CN108599994B - SDN slice construction method based on flow clustering - Google Patents

SDN slice construction method based on flow clustering Download PDF

Info

Publication number
CN108599994B
CN108599994B CN201810253899.6A CN201810253899A CN108599994B CN 108599994 B CN108599994 B CN 108599994B CN 201810253899 A CN201810253899 A CN 201810253899A CN 108599994 B CN108599994 B CN 108599994B
Authority
CN
China
Prior art keywords
slice
network
flow
sdn
qos
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
Application number
CN201810253899.6A
Other languages
Chinese (zh)
Other versions
CN108599994A (en
Inventor
田龙伟
史景伦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN201810253899.6A priority Critical patent/CN108599994B/en
Publication of CN108599994A publication Critical patent/CN108599994A/en
Application granted granted Critical
Publication of CN108599994B publication Critical patent/CN108599994B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • H04L41/0893—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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/16—Central resource management; Negotiation of resources or communication parameters, e.g. negotiating bandwidth or QoS [Quality of Service]

Abstract

The invention aims to respectively construct network slices suitable for each type of flow transmission by counting and clustering the service quality requirements of all network flows so as to better utilize network resources and guarantee the network service quality; in each slice update period T, the main steps include: collecting SLA (service level contract) of a user and the capacity C and the available R of the physical link resources of the whole network; counting the QoS characteristics of each flow demand to obtain OD flow characteristics; clustering OD flow characteristics to obtain typical QoS demand types and scales; assembling the flow of each type of QoS requirement into a physical link to obtain a network slice suitable for bearing the corresponding QoS type flow, and issuing slice information to a slice maintenance table of forwarding equipment by an SDN controller; when SDN forwarding equipment receives data flow of a client, slices suitable for transmission are searched for according to a QoS request for forwarding, and slice performance conditions such as resource utilization rate in the slices and SLA default are monitored.

