CN110198278A - A kind of Lyapunov optimization method in car networking cloud and the scheduling of edge Joint Task - Google Patents
A kind of Lyapunov optimization method in car networking cloud and the scheduling of edge Joint Task Download PDFInfo
- Publication number
- CN110198278A CN110198278A CN201910298270.8A CN201910298270A CN110198278A CN 110198278 A CN110198278 A CN 110198278A CN 201910298270 A CN201910298270 A CN 201910298270A CN 110198278 A CN110198278 A CN 110198278A
- Authority
- CN
- China
- Prior art keywords
- task
- queue
- lyapunov
- scheduling
- priority
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- 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
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
- H04L41/0833—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability for reduction of network energy consumption
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/50—Queue scheduling
- H04L47/56—Queue scheduling implementing delay-aware scheduling
- H04L47/564—Attaching a deadline to packets, e.g. earliest due date first
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/50—Queue scheduling
- H04L47/62—Queue scheduling characterised by scheduling criteria
- H04L47/625—Queue scheduling characterised by scheduling criteria for service slots or service orders
- H04L47/6255—Queue scheduling characterised by scheduling criteria for service slots or service orders queue load conditions, e.g. longest queue first
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/50—Queue scheduling
- H04L47/62—Queue scheduling characterised by scheduling criteria
- H04L47/625—Queue scheduling characterised by scheduling criteria for service slots or service orders
- H04L47/6275—Queue scheduling characterised by scheduling criteria for service slots or service orders based on priority
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
- H04W4/44—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for communication between vehicles and infrastructures, e.g. vehicle-to-cloud [V2C] or vehicle-to-home [V2H]
Abstract
The invention discloses the Lyapunov optimization methods that a kind of car networking cloud and edge Joint Task are dispatched, the following steps are included: the first stage, create vehicle Ad hoc network, the Ad hoc network has a Cloud Server, multiple Edge Servers, two RSU nodes and N number of node on behalf vehicle, place vehicle node using the sumo traffic flow model of Traci mobility model;Second stage: using the fuzzy queue scheduling based on Lyapunov as the vehicle node of task, it determines vehicle node being assigned to cloud and is also assigned to marginal end and carries out data transmission, fuzzy rule is exported by fuzzy Lyapunov Theory of Stability, stable queue scheduling is obtained, and maximizes the effectiveness of task schedule using Lyapunov penalty.The present invention realizes cloud, marginal end, the task schedule between terminal, rationally effective to greatly reduce the delay in task processes so that the speed for handling task becomes faster using cloud, the resource of marginal end.
Description
Technical field
The present invention relates to the Lyapunov optimization methods that a kind of car networking cloud and edge Joint Task are dispatched, and belong to movement
Calculating field.
Background technique
Currently, the epoch of all things on earth interconnection, numerous terminal devices are more strong to the dependence of cloud computing center.Its result is exactly
Huge data volume brings lot of challenges to original cloud computing model.The mass data processing of centralization has aggravated cloud data center
Burden, lead to network congestion, high delay, low service quality.It is related to secondly, working as data caused by network edge device
And when entity and individual's privacy, security of private data problem can also become especially prominent.Therefore edge calculations are come into being.Edge
Calculate the side referred to close to object or data source header, the opening being integrated using network, calculating, storage, application core ability
Platform provides most proximal end service nearby.The purpose is to original cloud computing task portion based on center is moved to network edge
In equipment, to improve the network transmission performance of data, network bandwidth consumption is reduced, guarantees the real-time of data processing, drops simultaneously
The computational load of low cloud computing center.The end of the year 2016 by Huawei Tech Co., Ltd, Shenyang Inst of Automation, Chinese Academy of Sciences,
Chinese information Communication Studies institute, Intel company, ARM and soft logical dynamic Information technology (group) Deng Duo company, which combine, to be set up
Edge calculations alliance.Edge calculations are in cloud computing task immigration, edge video analysis, smart home, smart city, including intelligence
Also there is certain application in the fields such as signal lamp, ParkSense system, pipe detection, smart grid, secondly, in car networking, can wear
Wearing equipment, wisdom farm etc. is also to be widely used.
