CN107896374A - A kind of cloudlet dynamic deployment method of facing moving terminal equipment - Google Patents
A kind of cloudlet dynamic deployment method of facing moving terminal equipment Download PDFInfo
- Publication number
- CN107896374A CN107896374A CN201711234702.6A CN201711234702A CN107896374A CN 107896374 A CN107896374 A CN 107896374A CN 201711234702 A CN201711234702 A CN 201711234702A CN 107896374 A CN107896374 A CN 107896374A
- Authority
- CN
- China
- Prior art keywords
- cloudlet
- mobile device
- mrow
- center
- region
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/18—Network planning tools
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/60—Software deployment
- G06F8/61—Installation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
Abstract
The invention discloses a kind of cloudlet dynamic deployment method of facing moving terminal equipment, comprise the following steps:Step 1, according to mobile device current location, K mobile device can be obtained by K means algorithms and assemble center;Step 2, the position at mobile device aggregation center is abstracted into a width path profile and handled;Step 3, according to above-mentioned aggregation center, ineligible mobile device is filtered out by cloudlet Placement Strategy and assembles center;Step 4, aggregation center and cloudlet current location are subjected to location matches, obtain each cloudlet position to be reached;Step 5, the mobile route of cloudlet is obtained by dijkstra's algorithm, cloudlet is moved to target location.
Description
Technical field
The present invention relates to mobile field of cloud calculation, the cloudlet dynamic state part management side of particularly a kind of facing moving terminal equipment
Method.
Background technology
With the popularization of mobile device, people are also more and more extensive to the demand of Mobile solution.It is current to have occurred being permitted
The exigent Mobile solution of multipair computing capability, operating lag, now, using the demand proposed remote ultra-mobile device
The calculating disposal ability of itself.So introduce the concept of mobile cloud computing in field of mobile equipment, so as to come solve it is current should
The problem of field face.Mobile cloud computing can move to the application on mobile device high in the clouds progress by way of remotely migrating
Processing, disposal ability is calculated so as to greatly improve.But the distance of high in the clouds and mobile terminal is all distant at present, is being migrated
During, the high transition delay of comparison may be brought.For some applications stricter to response time requirement, move
It is more short better that shifting delay needs.Such as the application such as recognition of face, interactive game, once transition delay is too high, it will influence user
Use feel.
For the huge delay for overcoming remote application migration to bring, one kind be called " cloudlet " small-sized cloud be deployed to from
The closer place in family, provides the user the cloud service of enhancing.On cloudlet, two types are can be generally divided into:First, network
Operator provides small server, then by connected reference WAP (Access Points, AP), self-organizing is formed;Separately
A kind of mode is by the mobile device under multiple P2P networks, there is provided idling-resource organizes the formation of.Pass through cloudlet, mobile device
It can will require that high application moves to nearest cloudlet and handled to calculating disposal ability.So, can not only subtract significantly
Huge delay caused by few remotely migrating, reaches the response time needed for application-specific, can also increase the processing speed of application
Degree.
This resource of cloudlet is rare, and cost is very high, therefore can not possibly be deployed in institute place in need.In order to
Strengthen cloud service experience to user as much as possible, how to effectively utilize limited cloudlet becomes current and its popular grind
Study carefully topic.Specifically, that is, refer in a specific region, the cloudlet of certain amount is given, how inside this block region
Cloudlet is affixed one's name to, can make it that the mobile device number that cloudlet covers is most, so as to reach highest cloudlet utilization rate.
The research that current research personnel do, which is directed to how designing efficient algorithm, makes cloudlet be covered in placement process to the greatest extent
Network node more than possible, so as to improve the utilization rate of cloudlet.For example, Z.Xu et al. is in " Efficient Algorithms
The large-scale wireless metropolitan area being made up of many WAPs is have studied in for Capacitated Cloudlet Placement "
The Placement Problems of cloudlet in net (Wireless Metropolitan Area Network, WMAN).M.Jia etc. is in " Optimal
cloudlet placement and user to cloudlet allocation in wireless metropolitan
A kind of cloudlet Placement is devised in area networks ", the user's close quarters placed it in wireless MAN, is come
Balance WMAN workload.In general, user's close quarters can dynamically produce very big change, and this change is sent out sometimes
Life is within the extremely short time.So we can only meet the needs of user by increasing cloudlet number constantly elsewhere,
Therefore a large amount of energy consumptions can be increased.In addition, the sparse regional cloudlet resource utilization of user is very low, therefore a large amount of cloudlets
The energy is wasted.