Description

SDN slice construction method based on flow clustering
Technical Field
The invention relates to the technical field of communication, in particular to a SDN slice construction method based on flow clustering.
Background
With the continuous enrichment of internet service types, the service reliability categories required by network applications increase day by day. Quality of Service (QoS) guarantees require that different network flows perform reliable contract transmission, especially end-to-end reliable contract transmission, which is one of the hot spots of continuous interest in academia and industry. This requires analysis of the characteristics of each traffic flow and defines a good QoS mechanism in the network. However, conventional networks provide best effort services and do not guarantee reliable delivery of traffic. QoS modes represented by IntServ and DiffServ often only consider bandwidth, resource reservation has blindness, and a network flow with resources reserved in an early stage easily blocks a resource request of a subsequent network flow. In addition, the natural distributed characteristics of the traditional network determine that network resources cannot be allocated globally, and it is difficult to accurately meet QoS requirements of different services or customer Service Level contracts (SLAs). In addition, the agreement support of the equipment provider in the traditional network is not uniform, and the policy and benefit of the operator or the country for treating the network service are inconsistent, thereby hindering the popularization of the network QoS guarantee.
The Software Defined Network (SDN) emerging network architecture provides a new idea and favorable infrastructure conditions for QoS guarantee. The system has the advantages that the control plane and the forwarding plane are decoupled, the whole network topology and the forwarding resources can be controlled in a centralized programming mode, and the system is used for dynamically optimizing flow and managing the resources. In addition, in the 5G planning proposed by the next generation mobile Network alliance (NGMN), a Network Slicing concept is introduced, the Network is divided into different subnets according to the physical bearable resource characteristics, each subnet has an independent logical topology view, so that different services are transmitted by using different slices, and the end-to-end transmission performance is ensured. However, the current treatises and forum data on network slicing still remain in the concept definition discussion of slicing and the underlying technical implementation details, and the overall slicing construction scheme and scientific analysis are lacked, so it is very important to solve the problem.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a flow clustering-based SDN slice construction method, which is used for counting the network flow requirements of users on different service quality types in real time, dynamically adjusting the settings of network slices, optimizing the resource utilization and realizing the perfect management of the network slices.
In order to solve the above problem, the present invention provides a method for constructing an SDN slice based on traffic clustering, which is characterized in that in each slice update period T, the method includes the following steps: the method comprises the following steps: registering a network service quality level contract (SLA) of each access network terminal user, wherein the contract comprises network data transmission characteristics such as access bandwidth, highest time delay, packet loss rate, jitter rate and the like;
step two: counting the resource information of the whole network physical link in the SDN control domain to obtain a physical resource capacity matrix C and a physical remaining available resource matrix R;
step three: acquiring real-time flow requirements of the whole network in an SDN control domain to obtain QoS characteristics of OD flow requirements of the whole network, and summarizing the QoS characteristics into an OD flow Set;
step four: calculating hot spot flow in OD flow, clustering the flow in the Set to obtain typical QoS requirement category and scale in the current network;
step five: dividing network slices which are most suitable for bearing the QoS requirements of each type obtained by clustering in a physical link by utilizing a self-designed resource assembly algorithm, issuing a flow table to a slice maintenance table of forwarding equipment by an SDN controller, wherein the slices bearing the QoS requirements of each type can be parallelly and independently calculated and then issued without waiting;
step six: after receiving the data stream of the client, the SDN forwarding device searches a network slice suitable for transmission according to the QoS request characteristic of the client for forwarding, monitors the resource utilization rate and SLA default condition in the slice, and can determine whether to repeat the first step to the sixth step to adjust the slice division according to the slice resource utilization rate and the SLA default rate.
The further improvement lies in that: in step one, a network in the SDN control domain, which is formed by all forwarding device nodes and user access nodes, is denoted as graph G ═ V, E, where the number of nodes is | V | ═ N, and the number of links is | E | ═ M.
The further improvement lies in that: in the second step, the QoS performance that can be borne on each link and the resource use condition of each link are reported to the SDN controller by forwarding equipment of the SDN network, and the physical available resources of the whole network are recorded as a performance matrix c (h):
Figure BDA0001608507150000031
wherein, c (h) is a QoS performance boundary constraint that the h-th dimension (h is greater than or equal to 1 and less than or equal to l) physical link can bear, specifically, the QoS performance boundary constraint may be QoS indexes such as bandwidth (upper bound), delay (lower bound), frame length, packet loss rate, and the like, and at least needs bandwidth characteristics and generally also delay characteristics. c. Cuv(h) Also can be written as c (h)<u,v>Is a link e<u,v>H-dimension QoS characteristics above; and simultaneously, the meaning of each element is the same as that of C (h), and initially, R (h) is equal to C (h).