Compared with cloud computing, the original intention of edge calculations be in order to which computing capability to be taken back to the place closer from data source, it
One local side operation net is designed in the operation in the local environment network where data source using distributed computing architecture
Network environment can disperse operation to handle close to the proximal device in ground data source, so as to share work originally beyond the clouds
It measures, centralized processing is carried out without data all to be passed back to cloud, so that the speed for handling task becomes faster.This
Outside, close due to being separated by between the arithmetic facility of marginal end and equipment, network transmission is more direct, thus data transmit also phase
To very fast, for calculating the response speed of service also quickly.The vision driving of existing technical research Internet of Things and 5G communication
Under, transformation of the mobile computing field from centralization mobile cloud computing (MCC) to distribution mobile edge calculations (MEC), thus
Realize that computation-intensive and the crucial application of delay time and mobile edge cloud can be by mentioning in resource-constrained mobile device
The high delay of conventional central cloud is overcome for radio network information and home environment perception and low latency and bandwidth protection, is calculated
The obstacle of scarce capacity.Also have and realize mobile side using the method for game theory under technical research channel wireless radio multi interference environment
The distributed efficient calculating unloading of edge cloud computing, and research is further expanded into the multiple terminals in channel wireless radio multi contention environment
Calculate unloading scene.However, the research that these methods handle data is all relatively simple, the explosive of data traffic increases
Long, the security and privacy problem of brought network should not be underestimated, and propose high performance requirements for future network calculating.Edge meter
Grow up on the basis of the relevant technologies at last, is complementary relationship with cloud computing, but at present for edge calculations-
The research that cloud computing combines is also relatively fewer, therefore cloud-edge-terminal method for allocating tasks will be that the important of future is ground
Study carefully direction.The target of Lyapunov function is stabilizing network queue while optimizing some performance objectives, such as: Lyapunov optimization
Refer to and carry out optimal control dynamical system using Lyapunov function, Lyapunov function is widely used in control theory, to ensure
Various forms of system stability.Lyapunov drift is to study the key problem of queuing network optimum control.One typical
Target is to stablize all-network queue, while optimizing some performance objectives, such as minimizes average energy or maximizes average throughput
Amount.Lyapunov penalty can be minimized time mean power while stabilizing network.Therefore, it is managed in conjunction with Lyapunov
By edge, cloud, the task schedule between terminal is studied for improving the network transmission performances of data, guarantee data processing
Real-time, while reducing for the computational load of cloud computing center is a kind of critically important research direction.
Summary of the invention
The technical problem to be solved by the present invention is in view of the shortcomings of the prior art, provide a kind of car networking cloud and edge
The Lyapunov optimization method of Joint Task scheduling, the priority of task based access control realize cloud, marginal end, appointing between terminal
Business scheduling problem, distributes to marginal end for partial task and handles, collect it is not necessary that all tasks are all transferred to cloud
Chinese style processing, so that the speed for handling task becomes faster, it is rationally effective using cloud, the resource of marginal end, reduce and appoints
The latency issue being engaged in scheduling process.
Technical principle of the invention is, Ad hoc net is a kind of multi-hop, acentric, ad hoc deployed wireless networks, again
Referred to as multihop network (Multi-hop Network), foundation-free facility network (Infrastructureless Network) or from group
Knitmesh (Self-organizing Network).The infrastructure that whole network is not fixed, each node be it is mobile,
And it can dynamically keep contacting with other nodes in any way.In such networks, since terminal wireless covering takes
It is worth the finiteness of range, two user terminals that can not be directly communicated can be grouped forwarding by other nodes.Often
One node is simultaneously a router, they can complete the function of finding and maintain to arrive other node-routings.In Ad Hoc
In network, when two mobile hosts are in mutual communication coverage, they can be with direct communication.But it is led due to mobile
The communication coverage of machine is limited, if two hosts apart from each other will be communicated, needs through the shifting between them
The forwarding of dynamic host B is just able to achieve.Therefore in Ad Hoc network, host simultaneously or router, be responsible for searching routing and
The work to E-Packet.In Ad Hoc network, the communication range of each host is limited, therefore routes general all by multi-hop group
At data get to destination by the forwarding of multiple main frames.Therefore Ad Hoc network is also referred to as multi-hop wireless network.Ad
Hoc network can be regarded as the intersection of mobile communication and computer network.In Ad Hoc network, point of computer network is used
Group exchanging mechanism, rather than circuit switching mechanism.The host of communication is usually portable computer, personal digital assistant (PDA)
Equal mobile terminal devices.Ad Hoc network is different from the mobile IP network in mesh internet environment.It is mobile in mobile IP network
Host can access network by modes such as fixed cable network, Radio Link and dial lines, and in Ad Hoc network only
There are a kind of connection types of Radio Link.In mobile IP network, branch that mobile host passes through wired facility such as adjacent base station
Holding could communicate, and be cable network between base station and base station (agency and agency), still use the traditional routing of internet
Agreement.And Ad Hoc network is without the support of these facilities.In addition, mobile host does not have routing function in mobile IP network
Can, an only common communication terminal.Do not change network topology when mobile host is moved to another area from an area
Structure, and the movement of mobile host will will lead to the change of topological structure in Ad Hoc network.
To solve problems of the prior art, the technical solution adopted in the present invention the following steps are included:
Step 1, vehicle Ad hoc network is created, the Ad hoc network has a Cloud Server, multiple Edge Servers, and two
RSU node and N number of node on behalf vehicle, place vehicle node using the sumo traffic flow model of Traci mobility model;Its
In, N >=1;
Step 2, using the fuzzy queue scheduling based on Lyapunov as the vehicle node of task, decision distributes vehicle node
It is also assigned to marginal end to cloud to carry out data transmission, fuzzy rule is exported by fuzzy Lyapunov Theory of Stability, is obtained
To stable queue scheduling, and maximize using Lyapunov penalty the effectiveness of task schedule.
In the step 2, using mac/phy and IEEE802.11p carry out vehicle node and edge, Cloud Server it is logical
Letter.