According to above-mentioned problem, invention introduces the concept of a mobile cloudlet, proposes a kind of towards mobile whole
The cloudlet dynamic deployment method of end equipment.Change huge occasion for the stream of people, can be according to the change in location reality of mobile device
When adjust cloudlet position, while can effectively strengthen user cloud service experience.
The content of the invention
Goal of the invention:The present invention changes huge occasion for the stream of people, can according to mobile device change in location it is real-time
The position of cloudlet is adjusted, while can effectively strengthen the cloud service experience of user
In order to solve the above-mentioned technical problem, the invention discloses a kind of cloudlet dynamic state part management side of facing moving terminal equipment
Method, including following 5 steps:
Step 1, according to mobile device current location, can obtain all mobile devices by K-means algorithms K gather
Collect center.
Whole mobile device scope of activities is defined as R, is expressed as in an x-y axial plane:
R=(x, y) | 0≤x≤W, 0≤y≤H }.
In this device activity region, random distribution N number of mobile device, and these mobile devices are all in dynamic
In movement.Mobile device representation is D={ d1,d2,…,dN}。dnN-th of mobile device (1≤n≤N) in the R of region is represented,
The position of each mobile device can be represented with coordinate.In t, dnPositional representation be dpn,t=(dpxn,t,
dpyn,t).Wherein dpxn,tRepresent t dnAbscissa positions in the R of region, dpyn,tRepresent t dnIt is vertical in the R of region
Coordinate position.
The quantity that the mobile mobile device in moment t region R assembles center can be obtained by K-means clustering algorithms
With position, wherein DC represents mobile device aggregation centralization, DC={ dc1,dc2,…,dcK, dckRepresent in the R of region k-th
Mobile device assembles center (1≤k≤K).In t, dckPosition with being expressed as dcpk,t=(dcpxk,t,dcpyk,t).Its
Middle dcpxk,tRepresent t dckAbscissa positions in the R of region, dcpyk,tRepresent t dckOrdinate in the R of region
Position.
Step 2, represent that mobile device assembles center using path profile, path profile is described as:G=(E, V, W), according to
According to above-mentioned path profile, mobile device can be assembled center and be adjusted on the node V of nearest figure;
According to step 1, when obtaining moment t, there is K mobile device aggregation center in the R of region.Cloudlet meeting in moving process
Many obstacles are run into, therefore whole equipment zone of action is abstracted into a path profile, are expressed as G=(E, V, W).Wherein, E generations
The mobile route of table cloudlet, V represent the position that cloudlet should be placed, and W represents the weight on all positions.The point and side table of figure
It is shown as V={ v1,v2,…,vF, E={ e1,e2,…,eF}。vfAnd efRepresent and side (1≤f≤F) respectively at f-th point.
According to above-mentioned path profile, the mobile device measured can be assembled to the section that center DC is adjusted to nearest figure
On point V.So as to obtain the position that cloudlet should be actually placed.It is as follows to adjust mobile device aggregation center method:
Assemble center dc for k-th of mobile devicekFor, it is necessary to when calculating moment t the mobile device assemble center
It is as follows with the distance of all node of graph, calculation:
Wherein, vxf,tAnd vyf,tWhen being illustrated respectively in moment t, node of graph v in the R of regionfAbscissa and ordinate position
Put.So dckIt is expressed as with the distance set of all node of graph:
If gatheringIn, dckWith vfDistance it is minimum, then dckNode of graph v will be adjusted tofOn.By this
Method, the position that cloudlet should be actually placed can be obtained.
Step 3, ineligible mobile device is filtered out by cloudlet Placement Strategy and assembles center;
Assuming that there is M cloudlet in the R of region, cloudlet expression way is CL={ cl1,cl2,…,clM}。clmRepresent region R
In m-th of cloudlet (1≤m≤M).For clmFor, each moment has where a center i.e. wireless router
Position (cloudlet is generally mounted on wireless router to strengthen the cloud service of user experience).In t, cloudlet position can be used
Coordinate representation is clpm,t=(clpxm,t,clpym,t), wherein clpxm,tRepresent t clmAbscissa positions in the R of region,
clpym,tRepresent t clmOrdinate position in the R of region.
According to cloudlet position, the distance of cloudlet and mobile device can be calculated, calculation is as follows:
Assuming that cloudlet clmRadius be rm, then the cluster tool of cloudlet covering is expressed as:
dclm(t)={ dn,t|dis(dpn,t,clpm,t)≤rm,1≤n≤N}。
In order that the number for obtaining cloudlet covering maximizes, the present invention proposes a kind of cloudlet Placement Strategy.The strategy is:
ρ represent cloudlet place needed for covering minimal equipment number (judge cloudlet whether the threshold value that can be placed).Such as
Fruit is in moment t, cloudlet clmMeet coverage condition on position P (x, y):dclm(t) >=ρ, then cloudlet can be placed on position P
On (x, y), the position is otherwise removed.