The further improvement lies in that: the step three of collecting the whole network OD flow demand Set is that a QoS guarantee request message is sent to an adjacent SDN forwarding device by a user side of an access network, the SDN controller is submitted by the forwarding device through a southbound link between the SDN controller and the forwarding device, and a QoS performance index of a data flow to be initiated is registered; the SDN controller checks whether the client identity is legitimate and whether the QoS request exceeds the SLA contract. One OD network flow from network node u to node v is denoted as:
q<u,v>＝[q1,q2,…ql],u,v∈V (2)
wherein, the nodes u and v represent the source node and sink node of the flow respectively, qhIs the h-th QoS index of the current flow.
During the current slice management period T, the mth network flow detected by the network controller can be denoted as qm<u,v>And m is more than or equal to 1 and less than or equal to FS, and adding a Set, wherein FS is the number of network streams counted in the current slice management period:
Set＝{qm(u,v)|1≤m≤FS,u,v∈V} (3)
the further improvement lies in that: in the fourth step, an unlimited mainstream clustering method can be selected to cluster the OD flow Set, and the clustering number K is determinedopt. Optionally, the invention can cluster vectors in the Set by using K-means, and automatically determine K by using a Gap statistical method (Gap statistical)opt. In the clustering result, the kth (K is more than or equal to 1 and less than or equal to K)opt) Class traffic subset is denoted as Set(k)The number of the collection elements is FSkIts cluster centroid reflects the typical QoS requirements of such traffic
Figure BDA0001608507150000041
And the upper and lower bounds of the attribute
Figure BDA0001608507150000042
Besides, the flow of each category is arranged according to the descending order of the occurrence frequency. The frequency can also reflect the scale of the flow, and the calculation mode is as follows:
Figure BDA0001608507150000043
the performance of the kth class flow component in the OD flow demand is then:
Figure BDA0001608507150000044
recording as a matrix OD per QoS dimension h(k)(h)：
Figure BDA0001608507150000051
Wherein
Figure BDA0001608507150000052
ehIs a column vector with only the h-th element being 1 and the remaining elements being 0.
The further improvement lies in that: and step five, finding the most suitable network slice for bearing the network slice for each type of QoS requirement obtained by clustering by using a self-designed resource assembly algorithm, wherein the specific method comprises the following steps:
(1) each type of traffic is processed in turn according to the descending order of equation (4). For the k-th type traffic OD after clustering(k)And (3) a proper resource allocation scheme needs to be found, the flow is mapped to the physical link of the graph G, a pre-allocation scheme suitable for bearing the kth type flow slice is obtained, and the step (2) is carried out to readjust the flow. In the present invention, the objective of resource allocation is to satisfy the lowest slice deployment cost while ensuring that the slice QoS performance meets the intra-slice traffic load demand, that is, the following optimized mathematical model:
(a) an objective function:
Figure BDA0001608507150000053
wherein f (u, v) is link e<u,v>The cost per unit bandwidth;
Figure BDA0001608507150000054
for continuous decision variables: the ith flow is in link e<u,v>Bandwidth usage of.
(b) Constraint conditions are as follows:
from each link on which any target traffic in the slice depends, the constraints are:
Figure BDA0001608507150000055
Figure BDA0001608507150000061
from the viewpoint of ensuring end-to-end performance when the network slice operates, the constraints are as follows:
Figure BDA0001608507150000062
Figure BDA0001608507150000063
Figure BDA0001608507150000064
Figure BDA0001608507150000065
Figure BDA0001608507150000066
Figure BDA0001608507150000067
Figure BDA0001608507150000068
Figure BDA0001608507150000069
wherein
Figure BDA00016085071500000610
Belong to Set(k)The flow is the ith flow in the kth flow in the OD flow, originates from the node u and ends at the node v; bw, d, ω and data are respectively the bandwidth, delay, loss rate representing QoS performanceAnd a data stream byte length index; alpha is alpha(k)In order to adjust the resource allocation of the kth flow, the network administrator can combine the slice load settings of various flows to perform manual intervention of resource allocation, and the default is 1, 0<α(k)≤1； V(k)Is a subset of V, containing nodes involved in class k OD flows;
Figure BDA00016085071500000611
establishing the connection number of communication connection between source nodes and sink nodes related to the kth type of traffic; d(k)<u,v>The average frame length for the kth traffic passes through the link e<u,v>Time delay caused by transmission; pD(k)、PBW(k) And Pω(k) The lower bounds of the probability that the time delay, the bandwidth and the loss rate in the network slice QoS performance meet the performance requirements of the user are represented respectively, the ideal situation is 1, and the lower bounds are set by a network slice manager and stored in a slice maintenance table. When more types of QoS constraints are added to the model, contract probability constraints can be written for other QoS indexes by analogy with formulas (10) to (12).
(2) Obtaining a preliminary scheme suitable for bearing the kth class flow slice according to the optimization model solved in the step (1): by
Figure BDA0001608507150000071
The existence of the k-th flow carried by each link can be known, and the topological graph formed by the links can be obtained
Figure BDA0001608507150000072
Figure BDA0001608507150000073
However, it is necessary to examine the drawings
Figure BDA0001608507150000074
The number of connected components is calculated by using the characteristic value of the Laplace matrix:
drawing (A)
Figure BDA0001608507150000075
The laplacian matrix L of is:
Figure BDA0001608507150000076
the eigenvalues of the matrix are
Figure BDA0001608507150000077
L is a semi-positive constant and L is a negative constant,
Figure BDA0001608507150000078
λi≥0,λ0the number of 0 s in the eigenvalues of the laplace matrix is the number of connected components of the graph:
Figure BDA0001608507150000079
if A (k) is 1, the diagram
Figure BDA00016085071500000710
Namely the slice topology suitable for bearing the kth type flow; if not, then,
Figure BDA00016085071500000711
each connected component of (a) forms a slice, a common slice
Figure BDA00016085071500000712
Figure BDA00016085071500000713
Wherein for slicing
Figure BDA00016085071500000714
Edge e<u,v>The weight value of is
Figure BDA00016085071500000715
Representing the allocated bandwidth of the slice on this link.
(3) The SDN controller slices the data
Figure BDA00016085071500000716
Figure BDA00016085071500000716
1≤k≤Kopt. And sending the information to SDN forwarding equipment. Specifically, the slice maintenance table ts (v) of the forwarding device node v is a set:
Figure BDA0001608507150000081
the further improvement lies in that: after the flow request initiated by each user, the forwarding layer equipment judges the clustering type of the flow first, routes in the corresponding network slice, and monitors the resource utilization rate and SLA default conditions in the slice at the same time, so that the slice structure of the next slice structure period can be flexibly adjusted.
The invention has the beneficial effects that:
1. the QoS performance requirements of all users and the QoS bearing capacity of a link are considered, when slices are constructed, the main categories and the required quantity of the whole network QoS requirements are obtained through clustering, and the proper slices are planned for the flow of each category of QoS requirements in the algorithm instead of considering all the QoS requirements at one time, so that the flow performance of each expected slice is similar, the time complexity of the slice planning algorithm is lower, and the realization possibility is higher;
2. according to the network slice constructed by the algorithm, the end-to-end QoS characteristic in the slice can be effectively guaranteed;
3. the scale of the slice can be directly related to the scale of various flows, various demands of the network can be known by monitoring the performance of each slice, and the solving speed is higher;
4. the slices bearing each type of QoS requirement type can be calculated in parallel, flow tables can be issued to forwarding equipment by the SDN controller after each slice is planned, and the flow tables do not need to be issued after all the slices are calculated, so that the waiting time of slice construction is shortened, and the practicability of the technology is better.
Drawings
Fig. 1 is a schematic diagram of an SDN network architecture to which an embodiment of the present invention is applicable;
FIG. 2 is a flow chart of the operation of a network controller according to an embodiment of the network slice construction method of the present invention;
FIG. 3 is a flowchart illustrating operation of a network forwarding device according to an embodiment of the network slice construction method of the present invention;
FIG. 4 is a flowchart of an algorithm for finding a slice suitable for loading for each type of traffic in step five of the present invention;
fig. 5 is a schematic diagram of functional modules included in a user interface, an SDN controller, and an SDN forwarding device according to an embodiment of the present invention;
FIG. 6 is an example of a communication interaction in which embodiments of the present invention are applicable.
Detailed Description
In order to further understand the present invention, the following detailed description will be made with reference to the following examples, which are only used for explaining the present invention and are not to be construed as limiting the scope of the present invention.
As shown in fig. 1, the network includes an SDN controller and a plurality of SDN forwarding devices, where the controller has a centralized view and can uniformly collect and control network and device state information of the SDN forwarding devices, and the forwarding devices are responsible for receiving a flow table issued by the controller to perform routing forwarding of data. Different network terminal users are connected with SDN forwarding devices, each forwarding device can keep communication with an SDN controller through a safe southbound link of the SDN, and network link connection exists among the forwarding devices to transfer data streams.
Fig. 5 shows functional modules that a user-side interface, an SDN controller, and an SDN forwarding device should respectively have in an embodiment of the present invention. The user side interface has the functions of registering SLA, registering QoS requirements and the like; the SDN controller has the functions of topology resource management, a routing calculation module, flow characteristic analysis, slice performance monitoring, slice resource management, user QoS registration and the like; the SDN forwarding equipment has the functions of flow characteristic analysis, slice maintenance table and the like.
When the SDN slice construction method works, the SDN slice construction method is executed according to the following steps:
the method comprises the following steps: a network service level contract (SLA) is registered for each access network end-user. The network user sends an SLA registration request to an adjacent SDN forwarding device, the forwarding device sends the SLA registration request to an SDN controller through a safe southbound link, and a QoS registration function module of the controller checks the user identity and request validity and registers the SLA. The contract comprises network data transmission characteristics such as access bandwidth, highest time delay, packet loss rate, jitter rate and the like, and is registered once only when updating is needed;
