CN114095358B - Dynamic network-oriented virtual network function deployment method and device - Google Patents
Dynamic network-oriented virtual network function deployment method and device Download PDFInfo
- Publication number
- CN114095358B CN114095358B CN202111402460.3A CN202111402460A CN114095358B CN 114095358 B CN114095358 B CN 114095358B CN 202111402460 A CN202111402460 A CN 202111402460A CN 114095358 B CN114095358 B CN 114095358B
- Authority
- CN
- China
- Prior art keywords
- deployment
- virtual network
- network function
- base station
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- H04L41/0893—Assignment of logical groups to network elements
-
- 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
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a 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
- 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
-
- 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 a dynamic network-oriented virtual network function deployment method and device. The method comprises the following steps: and (3) establishing a model according with a real situation according to the required virtual network function and the time delay requirement of a vehicle terminal user facing to a mobile vehicle-mounted network, judging when the optimal virtual network function deployment scheme needs to be calculated by utilizing an optimal time-stopping theory and finishing scheduling optimization. The method solves the problem of optimization of virtual network deployment scheduling in a dynamic network environment, comprehensively considers the balance between delay violation and deployment scheduling, realizes minimized migration cost and delay violation, and maximizes the performance of the vehicle-mounted cloud system.
Description
Technical Field
The invention relates to a virtual network function deployment and scheduling technology, in particular to a virtual network function deployment method and equipment for a dynamic network.
Background
The vehicle-mounted cloud is that computing resources on a mobile intelligent vehicle are utilized and integrated to provide services for all users in a network. The mobile intelligent vehicle, the roadside unit and the central base station jointly form a vehicle-mounted cloud system. The vehicle and equipment on the vehicle are used as user terminals, roadside units close to the terminals and a central base station are used as edges, a core content storage and service providing platform is used as a data center, and a service transmission path from the cloud to the edges and then to the terminals is constructed together. In the structure, the mobile intelligent vehicle is not only a server under the management of the cloud platform, but also can be a user terminal for providing a service request, the service request provided by the mobile intelligent vehicle is sent to a roadside unit or a central base station with computing resources, and the result is sent back to the vehicle after the execution is finished.
With the rapid increase of the number of vehicles connected to the internet of vehicles and the continuous emergence of high-bandwidth and high-quality task types, the service quality of vehicle-mounted cloud mobile tasks faces huge challenges, and the contradiction between supply and demand of available resources of the network is more prominent. To solve these problems, network function virtualization technology is widely used. The main idea of network function virtualization is to separate the network function from the physical device it relies on in the past, so that the network function traditionally coupled to the dedicated network hardware device is mapped to the large capacity server, switch, memory, thereby avoiding a large amount of new hardware from being purchased and installed, and greatly saving resources. Due to the inherent structure of cloud traffic, the data bandwidth consumed by the cloud data link is larger the closer the cloud data link is, so that the bandwidth consumption of the whole system can be reduced to the maximum extent by deploying the virtual network function at the edge of the path. In addition, the consumed computing resources are huge, so that the deployment of the virtual network function at the roadside nodes faces a great computing bottleneck. Therefore, it is mainstream to research and deploy the virtual network function in the central base station.
However, vehicles are naturally dynamic, and when a vehicle moves from one central base station to another, a problem occurs in that virtual network functions are not serviced in time, resulting in service delays. Therefore, it is necessary to provide a method for deploying virtual network functions for a dynamic network.
Disclosure of Invention
The invention aims to provide a virtual network function deployment method and equipment for a dynamic network, which are used for realizing VNF deployment in a dynamic vehicle-mounted cloud scene, effectively reducing service delay caused by vehicle movement, minimizing migration cost and delay violation and improving service quality.
