CN115174403B - Method and device for resource scheduling and route management of multi-mode communication network in low-carbon park - Google Patents
Method and device for resource scheduling and route management of multi-mode communication network in low-carbon park Download PDFInfo
- Publication number
- CN115174403B CN115174403B CN202210775986.4A CN202210775986A CN115174403B CN 115174403 B CN115174403 B CN 115174403B CN 202210775986 A CN202210775986 A CN 202210775986A CN 115174403 B CN115174403 B CN 115174403B
- Authority
- CN
- China
- Prior art keywords
- vnf
- server
- embedding
- flow
- data
- 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
- 238000000034 method Methods 0.000 title claims abstract description 87
- 229910052799 carbon Inorganic materials 0.000 title claims abstract description 53
- 238000004891 communication Methods 0.000 title claims abstract description 48
- 238000005457 optimization Methods 0.000 claims abstract description 75
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 36
- 230000008447 perception Effects 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 40
- 238000004364 calculation method Methods 0.000 claims description 19
- 238000012545 processing Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 16
- 230000007774 longterm Effects 0.000 claims description 11
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 238000003860 storage Methods 0.000 claims description 9
- 230000009471 action Effects 0.000 claims description 8
- 230000001186 cumulative effect Effects 0.000 claims description 8
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 7
- 230000033228 biological regulation Effects 0.000 claims description 6
- 238000011217 control strategy Methods 0.000 claims description 3
- 230000006641 stabilisation Effects 0.000 claims description 3
- 238000011105 stabilization Methods 0.000 claims description 3
- 230000000452 restraining effect Effects 0.000 claims description 2
- 241000764238 Isis Species 0.000 claims 1
- 238000009826 distribution Methods 0.000 abstract description 3
- 230000008878 coupling Effects 0.000 abstract description 2
- 238000010168 coupling process Methods 0.000 abstract description 2
- 238000005859 coupling reaction Methods 0.000 abstract description 2
- 230000004069 differentiation Effects 0.000 abstract 1
- 238000007726 management method Methods 0.000 description 40
- 238000004808 supercritical fluid chromatography Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000012986 modification Methods 0.000 description 7
- 230000004048 modification Effects 0.000 description 7
- 230000005540 biological transmission Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 239000000969 carrier Substances 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
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
- H04L41/12—Discovery or management of network topologies
-
- 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
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/02—Topology update or discovery
- H04L45/08—Learning-based routing, e.g. using neural networks or artificial intelligence
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention relates to a method and a device for resource scheduling and route management of a multi-mode communication network in a low-carbon park, and belongs to the technical field of communication. The VNF embedding and flow scheduling combined optimization algorithm provided by the invention solves the problem of coupling between VNF embedding and flow scheduling by utilizing Lyapunov optimization, optimizes a long-time-scale embedding strategy and short-time-scale flow scheduling, realizes flexible scheduling and routing management of multi-mode communication network resources, and ensures energy management service differentiation QoS requirements; the problem of external property is solved by establishing a matching problem between a server and the VNF and utilizing a switching matching theory, and an embedding scheme between continuous switching flows is used for obtaining an embedding strategy with stable switching, so that the cost of embedding is reduced, and the high-efficiency utilization of multi-mode communication network resources is realized; under the condition that the information is incomplete due to the diversified regional characteristics of the park and the complex power distribution environment, the routing strategy is optimized through Q learning and back pressure perception, the network throughput is improved, and the low-carbon operation of the energy management service is supported.
Description
Technical Field
The invention relates to a method and a device for resource scheduling and route management of a multi-mode communication network in a low-carbon park, and belongs to the technical field of communication.
Background
The park is an industrial park with informatization and intelligence characteristics realized by new generation information technology means such as 'cloud object intelligence development'. The low-carbon operation construction of the park energy management business is developed, and the method is an important ring for realizing and pushing the novel electric power system construction by the aid of the double-carbon targets of '3060'. The low-carbon park energy management business is flexible and various, such as flexible load regulation and control, electric power spot market, carbon footprint monitoring and the like, and the demands of various business applications on time delay, bandwidth, throughput, reliability and the like are different. Therefore, there is a need for flexible scheduling of communication network resources according to differentiated energy management traffic demands. However, the conventional communication network is sealed and stiff, and the software and hardware are highly coupled, so that the energy management service requirement of the low-carbon park cannot be met. By utilizing network function virtualization (Network Function Virtualization, NFV), the virtualization of multidimensional physical resources such as a low-carbon park 5G, optical fibers, alternating-current and direct-current carriers, WLAN communication links, a multi-mode communication edge computing gateway, a server and the like is realized, and the advantages of software defined network (Software Defined Networking, SDN) control and forwarding separation are combined, so that the centralized management and flexible scheduling of the multi-mode communication network resources are realized. In addition, the diversified regional characteristics and the complex power distribution environment of the low-carbon park make the channel transmission state information unpredictable, so that the advanced artificial intelligence technology needs to be combined to realize the route management under the incomplete information so as to meet the differentiated service quality (Quality of Service, qoS) requirement of the low-carbon park energy management service.
Based on NFV technology, low-carbon park energy management services are identified by service function chains (Service Function Chain, SFC). SFCs are made up of a set of virtual network functions (Virtual Network Function, VNFs) in a specific order, and the data flow must traverse all VNFs, i.e. the software running on the server, to service a specific energy management service on the low-carbon park. The invention considers the multi-mode communication network resource scheduling and route management supporting the low-carbon park energy management service, and mainly focuses on two core problems of VNF embedding and stream scheduling. VNF embedding decides how to effectively embed VNFs on a server to reduce embedding costs. The flow scheduling means that on the basis of the VNF embedded server, adaptive routing is realized through routing management, the admitted data volume is determined, and an appropriate next-hop server is selected, so that throughput maximization is realized while the service level agreement (Service Level Agreement, SLA) is satisfied. But joint optimization of VNF embedding and flow scheduling still faces some challenges, first, campus low-carbon traffic has differentiated QoS requirements in throughput, VNF embedding cost; secondly, the VNF embedding and the flow scheduling strategies are mutually coupled under different time scales, and the interaction of the VNF embedding strategies among different flows causes externality, so that the solution of the joint optimization problem is very complex; finally, because the information in the practical application has unpredictability, the flow scheduling optimization needs to be performed under incomplete information.
Therefore, there is an urgent need to design a multi-mode communication network resource scheduling and routing management method capable of jointly optimizing VNF embedding and flow scheduling, so as to greatly reduce the embedding cost while maximizing throughput, and meet the energy management service differentiated QoS requirements of a low-carbon park.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a method and a device for scheduling and route management of multi-mode communication network resources in a low-carbon park.