1) the specific implementation process of the step 2 described in includes: that vehicle can generate corresponding data information in moving process, uses X
(t) indicate that the vehicle data information of aggregation, each vehicle data information in X (t) are divided into different priority, and
It is placed in corresponding queue;
2) length of each queue should be kept to prevent network congestion no more than capacity of queue during task schedule, while being also to be
The case where preventing the task of low priority from not having scheduled and queue insufficient space that vehicle data information is caused to be lost;
3) deadline and final coutoff tune is arranged in each vehicle data information during being scheduled as task
Spend the time limit;
4) energy consumption of being handled during task schedule by selection distribution for task should be in total energy consumption budget
In the range of;
5) during task schedule, the scheduling of task is carried out using Lyapunov fuzzy theory, maximizes the effect of task schedule
With while the high task of priority preferentially handled by selection scheduling;It maintains to appoint using Lyapunov candidate functions simultaneously
The stability of business scheduling, if Lyapunov candidate functions are the time-derivatives of local positive definite and Lyapunov candidate functions
It is that negative semidefinite (for function V (x), if x value in the ball that radius is Ro, and meets: 1.V (x) is scalar letter for part
Number;2.V (x) is continuous function;When 3.x=0, V (x)=0;When 4.x ≠ 0, V (x)>0 (or V (x)<0), then just claiming V (x)
It is local positive definite (negative definite).If Ro changes infinity into, two above condition still meets, then claiming V (x) is global positive definite
(negative definite).If 3 conditions are constant to go forward, the 4th condition becomes V (x) >=0 (or V (x)≤0), then V (x) is claimed to be
Locally or globally positive semidefinite (negative semidefinite).), then just illustrating that task schedule tends towards stability;And use Lyapunov
Penalty maximizes the effectiveness of task schedule.
The principle of task schedule process is preferentially to be allocated the high task of priority in queue to be transferred to cloud progress
Processing, the low task of priority are assigned to marginal end and are handled.The expression formula of Lyapunov candidate functions are as follows:Wherein Q (t) indicates that scheduling queue, virtual queue P (t) correspond to energy consumption EQ max,
The smaller state for just representing queue of the value of Lyapunov candidate functions is more stable, so ideally, Q (t) should keep bounded,
P (t) should be kept close to a constant k.As Δ L < 0, queue is in stable state.Lyapunov penalty expression formula are as follows:
minΔL(t)-v*Ut, v is selection as needed to control the control parameter weight of effectiveness, UtIt is the energy consumption of task distribution,
Control parameter v is balanced between the effectiveness and queue priority of task schedule, selects biggish v that can obtain bigger task tune
The effectiveness of degree.
Compared with prior art, the advantageous effect of present invention is that: the present invention realizes cloud, marginal end, terminal
Between task schedule, partial task is distributed into cloud and is handled, partial task is distributed to marginal end and handled, and is not necessarily to
All tasks are all transferred to cloud and carry out centralized processing, so that the speed for handling task becomes faster, are more rationally had
The resource for making full use of cloud, marginal end of effect, greatly reduces the delay in task processes.
Detailed description of the invention
Fig. 1 is cloud, marginal end, the structural framing schematic diagram between terminal.
Fig. 2 is the schematic diagram of the Lyapunov method of cloud and the scheduling of edge Joint Task.
Specific embodiment
1 to 2 pair of the preferred embodiment of the present invention is described further with reference to the accompanying drawing, and Fig. 1 is cloud, marginal end, end
Structural framing schematic diagram between end, wherein cloud represents Cloud Server, and marginal end represents Edge Server, in the present specification
Terminal device represents vehicle, and vehicle generates corresponding data information in moving process, these data informations as task come into
Row distributes to cloud or marginal end is handled, to devise a kind of car networking cloud and edge Joint Task dispatches
Lyapunov optimization method rationally effectively utilizes cloud, side for realizing cloud, marginal end, the task schedule between terminal
The resource of acies effectively reduces the delay in task processes so that the speed for handling task becomes faster.
Fig. 2 is the schematic diagram of the Lyapunov method of cloud and the scheduling of edge Joint Task.It is described as follows:
Step 1: vehicle can generate corresponding data information in moving process, indicated with X (t) aggregation vehicle number it is believed that
It ceases, each vehicle data information in X (t) is divided into different priority, and is placed in corresponding queue.Assuming that there is Q
A queue type, in these queue types comprising with the queue 1 from highest priority to the queue n's with lowest priority
Data are expressed as Q={ q1,q2,q3,...,qn}。
Step 2: should keep the length of each queue no more than capacity of queue in task scheduling process, use Si(t) carry out table
Show the queue length in i-th of queue of time slot t, SQ maxIndicate total capacity of queue, the length of each queue is less than or equal to queue
Capacity prevents network congestion, and for preventing that the task of low priority is not scheduled and queue insufficient space causes
The case where vehicle data information is lost.
Step 3: each vehicle data information can have deadline and most during being scheduled as task
The cut-off scheduling time limit afterwards, each task join the team at the beginning of for ts, the end time of team is t oute, when the completion of i-th of task
Between be expressed as Wi.The deadline date of corresponding each task is expressed as T={ T accordingly in Q queue type1,T2,...,Tn}。
If task is scheduled not yet within the deadline date, then the task is just increased its priority weight as untreated task
Newly it is put into the vehicle data information of aggregation.