In order to achieve the purpose of the invention, the invention adopts the following technical scheme:
in a first aspect, a dynamic network-oriented virtual network function deployment method is provided, which is applied to a vehicle-mounted cloud system, where the vehicle-mounted cloud system includes | V | mobile smart vehicles and | S | central base stations, and the mobile smart vehicles are connected with the central base stations via wireless communication, and the method includes:
acquiring a network function set N required by a mobile intelligent vehicle, wherein the network function N is i The calculation resource R needed by the N i Network function n i Maximum tolerated delay theta of i Computing resources C owned by the central base station s S Network bandwidth b owned by network connection e between central base stations e ；
Performing initial deployment of virtual network functions based on a pre-constructed virtual network deployment model and constraint conditions of the virtual network deployment model;
acquiring the number L (t) of virtual network functions violating the delay requirement at the time t, and calculating a delay violating sequence L from the time 0 0 ,L 1 ,…,L t ；
Calculating deployment at time t I t Optimal deployment I migrating to time t t′ The migration cost to be spent is expected to be E [ M ]]；
According to the delay violation sequence L starting from time 0 0 ,L 1 ,…,L t And migration cost expectation E [ M]The known calculation cost expectation C converts the virtual network function dynamic migration problem into an optimal time-out problem;
and solving the optimal time-out problem by utilizing a antecedent rule method, determining the time when the optimal virtual network function deployment scheme needs to be calculated, and finishing scheduling optimization.
Further, the pre-constructed virtual network deployment model is represented as:
wherein | N-]Is integer set {1,2, …, | N | }, [ | S | N | } |)]Is an integer set {1,2, …, | S | }, [ | V |]Is an integer set {1,2, …, | V | }; x ijk Is a mobile intelligent vehicle v k Required network function n i Whether or not to be deployed at a central base station s j The two-dimensional decision variables of (1); l. the ijk Indicating for deployment at a central base station s j Network function n of i Moving intelligent vehicle v k A perceived delay;
the constraint conditions of the virtual network deployment model are as follows:
formula (1) represents a mobile intelligent vehicle v k Required virtual network function n i Deployed on a central base station s;
formula (2) shows that for any central base station node S belonging to S, the deployment scheme meets the requirement of computing resources;
equation (3) indicates that for any link connection E, the deployment scenario satisfies the bandwidth resource constraint, where the network path P = { P = { P = 1 ,p 2 ,…,p |P| The method comprises the following steps that (1) a path from a center base station to a mobile intelligent vehicle is defined, and | P | is the number of the paths;
equation (4) represents each network function n i All having a service provider level agreement delay theta i ；
Equation (5) represents network functions that are not required for a mobile smart vehicle, not requiring deployment, where F k Indicating a vehicle v k A set of required virtual network functions.
Further, the migration cost expectation E [ M ] is calculated as:
m (t, t') denotes deployment I from time t t Optimal deployment I migrating to time t t′ The cost of the migration that needs to be spent,
where I (x, x') is an indicator function, which is defined as follows: />
Further, the optimal stopping time problem form for representing the virtual network function dynamic migration problem is as follows:
wherein f (L) t ) Is the reward function:
λ is a coefficient greater than 0.
Further, the antecedent rule method employs a 1-order forward rule:
In a second aspect, a virtual network function deployment device for a dynamic network is provided, which is applied to an in-vehicle cloud system, where the in-vehicle cloud system includes | V | mobile smart vehicles and | S | central base stations, and the mobile smart vehicles are connected with the central base stations via wireless communication, and the device includes:
a data acquisition module for acquiring the network function set N required by the mobile intelligent vehicle, the network function N i The calculation resource R needed by the N i Network function n i Maximum tolerated delay theta of i Computing resources C owned by the central base station s S Network bandwidth b owned by network connection e between central base stations e ；
The initial deployment module is used for performing initial deployment of virtual network functions based on a pre-constructed virtual network deployment model and constraint conditions of the virtual network deployment model;
a delay violation recording module for acquiring the number L (t) of virtual network functions violating the delay requirement at time t and calculating the delay violation sequence L from time 0 0 ,L 1 ,…,L t ；
An auxiliary computing module for computing deployment I at time t t Migration to optimal deployment at time s I s The migration cost to be spent is expected to be E [ M ]]；
An optimal stall problem establishment module for establishing an optimal stall problem based on the delay violation sequence L starting from time 0 0 ,L 1 ,…,L t Migration cost expectation E [ M]The known calculation cost expectation C converts the virtual network function dynamic migration problem into an optimal time-out problem;
and the problem solving module is used for solving the optimal time-out problem by utilizing the former rule method, determining the time when the optimal virtual network function deployment scheme needs to be calculated and finishing scheduling optimization.
In a third aspect, a computing device is presented, the device comprising:
one or more processors; a memory; and one or more programs stored in the memory and configured to be executed by the one or more processors, the programs, when executed by the processors, implementing the virtual network function deployment method of the first aspect of the invention.
Compared with the prior art, the invention has the following beneficial effects: the VNF deployment scheduling scheme based on the optimal time-parking theory in the dynamic vehicle-mounted cloud scene can effectively reduce service delay caused by vehicle movement, minimize migration cost and delay violation and improve service quality.