The invention relates to a resource scheduling and routing management method for a multi-mode communication network in a low-carbon park, which comprises the following specific management steps:
modeling a VNF embedding and flow scheduling process in a pre-built architecture model into a VNF embedding model and a flow scheduling model based on a multi-time scale model, wherein the embedding of the VNF to a data plane server in a set time period is established as the VNF embedding model, and the route of a flow of a source server to a destination server in a time slot is established as the flow scheduling model;
based on the VNF embedding model and the flow scheduling model, a VNF embedding and flow scheduling combined optimization problem model is established, and network performance is restrained;
based on the above-mentioned constraint result of network performance, the VNF embedding and flow scheduling joint optimization problem model is described as maximizing the weighting difference of network throughput and VNF embedding cost; and
The VNF embedding strategy optimization problem and the time slot scale stream scheduling optimization problem of a set time period are converted by Lyapunov optimization;
and performing VNF embedding strategy optimization in a set time period through a switching-matched VNF embedding algorithm, performing time slot flow scheduling optimization through admission control in time slots and a backpressure sensing routing algorithm based on Q learning, and performing multi-mode communication network resource scheduling and routing management in a low-carbon park.
Further, the pre-constructed architecture model comprises a data plane, a control plane and an application plane; wherein:
in a data plane, the multi-mode heterogeneous communication network resources are abstracted into a unified virtual resource pool, wherein the virtual resource pool comprises communication, calculation and storage virtual resources, the virtual resources are deployed to support different VNs, and the VNs are arranged in a specific sequence to form SFC;
in a control plane, the SDN controller utilizes a resource scheduling component to embed VNF, and utilizes a route management component to perform flow scheduling to realize route management optimization; wherein the VNF embedding is embedding VNFs of SFC in a server to support multiple low-carbon services; the flow scheduling refers to selecting a proper path for a flow from a source server to transmit to a destination server on the basis of embedding the VNF;
In the application plane, the system comprises the energy management business of a low-carbon intelligent park: carbon footprint monitoring, electric power spot market, flexible load regulation.
Further the VNF embedding model is: VNF embedding an indication variable for the h time period
Wherein the subscriptRepresenting an nth processing server; upper energizer->The kth VNF representing SFC;
the flow scheduling model is as follows: t-th time slot embedded VNF k Flow f queue backlog at server n of (2)The update expression is:
wherein,and->Embedded VNF for t and t+1 time slots respectively k Flow f queue backlog at server n, +.>Routing an indicator variable for flow f; />For embedding VNF k+1 Server set of->For embedding VNF k-1 Is a server set of (a); />Is the t-th time slot processed by server n and transmitted to the set +.>Data volume of the server in (a); />Is the t-th time slot is defined by the set->The amount of data transmitted by the server to server n; />Is admitted data.
Further, the establishing a VNF embedding and flow scheduling joint optimization problem model, which constrains network performance, includes:
constraints on streaming, for origin serversAdmission data of stream f->The constraints are:
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; at the same time, only VNF k+1 The server j is embedded in the server and, can be transmitted to the server j, wherein +.>Embedding an indication variable for the VNF;
the amount of data transmitted via the link (n, j) is constrained by the link capacity L at the t-th slot n,j (t) and calculation Capacity C n (t) constraint, the formula of which is:
wherein the method comprises the steps ofRepresenting VNF k Processing the unit complexity of stream f data; τ is the duration of one slot;
queue backlog constraint, expressed as long-term queue backlog constraint:
wherein the method comprises the steps ofIs the t-th time slot embedded VNF k A backlog of the flow f queue at server n;
throughput constraint, average throughput r of flow f f The constraint is expressed as:
wherein the method comprises the steps ofAnd->The minimum throughput and the maximum throughput of the flow f are indicated, respectively.
Further, based on the result of constraining the network performance, the method for describing the VNF embedding and flow scheduling joint optimization problem model as a weighted difference for maximizing the network throughput and the VNF embedding cost is as follows:
the optimization problem P1 is a maximization weighted cumulative utility function, and its formula is:
P1:
s.t.C 1 :
C 2 :
C 3 (2), (3), (4), and (5)
Wherein the method comprises the steps ofβ and λ represent weights of throughput and embedding cost, respectively, +.>Indicating VNF embedded indication variable,/->Admission data, e, representing the flow f on the server n k (h) Representing VNF k Is embedded cost of (a); y (h), r (t), x (t) are each +. > WhereinRouting an indicator variable for flow f; constraint C 1 Representing server selection constraints; constraint C 2 Representing VNF embedded constraints;
formula (2) in C3 is
Equation (3) is
Equation (4) is
Equation (5) is
Wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; />Is the amount of data transmitted by the t-th slot via the link (n, j); l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively; />Representing VNF k Processing the unit complexity of stream f data; τ is the duration of one slot; />Is the t-th time slot embedded VNF k A backlog of the flow f queue at server n; />And->The minimum throughput and the maximum throughput of the flow f are respectively represented; r is (r) f Is the average throughput of stream f.
Further, the specific steps of the VNF embedding policy optimization problem and the time slot scale flow scheduling optimization problem converted from the lyapunov optimization into the time slot are as follows:
by utilizing the virtual queue principle, the long-term queue stability constraint is converted into a virtual queue, and the optimization problem P1 is rewritten as follows:
P1.1:
s.t.C 1 ,C 2
C 4 (2) and (3)
C 5 :Y f (t) and Z f (t) average Rate stabilization
Wherein formula (2) in C4 is
Equation (3) is
In C5
Wherein W is a weighted cumulative utility function;representation server- >Up, the amount of data that stream f arrives at; />Admission data representing the flow f on the server n; />Is the amount of data transmitted by the t-th slot via the link (n, j); l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively; />Representing VNF k Processing the unit complexity of stream f data; τ is the duration of one slot; y is Y f (t) and Z f (t) is a virtual queue; />Admission data representing flow f;
based on Lyapunov optimization, P1.1 translates into an upper bound that minimizes drift minus rewards, expressed as:
wherein the method comprises the steps ofV is a non-negative weighting parameter of the utility function; Θ (t) = [ Q (t), Y (t), Z (t)]Wherein->Andfor virtual queue vector, ++>Is a queue vector;is Lyapunov drift in whichIs a lyapunov function.