The purpose of dynamic dispatching multiplexed transport is to reduce the total queue rank of all queue types of queue.Objective function table
It is shown as:
Subject to:Si(t)≤SQ max
Wi≤Ti
qi(t) it is dynamic queue's priority in i-th of queue of discrete time t.
Step 4: considering during task schedule scheduled within the final coutoff time limit and within the scope of capacity of queue
Task queue priority carry out task scheduling.Principle is preferentially to be allocated the high task of priority to be transferred to cloud
It is handled, the low task of priority is assigned to marginal end and is handled.Also it can be related to energy consumption during task schedule
The problem of.If task i carries out processing by selection distribution and is expressed as xi=1, then unselected distribution carries out processing and is expressed as xi=
0, task i are expressed as E by the energy consumption that selection distribution is handledi, EQ maxTotal energy consumption budget is indicated, by selection point
Energy consumption with being handled for task should be within total energy consumption budget space;
The purpose of task schedule is the queue type q for minimizing equation (1) and selecting to have maximum queue prioritys, it may be assumed that
The effectiveness of task schedule is maximized in the range of the budget of energy consumption simultaneously, the effectiveness of time slot t task i is expressed as Ui
(t), it may be assumed that
Subject to:
piFor the energy expenditure rate in task assignment procedure, deadline and energy of the energy consumption of task distribution by task
Consumption rate codetermines.If exceeding total energy consumption budget by the energy consumption for the task that selection distribution is handled, that
The task is also reduced the distribution of its energy consumption as untreated task and is reentered into the vehicle data information of aggregation;
In the case that queue length, deadline and energy consumption constraint all meet during task choosing, task is considered
The optimization of priority and effectiveness in scheduling process carries out the scheduling of task.
Step 5: during task schedule, the scheduling of task is carried out using Lyapunov fuzzy theory, because of fuzzy reason
By that can provide simple and effective task schedule, the high task of priority is preferential while maximizing the effectiveness of task schedule
It is handled by selection scheduling.In Q queue type, if q1It is bigger, q2,q3,...,qnIt is relatively all smaller, then
Queue type q with maximum queue prioritysSelection be exactly q1, indicate q1Priority ratio it is larger, q1It is preferred tune
Degree is handled.If doing so a problem being related to is exactly to be difficult to ensure the stability of queue and maximize simultaneously to appoint
The effectiveness of business scheduling, referring to fig. 2, it is therefore desirable to it selects to obscure stability of the Lyapunov comprehensive theory to maintain task schedule,
The effectiveness of task schedule is maximized using Lyapunov penalty;
Lyapunov candidate functions are used to maintain the stability of task schedule during task schedule, if Lyapunov is candidate
Function is that the time-derivative of local positive definite and Lyapunov candidate functions is that part is negative semidefinite, then just illustrating task tune
Degree tends towards stability.Define two queues: one is scheduling queue Q (t), and one is virtual queue P (t), and P (t) corresponds to
Energy consumption EQ max, the smaller state for just representing queue of the value of Lyapunov candidate functions is more stable, so ideally, Q
(t) bounded should be kept, P (t) should be kept close to a constant, and defining this constant is k.In order to minimize (1) and stablize queue,
Selected Lyapunov candidate functions are as follows:
Lyapunov drift is the core of Optimum Control Study in queuing network, and target is to stablize all-network queue.
Lyapunov drift is change of the Lyapunov function from a time slot to next time slot, and Δ (L (t)) indicates to minimize drift
It moves, indicates are as follows:
Δ (L (t))=E [L (t+1)-L (t)] (5)
Update over time is overstock in the queue of scheduling queue Q (t) to be indicated are as follows:
Q (t+1)=[Q (t)-Xs(t)+Xe(t)]+ (6)
Wherein XsIt (t) is being scheduled in queue for task, XeIt (t) is newly-increased vehicle node information.
The update of the energy consumption of virtual queue P (t) over time indicates are as follows:
P (t+1)=[P (t)-Xc(t)+e(t)]+ (7)
XcIt (t) is energy consumed by scheduled task, e (t) is according to being increased the case where energy consumption during task schedule
The energy budget added.
The purpose for minimizing drift delta (L (t)) is to ensure that Q (t) keeps stablizing, and P (t) is kept at any time close to constant k.Team
It is dispatched in column and newly-increased vehicle node information is all bounded, there are constant B > 0 to make for all t satisfactions:
Δ(L(t))≤B-E[Q(t)Xc(t)+(P(t)-k)Xc(t)] (8)
XcIt (t) is energy consumed by scheduled task.As Δ (L (t)) < 0, queue is in stable state.