Specifically, according to the strategy in the present invention, center dc is assembled for any one mobile devicekFor, need
It and distance of all mobile devices in moment t are calculated, calculation is:
If meet condition:dis(dck,t,dn,t)≤rm, then illustrate if cloudlet clmIt is placed in mobile device aggregation
Heart dckWhen upper, equipment d can be coveredn。
For dckFor, its mobile device number is cloudlet clmThe number of devices that can be covered is dclm,t, according to putting
Put strategy, cloudlet clmAssemble center dc in mobile devicekMeet coverage condition:dclm(t) >=ρ, then illustrate that mobile device is assembled
Center dckCloudlet can be placed, otherwise then removes mobile device aggregation center dck.By cloudlet Placement Strategy, institute can be excluded
There is the mobile device aggregation center for the condition of being unsatisfactory for.
Step 4, aggregation center and cloudlet are subjected to location matches, calculate cloudlet with assembling the distance of center, obtain
Obtain the mobile device aggregation center that cloudlet needs to reach;
In order that the utilization ratio highest of cloudlet is obtained, it is necessary to judge whether mobile by corresponding cloudlet shift strategy
Cloud.Cloudlet shift strategy is as follows in the present invention:
Assuming that in moment t cloudlet clmIt is placed on position P (x, y), the mobile device number set of now cloudlet covering
For dclm,t.Elapsed time section (t, t'], some mobile devices are moved in the R of region.To moment t', if position P
Mobile device set and position P'(x', y' on (x, y)) on mobile device set meet coverage condition:
dclm(P')≥dclm(p),
Wherein, dclmAnd dcl (p)m(P') moment t and moment t' cloudlet cl is represented respectivelymThe mobile device number collection of covering
Close.And clmWith position P'(x', y') it is closest, then cloudlet clmP' will be moved to from position P.
Specifically, according to the strategy in the present invention, in moment t1When, cloudlet clmA certain mobile device is placed on to gather
Concentrate in the heart, the mobile device number collection of now cloudlet covering is combined into dclm(t1).Elapsed time section (t1, t], some movements are set
For in the R of region, there occurs dynamic mobile, new mobile device aggregation center is generated, is represented with set DC.Mobile device gathers
Collection center dckOn mobile device number be dclm(t).If dclm(t)≥dclm(t1), then calculate cloudlet clmSet with movement
The distance of standby aggregation center.Calculation is as follows:
If cloudlet clmAssemble center dc with mobile devicekDistance be less than other cloudlets and dckThe distance between, then
Cloudlet can be from moment t1Mobile device assembles center dc when the position at moment is moved to moment tkPosition.
Step 5, shortest path is found using dijkstra's algorithm, and converts the result to the motion track of cloudlet, will
Cloudlet is moved to target location.
Understood according to step 3,4, the initial position of all cloudlets, target location, can be set at figure G=(E, V,
W in), the mobile route of cloudlet can be thus converted into and finds digraph shortest route problem.Therefore the present invention passes through
Dijkstra's algorithm finds shortest path, and converts the result to the mobile route of cloudlet, and according to the path is moved cloudlet
Move to target location.The cloudlet for not having destination keeps original position constant and closed.
Thought in the present invention is:First, the position at mobile device aggregation center is determined simultaneously according to K-means clustering algorithms
The position at mobile device aggregation center is abstracted into a width path profile and handled;Then excluded not according to cloudlet Placement Strategy
Qualified mobile device assembles center;Then mobile device aggregation the distance between center and cloudlet and root are calculated
Judge whether to move cloudlet according to by corresponding cloudlet shift strategy;Cloudlet and movement are finally found using dijkstra's algorithm
Equipment assembles the shortest path at center, mobile cloudlet to target location.
Compared with prior art, beneficial effects of the present invention are embodied in:
(1) target location of cloudlet changes according to position of mobile equipment dynamic in region so that cloudlet movement result is more objective
See credible.
(2) clustering method in data mining is incorporated into mobile device aggregation center partition problem, utilizes cluster
Algorithm clusters to each mobile device, obtains highdensity mobile device aggregation center, and then select the optimal mesh of cloudlet
Cursor position.
(3), it is necessary to calculate cloudlet and the distance of mobile device aggregation center during mobile cloudlet so that the movement of cloudlet
Distance is minimum.