Drawings
FIG. 1 is a schematic diagram illustrating a virtual network function service process provided by an embodiment of the present invention;
fig. 2 is a flowchart of a virtual network function deployment method according to an embodiment of the present invention.
Detailed Description
For clarity of description of the technical solution of the present invention, the problem to be solved by the present invention is first formalized and described.
A central base station generally implements scheduling, management, and access control of Network traffic by using a plurality of Network technologies including Network Function Virtualization (NFV), and provides a computing service for a plurality of mobile end devices by creating a plurality of virtual machines in parallel, where a service process is shown in fig. 1. The central base station and the roadside units form a service network, the virtual network function is deployed on the central base station, and the service is transmitted to the vehicle terminal through the central base station, the roadside units and the multi-hop network of the intelligent vehicle. Therefore, one key issue in implementing an on-board cloud is Virtual Network Function (VNF) deployment and scheduling. The VNF deployment is mainly aimed at fully utilizing node computation and bandwidth resources of a virtual network to provide reliable network services for users.
First, assuming that the environment is static, the Virtual Network Function Deployment Problem (DepVeC) in this scenario can be described as follows.
Inputting a question: network topology G = (S, E, V) where S = { S = { S 1 ,s 2 ,…,s |S| The central base station is used as the position of the central base station, | S | is the number of the central base stations; e = { E = { E) 1 ,e 2 ,…,e |E| The physical network connection is the physical connection between the central base station and the base station, wherein | E | is the number of the connections; v = { V = 1 ,v 2 ,…,v |V| And | V | is the number of the vehicle terminals. The computing resources (CPU resources, bandwidth resources, IO resources) owned by the base station s are C S The network bandwidth of the network connection e is b e Network delay of l e . Set of network functions required by the terminal N = { N = } 1 ,n 2 ,…,n |N| }. For network function n i E.n, with the required computational resource R i For the user of the network function, the maximum tolerated delay is θ i 。
Output of the question: decision variable X ijk Indicates the terminal vehicle v k Required virtual network function n i Whether or not to be deployed at a central base station s j The upper part, namely:
optimization objective of the problem: minimizing end-to-end delay, for deployments at central base station s j Network function n of i Terminal vehicle v k The perceived delay is l ijk . Formalized description optimization objectives are:
where [ | N | ] is an integer set {1,2, …, | N | }, [ | S | ] and [ | V | ] are similar.
The constraint conditions are as follows:
equation (1) represents the end vehicle v k Required virtual network function n i Must be deployed at some central base station s.
Equation (2) indicates that for any central base station node S e S, the deployment scenario should meet the computational resource requirements.
Equation (3) indicates that for any link connection E, the deployment scenario should meet the bandwidth resource constraint, where the network path P = { P = ∈ E 1 ,p 2 ,…,p |P| And P is the number of paths. The paths being formed by multi-hop connections, for a path p, a network function n i The required link bandwidth is b ijk Wherein the first hop of the path is s i Last hop is vehicle v k 。
Equation (4) represents each network function n i All having a service provider level agreement delay theta i 。
Equation (5) does not require deployment for network functions that are not required by the end vehicle. Wherein F k Indicating vehicle v k A set of required virtual network functions.
The above DepVeC problem is easily analyzed as an ILP problem (integer linear programming), is NP-hard, and has an exponential linear programming solution.
The deployment of virtual network functions in a dynamic environment is described next. Under the dynamic environment, due to the influence of the mobility of the terminal vehicle, environmental factors of the network condition along with weather, air humidity and the like, vehicle factors of the vehicle speed, the position and the like, and transient factors of transient network occupation, transient server breakdown and the like, the delay matrix l is influenced ijk In other words, the true value thereof changes with time. Therefore, a time-varying delay matrix is introduced
Then the optimal deployment at time t satisfies:
in a dynamic environment, over time, at t 0 Optimal deployment of time of day I 0 At t of 1 The time of day is likely to be less than optimal. Thus at t 1 The optimal deployment I needs to be recalculated once at any moment 1 . But consumes a large amount of computing resources if it is calculated frequently. Thus, the scheduler can typically tolerate some deviation of the total delay from the optimal solution in exchange for some "ease" without computation and migration. That is, the optimal solution calculation is performed only if a certain condition is satisfied. This condition can be described by migration cost and delay violation.