Secondly, the effectiveness of task schedule is maximized during task schedule using Lyapunov penalty.It is in queue
Under stable state, maximizing (3) can be represented as:
minΔL(t)-v*Ut (9)
Wherein v is selection as needed to control the control parameter weight of effectiveness, UtIt is the energy consumption of task distribution, control ginseng
Number v is balanced between the effectiveness and queue priority of task schedule, selects biggish v that can obtain the effect of bigger task schedule
With.In (9) middle application expression formula (8) and the maximization for replacing with objective function and negating will be minimized, under string stability state
Optimized:
maxE[Q(t)Xs(t)+(P(t)-k)Xc(t)]+v*Ut (10)
Wherein XsIt (t) is being scheduled in queue for task, XcIt (t) is energy consumed by scheduled task.That is Δ (L
(t)) when < 0, queue is in stable state, while in the case where selecting biggish v, obtains the effectiveness of bigger task schedule,
Task priority scheduling is selected to carry out the processing of task, in the scheduling of every round, team from scheduling queue in round scheduling
The state and energy budget of column also will do it update, and the scheduling of task is repeated with this.
The present invention considers task priority, queue size, deadline, deadline, energy consumption and task schedule
Effectiveness these factor combination Lyapunov methods devise a kind of cloud, marginal end, the method for scheduling task between terminal, make
Stablize scheduling queue with Lyapunov candidate functions, and maximize the effectiveness of task schedule using Lyapunov penalty,
Make the task with higher priority preferentially be assigned to cloud to be handled, the task of lower priority is assigned to marginal end
Handled, so as to avoid in task processes task concentrate on marginal end or the case where cloud handled, nothing
Method efficiently uses the resource in cloud, marginal end, so that the speed for handling task becomes faster, it is processed to substantially reduce task
Delay in journey.
Claims (6)
1. the Lyapunov optimization method in a kind of car networking cloud and the scheduling of edge Joint Task, which is characterized in that including following
Step:
Step 1, creation vehicle Ad hoc network, the Ad hoc network include that a Cloud Server and at least two edges take
Business device;Two RSU nodes and N number of node on behalf vehicle, place vehicle using the sumo traffic flow model of Traci mobility model
Node;Wherein, N >=1;
Step 2, using the fuzzy queue scheduling based on Lyapunov as the vehicle node of task, decision distributes vehicle node
It to cloud or is assigned to marginal end and carries out data transmission, fuzzy rule is exported by fuzzy Lyapunov Theory of Stability, is obtained
To stable queue scheduling, and maximize using Lyapunov penalty the effectiveness of task schedule.
2. the Lyapunov optimization method in a kind of car networking cloud according to claim 1 and the scheduling of edge Joint Task,
It is characterized in that, the specific implementation process in step 2 includes:
Step 2-1, indicate that the vehicle data information of aggregation, each vehicle data information in X (t) are divided into difference with X (t)
Priority, and be placed in corresponding queue;It is excellent from highest comprising having in these queue types equipped with Q queue type
The data of the queue n with lowest priority are arrived in the queue 1 of first grade, are expressed as Q={ q1,q2,q3,...,qn};
Step 2-2, the length of each queue should be kept to prevent network congestion no more than capacity of queue during task schedule, simultaneously
Also for prevent that the task of low priority is not scheduled and the case where queue insufficient space causes vehicle data information to be lost
Occur;Use Si(t) queue length in i-th of queue of time slot t, S are indicatedQ maxIndicate total capacity of queue, the length of each queue
Degree is less than or equal to capacity of queue to prevent network congestion;
Step 2-3, deadline and last is arranged in each vehicle data information during being scheduled as task
The cut-off scheduling time limit;Each task is t at the beginning of joining the teams, the end time of team is t oute, the deadline of i-th of task
It is expressed as Wi, the deadline date of corresponding each task is expressed as T={ T accordingly in Q queue type1,T2,...,Tn};If
If task is scheduled not yet within the deadline date, then the task is just increased its priority again as untreated task
It is put into the vehicle data information of aggregation;
The purpose of dynamic dispatching multiplexed transport is to reduce the total queue rank of all queue types of queue, and objective function indicates
Are as follows:
Subject to:Si(t)≤SQ max
Wi≤Ti
qi(t) it is dynamic queue's priority in i-th of queue of discrete time t;
Step 2-4, should be disappeared in total energy during task schedule by the energy consumption for the task that selection distribution is handled
It consumes in the range of budget;If task i carries out processing by selection distribution and is expressed as xi=1, then unselected distribution is handled
It is expressed as xi=0, task i are expressed as E by the energy consumption that selection distribution is handledi, EQ maxIndicate that total energy consumption is pre-
It calculates, the purpose of task schedule is the queue type q that selection has maximum queue prioritys, it may be assumed that
The effectiveness of task schedule is maximized in the range of the budget of energy consumption simultaneously, the effectiveness of time slot t task i is expressed as Ui
(t), it may be assumed that
Ei≤EQ maxi∈(1,n)
piFor the energy expenditure rate in task assignment procedure, deadline and energy of the energy consumption of task distribution by task
Consumption rate codetermines.If exceeding total energy consumption budget by the energy consumption for the task that selection distribution is handled, that
The task is also reduced the distribution of its energy consumption as untreated task and is reentered into the vehicle data information of aggregation;
Step 2-5, during task schedule, the scheduling of task is carried out using Lyapunov fuzzy theory, maximizes task tune
The high task of priority is preferentially handled by selection scheduling while the effectiveness of degree;Simultaneously using Lyapunov candidate functions come
The stability for maintaining task schedule, in Q queue type, if q1It is bigger, q2,q3,...,qnIt is relatively all smaller, that
Queue type q with maximum queue prioritysSelection be exactly q1, indicate q1Priority ratio it is larger, q1It is preferred
Scheduling is handled;If the time-derivative that Lyapunov candidate functions are local positive definite and Lyapunov candidate functions is
Part is negative semidefinite, then just illustrating that task schedule tends towards stability;And it is maximized using Lyapunov penalty
The effectiveness of task schedule.