step two: and counting the resource information of the physical link of the whole network in the SDN control domain to obtain a physical resource capacity matrix C and a physical remaining available resource matrix R. Reporting the QoS performance boundary and the residual QoS performance which can be loaded on each link to an SDN controller by forwarding equipment of the SDN;
step three: and collecting the real-time flow demand of the whole network in the SDN control domain to obtain the QoS characteristics of the OD flow demand of the whole network, and summarizing the QoS characteristics into an OD flow Set. Firstly, a user side of an access network sends a QoS guarantee request message to adjacent SDN forwarding equipment, the forwarding equipment submits an SDN controller through a safe southbound link, and QoS performance indexes of data streams to be initiated are registered; the SDN controller checks whether the client identity is legal and whether the QoS request exceeds an SLA contract;
step four: calculating hot spot flow in OD flow, clustering the flow in the Set to obtain typical QoS requirement category and scale in the current network;
and fifthly, dividing network slices which are most suitable for bearing the QoS requirements of each type obtained by clustering in a physical link by using a self-designed resource assembly algorithm, and issuing a flow table to a slice maintenance table of the forwarding equipment by the SDN controller. The slices bearing each type of QoS requirement can be parallelly and independently calculated, and are not required to be completely calculated and then issued;
and step six, after receiving the data flow of the client, the SDN forwarding equipment searches a network slice suitable for transmission according to the QoS request characteristic of the client for forwarding, and simultaneously monitors the resource utilization rate in the slice and the SLA default condition. Whether to repeat S1-S6 to adjust slice division can be determined according to slice resource utilization rate and SLA default rate.
In the network in the SDN control domain in the first step, a network formed by all forwarding device nodes and user access nodes is denoted as graph G ═ V, E, where the number of nodes is | V | ═ N, and the number of links is | E | ═ M.
In the second step, the QoS performance that can be borne on each link and the resource usage of each link are reported to the SDN controller by forwarding devices of the SDN network, and the physical available resources of the whole network are recorded as a performance matrix c (h):
Figure BDA0001608507150000111
wherein, C(h)The constraint is a QoS performance boundary constraint that the physical link in the h-th dimension (h is greater than or equal to 1 and less than or equal to l) can bear, specifically, the constraint can be QoS indexes such as bandwidth (upper bound), delay (lower bound), frame length, packet loss rate and the like, and at least bandwidth characteristics and generally delay characteristics are required. c. Cuv(h) Is a link e<u,v>H-dimension QoS characteristics above;
and simultaneously, the meaning of each element is the same as that of C (h), and initially, R (h) is equal to C (h).
Acquiring a Set of overall network OD (origin-destination) flow demands, namely, sending a QoS (quality of service) guarantee request message to an adjacent SDN forwarding device by a user side accessing a network, submitting the SDN controller by the forwarding device through a southbound link between the SDN controller and the forwarding device, and registering a QoS (quality of service) performance index of a data stream to be initiated; the SDN controller checks whether the client identity is legitimate and whether the QoS request exceeds the SLA contract. One OD network flow from network node u to node v is denoted as:
q<u,v>＝[q1,q2,…ql],u,v∈V (2)
wherein, the nodes u and v represent the source node and sink node of the flow respectively, qhIs the h-th QoS index of the current flow.
During the current slice management period T, the mth network flow detected by the network controller can be denoted as qm<u,v>，1≤m≤And adding an FS into the Set, wherein the FS is the number of network streams counted in the current slice management period:
Set＝{qm(u,v)|1≤m≤FS,u,v∈V} (3)
the fourth step can select an unlimited mainstream clustering method to cluster OD flow, wherein K-means is selected to cluster the vector in the Set, and the number K of clusters is automatically determined by using gap statisticopt. In the clustering result, the kth (K is more than or equal to 1 and less than or equal to K)opt) Class traffic subset is denoted as Set(k)The number of the collection elements is FSkIts cluster centroid reflects the typical QoS requirements of such traffic
Figure BDA0001608507150000121
And the upper and lower bounds of the attribute
Figure BDA0001608507150000122
Besides, the flow of each category is arranged according to the descending order of the occurrence frequency. The frequency can also reflect the scale of the flow, and the calculation mode is as follows:
Figure BDA0001608507150000123
the performance of the kth class flow component in the OD flow demand is then:
Figure BDA0001608507150000124
recording as a matrix OD per QoS dimension h(k)(h)：
Figure BDA0001608507150000131
Wherein
Figure BDA0001608507150000132
ehIs a column vector with only the h-th element being 1 and the remaining elements being 0.
And step five, finding the most suitable network slice for bearing the network slice for each type of QoS requirement obtained by clustering by using a self-designed resource assembly algorithm, wherein the specific method comprises the following steps:
(1) each type of traffic is processed in turn according to the descending order of equation (4). For the k-th type traffic OD after clustering(k)And (3) a proper resource allocation scheme needs to be found, the flow is mapped to the physical link of the graph G, a pre-allocation scheme suitable for bearing the kth type flow slice is obtained, and the step (2) is carried out to readjust the flow. In the present invention, the objective of resource allocation is to satisfy the lowest slice deployment cost while ensuring that the slice QoS performance meets the intra-slice traffic load demand, that is, the following optimized mathematical model:
(a) an objective function:
Figure BDA0001608507150000133
wherein f (u, v) is link e<u,v>The cost per unit of bandwidth is increased by the upper,
Figure BDA0001608507150000134
for continuous decision variables: the ith flow is in link e<u,v>Bandwidth usage of.
(b) Constraint conditions are as follows:
from each link on which any target traffic in the slice depends, the constraints are:
Figure BDA0001608507150000135
Figure BDA0001608507150000141
from the viewpoint of ensuring end-to-end performance when the network slice operates, the constraints are as follows:
Figure BDA0001608507150000142
Figure BDA0001608507150000143
Figure BDA0001608507150000144
Figure BDA0001608507150000145
Figure BDA0001608507150000146
Figure BDA0001608507150000147
Figure BDA0001608507150000148
Figure BDA0001608507150000149
wherein
Figure BDA00016085071500001410
Belong to Set(k)The flow is the ith flow in the kth flow in the OD flow, originates from the node u and ends at the node v; bw, d, ω and data are subscripts representing bandwidth, delay, loss rate and data stream byte length of the QoS performance, respectively; alpha is alpha(k)In order to adjust the resource allocation of the kth flow, the network administrator can combine the slice load settings of various flows to perform manual intervention of resource allocation, and the default is 1, 0<α(k)≤1； V(k)Is a subset of V, containing nodes involved in class k OD flows;
Figure BDA00016085071500001411
is as followsThe connection number of the communication connection between the source node and the sink node related to the k-type flow is established; d(k)<u,v>The average frame length for the kth traffic passes through the link e<u,v>Time delay caused by transmission; pD(k)、PBW(k) And Pω(k) The lower bounds of the probability that the time delay, the bandwidth and the loss rate in the network slice QoS performance meet the performance requirements of the user are represented respectively, the ideal situation is 1, and the lower bounds are set by a network slice manager and stored in a slice maintenance table. When more types of QoS constraints are added to the model, contract probability constraints can be written for other QoS indexes by analogy with formulas (10) to (12).
(2) Obtaining a preliminary scheme suitable for bearing the kth class flow slice according to the optimization model solved in the step (1): by
Figure BDA0001608507150000151
The existence of the k-th flow carried by each link can be known, and the topological graph formed by the links can be obtained
Figure BDA0001608507150000152
Figure BDA0001608507150000153
However, it is necessary to examine the drawings
Figure BDA0001608507150000154
The number of connected components is calculated by using the characteristic value of the Laplace matrix:
drawing (A)
Figure BDA0001608507150000155
The laplacian matrix L of is:
Figure BDA0001608507150000156
the eigenvalues of the matrix are
Figure BDA0001608507150000157
L is a semi-positive constant and L is a negative constant,
Figure BDA0001608507150000158
λi≥0,λ0the number of 0 s in the eigenvalues of the laplace matrix is the number of connected components of the graph:
Figure BDA0001608507150000159
if A (k) is 1, the diagram
Figure BDA00016085071500001510
Namely the slice topology suitable for bearing the kth type flow; if not, then,
Figure BDA00016085071500001511
each connected component of (a) forms a slice, a common slice
Figure BDA00016085071500001512
Figure BDA00016085071500001513
Wherein for slicing
Figure BDA00016085071500001514
Edge e<u,v>The weight value of is
Figure BDA00016085071500001515
Representing the allocated bandwidth of the slice on this link.
(3) The SDN controller slices the data
Figure BDA00016085071500001516
Figure BDA00016085071500001516
1≤k≤Kopt. And sending the information to SDN forwarding equipment. Specifically, the slice maintenance table ts (v) of the forwarding device node v is a set:
Figure BDA0001608507150000161
the above-mentioned process steps (1) - (3) are organized into a flow chart, as shown in FIG. 2.
After the flow request initiated by each user, the forwarding layer equipment firstly judges the clustering category of the flow, judges the segment attribution, marks the segment in the corresponding field in the network flow data packet, routes in the corresponding network segment, and simultaneously monitors the resource utilization rate in the segment and SLA default conditions.
In the sixth step, after the flow request initiated by each user, the forwarding layer device first determines the cluster type of the flow, routes in the corresponding network slice, and monitors the resource utilization rate and SLA default conditions in the slice, so that the slice structure of the next slice structure period can be flexibly adjusted.
Wherein, for a specific g slice, the resource utilization η (g) is the ratio of the used bandwidth of each link to the total bandwidth in the slice;
the SLA breach rate φ (g) is a ratio of the number of network flow performances sampled within a slice that do not reach their requested QoS performances;
when the method is specifically applied, the slices can be flexibly adjusted according to eta (g) and phi (g). This example illustrates one possible solution, but is not intended to limit the invention:
when phi (g) is greater than a certain threshold, which indicates that the slice has a heavy load and a high default, finding the traffic class k supported by the slice, should increase alpha in formula (9) in the next slice construction period(k)；
When phi (g) is less than a certain threshold and eta (g) is greater than a certain threshold, it indicates that the slice is working well, and finds the traffic class k supported by the slice, which can keep alpha(k)；
When phi (g) is less than a certain threshold value and eta (g) is also less than a certain threshold value, it indicates that the slice is in a light load condition, if the rest slices are too heavily loaded, alpha can be reduced(k)。
When the first to sixth steps are performed, a communication process among the network user, the SDN controller, and the SDN forwarding device is as shown in fig. 6.