Migration cost: the scheduler needs to minimize the migration cost per schedule. Deployment at time t I t Optimal deployment I migrating to time t t' The migration cost required to be spent is:
where I (x, x') is an indicator function, which is defined as follows:
for each scheduling, the cost of scheduling is expected to be:
wherein
Refers to the probability that a certain virtual network function needs to be migrated. The probability may be a priori or a posteriori depending on the system settings. It is considered herein that this probability can be obtained a priori.
Delay violation: during the balancing of migration and total delay, it is difficult to avoid that some network functions exceed SLA (Service-Level agent) constraints. In practical scenarios, the delay SLA of a real-time packet processing VNF is often within 10 ms. For those virtual network functions that exceed the delay constraint,defining an indicator variable
For the constraint condition satisfied by the virtual network function at time t:
this definition means that such a deployment strategy is considered unnecessary to schedule if only a few virtual network functions fail to meet the delay requirement and do not deviate from the delay for a long time. This is a common method to trade some constraint for system stability.
On this basis, the number of virtual network functions violating the delay requirement at time t is:
definition of the sequence L 0 ,L 1 ,…,L t For a delay violation sequence starting from time 0:
L t ＝max τ∈[|t|] {L(τ)}
optimizing the target: the scheduler wants to minimize the migration cost and delay violations. For a given delay tolerant vector
And migration cost expectation E [ M]The scheduler wants the delay not to exceed a threshold theta. When L is t When the sum is less than or equal to theta, the solution of the integer programming problem is not carried out and the optimal deployment is adjusted, so that the introduction of calculation cost and migration cost E [ M ] is avoided]. When L is t When the integer is more than theta, the integer programming problem needs to be recalculated and adjusted to the optimal deployment, and calculation cost and migration cost are introduced.
Therefore, the VNF deployment problem becomes one in which the sequence L is known 0 ,L 1 ,…,L t Then, it is determined whether to perform calculation and migration at time t +1, i.e. the scheduler seeksFind the interval that maximizes neighboring migration:
min t＞0 {t|L t+1 ＞Θ}
to sum up, the deployment Scheduling problem (FSDNC problem) facing the Dynamic Network environment can be described as follows:
wherein f (L) t ) Is the reward function:
if the sequence L t Not exceeding the threshold Θ, a prize L is introduced at time t t If L is t Beyond the threshold Θ, a penalty of- λ (E [ M ] is introduced]+ C), where λ is a constant greater than 0, C is a known computational cost expectation, E [ M ]]Is expected for the migration cost. It is clear that as the timing t increases, the transition cost tends to increase. But may be chaotic and chaotic due to the increase in migration costs and have little impact on the overall model. The scheduler wants to maximize the interval between two migrations to enjoy the longest "ease" time.
Solving the problem: in order to solve the FSDNC problem in the dynamic environment, the invention utilizes the technical scheme of the optimal time-out theory.
The optimal time-out theory is that a random variable sequence X is given 1 ,X 2 …, assuming the sequence joint distribution is known. And a sequence of true observed reward values: y is 0 ,y 1 (x 1 ),y 2 (x 1 ,x 2 ),…,y ∞ (x 1 ,x 2 …). For the timing sequences t =1,2, …, X can be observed 1 ＝x 1 ,X 2 ＝x 2 …. The optimal stopping time problem is that the optimal stopping time t is found * So that maximizing the return expectation:
a common solution to this problem is a method that employs the Forward rules (look-ahead rules). The logical meaning of the forward rule is that at time t, if the reward expectation can be found from t to t + k, if the largest of them is smaller than the observed value, then it should be stopped at that time. For the 1-degree forward rule, i.e., if the next reward for a given sequence of rewards is expected to be less than an observed value, then the observed value should be stopped and taken. If the 1-step forward rule fails to solve the problem, a more complex 2-step stopping rule should be employed, and so on until appropriate. From literature, it is known that for the optimal dwell time problem of monotonic finite phase, the 1-order forward rule is already optimal among all k-order forward rules. For the monotonic optimal settling problem and the finite-bound monotonic optimal settling problem, the 1-order forward rule is optimal. It can be shown that the first order forward rule of the FSDNC problem is:
wherein Pr [ L ] t+1 ＝l]Is L t+1 The distribution function of (2). I.e., from time 0, the scheduler needs to recalculate the optimal deployment scenario once the inequality is satisfied.