Further, the method for optimizing the VNF embedding policy under the time period by using the exchange-matched VNF embedding algorithm includes:
by solving the VNF embedding sub-problem SP1 in the set period of time, specific embedding positions of the K VNFs on the N servers are determined, and the formula is as follows:
SP1:
s.t.C 2
wherein y (h) is the VNF embedding policy for the h time period;
based on SP1, the utility function of flow f is:
wherein V is a non-negative weighting parameter of the utility function;β and λ represent weights for throughput and embedding cost, respectively; Representing VNF embedded indication variables; e, e k (h) Representing VNF k Is embedded cost of (a); />Routing an indicator variable for flow f;the embedding condition of the h time period; /> Respectively isL n,j (t),C n Empirical value of (t)/(t)>Is admitted data; /> Is the queue backlog of the t-th time slot under a given embedding strategy;
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively;
the flow calculates utility value according to the formula and establishes a preference list according to descending order; the provided VNF embedding algorithm based on the exchange matching set time period has the following specific flow:
giving satisfaction of constraint C 2 Is embedded policy of (a)Calculating the utility of the flow;
in each iteration, all servers randomly select one serverAnd another subset->If->The number of servers in (a) is greater than 1, i.e. +.>The server n leaves +.>Add->Form a new embedding strategy->Then, the utility of the stream is recalculated; if->Is->Exchange matching of (a) by embedding the original policy +.>Replaced by->OtherwiseRemain unchanged;
according toThen->Otherwise->Will->Conversion to y * (h)。
Further, the admission control method comprises the following steps:
determining a slot admission control subproblem SP2:
SP2:
s.t.C 6 :(2)
Wherein formula (2) in C6 isV is a non-negative weighting parameter of the utility function; beta is throughput weight; y is Y f (t) and Z f (t) is a virtual queue; />Backlog for the t-th slot queue; />Admission data representing the flow f on the server n; />Representation server->Up, the amount of data that stream f arrives at;
adopting admission control method to solve SP2, expressed as
Further, the back pressure perception routing algorithm based on Q learning comprises the following specific procedures:
determining a slot route selection sub-problem SP3, and determining an optimal next-hop server:
SP3:
s.t.C 1 :
C 7 :(3)
wherein formula (3) in C7 is Is the t-th time slot embedded VNF k A backlog of the flow f queue at server n; />Is the amount of data transmitted by the t-th slot via the link (n, j); />Routing an indicator variable for flow f; x (t) is a routing policy;
the back pressure perception routing algorithm based on Q learning is provided by adopting the Q learning to solve SP2, and the specific flow is as follows:
initializing Q value, setting Q (S n (t),a n (t))＝0；
At the beginning of the t time slot, each server selects actions according to an epsilon greedy strategy; server j will select B n (t) the largest server n and rejecting others, resolving server selection conflicts; all servers update rewards and queue information and go to the next state S n (t+1); and updating the Q value;
according to a n (t) =jOtherwise->Will { a } n (t) } conversion to x * (t)；
Wherein the state isAction as->The rewards areQ value Q (S) n (t),a n (t)) estimating state S for server n (t) selecting action a n A value function of (t).
Further, the algorithm of the VNF embedding and flow scheduling joint optimization is as follows:
initializing queue backlog of all queues, and embedding, admittance control and routing strategy indication functions of the VNF;
in each set time period, the VNs obtain an optimal VNF embedding strategy y (t) according to a VNF embedding algorithm of the set time period based on exchange matching;
in each time slot, each source server obtains an optimal admission control strategy r (t) according to a formula, and the source server and the processing server obtain an optimal routing strategy x (t) according to a backpressure sensing routing algorithm based on Q learning.
A low-carbon park multi-mode communication network resource scheduling and route management device specifically comprises:
modeling module: the method comprises the steps of modeling a VNF embedding and flow scheduling process in a pre-built architecture model into a VNF embedding model and a flow scheduling model based on a multi-time scale model, wherein the embedding of the VNF to a data plane server in a time period is established as the VNF embedding model, and the route of a flow of a source server to a destination server in a time slot is established as the flow scheduling model;
Constraint module: the method is used for establishing a VNF embedding and flow scheduling joint optimization problem model by utilizing the VNF embedding model and the flow scheduling model, and restraining network performance;
and a conversion module: using the constraint, describing the VNF embedding and flow scheduling joint optimization problem model as a weighted difference for maximizing network throughput and VNF embedding cost; and
the VNF embedding strategy optimization problem and the time slot scale stream scheduling optimization problem of the time period are converted by Lyapunov optimization;
and performing VNF embedding strategy optimization under a time period through a switching-matched VNF embedding algorithm, and performing time slot flow scheduling optimization through admission control under time slots and a backpressure sensing routing algorithm based on Q learning, thereby realizing multi-mode communication network resource scheduling and routing management of the low-carbon park.
Further, the modeling module includes:
the system comprises a data plane, a control plane and an application plane; wherein:
in a data plane, the multi-mode heterogeneous communication network resources are abstracted into a unified virtual resource pool, wherein the virtual resource pool comprises communication, calculation and storage virtual resources, different VNs are supported by deploying the virtual resources, and the VNs are arranged in a specific sequence to form SFC;
In a control plane, the SDN controller utilizes a resource scheduling component to embed VNF, and utilizes a route management component to perform flow scheduling to realize route management optimization; wherein the VNF embedding is embedding VNFs of SFC in a server to support multiple low-carbon services; the flow scheduling refers to selecting a proper path for a flow from a source server to transmit to a destination server on the basis of embedding the VNF;
in the application plane, the system comprises the energy management business of a low-carbon intelligent park: carbon footprint monitoring, electric power spot market, flexible load regulation.
Further, the constraint model is used for constraint of the constraint model for streaming, and for an origin serverAdmission data of stream f->The constraints are:
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; at the same time, only VNF k+1 The server j is embedded in the server and, can be transmitted to the server j, wherein +.>Embedding an indication variable for the VNF;
the amount of data transmitted via the link (n, j) is constrained by the link capacity L at the t-th slot n,j (t) and calculation Capacity C n (t) constraint, the formula of which is:
wherein the method comprises the steps ofRepresenting VNF k Processing the unit complexity of stream f data; τ is the duration of one slot;
queue backlog constraint, expressed as long-term queue backlog constraint:
Wherein the method comprises the steps ofIs the t-th time slot embedded VNF k A backlog of the flow f queue at server n;
throughput constraint, average throughput r of flow f f The constraint is expressed as:
wherein the method comprises the steps ofAnd->The minimum throughput and the maximum throughput of the flow f are indicated, respectively.
By means of the scheme, the invention has at least the following advantages:
(1) The VNF embedding and flow scheduling combined optimization algorithm VEFS provided by the invention solves the problem of coupling between VNF embedding and flow scheduling by utilizing Lyapunov optimization, optimizes the VNF embedding strategy of a long time scale and the flow scheduling of a short time scale, realizes flexible scheduling and route management of multi-mode communication network resources, and ensures the differentiated QoS requirements of energy management services.
(2) The long-time scale VNF embedding algorithm based on exchange matching solves the problem of externality by establishing the matching problem between a server and the VNF and utilizing the exchange matching theory, obtains an exchange-stable embedding strategy by continuously exchanging embedding schemes, reduces the cost and expense of embedding, and realizes the efficient utilization of multi-mode communication network resources.
(3) The backpressure sensing routing algorithm based on Q learning provided by the invention can optimize the routing strategy through Q learning and backpressure sensing under the condition that the information is incomplete due to the diversified regional characteristics and the complex power distribution environment of a park, improves the network throughput, supports the low-carbon operation of energy management services such as carbon footprint monitoring and the like.
The foregoing description is only an overview of the present invention, and is intended to provide a better understanding of the present invention, as it is embodied in the following description, with reference to the preferred embodiments of the present invention and the accompanying drawings.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate a certain embodiment of the present invention and therefore should not be considered as limiting the scope, and that other related drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of a low-carbon intelligent park multi-modal communication network resource scheduling and routing management architecture based on SDN/NFV of the present invention;
fig. 2 is a schematic diagram of a time slot model of the present invention.
Detailed Description
The following describes in further detail the embodiments of the present invention with reference to the drawings and examples. The following examples are illustrative of the invention and are not intended to limit the scope of the invention.