Claims (6)

1. An SDN slice construction method based on traffic clustering is characterized by comprising the following steps in each slice updating period T:
the method comprises the following steps: registering a network service level contract (SLA) of each access network terminal user, wherein the contract comprises the network data transmission characteristics of access bandwidth, highest time delay, packet loss rate and jitter rate;
step two: counting the resource information of the whole network physical link in the SDN control domain to obtain a physical resource capacity matrix C and a physical remaining available resource matrix R;
step three: acquiring real-time flow requirements of the whole network in an SDN control domain to obtain QoS characteristics of OD flow requirements of the whole network, and summarizing the QoS characteristics into an OD flow Set;
step four: calculating hot spot flow in OD flow, clustering the flow in the Set to obtain typical QoS requirement category and scale in the current network;
step five: dividing network slices which are most suitable for bearing the QoS requirements of each type obtained by clustering in a physical link by using a self-designed resource assembly algorithm, issuing a flow table to a slice maintenance table of forwarding equipment by an SDN controller, wherein the slices bearing the QoS requirements of each type can be parallelly and independently calculated;
step six: after receiving the data stream of the client, the SDN forwarding device searches a network slice suitable for transmission according to the QoS request characteristic of the client for forwarding, monitors the resource utilization rate and SLA default condition in the slice, and can determine whether to repeat the first step to the sixth step to adjust the slice division according to the slice resource utilization rate and the SLA default rate.
2. The SDN slice construction method based on traffic clustering of claim 1, wherein: and in the second step, the QoS performance bearable on each link and the resource use condition of each link are reported to the SDN controller by the forwarding device of the SDN network.
3. The SDN slice construction method based on traffic clustering of claim 1, wherein: the step of collecting the total network OD flow Set in the third step is that a QoS guarantee request message is sent to an adjacent SDN forwarding device by a user side of an access network, the forwarding device submits an SDN controller through a southbound link between the SDN controller and the forwarding device, a QoS performance index of a data flow to be initiated is registered, and the SDN controller checks whether the identity of the client side is legal and whether a QoS request exceeds an SLA contract.
4. The SDN slice construction method based on traffic clustering of claim 1, wherein: selecting an unlimited main stream clustering method in the fourth step to cluster the OD flow Set and determining the good clustering number KoptIn the clustering result, the K (1. ltoreq. K. ltoreq.K)opt) Class traffic subset is denoted as Set(k)The number of the collection elements is FSkThe cluster centroid reflects the typical QoS requirements and the upper and lower bounds of the attributes of the flows, and the flows of each category are arranged in descending order according to the occurrence frequency of the flows.
5. The SDN slice construction method based on traffic clustering of claim 1, wherein: and step five, calculating the most suitable network slice for bearing the network slice in the SDN controller according to each type of QoS requirement obtained by clustering by using a self-designed resource assembly algorithm, issuing slice information to SDN forwarding equipment, and storing the slice information in a slice maintenance table.
6. The SDN slice construction method based on traffic clustering of claim 1, wherein: after each user initiates a flow request in the sixth step, the forwarding layer equipment firstly judges the clustering category of the flow, routes in the corresponding network slice, and simultaneously monitors the resource utilization rate and SLA default conditions in the slice to flexibly adjust the slice structure of the next slice structure period.
CN201810253899.6A 2018-03-26 2018-03-26 SDN slice construction method based on flow clustering Active CN108599994B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810253899.6A CN108599994B (en) 2018-03-26 2018-03-26 SDN slice construction method based on flow clustering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810253899.6A CN108599994B (en) 2018-03-26 2018-03-26 SDN slice construction method based on flow clustering