Based on the above knowledge, the method for solving the FSDNC problem by using the optimal dead time theory comprises the following steps:
inputting a question: observations at time t of L (0), L (1), …, L (t), E [ M ], C, λ
And (4) problem output: time of optimal stop at time t
The method comprises the following steps:
(1) The distribution in the delay violation sequence at time L (t + 1) at t +1 is first obtained. Considering a real-world scenario, violating the delay at time t +1 is not likely to exhibit an identical distribution. In general, the smaller t +1, the less likely it is to violate the delay, so the expectation and variance of the distribution at that time L (t + 1) are small. As t +1 becomes larger, the system becomes more disordered. The expectation and variance of the distribution at this time L (t + 1) are large. The L (t + 1) distribution is therefore calculated using a first order incremental mean algorithm.
(2) Calculating L t ＝max τ∈[|t|] Probability distribution of { L (τ) }. .
(3) Calculating the D value according to the optimal stopping theory:
and judging whether the deployment is less than 0, if so, stopping, and calculating new deployment by utilizing an ILP algorithm of the DepVeC problem and scheduling.
The first-order incremental mean algorithm is specifically described as follows:
inputting: observed value at t-th time L (t), increment mean [ m ] t ,n t ]Incremental probability p t ，m 0 ＝0,n 0 ＝0，p 0 ＝1；
And (3) outputting: the distribution of L (t + 1) at time t +1, and updating the incremental mean and probability;
and (3) calculating:
(2)A＝{a,a+1,…,L(t)},B＝{L(t)+m t }
(3) Return to
Wherein i belongs to A, j belongs to B
(4) A value of L (t + 1) was observed;
(5) If L (t + 1) > L (t), update
Update>
(6) Otherwise, update
Update>
Based on the above-described model that meets the real situation according to the required virtual network function and the delay requirement of the vehicle terminal user, and the process of judging when the optimal virtual network function deployment scheme needs to be calculated and completing the scheduling optimization based on the optimal time-out theory, it can be known that, in the actual application, the dynamic network-oriented virtual network function deployment method comprises the following steps, referring to fig. 2,
s1, acquiring a network function set N required by a mobile intelligent vehicle, and acquiring a network function N i The calculation resource R needed by the element N i Network function n i Maximum tolerated delay theta of i Computing resources C owned by the central base station s S Network bandwidth b owned by network connection e between central base stations e ；
S2, performing initial deployment of virtual network functions based on a pre-constructed virtual network deployment model and constraint conditions of the virtual network deployment model;
s3, acquiring the number L (t) of virtual network functions violating the delay requirement at the time t, and calculating a delay violating sequence L starting from the time 0 0 ,L 1 ,…,L t ；
S4, calculating deployment I at t moment t Optimal deployment I migrating to time t t′ The migration cost to be spent is expected to be E [ M ]]；
S5, according to the delay violation sequence L starting from the time 0 0 ,L 1 ,…,L t And migration cost expectation E [ M]The known calculation cost expectation C converts the dynamic migration problem of the virtual network function into the optimal time-out problem;
and S6, solving the optimal stopping time problem by using the antecedent rule method, determining the time when the optimal virtual network function deployment scheme needs to be calculated, and finishing scheduling optimization.
According to the same technical concept of the embodiment of the method, the invention also provides a dynamic network-oriented virtual network function deployment device, which is applied to a vehicle-mounted cloud system, wherein the vehicle-mounted cloud system comprises | V | mobile intelligent vehicles and | S | central base stations, the mobile intelligent vehicles are connected with the central base stations through wireless communication, and the device specifically comprises:
a data acquisition module for acquiring the network function set N required by the mobile intelligent vehicle, the network function N i The calculation resource R needed by the element N i Network function n i Maximum tolerated delay theta of i Computing resources C owned by the central base station s S Network bandwidth b owned by network connection e between central base stations e ；
The initial deployment module is used for performing initial deployment of virtual network functions based on a pre-constructed virtual network deployment model and constraint conditions of the virtual network deployment model;
a delay violation recording module for acquiring the number L (t) of virtual network functions violating the delay requirement at the time t and calculating the delay violation sequence L from the time 0 0 ,L 1 ,…,L t ；
An auxiliary computing module for computing deployment I at time t t Migration to optimal deployment at time s I s The migration cost to be spent is expected to be E [ M ]]；
An optimal stall problem establishing module for establishing an optimal stall problem according to the delay violation sequence L starting from 0 0 ,L 1 ,…,L t And migration cost expectation E [ M]The known calculation cost expectation C converts the virtual network function dynamic migration problem into an optimal time-out problem;
and the problem solving module is used for solving the optimal stopping time problem by utilizing the antecedent rule method, determining the time when the optimal virtual network function deployment scheme needs to be calculated and finishing scheduling optimization.