The whole technical scheme of the invention mainly comprises five steps: (1) constructing a system model framework; (2) model refinement; (3) providing VNF embedding and stream scheduling constraint and optimization problems; (4) optimization problem transformation based on Lyapunov; (5) A multi-mode communication network resource scheduling and route management method is provided. . The concrete introduction is as follows:
1. Construction of System model framework
The invention constructs a low-carbon park multi-mode communication network resource scheduling and routing management framework based on SDN/NFV, which comprises a data plane, a control plane and an application plane as shown in figure 1. The data plane includes multimode heterogeneous communication network resources, namely 'source charge storage' terminal resources such as photovoltaic panels, air conditioners, user energy storage and the like, communication resources such as WLAN, 5G/6G, optical fibers, power line carriers (Power Line Carrier, PLC) and the like, and resource servers and the like, and abstracts the multimode heterogeneous communication network resources into a unified virtual resource pool based on the NFV technology, wherein the communication, calculation and storage virtual resources are included. By flexibly deploying virtual resources to support SFC, the SFC is composed of diversified VNs arranged according to a specific sequence and is used for identifying various low-carbon park energy management services. The control plane is composed of an SDN controller, the SDN controller comprises a resource scheduling component and a routing management component, the resource scheduling component and the routing management component are respectively used for coordinating VNF embedding and configuring flow scheduling, the control plane SDN controller utilizes the resource scheduling component to carry out VNF embedding so as to realize network resource scheduling optimization, and the routing management component is utilized to carry out flow scheduling so as to realize routing management optimization, so that the application plane is supported to comprise park management business low-carbon operation such as carbon footprint monitoring, power spot market, flexible load regulation and control and the like.
2. Model refinement
(1) Multi-time scale time slot model
The present invention adopts a multi-time scale slot model, as shown in fig. 2, optimizes VNF embedding policies at long time scales (time periods), and optimizes flow scheduling policies at short time scales (time slots), wherein the flow scheduling policies are implemented by optimizing admission control and routing. The invention divides the whole optimization timeIs divided into T Time slots, using setsDenoted where each slot has a length τ. Next, will T Equally dividing a time slot into H For each time period, use the set->A representation in which each time period is of length T 0 I.e. t=t 0 H. The h time period is denoted by->
(2) VNF embedded model
The invention describes the underlying physical network as an undirected graphWherein->Representing a collection of N processing servers, N D Representing the destination server and epsilon representing the link set. The set of origin servers is denoted as N S And (2) andeach SFC is composed of K The VNs are composed, the set of which is denoted +.>Defining VNF embedded indication variable to +.>When->Time-dependent presentation of VNF k Successfully embedding the server n in the h time period, otherwise +.>The invention can only support one VNF embedding at most in one server per time period.
(3) Flow scheduling model
On the basis of embedding the VNF into the underlying server, stream scheduling optimization is carried out, and for embedding the VNF k Is a server of (a) n Definition of the inventionFor embedding VNF k+1 Server set of->For embedding VNF k-1 Is a server set of the server. And processing the flow from the source server through the routing sequence of the VNs 1-K through flow scheduling optimization in the t time slot, and finally reaching the destination server. Define the set of streams as +.>The routing indication variable defining flow f is +.>When->At the time, the data of the flow f is embedded into the VNF at the t-th time slot k Is transmitted to the server +.>Otherwise->At the end of the t-th time slot, at the VNF k The unprocessed data of the upper stream f will be stored in the buffer of the server n, which is modeled as a queue with backlog +.>The method is characterized in that:
wherein the method comprises the steps ofRepresenting the embedding of a VNF by a tth time slot k Server n transmission to embedded VNF k+1 Is a server of (a)Is a data amount of (a) in the data stream. />Representing processing by server n and transmission to the collection at the t-th slotData volume of the server in (a). />Representing the time slot at t by the set +.>Where k=1, the value is 0./>Representing admitted data, where k=1, 1 {k＝1} Identical to 1, otherwise 1 {k＝1} ＝0。
VNF embedding and stream scheduling constraint and optimization problem
(1) VNF embedding and flow scheduling constraints
1) Stream transmission constraint: for source server Admission data for flow f->Data not higher than arrivalNamely:
at the same time, only VNF k+1 The server j is embedded in the server and,i.e. < ->Can be transmitted to server j.
2) Resource constraint: at the t-th time slot, the amount of data transmitted via the link (n, j) is subject to the link capacity L n,j (t) and calculation Capacity C n (t) constraint, the formula of which is:
wherein the method comprises the steps ofRepresenting VNF k The unit complexity of processing stream f data.
3) Network performance constraints: to ensure network stability, long-term queue backlog constraints should satisfy:
to meet SLA, average throughput r of flow f f Should meet maximum throughputAnd minimum throughput->Constraint, namely:
(2) Problem of optimization
The invention solves the optimization problems that: under the multi-time scale, through combining and optimizing VNF embedding and stream scheduling, the network throughput is maximized, meanwhile, the VNF embedding cost is reduced, and a weighted cumulative utility function is defined as
Where β and λ represent weights of throughput and embedding cost, e, respectively k (t) represents VNF k Is embedded cost of (a). Therefore, the optimization problem P1 is a maximization weighted cumulative utility function, and its formula is:
wherein y (h), r (t), x (t) each representIs a set of (3). Constraint C 1 Representing a server selection constraint that each server can only select one next-hop server processing task in each time slot; constraint C 2 Representing VNF embedding constraints, i.e. each VNF can only be embedded on one server per time period.
4. Transformation of optimization problem and algorithm design
(1) Conversion of optimization problem
The optimization problem P1 is a long-term random mixed integer nonlinear programming problem, and is difficult to directly solve. Therefore, the invention converts the long-term queue stability constraint (5) into a virtual queue based on the virtual queue principle, and the formula is as follows:
wherein the method comprises the steps ofIf Y f (t+1) and Z f (t+1) is average rate stable, then constraint (5) is automatically satisfied, and therefore, optimization problem P1 can be rewritten as:
definition of the definitionAnd->Is a virtual queue vector. Definition Θ (t) = [ Q (t), Y (t), Z (t)]The lyapunov function is expressed as:
the single slot conditional lyapunov drift is defined as:
thus, P1.1 can be translated into an upper bound that minimizes drift minus rewards, expressed as:
wherein the method comprises the steps ofV is the utilityNon-negative weighting parameters of the function.
Thus, P1.1 is equivalent to when constraint C is satisfied 1 、C 2 、C 4 Maximizing under the condition of (2) Γ . Further maximize Γ The problem is translated into three sub-problems, namely a large time scale VNF embedded sub-problem SP1, a short time scale admission control sub-problem SP2, a routing sub-problem SP3.
VNF embedding and flow scheduling joint optimization algorithm
(1) Long time scale VNF embedding optimization based on exchange matching
Determination by solving large time scale VNF embedding sub-problem SP1 K The specific embedding positions of the VNs on the N servers are as follows:
due toL n,j (t),C n (t) dynamically changing and unknown, thus, using its empirical value +.> Wherein->It can be calculated as: />
Therefore, the same principle can be foundThe empirical value of the amount of data transmitted via link (n, j) can be further found based on equation (3)/>
The invention solves the problem of SP1 by using the matching theory, and the stream and the embedding strategy are both matched. Defining the embedding strategy of the h time period asWherein->Representing VNF k An embedded set of servers. If it isThen->
Based on SP1 we define the utility function of flow f as
Wherein the method comprises the steps ofIs admitted data, is admitted>Is the queue backlog for the t-th slot given the embedding strategy, which can be deduced from the short time scale admission control and routing in the following section.