3. the Lyapunov optimization method in a kind of car networking cloud according to claim 2 and the scheduling of edge Joint Task,
It is characterized in that, the principle of task schedule process be the high task of priority in queue is preferentially allocated be transferred to cloud into
Row processing, the low task of priority are assigned to marginal end and are handled.
4. the optimization side Lyapunov in a kind of car networking cloud according to claim 1 or 2 and the scheduling of edge Joint Task
Method, which is characterized in that
Two queues: scheduling queue Q (t), virtual queue P (t) are defined in step 2-5, P (t) corresponds to energy consumption EQ max,
The smaller state for just representing queue of the value of Lyapunov candidate functions is more stable, so ideally, Q (t) should keep bounded,
P (t) should be kept close to a constant, if the constant is k;
The expression formula of the Lyapunov candidate functions are as follows:
Lyapunov drift is the core of Optimum Control Study in queuing network, and target is to stablize all-network queue.
Lyapunov drift is defined as change of the Lyapunov function from a time slot to next time slot, and Δ (L (t)) indicates most
Smallization drift, indicates are as follows:
Δ (L (t))=E [L (t+1)-L (t)]
Update over time is overstock in the queue of scheduling queue Q (t) to be indicated are as follows:
Q (t+1)=[Q (t)-Xs(t)+Xe(t)]+
Wherein XsIt (t) is being scheduled in queue for task, XeIt (t) is newly-increased vehicle node information;
The update of the energy consumption of virtual queue P (t) over time indicates are as follows:
P (t+1)=[P (t)-Xc(t)+e(t)]+
XcIt (t) is energy consumed by scheduled task, e (t) is according to being increased the case where energy consumption during task schedule
Energy budget;
Minimize driftΔThe purpose of (L (t)) is to ensure that Q (t) keeps stablizing, and P (t) is kept at any time close to constant k, in queue
Scheduling and newly-increased vehicle node information are all bounded, and there are constant B > 0 to make for all t satisfactions:
Δ(L(t))≤B-E[Q(t)Xc(t)+(P(t)-k)Xc(t)]
XcIt (t) is energy consumed by scheduled task, as Δ (L (t)) < 0, queue is in stable state.
5. the optimization side Lyapunov in a kind of car networking cloud according to claim 1 or 2 and the scheduling of edge Joint Task
Method, which is characterized in that
In step 2-5, Lyapunov penalty expression formula are as follows:
minΔL(t)-v*Ut
Optimized under string stability state:
maxE[Q(t)Xs(t)+(P(t)-k)Xc(t)]+v*Ut
Wherein XsIt (t) is being scheduled in queue for task, XcIt (t) is energy consumed by scheduled task, v is basis
It needs to select to control the control parameter weight of effectiveness, UtIt is the energy consumption of task distribution, control parameter v is in task schedule
It is balanced between effectiveness and queue priority, selects biggish v that can obtain the effectiveness of bigger task schedule.
6. the Lyapunov optimization method in a kind of car networking cloud according to claim 1 and the scheduling of edge Joint Task,
It is characterized in that, carrying out vehicle node and Cloud Server or edge service using mac/phy and IEEE802.11p in step 2
The communication of device.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910298270.8A CN110198278B (en) | 2019-04-15 | 2019-04-15 | Lyapunov optimization method for vehicle networking cloud and edge joint task scheduling |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910298270.8A CN110198278B (en) | 2019-04-15 | 2019-04-15 | Lyapunov optimization method for vehicle networking cloud and edge joint task scheduling |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110198278A true CN110198278A (en) | 2019-09-03 |
CN110198278B CN110198278B (en) | 2022-05-10 |
Family
ID=67751950
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910298270.8A Active CN110198278B (en) | 2019-04-15 | 2019-04-15 | Lyapunov optimization method for vehicle networking cloud and edge joint task scheduling |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110198278B (en) |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110647398A (en) * | 2019-09-18 | 2020-01-03 | 北京航空航天大学 | Intersection control task scheduling method facing edge calculation and based on task criticality and timeliness |
CN110662221A (en) * | 2019-09-10 | 2020-01-07 | 杭州电子科技大学 | Resource allocation method for security and performance perception of enterprise multimedia in MEC |
CN111311091A (en) * | 2020-02-13 | 2020-06-19 | 中国人民解放军国防科技大学 | Expressway task detection and scheduling method and system based on vehicle-mounted cloud and unmanned aerial vehicle |
CN111538571A (en) * | 2020-03-20 | 2020-08-14 | 重庆特斯联智慧科技股份有限公司 | Method and system for scheduling task of edge computing node of artificial intelligence Internet of things |