The virtual network deployment model constructed in advance is represented as follows:
wherein | N-]Is an integer set {1,2, …, | N | }, [ | S |]Is integer set {1,2, …, | S | }, [ | V |]Is an integer set {1,2, …, | V | }; x ijk Is a representation of a moving intelligent vehicle v k Required network function n i Whether or not to be deployed at a central base station s j The two-dimensional decision variables of (1); l ijk Indicating for deployment at a central base station s j Network function n of i Moving intelligent vehicle v k A perceived delay;
the constraint conditions of the virtual network deployment model are as follows:
formula (1) represents a mobile intelligent vehicle v k Required virtual network function n i Deployed on a central base station s;
formula (2) shows that for any central base station node S belonging to S, the deployment scheme meets the requirement of computing resources;
equation (3) shows that for any link connection E E, the deployment scenario satisfies the bandwidth resource constraint, where the networkPath P = { P 1 ,p 2 ,…,p |P| The method comprises the following steps that (1) a path from a center base station to a mobile intelligent vehicle is defined, and | P | is the number of the paths;
equation (4) represents each network function n i All having a service provider level agreement delay theta i ；
Equation (5) represents network functions that are not required for a mobile smart vehicle, not requiring deployment, where F k Indicating a vehicle v k A set of required virtual network functions.
Further, the migration cost expectation E [ M ] is calculated as:
m (t, t') denotes deployment I from time t t Optimal deployment I migrating to time t t′ The cost of the migration that needs to be expended,
where I (x, x') is an indicator function, which is defined as follows:
Further, the optimal stopping time problem form for representing the virtual network function dynamic migration problem is as follows:
wherein f (L) t ) Is the reward function:
λ is a coefficient greater than 0.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: although the present invention is described in detail with reference to the above embodiments, the interaction manner and the online scheduling method of the network information collecting and scheduling device (apparatus) in the present invention are applicable to each system, and it should be understood by those skilled in the art that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (4)
1. A dynamic network-oriented virtual network function deployment method is applied to a vehicle-mounted cloud system, the vehicle-mounted cloud system comprises | V | mobile intelligent vehicles and | S | central base stations, and the mobile intelligent vehicles are connected with the central base stations through wireless communication, and the method comprises the following steps:
acquiring a network function set N required by a mobile intelligent vehicle, wherein the network function N is i The calculation resource R needed by the N i Network function n i Maximum tolerated delay theta of i Computing resources C owned by the central base station s S Network bandwidth b owned by network connection e between central base stations e ；
Performing initial deployment of virtual network functions based on a pre-constructed virtual network deployment model and constraint conditions of the virtual network deployment model;
acquiring the number L (t) of virtual network functions violating the delay requirement at the time t, and calculating a delay violating sequence L from the time 0 0 ,L 1 ,…,L t ；
Calculating deployment at time t I t Optimal deployment I migrating to time t t′ The migration cost to be spent is expected to be E [ M ]]；
According to the delay violation sequence L starting from time 0 0 ,L 1 ,…,L t Migration cost expectation E [ M]The known calculation cost expectation C converts the virtual network function dynamic migration problem into an optimal time-out problem;
solving the optimal time-out problem by utilizing a forward rule method, determining the time when the optimal virtual network function deployment scheme needs to be calculated, and finishing scheduling optimization;
the pre-constructed virtual network deployment model is represented as follows:
wherein | N-]Is integer set {1,2, …, | N | }, [ | S | N | } |)]Is an integer set {1,2, …, | S | }, [ | V |]Is an integer set {1,2, …, | V | }; x ijk Is a mobile intelligent vehicle v k Required network function n i Whether or not to be deployed at a central base station s j The two-dimensional decision variables of (1); l ijk Indicating for deployment at a central base station s j Network function n of i Moving intelligent vehicle v k A perceived delay;
the constraint conditions of the virtual network deployment model are as follows:
formula (1) represents a mobile intelligent vehicle v k Required virtual network function n i Deployed on a central base station s;
formula (2) shows that for any central base station node S E S, the deployment scheme meets the requirement of computing resources;
equation (3) indicates that for any link connection E, the deployment scenario satisfies the bandwidth resource constraint, where network path P = { P = } 1 ,p 2 ,…,p |P| The position of the center base station is the position of the center base station, and the position of the center base station is the position of the center base station;
equation (4) represents each network function n i All having a service provider level agreement delay theta i ；
Equation (5) represents network functions not required for a mobile smart vehicle, not requiring deployment, where F k Indicating a vehicle v k A set of required virtual network functions;
the migration cost expectation E [ M ] is calculated as:
m (t, t') denotes deployment I from time t t Optimal deployment I migrating to time t t′ The cost of the migration that needs to be expended,
where I (x, x') is an indicator function, which is defined as follows:
the optimal stopping time problem form for representing the dynamic migration problem of the virtual network function is as follows:
wherein f (L) t ) Is the reward function:
λ is a coefficient greater than 0.