(4) need to judge whether the quantity of cloudlet overlay device meets cloudlet shift strategy during mobile cloudlet so that cloudlet
The mobile device number of covering maximizes.
Brief description of the drawings
Fig. 1 is flow chart of steps of the present invention.
Fig. 2, Fig. 3 are t respectively1Moment and t2Moment mobile device distribution situation and cloudlet position.
Embodiment
The present invention is illustrated below in conjunction with the accompanying drawings.It is noted that described embodiment merely to explanation
Purpose, without limiting the scope of the present invention.
The invention discloses a kind of cloudlet dynamic deployment method of facing moving terminal equipment, this method flow chart of steps is such as
Shown in Fig. 1, comprise the following steps:
Step 1, according to mobile device current location, K shifting of all mobile devices can be obtained by K-means algorithms
Dynamic equipment aggregation center.
Whole mobile device scope of activities is defined as R, is expressed as in an x-y axial plane:
R=(x, y) | 0≤x≤W, 0≤y≤H }.
In this device activity region, random distribution N number of mobile device, and these mobile devices are all in dynamic
In movement.Mobile device representation is D={ d1,d2,…,dN}。dnN-th of mobile device (1≤n≤N) in the R of region is represented,
The position of each mobile device can be represented with coordinate.In t, dnPositional representation be dpn,t=(dpxn,t,
dpyn,t).Wherein dpxn,tRepresent t dnAbscissa positions in the R of region, dpyn,tRepresent t dnIt is vertical in the R of region
Coordinate position.
Quantity and position that the mobile device in moment t region R assembles center can be obtained by K-means clustering algorithms
Put, wherein DC represents mobile device aggregation centralization, DC={ dc1,dc2,…,dcK, dckRepresent k-th of movement in the R of region
Equipment assembles center (1≤k≤K).In t, dckPosition with being expressed as dcpk,t=(dcpxk,t,dcpyk,t).Wherein
dcpxk,tRepresent t dckAbscissa positions in the R of region, dcpyk,tRepresent t dckOrdinate position in the R of region
Put.
Step 2, represent that mobile device assembles center using path profile, path profile is described as:G=(E, V, W), according to
According to above-mentioned path profile, mobile device can be assembled center and be adjusted on the node V of nearest figure;
According to step 1, when obtaining moment t, there is K mobile device aggregation center in the R of region.Cloudlet meeting in moving process
Many obstacles are run into, therefore whole equipment zone of action is abstracted into a path profile, are expressed as G=(E, V, W).Wherein, E generations
The mobile route of table cloudlet, V represent the position that cloudlet should be placed, and W represents the weight on all positions.The point and side table of figure
It is shown as V={ v1,v2,…,vF, E={ e1,e2,…,eF}。vfAnd efRepresent and side (1≤f≤F) respectively at f-th point.
According to above-mentioned path profile, the mobile device measured can be assembled to the section that center DC is adjusted to nearest figure
On point V.So as to obtain the position that cloudlet should be actually placed.It is as follows to adjust mobile device aggregation center method:
Assemble center dc for k-th of mobile devicekFor, it is necessary to when calculating moment t the mobile device assemble center
It is as follows with the distance of all node of graph, calculation:
Wherein, vxf,tAnd vyf,tWhen being illustrated respectively in moment t, node of graph v in the R of regionfAbscissa and ordinate position
Put.So dckIt is expressed as with the distance set of all node of graph:
If gatheringIn, dckWith vfDistance it is minimum, then dckNode of graph v will be adjusted tofOn.By this
Method, the position that cloudlet should be actually placed can be obtained.
Step 3, ineligible mobile device is filtered out by cloudlet Placement Strategy and assembles center;
Assuming that there is M cloudlet in the R of region, cloudlet expression way is CL={ cl1,cl2,…,clM}。clmRepresent region R
In m-th of cloudlet (1≤m≤M).For clmFor, each moment has where a center i.e. wireless router
Position (cloudlet is generally mounted on wireless router to strengthen the cloud service of user experience).In t, cloudlet position can be used
Coordinate representation is clpm,t=(clpxm,t,clpym,t), wherein clpxm,tRepresent t clmAbscissa positions in the R of region,
clpym,tRepresent t clmOrdinate position in the R of region.
According to cloudlet position, the distance of cloudlet and mobile device can be calculated, calculation is as follows:
Assuming that cloudlet clmRadius be rm, then the cluster tool of cloudlet covering is expressed as:
dclm(t)={ dn,t|dis(dpn,t,clpm,t)≤rm,1≤n≤N}。
In order that the number for obtaining cloudlet covering maximizes, the present invention proposes a kind of cloudlet Placement Strategy.The strategy is:
ρ represent cloudlet place needed for covering minimal equipment number (judge cloudlet whether the threshold value that can be placed).Such as
Fruit is in moment t, cloudlet clmMeet coverage condition on position P (x, y):dclm(t) >=ρ, then cloudlet can be placed on position P
On (x, y), the position is otherwise removed.