The flow calculates utility values according to equation (15) and builds a favorites list in descending order. Because different embedding strategies affect the routing of flows and their utility, i.e. VNF embedding is external, exchange matching is used to solve this problem. Exchange matching refers to embedding policies in a given Next, if VNF embedded on server n changes, original embedding policy +.>Will be surrounded by new embedding strategies>And (3) substitution. If it isDefinition +.>Is->Is a swap match.
The invention provides a long-time scale VNF embedding algorithm based on exchange matching, which comprises the following specific processes:
1) Initializing: giving satisfaction of constraint C 2 Is embedded policy of (a)The utility of the flow is calculated.
2) Exchange matching: in each iteration, all servers randomly select one serverAnd another subset->If->The number of servers in (1) is greater thanThe server n leaves +.>Add->Forming new embedding strategiesThe utility of the stream is then recalculated. If->Is->Exchange matching of (a) by embedding the original policy +.>Replaced by->Otherwise->Remain unchanged.
3) VNF embedding: according toThen->Otherwise->Will->Conversion to y * (h)。
(2) Admission control and Q learning based backpressure aware routing algorithm
By solving the short time scale admission control sub-problem SP2, the admission data of the flow f is determined to keep the queue stable, and the formula is:
/>
utilizing a strategy based on a queue backlog threshold to solve the optimization problem SP2 and complete an admission control decisionExpressed as:
to determine the optimal next-hop server, the short-time-scale routing sub-problem SP3 can be expressed as:
Based on a markov decision process (Markov decision process, MDP), the following definition 1) states are given: the network state is defined as a function of throughput and queue information, i.e2) The actions are as follows: server n next hop server selection action space is +.>3) Rewarding: for embedded VNF k The potential next hop server queue backlog should be taken into account, rewriting SP3 to optimize rewards +.>4) Definition of Q value->Estimating state S for a server n (t) selecting action a n A function of the value of (t),where ψ is the learning rate and γ is the decay factor.
The invention provides a backpressure sensing routing algorithm based on Q learning, which comprises the following specific processes:
1) Initializing: q (S) n (t),a n (t))＝0。
2) Learning: at the beginning of the t-th slot, each server selects actions according to an ε greedy policy. If multiple servers select the same server j, server j will select B n (t) the largest server n and reject others. Then, all servers update the rewards and queue information and go to the next state S n (t+1). Finally, the Q value is updated.
3) Routing: according to a n (t) =jOtherwise->Will { a } n (t) } conversion to x * (t)。
(3) VNF embedding and flow scheduling combined optimization algorithm
The VNF embedding and flow scheduling (VNF Embedding and Flow Scheduling, VEFS) joint optimization algorithm provides a flexible solution to the multi-time-scale VNF embedding and flow scheduling problem, and the VEFS algorithm can implement an admission control and routing strategy for optimizing a small time scale based on a large time scale VNF embedding result, which is specifically described as follows:
1) Initializing: initializing queue backlog of all queues, and indicating functions such as VNF embedding, admission control, routing policy and the like.
2) VNF embedding: at each time period, the VNFs obtain an optimal VNF embedding strategy y (t) according to a large time scale VNF embedding algorithm based on exchange matching.
3) Admission control and routing: in each time slot, each source server obtains an optimal admission control strategy r (t) according to a formula (17), and the source server and the processing server obtain an optimal routing strategy x (t) according to a backpressure sensing routing algorithm based on Q learning.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, and it should be noted that it is possible for those skilled in the art to make several improvements and modifications without departing from the technical principle of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application 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 solutions in the embodiments of the present application may be implemented in various computer languages, for example, object-oriented programming language Java, and an transliterated scripting language JavaScript, etc.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations 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.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the spirit or scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims and the equivalents thereof, the present application is intended to cover such modifications and variations.
Claims (9)
1. A method for managing multi-mode communication network resources of a low-carbon park and a route is characterized by comprising the following specific management steps:
Modeling a VNF embedding and flow scheduling process in a pre-built architecture model into a VNF embedding model and a flow scheduling model based on a multi-time scale model, wherein the embedding of the VNF to a data plane server in a set time period is established as the VNF embedding model, and the route of a flow of a source server to a destination server in a time slot is established as the flow scheduling model; the VNF embedding model is: VNF embedding an indication variable for the h time period
Wherein the subscriptRepresenting an nth processing server; upper energizer->The kth VNF representing SFC;
the flow scheduling model is as follows: flow f queue backlog at server n with t-th slot embedded VNF kThe update expression is:
wherein,and->Flow f queue backlog at server n embedded VNF k for the t-th and t+1-th time slots, respectively, +.>Routing an indicator variable for flow f; />For the server set embedded with VNF k+1,/for the server set embedded with VNF k+1>To embed VNF k - 1, a server set; />Is the t-th time slot processed by server n and transmitted to the set +.>Data volume of the server in (a); />Is the t-th time slot is defined by the set->The amount of data transmitted by the server to server n; />Is admitted data;
based on the VNF embedding model and the flow scheduling model, establishing a VNF embedding and flow scheduling joint optimization problem model, and constraining network performance, including:
Constraints on streaming, for origin serversAdmission data of stream f->The constraints are:
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; meanwhile, only VNF k+1 is embedded in server j,can be transmitted to the server j, wherein/>Embedding an indication variable for the VNF;
the amount of data transmitted via the link (n, j) is constrained by the link capacity L at the t-th slot n,j (t) and calculation Capacity C n (t) constraint, the formula of which is:
wherein the method comprises the steps ofRepresenting the unit complexity of VNF k to process flow f data; τ is the duration of one slot;
queue backlog constraint, expressed as long-term queue backlog constraint:
wherein the method comprises the steps ofIs the flow f queue backlog at server n where the t-th slot is embedded in VNF k;
throughput constraint, average throughput r of flow f f The constraint is expressed as:
wherein the method comprises the steps ofAnd->The minimum throughput and the maximum throughput of the flow f are respectively represented;
based on the above constraint result on network performance, the VNF embedding and flow scheduling joint optimization problem model is described as a weighted difference between the maximized network throughput and the VNF embedding cost, and the specific method is as follows:
the optimization problem P1 is a maximization weighted cumulative utility function, and its formula is:
P1:
C 3 (2), (3), (4), and (5)
Wherein the method comprises the steps ofβ and λ represent weights of throughput and embedding cost, respectively, +. >Indicating VNF embedded indication variable,/->Admission data, e, representing the flow f on the server n k (h) Representing the embedding cost of VNF k; y (h), r (t), x (t) are each k +.> Is a set of (1)Middle->Routing an indicator variable for flow f; constraint C 1 Representing server selection constraints; constraint C 2 Representing VNF embedded constraints;
formula (2) in C3 is
Equation (3) is
Equation (4) is
Equation (5) is
Wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; />Is the amount of data transmitted by the t-th slot via the link (n, j); l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively; />Representing the unit complexity of VNF k to process flow f data; τ is the duration of one slot; />Is the flow f queue backlog at server n where the t-th slot is embedded in VNF k; />And->The minimum throughput and the maximum throughput of the flow f are respectively represented; r is (r) f Average throughput for flow f;
and
The VNF embedding strategy optimization problem and the time slot scale stream scheduling optimization problem converted from Lyapunov optimization into a set time period are specifically implemented by the following steps:
by utilizing the virtual queue principle, the long-term queue stability constraint is converted into a virtual queue, and the optimization problem P1 is rewritten as follows:
P1.1:
s.t.C 1 ,C 2
C 4 (2) and (3)
C 5 :Y f (t) and Z f (t) average Rate stabilization
Wherein formula (2) in C4 is
Equation (3) is
In C5
Wherein W is a weighted cumulative utility function;representation server->Up, the amount of data that stream f arrives at; />Admission data representing the flow f on the server n; />Is the amount of data transmitted by the t-th slot via the link (n, j); l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively; />Representing the unit complexity of VNF k to process flow f data; τ is the duration of one slot; y is Y f (t) and Z f (t) is a virtual queue; />Admission data representing flow f;
based on Lyapunov optimization, P1.1 translates into an upper bound that minimizes drift minus rewards, expressed as:
wherein the method comprises the steps ofV is a non-negative weighting parameter of the utility function; Θ (t) = [ Q (t), Y (t), Z (t)]Wherein->And->For virtual queue vector, ++>Is a queue vector; />Is Lyapunov drift, wherein +.>Is a Lyapunov function;
and performing VNF embedding strategy optimization in a set time period through a switching-matched VNF embedding algorithm, performing time slot flow scheduling optimization through admission control in time slots and a backpressure sensing routing algorithm based on Q learning, and performing multi-mode communication network resource scheduling and routing management in a low-carbon park.