CN111538595A (en) * | 2020-04-23 | 2020-08-14 | 电子科技大学中山学院 | MEC server task scheduling method based on task cutting |
CN111586568A (en) * | 2020-05-26 | 2020-08-25 | 多伦科技股份有限公司 | Vehicle-mounted communication emergency message transmission optimization method based on edge calculation |
CN111597043A (en) * | 2020-05-14 | 2020-08-28 | 行星算力(深圳)科技有限公司 | Method, device and system for calculating edge of whole scene |
CN111597025A (en) * | 2020-05-14 | 2020-08-28 | 行星算力(深圳)科技有限公司 | Edge calculation scheduling algorithm and system |
CN111866811A (en) * | 2020-08-07 | 2020-10-30 | 中国联合网络通信集团有限公司 | Information transmission method and system |
CN112333659A (en) * | 2020-10-13 | 2021-02-05 | 联通智网科技有限公司 | Internet of vehicles-based over-the-air downloading method, system, equipment and medium |
CN112434789A (en) * | 2020-11-06 | 2021-03-02 | 中山大学 | Distributed neural network model partitioning method for edge video analysis |
CN112650946A (en) * | 2020-12-30 | 2021-04-13 | 中国联合网络通信集团有限公司 | Product information recommendation method, device and system and storage medium |
CN112882809A (en) * | 2021-02-23 | 2021-06-01 | 国汽(北京)智能网联汽车研究院有限公司 | Method and device for determining computing terminal of driving task and computer equipment |
CN113051046A (en) * | 2020-12-09 | 2021-06-29 | 南京航空航天大学 | Task scheduling method based on dynamic priority under vehicle-mounted cloud environment |
CN113783798A (en) * | 2021-09-24 | 2021-12-10 | 上海明胜品智人工智能科技有限公司 | Data transmission method and system and edge service equipment |
WO2023044269A1 (en) * | 2021-09-17 | 2023-03-23 | Harman International Industries, Incorporated | Methods and systems for vehicle communication |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170164237A1 (en) * | 2015-12-03 | 2017-06-08 | The Trustees Of The Stevens Institute Of Technology | System Apparatus And Methods For Cognitive Cloud Offloading In A Multi-Rat Enabled Wireless Device |
CN108809695A (en) * | 2018-04-28 | 2018-11-13 | 国网浙江省电力有限公司电力科学研究院 | A kind of distribution uplink unloading strategy towards mobile edge calculations |
CN109343904A (en) * | 2018-09-28 | 2019-02-15 | 燕山大学 | A kind of mist calculating dynamic offloading method based on Lyapunov optimization |
CN109591804A (en) * | 2018-11-22 | 2019-04-09 | 湖南大学 | Consider the vehicle platoon stability control method of communication delay |
-
2019
- 2019-04-15 CN CN201910298270.8A patent/CN110198278B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170164237A1 (en) * | 2015-12-03 | 2017-06-08 | The Trustees Of The Stevens Institute Of Technology | System Apparatus And Methods For Cognitive Cloud Offloading In A Multi-Rat Enabled Wireless Device |
CN108809695A (en) * | 2018-04-28 | 2018-11-13 | 国网浙江省电力有限公司电力科学研究院 | A kind of distribution uplink unloading strategy towards mobile edge calculations |
CN109343904A (en) * | 2018-09-28 | 2019-02-15 | 燕山大学 | A kind of mist calculating dynamic offloading method based on Lyapunov optimization |
CN109591804A (en) * | 2018-11-22 | 2019-04-09 | 湖南大学 | Consider the vehicle platoon stability control method of communication delay |
Non-Patent Citations (2)
Title |
---|
XINHOU WANG ETAL: "Dynamic Resource Scheduling in Mobile Edge Cloud with Cloud Radio Access Network", 《IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS》 * |
于博文等: "移动边缘计算任务卸载和基站关联协同决策问题研究", 《计算机研究与发展》 * |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110662221A (en) * | 2019-09-10 | 2020-01-07 | 杭州电子科技大学 | Resource allocation method for security and performance perception of enterprise multimedia in MEC |
CN110647398A (en) * | 2019-09-18 | 2020-01-03 | 北京航空航天大学 | Intersection control task scheduling method facing edge calculation and based on task criticality and timeliness |
CN111311091A (en) * | 2020-02-13 | 2020-06-19 | 中国人民解放军国防科技大学 | Expressway task detection and scheduling method and system based on vehicle-mounted cloud and unmanned aerial vehicle |
CN111311091B (en) * | 2020-02-13 | 2023-05-02 | 中国人民解放军国防科技大学 | Expressway task detection and scheduling method and system based on vehicle-mounted cloud and unmanned aerial vehicle |
CN111538571A (en) * | 2020-03-20 | 2020-08-14 | 重庆特斯联智慧科技股份有限公司 | Method and system for scheduling task of edge computing node of artificial intelligence Internet of things |
CN111538595A (en) * | 2020-04-23 | 2020-08-14 | 电子科技大学中山学院 | MEC server task scheduling method based on task cutting |
CN111538595B (en) * | 2020-04-23 | 2023-05-02 | 电子科技大学中山学院 | MEC server task scheduling method based on task cutting |
CN111597043A (en) * | 2020-05-14 | 2020-08-28 | 行星算力(深圳)科技有限公司 | Method, device and system for calculating edge of whole scene |
CN111597025A (en) * | 2020-05-14 | 2020-08-28 | 行星算力(深圳)科技有限公司 | Edge calculation scheduling algorithm and system |