Specifically, according to the strategy in the present invention, center dc is assembled for any one mobile devicekFor, need
It and distance of all mobile devices in moment t are calculated, calculation is:
If meet condition:dis(dck,t,dn,t)≤rm, then illustrate if cloudlet clmIt is placed in mobile device aggregation
Heart dckWhen upper, equipment d can be coveredn。
For dckFor, its mobile device number is cloudlet clmThe number of devices that can be covered is dclm,t, according to putting
Put strategy, cloudlet clmAssemble center dc in mobile devicekMeet coverage condition:dclm(t) >=ρ, then illustrate that mobile device is assembled
Center dckCloudlet can be placed, otherwise then removes mobile device aggregation center dck.By cloudlet Placement Strategy, institute can be excluded
There is the mobile device aggregation center for the condition of being unsatisfactory for.
Step 4, aggregation center and cloudlet are subjected to location matches, calculate cloudlet with assembling the distance of center, obtain
Obtain the mobile device aggregation center that cloudlet needs to reach;
In order that the utilization ratio highest of cloudlet is obtained, it is necessary to judge whether mobile by corresponding cloudlet shift strategy
Cloud.Cloudlet shift strategy is as follows in the present invention:
Assuming that in moment t cloudlet clmIt is placed on position P (x, y), the mobile device number set of now cloudlet covering
For dclm,t.Elapsed time section (t, t'], some mobile devices are moved in the R of region.To moment t', if position P
Mobile device set and position P'(x', y' on (x, y)) on mobile device set meet coverage condition:
dclm(P')≥dclm(p),
Wherein, dclmAnd dcl (p)m(P') moment t and moment t' cloudlet cl is represented respectivelymThe mobile device number collection of covering
Close.And clmWith position P'(x', y') it is closest, then cloudlet clmP' will be moved to from position P.
Specifically, according to the strategy in the present invention, in moment t1When, cloudlet clmA certain mobile device is placed on to gather
Concentrate in the heart, the mobile device number collection of now cloudlet covering is combined into dclm(t1).Elapsed time section (t1, t], some movements are set
For in the R of region, there occurs dynamic mobile, new mobile device aggregation center is generated, is represented with set DC.Mobile device gathers
Collection center dckOn mobile device number be dclm(t).If dclm(t)≥dclm(t1), then calculate cloudlet clmSet with movement
The distance of standby aggregation center.Calculation is as follows:
If cloudlet clmAssemble center dc with mobile devicekDistance be less than other cloudlets and dckThe distance between, then
Cloudlet can be from moment t1Mobile device assembles center dc when the position at moment is moved to moment tkPosition.
Step 5, shortest path is found using dijkstra's algorithm, and converts the result to the motion track of cloudlet, will
Cloudlet is moved to target location.
Understood according to step 3,4, the initial position of all cloudlets, target location, can be set at figure G=(E, V,
W in), the mobile route of cloudlet can be thus converted into and finds digraph shortest route problem.Therefore the present invention passes through
Dijkstra's algorithm finds shortest path, and converts the result to the mobile route of cloudlet, and according to the path is moved cloudlet
Move to target location.The cloudlet for not having destination keeps original position constant and closed.
Embodiment
In an experiment, we create hadoop cluster with two node master and slave.It is simple in order to test
Change, tested here using the method for analogue simulation.Zone of action is shaped to a square, length of side 280m.
In this region, there are 4 cloudlets and random distribution 350 mobile devices.While in order to closer to truth, move
The distribution of equipment meets the rule of Gaussian Profile.
In t1Moment, position of mobile equipment and cloudlet position in region are set to the initial bit of mobile device and cloudlet
Put, therefore t can be obtained1The quantity at the mobile device aggregation center at moment and its position, as shown in table 1.Mobile device is assembled
The position at center both cloudlet position.
Table 1.t1Moment mobile device assembles center
DC | dc1 | dc2 | dc3 | dc4 |
Coordinate position | (70,200) | (80,60) | (190,115) | (240,220) |
Assemble the position at center according to mobile device, t can be obtained1The number of devices of moment cloudlet covering, such as the institute of table 2
Show.