2. The dynamic network-oriented virtual network function deployment method of claim 1, wherein the forward rule method employs a 1-order forward rule:
3. A virtual network function deployment device oriented to a dynamic network, applied to a vehicle-mounted cloud system including | V | mobile smart vehicles and | S | central base stations, the mobile smart vehicles being connected with the central base stations via wireless communication, the device comprising:
the data acquisition module is used for acquiring a network function set N required by the mobile intelligent vehicle, and the network function N i The calculation resource R needed by the N i Network function n i Maximum tolerated delay theta of i Computing resources C owned by the central base station s S Network bandwidth b owned by network connection e between central base stations e ；
The initial deployment module is used for performing initial deployment of virtual network functions based on a pre-constructed virtual network deployment model and constraint conditions of the virtual network deployment model;
a delay violation recording module for acquiring the number L (t) of virtual network functions violating the delay requirement at the time t and calculating the delay violation sequence L from the time 0 0 ,L 1 ,…,L t ；
An auxiliary computing module for computing deployment I at time t t Migration to optimal deployment at time s I s The migration cost to be spent is expected to be E [ M ]]；
An optimal stall problem establishing module for establishing an optimal stall problem according to the delay violation sequence L starting from 0 0 ,L 1 ,…,L t And migration cost expectation E [ M]The known calculation cost expectation C converts the dynamic migration problem of the virtual network function into the optimal time-out problem;
the problem solving module is used for solving the optimal time-out problem by utilizing a forward rule method, determining the time when the optimal virtual network function deployment scheme needs to be calculated and finishing scheduling optimization;
the pre-constructed virtual network deployment model is represented as follows:
wherein | N-]Is integer set {1,2, …, | N | }, [ | S | N | } |)]Is integer set {1,2, …, | S | }, [ | V |]Is an integer set {1,2, …, | V | }; x ijk Is a mobile intelligent vehicle v k Required network function n i Whether or not to be deployed at a central base station s j The two-dimensional decision variables of (1); l. the ijk Indicating for deployment at a central base station s j Network function n of i Moving intelligent vehicle v k A perceived delay;
the constraint conditions of the virtual network deployment model are as follows:
formula (1) represents a mobile intelligent vehicle v k Required virtual network function n i Deployed on a central base station s;
formula (2) shows that for any central base station node S belonging to S, the deployment scheme meets the requirement of computing resources;
equation (3) indicates that for any link connection E, the deployment scenario satisfies the bandwidth resource constraint, where network path P = { P = } 1 ,p 2 ,…,p |P| The position of the center base station is the position of the center base station, and the position of the center base station is the position of the center base station;
equation (4) represents each network function n i All having a service provider level agreement delay theta i ；
Equation (5) represents network functions that are not required for a mobile smart vehicle, not requiring deployment, where F k Indicating a vehicle v k A set of required virtual network functions;
the migration cost expectation E [ M ] is calculated as:
m (t, t') denotes deployment I from time t t Optimal deployment I migrating to time t t′ The cost of the migration that needs to be expended,
where I (x, x') is an indicator function, which is defined as follows:
the optimal stopping time problem form for representing the dynamic migration problem of the virtual network function is as follows:
wherein f (L) t ) Is the reward function:
λ is a coefficient greater than 0.