CN111597025B (en) * | 2020-05-14 | 2024-02-09 | 行星算力(深圳)科技有限公司 | Edge calculation scheduling algorithm and system |
CN111586568A (en) * | 2020-05-26 | 2020-08-25 | 多伦科技股份有限公司 | Vehicle-mounted communication emergency message transmission optimization method based on edge calculation |
CN111866811A (en) * | 2020-08-07 | 2020-10-30 | 中国联合网络通信集团有限公司 | Information transmission method and system |
CN111866811B (en) * | 2020-08-07 | 2023-04-07 | 中国联合网络通信集团有限公司 | Information transmission method and system |
CN112333659A (en) * | 2020-10-13 | 2021-02-05 | 联通智网科技有限公司 | Internet of vehicles-based over-the-air downloading method, system, equipment and medium |
CN112434789A (en) * | 2020-11-06 | 2021-03-02 | 中山大学 | Distributed neural network model partitioning method for edge video analysis |
CN112434789B (en) * | 2020-11-06 | 2023-07-07 | 中山大学 | Distributed neural network model partitioning method for edge video analysis |
CN113051046A (en) * | 2020-12-09 | 2021-06-29 | 南京航空航天大学 | Task scheduling method based on dynamic priority under vehicle-mounted cloud environment |
CN112650946B (en) * | 2020-12-30 | 2023-06-09 | 中国联合网络通信集团有限公司 | Product information recommendation method, device, system and storage medium |
CN112650946A (en) * | 2020-12-30 | 2021-04-13 | 中国联合网络通信集团有限公司 | Product information recommendation method, device and system and storage medium |
CN112882809A (en) * | 2021-02-23 | 2021-06-01 | 国汽(北京)智能网联汽车研究院有限公司 | Method and device for determining computing terminal of driving task and computer equipment |
WO2023044269A1 (en) * | 2021-09-17 | 2023-03-23 | Harman International Industries, Incorporated | Methods and systems for vehicle communication |
CN113783798A (en) * | 2021-09-24 | 2021-12-10 | 上海明胜品智人工智能科技有限公司 | Data transmission method and system and edge service equipment |
Also Published As
Publication number | Publication date |
---|---|
CN110198278B (en) | 2022-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110198278A (en) | A kind of Lyapunov optimization method in car networking cloud and the scheduling of edge Joint Task | |
Mao et al. | Energy efficiency and delay tradeoff for wireless powered mobile-edge computing systems with multi-access schemes | |
Tabibi et al. | Energy-efficient routing mechanism for mobile sink in wireless sensor networks using particle swarm optimization algorithm | |
Gu et al. | Task offloading in vehicular mobile edge computing: A matching-theoretic framework | |
CN111930436B (en) | Random task queuing unloading optimization method based on edge calculation | |
Liu et al. | An agent-assisted QoS-based routing algorithm for wireless sensor networks | |
Bozorgchenani et al. | Centralized and distributed architectures for energy and delay efficient fog network-based edge computing services | |
CN112600921B (en) | Heterogeneous mobile edge network-oriented dynamic task unloading method | |
US10009287B2 (en) | Hierarchical software-defined network traffic engineering controller | |
CN103369599B (en) | A kind of many radio frequencies multi-Channel Wireless Mesh Network resource cross-layer optimizing method | |
CN105224393B (en) | A kind of scheduling virtual machine mechanism of JT-CoMP under C-RAN frameworks | |
Tham et al. | A load balancing scheme for sensing and analytics on a mobile edge computing network | |
Mirzaei et al. | Towards optimal configuration in MEC Neural networks: deep learning-based optimal resource allocation | |
CN103888976A (en) | Link selection method combined with network scheduling and routing | |
CN105636062A (en) | Cognitive radio network transmission learning method for moderate business services | |
Wang et al. | Distributed clustering-based cooperative vehicular edge computing for real-time offloading requests | |
Zhao et al. | Online distributed optimization for energy-efficient computation offloading in air-ground integrated networks | |
Cai et al. | Mobile edge computing network control: Tradeoff between delay and cost | |
Liu et al. | Genetic algorithm for delay efficient computation offloading in dispersed computing | |
Selvakumar et al. | Escalating quality of services with channel assignment and traffic scheduling in wireless mesh networks | |
CN103618674B (en) | A united packet scheduling and channel allocation routing method based on an adaptive service model | |
CN112235387B (en) | Multi-node cooperative computing unloading method based on energy consumption minimization | |
Zhu et al. | Efficient hybrid multicast approach in wireless data center network | |
Yin et al. | A cooperative edge computing scheme for reducing the cost of transferring big data in 5G networks | |
Yu et al. | Coordinated parallel resource allocation for integrated access and backhaul 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 |