Table 2.t1Moment cloudlet overlay device quantity
CL | cl1 | cl2 | cl3 | cl4 |
Overlay device quantity | 48 | 100 | 72 | 50 |
Elapsed time section (t1,t2], mobile device has dynamically carried out irregular movement in the region, therefore, in t2
The position at the moment region mobile device aggregation center occurs dynamic and changed, it is therefore desirable to which mobile cloudlet meets the need of mobile device
Ask.According to method provided by the invention, we can determine whether t2The quantity at the mobile device aggregation center at moment and its position, such as
Shown in table 3.
Table 3.t2Moment mobile device assembles center
DC | dc1 | dc2 | dc3 | dc4 |
Coordinate position | (95,180) | (80,60) | (215,70) | (240,220) |
Assemble the position at center according to mobile device, t can be obtained2The number of devices of moment cloudlet covering, such as the institute of table 2
Show.
Table 4.t2Moment cloudlet overlay device quantity
CL | cl1 | cl2 | cl3 | cl4 |
Overlay device quantity | 50 | 103 | 84 | 52 |
From table 2 and table 4, t1Moment and t2The mobile device sum of moment cloudlet covering is respectively 270 and 289.Cause
This can be seen that method proposed by the present invention can make it that cloudlet covering efficiency is higher.
Fig. 2, Fig. 3 more intuitively embody the situation of cloudlet dynamic mobile.In figure, black dot represents random distribution
Mobile device, black large circle point are that A, B, C and D represent cloudlet t respectively1The position at moment, A', B, C and D' represent cloudlet t respectively2
The position at moment, black arrow represent the moving direction of cloudlet, and represented by dashed circles is loaded with the letter of the wireless router of cloudlet
Number coverage (namely cloudlet signal cover).
In t1Moment, each cloudlet are in current time Optimal coverage position.T is arrived2Moment, a large amount of mobile device positions
Put and change, according to the cloudlet dynamic deployment method of the facing moving terminal equipment proposed in the present invention, it is determined that two need
The cloudlet A and D of dynamic mobile are carried out, and piece being moved on moment Optimal coverage position A' and D', B and location of C
Cloud remains stationary as.
The invention provides a kind of cloudlet dynamic deployment method of facing moving terminal equipment, the technical scheme is implemented
Method and approach it is a lot, described above is only the preferred embodiments of the invention, it is noted that for the general of the art
For logical technical staff, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improve and
Retouching also should be regarded as protection scope of the present invention.The available prior art of each part being not known in the present embodiment is subject to reality
It is existing.
Claims (6)
1. a kind of cloudlet dynamic deployment method of facing moving terminal equipment, it is characterised in that comprise the following steps:
Step 1, according to mobile device current location, K mobile device can be obtained by K-means algorithms and assemble centre bit
Put;
Step 2, represent that mobile device assembles center using path profile, path profile is described as:G=(E, V, W), according to upper
The path profile stated, mobile device can be assembled center and be adjusted on the node V of nearest figure;
Step 3, ineligible mobile device is filtered out by cloudlet Placement Strategy and assembles center;
Step 4, location matches are done at aggregation center and cloudlet, calculates cloudlet and assemble the minimum distance of center, obtain
Cloudlet needs the mobile device aggregation center reached;
Step 5, shortest path is found by dijkstra's algorithm, and converts the result to the motion track of cloudlet, by cloudlet
It is moved to target location.
A kind of 2. cloudlet dynamic deployment method of facing moving terminal equipment according to claim 1, it is characterised in that step
In rapid 1, whole equipment scope of activities is defined as R in an x-y axial plane, be expressed as R=(x, y) | 0≤x≤W, 0≤y≤
H};
In this device activity region, random distribution N number of mobile device, and these mobile devices are all in dynamic mobile
In;Mobile device representation is D={ d1,d2,…,dN};dnN-th of mobile device (1≤n≤N) in the R of region is represented, it is each
The position of individual mobile device is all represented with coordinate;In t, dnPositional representation be dpn,t=(dpxn,t,dpyn,t);Wherein
dpxn,tRepresent t dnAbscissa positions in the R of region, dpyn,tRepresent t dnOrdinate position in the R of region;
Quantity and position that the mobile device in moment t region R assembles center can be obtained by K-means clustering algorithms,
Wherein DC represents mobile device aggregation center, DC={ dc1,dc2,…,dcK, dckRepresent that k-th of mobile device gathers in the R of region
Collection center (1≤k≤K);In t, dckPosition with being expressed as dcpk,t=(dcpxk,t,dcpyk,t);Wherein dcpxk,tTable
Show t dckAbscissa positions in the R of region, dcpyk,tRepresent t dckOrdinate position in the R of region.