2. The method for resource scheduling and routing management of a multi-mode communication network in a low-carbon park according to claim 1, wherein the method comprises the following steps: the pre-constructed architecture model comprises a data plane, a control plane and an application plane; wherein:
in a data plane, the multi-mode heterogeneous communication network resources are abstracted into a unified virtual resource pool, wherein the virtual resource pool comprises communication, calculation and storage virtual resources, the virtual resources are deployed to support different VNs, and the VNs are arranged in a specific sequence to form SFC;
in a control plane, the SDN controller utilizes a resource scheduling component to embed VNF, and utilizes a route management component to perform flow scheduling to realize route management optimization; wherein the VNF embedding is embedding VNFs of SFC in a server to support multiple low-carbon services; the flow scheduling refers to selecting a proper path for a flow from a source server to transmit to a destination server on the basis of embedding the VNF;
in the application plane, the system comprises the energy management business of a low-carbon intelligent park: carbon footprint monitoring, electric power spot market, flexible load regulation.
3. The method for resource scheduling and routing management of a multi-mode communication network in a low-carbon park according to claim 1, wherein the method comprises the following steps: the method for optimizing the VNF embedding strategy under the time period by the exchange matching VNF embedding algorithm comprises the following steps:
By solving the VNF embedding sub-problem SP1 in the set period of time, specific embedding positions of the K VNFs on the N servers are determined, and the formula is as follows:
SP1:
s.t.C 2
wherein y (h) is the VNF embedding policy for the h time period;
based on SP1, the utility function of flow f is:
wherein V is a non-negative weighting parameter of the utility function; β and λ represent weights for throughput and embedding cost, respectively;representing VNF embedded indication variables; e, e k (h) Representing the embedding cost of VNF k; />Routing an indicator variable for flow f;the embedding condition of the h time period; /> Respectively isL n,j (t),C n Empirical value of (t)/(t)>Is admitted data; /> Is the queue backlog of the t-th time slot under a given embedding strategy;
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively;
the flow calculates utility value according to the formula and establishes a preference list according to descending order; the provided VNF embedding algorithm based on the exchange matching set time period has the following specific flow:
giving satisfaction of constraint C 2 Is embedded policy of (a)Calculating the utility of the flow;
in each iteration, all servers randomly select one serverAnd another subset->If->The number of servers in (a) is greater than 1, i.e. +. >The server n leaves +.>Adding inForm a new embedding strategy->Then, the utility of the stream is recalculated; if->Is->Exchange matching of (a) by embedding the original policy +.>Replaced by->Otherwise->Remain unchanged;
according toThen->Otherwise->Will->Conversion to y * (h)。
4. The method for resource scheduling and routing management of a multi-mode communication network in a low-carbon park according to claim 1, wherein the method comprises the following steps: the admission control method comprises the following steps:
determining a slot admission control subproblem SP2:
SP2:
s.t.C 6 :(2)
wherein formula (2) in C6 isV is a non-negative weighting parameter of the utility function; beta is throughput weight; y is Y f (t) and Z f (t) is a virtual queue; />Backlog for the t-th slot queue; />Admission data representing the flow f on the server n; />Representation server->Up, the amount of data that stream f arrives at;
adopting admission control method to solve SP2, expressed as
5. The method for resource scheduling and routing management of a multi-mode communication network in a low-carbon park according to claim 1, wherein the method comprises the following steps: the backpressure sensing routing algorithm based on Q learning comprises the following specific flow:
determining a slot route selection sub-problem SP3, and determining an optimal next-hop server:
SP3:
C 7 :(3)
wherein formula (3) in C7 isIs the flow f queue backlog at server n where the t-th slot is embedded in VNF k; / >Is the amount of data transmitted by the t-th slot via the link (n, j); />Routing an indicator variable for flow f; x (t) is a routing policy;
the back pressure perception routing algorithm based on Q learning is provided by adopting the Q learning to solve SP2, and the specific flow is as follows:
initializing Q value, setting Q (S n (t),a n (t))＝0；
At the beginning of the t time slot, each server selects actions according to an epsilon greedy strategy; server j will select B n (t) the largest server n and rejecting others, resolving server selection conflicts; all servers update rewards and queue information and go to the next state S n (t+1); and updating the Q value;
according to a n (t) =jOtherwise->Will { a } n (t) } conversion to x * (t)；
Wherein the state isAction as->The rewards areQ value Q (S) n (t),a n (t)) estimating state S for server n (t) selecting action a n A value function of (t).
6. The method for resource scheduling and routing management of a multi-mode communication network in a low-carbon park according to claim 1, wherein the method comprises the following steps: the algorithm of the VNF embedding and flow scheduling joint optimization is as follows:
initializing queue backlog of all queues, and embedding, admittance control and routing strategy indication functions of the VNF;
in each set time period, the VNs obtain an optimal VNF embedding strategy y (t) according to a VNF embedding algorithm of the set time period based on exchange matching;
In each time slot, each source server obtains an optimal admission control strategy r (t) according to a formula, and the source server and the processing server obtain an optimal routing strategy x (t) according to a backpressure sensing routing algorithm based on Q learning.
