the delay and the calculation of the energy consumption, respectively, wherein &>
And->
(2.3) if the user i chooses to calculate in the cloud, calculating the transmission rate of the user i on the jth radio frequency
The invention adopts a wireless competition channel transmission model to transmit data:
wherein the content of the first and second substances,
represents the proportion of user i that performs task transmission with radio frequency j, based on the comparison result>
(2.4) respectively calculating the data uploading time of the user i on the radio frequency j
And the energy consumed by uploading data by the user i through the radio frequency j>
Wherein D is i Indicating user i completed task J i Amount of data to be uploaded, P i j The transmission power of radio frequency j on the mobile device representing user i;
(2.5) computing task J i Calculating the required time T at the cloud i,cloud And the energy consumed by the equipment in idle state
Wherein f is i (c) Representing the computing power, P, allocated by user i in the cloud i (IDLE) Represents the power of the device of user i in the idle state;
(2.6) calculating task Transmission time T i (mtc) And sum of energy consumed by users at each radio frequency
Wherein the content of the first and second substances,
uploading the energy consumed by data for a user i through a radio frequency j;
(2.7) calculating the calculation cost of the user i in the cloud
Wherein the content of the first and second substances,
energy consumed in an idle state of the equipment;
(2.8) since the user benefits only if the cost calculated by the user in the cloud is less than the cost calculated locally, the migration is performed, so the following constraints are obtained:
(2.9) due to
And &>
Represents whether the user has migrated, so there are the following constraints:
(2.10) will benefit in generalNumber of people
As a target, the constraint of the target is +>
Satisfy a constraint by solving>
Obtaining the maximum beneficial number of people;
the step (1.2) comprises:
(3.1) mixing
Conversion to the following form to remove the non-linear max function:
(3.2) introduction of variables
Will->
Equivalent replacement is by ^ or ^>
The following four conditions need to be satisfied:
(1) if it is not
Then there should be->
(2) If it is not
Then there should be->
(3) If it is used
Then there should be->
(4) If it is not
Then there should be->
Two constraints are as follows:
(3.3) to be restricted
Equivalent substitutions are made with the following constraints:
(3.4) introduction of New variables
Because +>
There are therefore the following constraints:
①
②
③
④
(3.5) simplifying the constraint, and obtaining the following constraint after replacement:
(3.6) due to
And according to the physical meaning: />
Thus using>
Replacement->
I.e. is>
Get the target function as->
Constraining the convex optimization problem as described above;
the step (1.3) comprises:
(4.1) setting an upper bound UB 0 = + ∞, lower bound LB = - ∞, objective function value obj =0, wait queue wlist = Φ, iteration count variable k =0;
(4.2) set relaxation sub-problem SP (ω) kth ,β kth ) Where kth represents the kth sub-problem and the constraint β = {0,1} is relaxed to 0 ≦ β ≦ 1, adding the relaxed sub-problem to the queue, wlist
(4.3) judging whether the queue is empty, if not, executing the following steps;
(4.4) selecting a sub-problem from the queue, removing the queue and solving the sub-problem;
(4.5) if the sub-problem is not solved, pruning the branch;
(4.6) if the sub-problem optimal solution is obtained and is complete
All integers, set LB = obj, and remove all UBs in the queue k A node < LB;
(4.7) if the optimal solution of the subproblem is obtained
Not all integers, pair sub-problem SP (ω) k ,β k ) Adding constraints
Get the sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k = obj; then, the sub-problem SP (ω) k ,β k ) Adding a restraint->
To sub-problem SP (omega) k+1 ,β k+1 ) Set k = k +1, ub k ＝obj；
(4.8) jumping to the step (4.3) until an optimal solution is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811017232.2A CN109376374B (en) | 2018-09-01 | 2018-09-01 | Multi-user computing migration method based on multi-radio frequency communication |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811017232.2A CN109376374B (en) | 2018-09-01 | 2018-09-01 | Multi-user computing migration method based on multi-radio frequency communication |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109376374A CN109376374A (en) | 2019-02-22 |
| CN109376374B true CN109376374B (en) | 2023-04-07 |
Family
ID=65404088