3. a kind of cloudlet dynamic deployment method of facing moving terminal equipment according to claim 1 or 2, its feature exist
In in step 2, according to step 1, when obtaining moment t, there is K mobile device aggregation center in the R of region;Cloudlet is in moving process
In can run into many obstacles, therefore whole equipment zone of action is abstracted into a path profile, is expressed as G=(E, V, W);Its
In, V represents the position that cloudlet should be placed, and E represents the mobile route of cloudlet, and W represents the weight on all positions;The point of figure
V={ v are expressed as with side1,v2,…,vF, E={ e1,e2,…,eF};vfAnd efRepresent and side (1≤f≤F) respectively at f-th point;
According to above-mentioned path profile, the mobile device measured can be assembled to the node V that center DC is adjusted to nearest figure
On;So as to obtain the position that cloudlet should be actually placed.
A kind of 4. cloudlet dynamic deployment method of facing moving terminal equipment according to claim 1, it is characterised in that step
In rapid 3, it is assumed that have M cloudlet in the R of region, cloudlet expression way is CL={ cl1,cl2,…,clM};clmRepresent in the R of region
M-th of cloudlet (1≤m≤M);For clmFor, there be position of the center i.e. where wireless router at each moment
Put, cloudlet is generally mounted on wireless router to strengthen the cloud service of user experience;In t, cloudlet position can be used and sit
Mark is expressed as clpm,t=(clxm,t,clym,t), wherein clpxm,tRepresent t clmAbscissa positions in the R of region,
clpym,tRepresent t clmOrdinate position in the R of region;
According to cloudlet position, the distance of cloudlet and mobile device is calculated, calculation is as follows:
<mrow>
<mi>d</mi>
<mi>i</mi>
<mi>s</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>dp</mi>
<mrow>
<mi>n</mi>
<mo>,</mo>
<mi>t</mi>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>clp</mi>
<mrow>
<mi>m</mi>
<mo>,</mo>
<mi>t</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msqrt>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>dpx</mi>
<mrow>
<mi>n</mi>
<mo>,</mo>
<mi>t</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>clpx</mi>
<mrow>
<mi>m</mi>
<mo>,</mo>
<mi>t</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>dpy</mi>
<mrow>
<mi>n</mi>
<mo>,</mo>
<mi>t</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>clpy</mi>
<mrow>
<mi>m</mi>
<mo>,</mo>
<mi>t</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
<mo>;</mo>
</mrow>
Assuming that cloudlet clmRadius be rm, then the cluster tool of cloudlet covering is expressed as:
dclm,t={ dn,t|dis(dpn,t,clpm,t)≤rm,1≤n≤N}。
In order that the number for obtaining cloudlet covering maximizes, represent that cloudlet places the required minimal equipment number covered with ρ, that is, judge
The threshold value whether cloudlet is placed;If in moment t, cloudlet clmMeet coverage condition on position P (x, y):dclm,t>=ρ, that
Cloudlet is placed on position P (x, y), otherwise removes mobile device aggregation center.
A kind of 5. cloudlet dynamic deployment method of facing moving terminal equipment according to claim 1, it is characterised in that step
In rapid 4, in order to keep the utilization ratio highest of cloudlet, it is necessary to judge whether to move cloudlet by corresponding cloudlet shift strategy;
Cloudlet shift strategy is as follows:
Assuming that in moment t cloudlet clmIt is placed on position P (x, y), the mobile device number collection of now cloudlet covering is combined into
dclm,t;Elapsed time section (t, t'], some mobile devices are moved in the R of region;To moment t', if position P
Mobile device set and position P'(x', y' on (x, y)) on mobile device set meet coverage condition:
dclm(p)≥dclm(P'),
Wherein, dclmAnd dcl (p)m(P') moment t and moment t' cloudlet cl is represented respectivelymThe mobile device number set of covering.
And clmWith position P'(x', y') it is closest, then cloudlet clmP' will be moved to from position P.