4. A computer device, comprising:
one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, which when executed by the processors implement the virtual network function deployment method of any of claims 1-2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111402460.3A CN114095358B (en) | 2021-11-23 | 2021-11-23 | Dynamic network-oriented virtual network function deployment method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111402460.3A CN114095358B (en) | 2021-11-23 | 2021-11-23 | Dynamic network-oriented virtual network function deployment method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114095358A CN114095358A (en) | 2022-02-25 |
| CN114095358B true CN114095358B (en) | 2023-04-11 |
Family
ID=80303839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111402460.3A Active CN114095358B (en) | 2021-11-23 | 2021-11-23 | Dynamic network-oriented virtual network function deployment method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114095358B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115086223B (en) * | 2022-05-06 | 2023-07-25 | 南京邮电大学 | Vehicle-mounted delay tolerant network data forwarding method based on dynamic rewarding mechanism |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018178033A1 (en) * | 2017-03-27 | 2018-10-04 | British Telecommunications Public Limited Company | Virtualised network function deployment |
| WO2020026140A1 (en) * | 2018-07-30 | 2020-02-06 | Telefonaktiebolaget Lm Ericsson (Publ) | Machine learning method for adaptive virtual network functions placement and readjustment |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10678579B2 (en) * | 2017-03-17 | 2020-06-09 | Vmware, Inc. | Policy based cross-cloud migration |
| CN109348430B (en) * | 2018-10-29 | 2020-05-12 | 电子科技大学 | Multicast scheduling method for multi-channel multi-content base station cell |
| CN111694636B (en) * | 2020-05-11 | 2022-08-23 | 国网江苏省电力有限公司南京供电分公司 | Electric power Internet of things container migration method oriented to edge network load balancing |
-
2021
- 2021-11-23 CN CN202111402460.3A patent/CN114095358B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018178033A1 (en) * | 2017-03-27 | 2018-10-04 | British Telecommunications Public Limited Company | Virtualised network function deployment |
| WO2020026140A1 (en) * | 2018-07-30 | 2020-02-06 | Telefonaktiebolaget Lm Ericsson (Publ) | Machine learning method for adaptive virtual network functions placement and readjustment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114095358A (en) | 2022-02-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Online adaptive interference-aware VNF deployment and migration for 5G network slice | |
| Zhou et al. | CE-IoT: Cost-effective cloud-edge resource provisioning for heterogeneous IoT applications | |
| Kim et al. | Multi-agent reinforcement learning-based resource management for end-to-end network slicing | |
| CN108667657B (en) | SDN-oriented virtual network mapping method based on local feature information | |
| Wang et al. | Application-aware offloading policy using SMDP in vehicular fog computing systems | |
| CN110069341A (en) | What binding function configured on demand has the dispatching method of dependence task in edge calculations | |
| CN114095358B (en) | Dynamic network-oriented virtual network function deployment method and device | |
| Chen et al. | A hybrid task scheduling scheme for heterogeneous vehicular edge systems | |
| Gonçalves et al. | Dynamic network slicing in fog computing for mobile users in mobfogsim | |
| Shadi et al. | Ready-time partitioning algorithm for computation offloading of workflow applications in mobile cloud computing | |
| Maray et al. | Dependent task offloading with deadline-aware scheduling in mobile edge networks | |
| Meneguette et al. | An efficient green-aware architecture for virtual machine migration in sustainable vehicular clouds | |
| Liu et al. | RFID: Towards low latency and reliable DAG task scheduling over dynamic vehicular clouds | |
| Xu et al. | Online learning algorithms for offloading augmented reality requests with uncertain demands in MECs | |
| De Mendoza et al. | Near optimal VNF placement in edge-enabled 6G networks | |
| CN111159859A (en) | Deployment method and system of cloud container cluster | |
| CN113190342A (en) | Method and system architecture for multi-application fine-grained unloading of cloud-edge cooperative network | |
| Nguyen et al. | EdgePV: collaborative edge computing framework for task offloading | |
| CN110290206A (en) | A kind of distributed computing system and method for cafe environment | |
| CN110677301A (en) | Software defined transmission control method for single controller with multiple switches in 5G network | |
| Wang et al. | Cost-driven data caching in edge-based content delivery networks | |
| CN106789289B (en) | Method and apparatus for virtual network mapping | |
| Ge et al. | Dynamic hierarchical caching resource allocation for 5G-ICN slice | |
| Tang et al. | Edge computing energy-efficient resource scheduling based on deep reinforcement learning and imitation learning | |
| He et al. | QoS-Aware and Resource-Efficient Dynamic Slicing Mechanism for Internet of Things. |
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 |