7. A low-carbon park multi-mode communication network resource scheduling and routing management device is characterized by comprising the following specific components:
modeling module: the method comprises the steps of modeling a VNF embedding and flow scheduling process in a pre-built architecture model into a VNF embedding model and a flow scheduling model based on a multi-time scale model, wherein the embedding of the VNF to a data plane server in a time period is established as the VNF embedding model, and the route of a flow of a source server to a destination server in a time slot is established as the flow scheduling model; the VNF embedding model is: VNF embedding an indication variable for the h time period
Wherein the subscriptRepresenting an nth processing server; upper energizer->The kth VNF representing SFC;
the flow scheduling model is as follows: flow f queue backlog at server n with t-th slot embedded VNF kThe update expression is:
wherein,and->Flow f queue backlog at server n embedded VNF k for the t-th and t+1-th time slots, respectively, +.>Routing an indicator variable for flow f; / >For the server set embedded with VNF k+1,/for the server set embedded with VNF k+1>Is a server set embedded with VNF k-1; />Is the t-th time slot processed by server n and transmitted to the set +.>Data volume of the server in (a); />Is the t-th time slot is defined by the set->The amount of data transmitted by the server to server n; />Is admitted data;
constraint module: the method is used for establishing a VNF embedding and flow scheduling joint optimization problem model by utilizing the VNF embedding model and the flow scheduling model, and restraining network performance; comprising the following steps:
constraints on streaming, for origin serversQuasi-flow fEntry data->The constraints are:
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; meanwhile, only VNF k+1 is embedded in server j,can be transmitted to the server j, wherein +.>Embedding an indication variable for the VNF;
the amount of data transmitted via the link (n, j) is constrained by the link capacity L at the t-th slot n,j (t) and calculation Capacity C n (t) constraint, the formula of which is:
wherein the method comprises the steps ofRepresenting the unit complexity of VNF k to process flow f data; τ is the duration of one slot;
queue backlog constraint, expressed as long-term queue backlog constraint:
wherein the method comprises the steps ofIs the flow f queue backlog at server n where the t-th slot is embedded in VNF k;
Throughput constraint, average throughput r of flow f f The constraint is expressed as:
wherein the method comprises the steps ofAnd->The minimum throughput and the maximum throughput of the flow f are respectively represented;
and a conversion module: using the constraint, describing the VNF embedding and flow scheduling joint optimization problem model as a weighted difference for maximizing network throughput and VNF embedding cost; the specific method comprises the following steps:
the optimization problem P1 is a maximization weighted cumulative utility function, and its formula is:
P1:
C 3 (2), (3), (4), and (5)
Wherein the method comprises the steps ofβ and λ represent weights of throughput and embedding cost, respectively, +.>Indicating VNF embedded indication variable,/->Admission data, e, representing the flow f on the server n k (h) Representing the embedding cost of VNF k; y (h), r (t), x (t) are each k +.> WhereinRouting an indicator variable for flow f; constraint C 1 Representing server selection constraints; constraint C 2 Representing VNF embedded constraints;
formula (2) in C3 is
Equation (3) is
Equation (4) is
Equation (5) is
Wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; />Is the amount of data transmitted by the t-th slot via the link (n, j); l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively; />Representing the unit complexity of VNF k to process flow f data; τ is the duration of one slot; / >Is the flow f queue backlog at server n where the t-th slot is embedded in VNF k; />And->The minimum throughput and the maximum throughput of the flow f are respectively represented; r is (r) f Average throughput for flow f;
and
The VNF embedding strategy optimization problem and the time slot scale stream scheduling optimization problem of the time period are converted by Lyapunov optimization; the method comprises the following specific steps:
by utilizing the virtual queue principle, the long-term queue stability constraint is converted into a virtual queue, and the optimization problem P1 is rewritten as follows:
P1.1:
s.t.C 1 ,C 2
C 4 (2) and (3)
C 5 :Y f (t) and Z f (t) average Rate stabilization
Wherein formula (2) in C4 is
Equation (3) is
In C5
Wherein W is a weighted cumulative utility function;representation server->Up, the amount of data that stream f arrives at; />Admission data representing the flow f on the server n; />Is the amount of data transmitted by the t-th slot via the link (n, j); l (L) n,j (t) and C n (t) represents the link capacity of the link (n, j) and the calculation capacity of the server n, respectively; />Representing the unit complexity of VNF k to process flow f data; τ is the duration of one slot; y is Y f (t) and Z f (t) is a virtual queue; />Admission data representing flow f;
based on Lyapunov optimization, P1.1 translates into an upper bound that minimizes drift minus rewards, expressed as:
wherein the method comprises the steps ofV is a non-negative weighting parameter of the utility function; Θ (t) = [ Q (t), Y (t), Z (t) ]Wherein->And->For virtual queue vector, ++>Is a queue vector; />Is Lyapunov drift, wherein +.>Is a Lyapunov function;
and performing VNF embedding strategy optimization under a time period through a switching-matched VNF embedding algorithm, and performing time slot flow scheduling optimization through admission control under time slots and a backpressure sensing routing algorithm based on Q learning, thereby realizing multi-mode communication network resource scheduling and routing management of the low-carbon park.
8. The low-carbon park multi-modal communication network resource scheduling and routing management apparatus of claim 7, wherein:
the modeling module includes:
the system comprises a data plane, a control plane and an application plane; wherein:
in a data plane, the multi-mode heterogeneous communication network resources are abstracted into a unified virtual resource pool, wherein the virtual resource pool comprises communication, calculation and storage virtual resources, different VNs are supported by deploying the virtual resources, and the VNs are arranged in a specific sequence to form SFC;
in a control plane, the SDN controller utilizes a resource scheduling component to embed VNF, and utilizes a route management component to perform flow scheduling to realize route management optimization; wherein the VNF embedding is embedding VNFs of SFC in a server to support multiple low-carbon services; the flow scheduling refers to selecting a proper path for a flow from a source server to transmit to a destination server on the basis of embedding the VNF;
In the application plane, the system comprises the energy management business of a low-carbon intelligent park: carbon footprint monitoring, electric power spot market, flexible load regulation.