A kind of 6. cloudlet dynamic deployment method of facing moving terminal equipment according to claim 1, it is characterised in that step
In rapid 5, according to step 3 and step 4, the initial position of all cloudlets, target location, figure G=(E, V, W) can be set at
In, thus the mobile route of cloudlet is converted into and finds digraph shortest route problem;Therefore by dijkstra's algorithm come
Shortest path is found, and converts the result to the mobile route of cloudlet, and cloudlet is moved to by target location according to the path;Not yet
Cloudlet purposefully keeps original position constant.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711234702.6A CN107896374A (en) | 2017-11-30 | 2017-11-30 | A kind of cloudlet dynamic deployment method of facing moving terminal equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711234702.6A CN107896374A (en) | 2017-11-30 | 2017-11-30 | A kind of cloudlet dynamic deployment method of facing moving terminal equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107896374A true CN107896374A (en) | 2018-04-10 |
Family
ID=61806984
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711234702.6A Pending CN107896374A (en) | 2017-11-30 | 2017-11-30 | A kind of cloudlet dynamic deployment method of facing moving terminal equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107896374A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102740312A (en) * | 2012-07-12 | 2012-10-17 | 北京邮电大学 | Coverage control method of wireless sensor network |
CN103297983A (en) * | 2013-05-06 | 2013-09-11 | 南京邮电大学 | Wireless sensor network node dynamic deployment method based on network flow |
CN104981783A (en) * | 2013-03-07 | 2015-10-14 | 思杰***有限公司 | Dynamic configuration in cloud computing environments |
US20170289002A1 (en) * | 2016-03-31 | 2017-10-05 | Mrittika Ganguli | Technologies for deploying dynamic underlay networks in cloud computing infrastructures |
-
2017
- 2017-11-30 CN CN201711234702.6A patent/CN107896374A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102740312A (en) * | 2012-07-12 | 2012-10-17 | 北京邮电大学 | Coverage control method of wireless sensor network |
CN104981783A (en) * | 2013-03-07 | 2015-10-14 | 思杰***有限公司 | Dynamic configuration in cloud computing environments |
CN103297983A (en) * | 2013-05-06 | 2013-09-11 | 南京邮电大学 | Wireless sensor network node dynamic deployment method based on network flow |
US20170289002A1 (en) * | 2016-03-31 | 2017-10-05 | Mrittika Ganguli | Technologies for deploying dynamic underlay networks in cloud computing infrastructures |
Non-Patent Citations (3)
Title |
---|
HAOLONG XIANG,XIAOLONG XU,ETC.: "An Adaptive Cloudlet Placement Method for Mobile Applications over GPS Big Data", 《2016 IEEE GLOBAL COMMUNICATIONS CONFERENCE (GLOBECOM)》 * |
薛胜军,胡敏达,许小龙: "云环境下公平性优化的资源分配方法", 《CNKI 计算机应用》 * |
许小龙: "支持绿色云计算的资源调度方法及关键技术研究", 《中国博士学位论文全文数据库 信息科技辑(月刊)》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Xu et al. | Uav-assisted task offloading for iot in smart buildings and environment via deep reinforcement learning | |
CN103179052B (en) | A kind of based on the central virtual resource allocation method and system of the degree of approach | |
CN107277889B (en) | Wireless sensor network clustering method based on k-means | |
CN108632831B (en) | Unmanned aerial vehicle cluster frequency spectrum resource allocation method based on dynamic flight path | |
Patel et al. | A hybrid ACO/PSO based algorithm for QoS multicast routing problem | |
Ji et al. | Particle swarm optimization for mobile ad hoc networks clustering | |
CN110794965B (en) | Virtual reality language task unloading method based on deep reinforcement learning | |
CN103971160A (en) | Particle swarm optimization method based on complex network | |
Yao et al. | Coverage enhancement strategy for WSNs based on virtual force-directed ant lion optimization algorithm | |
Qu et al. | Empowering edge intelligence by air-ground integrated federated learning | |
CN113573333B (en) | Particle swarm heterogeneous WSNs coverage optimization algorithm based on virtual force | |
CN104837155B (en) | The cluster-dividing sensing network method of data capture of class BP neutral net | |
CN104219319A (en) | Method for distributed network flow self-organizing scheduling | |
Mi et al. | Software-defined green 5G system for big data | |
CN105205216B (en) | The cross-cutting emulation mode of extensive pedestrian movement and device in hinge | |
CN103402265B (en) | Based on fuzzy logic and the frequency spectrum distributing method of communication priority | |
Wang et al. | Towards fairer and more efficient federated learning via multidimensional personalized edge models | |
CN106656612B (en) | A kind of approximation method for super-intensive network system traversal and rate | |
CN107896374A (en) | A kind of cloudlet dynamic deployment method of facing moving terminal equipment | |
Annauth et al. | OFDM systems resource allocation using multi-objective particle swarm optimization | |
CN106656797B (en) | A method of considering the AFDX network path optimization of risk balance | |
CN115510652A (en) | Crowd gathering and evacuating simulation system and method based on digital twinning technology | |
CN115118591A (en) | Cluster federation learning method based on alliance game | |
CN101655712B (en) | Cluster control method of mobile robot | |
CN109412882A (en) | Feature extracting method, device, server and computer readable storage medium |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180410 |
|
WD01 | Invention patent application deemed withdrawn after publication |