Publications (2)

Publication Number Publication Date
CN108599994A CN108599994A (en) 2018-09-28
CN108599994B true CN108599994B (en) 2021-04-20

Family

ID=63624626

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810253899.6A Active CN108599994B (en) 2018-03-26 2018-03-26 SDN slice construction method based on flow clustering

Country Status (1)

Country Link
CN (1) CN108599994B (en)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10897423B2 (en) 2019-05-14 2021-01-19 Vmware, Inc. Congestion avoidance in a slice-based network
CN111092751B (en) * 2019-11-26 2022-04-19 中盈优创资讯科技有限公司 Monitoring data processing method and device
CN113347641B (en) * 2020-03-02 2023-05-16 中国电信股份有限公司 Network deployment method, device and computer readable storage medium
CN113498084B (en) * 2020-04-07 2022-07-29 烽火通信科技股份有限公司 Hierarchical slice-based 5G service bearing method and system
CN113965475B (en) * 2020-07-02 2023-09-05 ***通信集团设计院有限公司 Network slice engineering acceptance method and system
CN113905448B (en) * 2020-07-06 2023-09-08 ***通信集团重庆有限公司 Wireless network resource scheduling method, device and equipment
CN112583636B (en) * 2020-11-24 2023-07-04 华南师范大学 Construction method of government network slice, electronic equipment and storage medium
CN112954807B (en) * 2021-01-28 2022-11-18 哈尔滨工程大学 Heterogeneous Massive MIMO system multi-element resource configuration method based on network slices
CN113556761B (en) * 2021-06-25 2024-03-22 中国联合网络通信集团有限公司 Slice resource adjustment method, system, terminal equipment and storage medium
CN113746677B (en) * 2021-09-01 2023-09-01 之江实验室 5G new air interface-oriented network intra-slice resource allocation method
CN114465723B (en) * 2022-02-11 2023-10-20 西安电子科技大学 Quantum encryption communication system and method based on software defined network and slice
WO2024007208A1 (en) * 2022-07-06 2024-01-11 Huawei Technologies Co., Ltd. Systems and methods to minimize resource utilization across multiple domains to assure quality of service
CN116318434B (en) * 2023-02-03 2023-09-12 军事科学院***工程研究院网络信息研究所 Dynamic matching method for space terahertz communication network slice and clustering flow

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103905523A (en) * 2013-12-23 2014-07-02 浪潮(北京)电子信息产业有限公司 Cloud computing network virtualization method and system based on SDN
CN105516312B (en) * 2015-12-09 2019-02-22 重庆邮电大学 A kind of software defined network load balancing apparatus and method
CN107659419B (en) * 2016-07-25 2021-01-01 华为技术有限公司 Network slicing method and system
CN106851705B (en) * 2017-02-22 2019-12-27 重庆邮电大学 Wireless network slicing method based on slice flow table
CN107070700B (en) * 2017-03-07 2020-01-21 浙江工商大学 Network service providing method based on automatic identification of identity

Also Published As

Publication number Publication date
CN108599994A (en) 2018-09-28

Similar Documents

Publication Publication Date Title
CN108599994B (en) SDN slice construction method based on flow clustering
US11281486B2 (en) Bandwidth control in multi-tenant virtual networks
US9838268B1 (en) Distributed, adaptive controller for multi-domain networks
US6594268B1 (en) Adaptive routing system and method for QOS packet networks
US7599290B2 (en) Methods and systems for providing quality of service in packet-based core transport networks
JP3987394B2 (en) How to allocate bandwidth within a network domain with a bandwidth broker
US8385194B2 (en) Quality of service admission control network
EP1844582B1 (en) Routing method and system
US7969886B1 (en) Bandwidth allocation for hierarchical telecommunications networks
GB2542870A (en) Local and demand driven QoS models
CN108476175B (en) Transfer SDN traffic engineering method and system using dual variables
US20140101316A1 (en) Apparatus and method for provisioning
Liu et al. Cost-efficient virtual network function placement and traffic steering
Tomovic et al. Toward a scalable, robust, and QoS-aware virtual-link provisioning in SDN-based ISP networks
CN107995121B (en) Method for counting flow in software defined network based on wildcard
Liao et al. Dynamic load-balancing mechanism for software-defined networking
US7836201B2 (en) Methods and systems for providing efficient provisioning of data flows
WO2024055780A1 (en) Computing power network information announcement and routing decision-making method and apparatus, and medium
EP1834448B1 (en) Admission control in a telecommunication network
Saadatpour et al. QoS improvement in SDN using centralized routing based on feedback
Majdabadi et al. Soda-stream: Sdn optimization for enhancing qoe in dash streaming
US6785245B1 (en) Method and apparatus for controlling fanout of a multicast tree
US20050171797A1 (en) Intelligent access control and warning system for operations management personnel
JP4014889B2 (en) Network management device
Kostiv et al. Metaheuristics in cloud platform traffic engineering

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