9. The low-carbon park multi-modal communication network resource scheduling and routing management apparatus of claim 8, wherein: constraints for streaming of the constraint model for the origin serverAdmission data of stream f->The constraints are:
wherein the method comprises the steps ofRepresentation server->Up, the amount of data that stream f arrives at; meanwhile, only VNF k+1 is embedded in server j,can be transmitted to the server j, wherein +.>Embedding an indication variable for the VNF;
the amount of data transmitted via the link (n, j) is constrained by the link capacity L at the t-th slot n,j (t) and calculation Capacity C n (t) constraint, the formula of which is:
wherein the method comprises the steps ofRepresenting the unit complexity of VNF k to process flow f data; τ is the duration of one slot;
queue backlog constraint, expressed as long-term queue backlog constraint:
wherein the method comprises the steps ofIs the flow f queue backlog at server n where the t-th slot is embedded in VNF k;
throughput constraint, average throughput r of flow f f The constraint is expressed as:
wherein the method comprises the steps ofAnd->The minimum throughput and the maximum throughput of the flow f are indicated, respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210775986.4A CN115174403B (en) | 2022-07-02 | 2022-07-02 | Method and device for resource scheduling and route management of multi-mode communication network in low-carbon park |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210775986.4A CN115174403B (en) | 2022-07-02 | 2022-07-02 | Method and device for resource scheduling and route management of multi-mode communication network in low-carbon park |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115174403A CN115174403A (en) | 2022-10-11 |
| CN115174403B true CN115174403B (en) | 2024-03-12 |
Family
ID=83488569
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210775986.4A Active CN115174403B (en) | 2022-07-02 | 2022-07-02 | Method and device for resource scheduling and route management of multi-mode communication network in low-carbon park |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115174403B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015157947A1 (en) * | 2014-04-16 | 2015-10-22 | 华为技术有限公司 | Software defined network based networking method and device |
| CN105847151A (en) * | 2016-05-25 | 2016-08-10 | 安徽大学 | Multi-constrained QoS (Quality of Service) routing strategy designing method for software defined network |
| EP3058464A1 (en) * | 2013-12-06 | 2016-08-24 | Huawei Technologies Co., Ltd. | Method and controller for chaining applications in a software defined network |
| WO2016188548A1 (en) * | 2015-05-22 | 2016-12-01 | Huawei Technologies Co., Ltd. | Telecommunication network with automated control and data plane instantiation |
| CN108206780A (en) * | 2016-12-19 | 2018-06-26 | 中国电信股份有限公司 | Flow scheduling strategy report method, autonomy field system and SDN network system |
| CN109618312A (en) * | 2019-01-18 | 2019-04-12 | 华北电力大学 | A kind of low complex degree online resource distribution optimization algorithm towards D2D junction network |
| CN110765595A (en) * | 2019-10-10 | 2020-02-07 | 内蒙古农业大学 | SDN data center network flow scheduling algorithm based on multi-layer virtual topology energy saving |
| CN111475252A (en) * | 2020-03-06 | 2020-07-31 | 重庆邮电大学 | Virtual network function deployment optimization algorithm based on deep reinforcement learning |
| CN112787943A (en) * | 2019-11-08 | 2021-05-11 | 中兴通讯股份有限公司 | Traffic scheduling method and device and computer readable storage medium |
| CN114501355A (en) * | 2022-03-16 | 2022-05-13 | 华北电力大学 | Privacy capacity guaranteeing system and method suitable for electric carbon information conversion of park |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104954226A (en) * | 2015-07-28 | 2015-09-30 | 上海斐讯数据通信技术有限公司 | SDN-based QoS-supported communication tunnel establishment method and system |
-
2022
- 2022-07-02 CN CN202210775986.4A patent/CN115174403B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3058464A1 (en) * | 2013-12-06 | 2016-08-24 | Huawei Technologies Co., Ltd. | Method and controller for chaining applications in a software defined network |
| WO2015157947A1 (en) * | 2014-04-16 | 2015-10-22 | 华为技术有限公司 | Software defined network based networking method and device |
| WO2016188548A1 (en) * | 2015-05-22 | 2016-12-01 | Huawei Technologies Co., Ltd. | Telecommunication network with automated control and data plane instantiation |
| CN105847151A (en) * | 2016-05-25 | 2016-08-10 | 安徽大学 | Multi-constrained QoS (Quality of Service) routing strategy designing method for software defined network |
| CN108206780A (en) * | 2016-12-19 | 2018-06-26 | 中国电信股份有限公司 | Flow scheduling strategy report method, autonomy field system and SDN network system |
| CN109618312A (en) * | 2019-01-18 | 2019-04-12 | 华北电力大学 | A kind of low complex degree online resource distribution optimization algorithm towards D2D junction network |
| CN110765595A (en) * | 2019-10-10 | 2020-02-07 | 内蒙古农业大学 | SDN data center network flow scheduling algorithm based on multi-layer virtual topology energy saving |
| CN112787943A (en) * | 2019-11-08 | 2021-05-11 | 中兴通讯股份有限公司 | Traffic scheduling method and device and computer readable storage medium |
| CN111475252A (en) * | 2020-03-06 | 2020-07-31 | 重庆邮电大学 | Virtual network function deployment optimization algorithm based on deep reinforcement learning |
| CN114501355A (en) * | 2022-03-16 | 2022-05-13 | 华北电力大学 | Privacy capacity guaranteeing system and method suitable for electric carbon information conversion of park |
Non-Patent Citations (3)
| Title |
|---|
| 《Routing cost minimization and throughput maximization of NFV-enables unicasting in software-defined networks》;Jia M;《IEEE Transaction》;全文 * |
| 虚拟网络中功能节点部署和流调度研究;姚毅夫;《中国优秀硕士学位论文全文数据库》;全文 * |
| 面向轻量云的嵌入式系统容器化资源管理技术研究;严亮亮;《中国优秀硕士学位论文全文数据库》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115174403A (en) | 2022-10-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Qi et al. | Knowledge-driven service offloading decision for vehicular edge computing: A deep reinforcement learning approach | |
| Hussein et al. | Efficient task offloading for IoT-based applications in fog computing using ant colony optimization | |
| Rahbari et al. | Task offloading in mobile fog computing by classification and regression tree | |
| Zhang et al. | Virtual network embedding based on modified genetic algorithm | |
| Kumari et al. | Task offloading in fog computing: A survey of algorithms and optimization techniques | |
| Chowdhury et al. | DA-DRLS: Drift adaptive deep reinforcement learning based scheduling for IoT resource management | |
| Fan et al. | Multi-objective optimization of container-based microservice scheduling in edge computing | |
| Hussin et al. | Improving reliability in resource management through adaptive reinforcement learning for distributed systems | |
| Xie et al. | Workflow scheduling in serverless edge computing for the industrial internet of things: A learning approach | |
| US20230145097A1 (en) | Autonomous traffic (self-driving) network with traffic classes and passive and active learning | |
| Perin et al. | Towards sustainable edge computing through renewable energy resources and online, distributed and predictive scheduling | |
| Qi et al. | Vehicular edge computing via deep reinforcement learning | |
| Zahedi et al. | A power-efficient and performance-aware online virtual network function placement in SDN/NFV-enabled networks | |
| Ullah et al. | Optimizing task offloading and resource allocation in edge-cloud networks: a DRL approach | |
| CN115174403B (en) | Method and device for resource scheduling and route management of multi-mode communication network in low-carbon park | |
| Li et al. | Optimal service selection and placement based on popularity and server load in multi-access edge computing | |
| Kim et al. | Collaborative policy learning for dynamic scheduling tasks in cloud-edge-terminal IoT networks using federated reinforcement learning | |
| Chen et al. | Traffic engineering based on deep reinforcement learning in hybrid IP/SR network | |
| Ghebleh et al. | A multi-criteria method for resource discovery in distributed systems using deductive fuzzy system | |
| Teng et al. | Joint Optimization of Sequential Task Offloading and Service Deployment in End-Edge-Cloud System for Energy Efficiency | |
| Zhu et al. | Deep reinforcement learning-based edge computing offloading algorithm for software-defined IoT | |
| Hsieh et al. | Deep reinforcement learning-based task assignment for cooperative mobile edge computing | |
| Bekhit et al. | Multi-objective VNF Placement Optimization with NSGA-III | |
| CN114785692A (en) | Virtual power plant aggregation regulation and control communication network flow balancing method and device | |
| Xu et al. | A Graph reinforcement learning based SDN routing path selection for optimizing long-term revenue |
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 |