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811017232.2A Active CN109376374B (en) | 2018-09-01 | 2018-09-01 | Multi-user computing migration method based on multi-radio frequency communication |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109376374B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110535700B (en) * | 2019-08-30 | 2022-07-15 | 哈尔滨工程大学 | Calculation unloading method under multi-user multi-edge server scene |
| CN112188551B (en) * | 2020-09-29 | 2023-04-07 | 广东石油化工学院 | Computation migration method, computation terminal equipment and edge server equipment |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106912079A (en) * | 2017-02-20 | 2017-06-30 | 北京邮电大学 | Federated user accesses selection and resource allocation methods in one kind caching heterogeneous network |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10492102B2 (en) * | 2009-01-28 | 2019-11-26 | Headwater Research Llc | Intermediate networking devices |
| US9288744B2 (en) * | 2011-05-27 | 2016-03-15 | Nokia Technologies Oy | Method and apparatus for sharing connectivity settings via social networks |
| CN105245265B (en) * | 2014-06-20 | 2021-01-01 | 中兴通讯股份有限公司 | User service mode selection method under 3D wave beam, pilot frequency information sending method and device |
| CN107919986B (en) * | 2017-11-09 | 2020-09-08 | 北京邮电大学 | VM migration optimization method among MEC nodes in ultra-dense network |
-
2018
- 2018-09-01 CN CN201811017232.2A patent/CN109376374B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106912079A (en) * | 2017-02-20 | 2017-06-30 | 北京邮电大学 | Federated user accesses selection and resource allocation methods in one kind caching heterogeneous network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109376374A (en) | 2019-02-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109240818B (en) | Task unloading method based on user experience in edge computing network | |
| CN110099384B (en) | Multi-user multi-MEC task unloading resource scheduling method based on edge-end cooperation | |
| CN109814951B (en) | Joint optimization method for task unloading and resource allocation in mobile edge computing network | |
| CN109710336B (en) | Mobile edge computing task scheduling method based on joint energy and delay optimization | |
| CN107682443A (en) | Joint considers the efficient discharging method of the mobile edge calculations system-computed task of delay and energy expenditure | |
| CN110351754B (en) | Industrial Internet machine equipment user data calculation unloading decision method based on Q-learning | |
| CN108880893A (en) | A kind of mobile edge calculations server consolidation collection of energy and task discharging method | |
| CN104619029B (en) | It is a kind of centralization cellular network architecture under baseband pool resource allocation methods and device | |
| Shi et al. | Toward energy-efficient federated learning over 5g+ mobile devices | |
| CN108924938B (en) | Resource allocation method for calculating energy efficiency of wireless charging edge computing network | |
| CN103905343B (en) | Method for dynamically allocating broadband resources | |
| CN112286677A (en) | Resource-constrained edge cloud-oriented Internet of things application optimization deployment method | |
| CN113286317B (en) | Task scheduling method based on wireless energy supply edge network | |
| CN109376374B (en) | Multi-user computing migration method based on multi-radio frequency communication | |
| CN109905888A (en) | Combined optimization migration decision and resource allocation methods in mobile edge calculations | |
| CN109358953B (en) | Multitask application unloading method in micro cloud | |
| CN113286329B (en) | Communication and computing resource joint optimization method based on mobile edge computing | |
| CN110191489B (en) | Resource allocation method and device based on reinforcement learning in ultra-dense network | |
| CN109246761A (en) | Consider the discharging method based on alternating direction multipliers method of delay and energy consumption | |
| CN108990067B (en) | Energy efficiency control method applied to ultra-dense heterogeneous network | |
| CN112867066A (en) | Edge calculation migration method based on 5G multi-cell deep reinforcement learning | |
| CN115473896A (en) | Electric power internet of things unloading strategy and resource configuration optimization method based on DQN algorithm | |
| CN109981340B (en) | Method for optimizing joint resources in fog computing network system | |
| Dou et al. | Mobile edge computing based task offloading and resource allocation in smart grid | |
| CN110139282A (en) | A kind of energy acquisition D2D communication resource allocation method neural network based |
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 |