CN106302226B - A kind of service dynamic dispatching method of QoS perception - Google Patents
A kind of service dynamic dispatching method of QoS perception Download PDFInfo
- Publication number
- CN106302226B CN106302226B CN201610920313.8A CN201610920313A CN106302226B CN 106302226 B CN106302226 B CN 106302226B CN 201610920313 A CN201610920313 A CN 201610920313A CN 106302226 B CN106302226 B CN 106302226B
- Authority
- CN
- China
- Prior art keywords
- node
- service
- flow
- user
- algorithm
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/80—Actions related to the user profile or the type of traffic
- H04L47/805—QOS or priority aware
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
- H04L67/61—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources taking into account QoS or priority requirements
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a kind of service dynamic dispatching methods of QoS perception, comprising the following steps: S1: determining role and the problem of scheduling;S2: case study and formal definitions;S3: analysis of complexity and problem reduction;S4: algorithm design;S5: it executes algorithm and obtains scheduling result.The present invention is directed to how accurately and efficiently to schedule users to most suitable service node, not only allow user can not congestion, do not obtain required service with not conflicting, and the problem of can guarantee the QoS demands such as its response time, the scheduling strategy for meeting all user's QoS requirements is provided, to meet the service dispatch demand of Military Application.
Description
Technical field
The present invention relates to a kind of service dynamic dispatching methods of QoS perception.
Background technique
Commercial most of service dispatch generallys use polling method etc. and carries out load balancing at present.These service dispatch
Method is usually user's access ultra-large on wide area network-oriented network, to realize high load, low time delay, using load balancing
Device/reverse proxy+cluster mode, using the low load balancing of complexity, to support large-scale user to access needs,
When high-quality service can not be provided for all users, it can lose a collection of user's using modes such as admission control, delay waitings
Service quality is kept the system running continuously.But in certain dedicated application fields, especially internal network service, although with
Family scale is compared will relatively rarely, but user has different priority, some proprietary applications according to different application levels
The service quality guarantee of middle user requires strictly, and the quality of service requirement even denied access for losing certain customers can be brought not
The cost that can be estimated, therefore the promotion for the entirety or average service quality usually pursued on internet is not able to satisfy above-mentioned application and needs
It asks.In some internal proprietary applications, the priority of user has difference, and the QoS requirement of user is that the important of business is wanted again
Element, it is necessary to be protected.It is different department (emergency ward, Medicine and Surgery etc.), common for example, the medical information service of medical system
User responses to which that time requirement respectively has difference to service;The related service of financial system, common employee and financial staff are to clothes
The quality requirement of business also has difference, and violating for these quality of service requirement can not put up with.It is also noted that being used on internet
Family scale is very huge, therefore poll etc. is simple, time complexity (being increased according to number of users at series) is low, overall performance optimization
Algorithm comparison be applicable in, and user is relatively fewer in profession or internal applications, therefore using time complexity is slightly higher, service quality
Ensure that stringent optimized scheduling algorithm is more appropriate.From this angle, the present invention designs the service dispatch of multiple constraint
Strategy meets the service dispatch demand of these applications.
Summary of the invention
Goal of the invention: the object of the present invention is to provide a kind of QoS perception for being able to solve defect existing in the prior art
Service dynamic dispatching method.
Technical solution:
The service dynamic dispatching method of QoS perception of the present invention, comprising the following steps:
S1: determine role and the problem of scheduling: model character includes user node, service dispatch node and service node,
The purpose of scheduling is that the request of user is sent to corresponding service node by service dispatch node, it is desirable that can satisfy the QoS of user
Demand, service node are unable to excess load, and guarantee that the user met is as more as possible;
S2: case study and formal definitions: considering object, target, relevant parameter and the constraint condition of service dispatch, right
It services dynamic scheduling problem and carries out Formal Modeling;
S3: analysis of complexity and problem reduction: according to model built, the complexity of problem analysis, find can reduction shape
Formula model;
S4: it algorithm design: proposes the service dynamic dispatching algorithm of QoS perception, provides the execution sequence of algorithm;
S5: it executes algorithm and obtains scheduling result.
Further, the step S4 the following steps are included:
S4.1: parameter needed for acquisition algorithm: the major parameter of the dispatching algorithm include: service node distance distance,
The processing capacity parameter of service node dependability parameter and service node;
S4.2: algorithm is executed: one figure G=(V of construction0∪VS∪VC∪V1, E), wherein V0Corresponding to starting point vs, V1It is corresponding
In starting point vs, each server node V ∈ VSCorresponding to an element in S, each user node V ∈ VCIt also corresponds in C
One element;Network is Capacity-network;In V0With i ∈ VSBetween there are a line, the maximum stream flow that allows to flow through is min
{Ui,Bi};In i ∈ VSWith j ∈ VCBetween there are a lines, i.e. (i, j) ∈ E, and if only if SiIt is able to satisfy CjService quality need
It asks, the maximum stream flow allowed to flow through is Rj;In j ∈ VCAnd V1Between there are a line, the maximum stream flow allowed to flow through is
Rj;From any one feasible flow x, judge in network G whether about x s-t augmenting path: if without such augmentation
Road, then x is exactly max-flow;If there is augmenting path, then by carrying out augmentation to flow on augmenting path, a new flow is obtained
Bigger feasible flow x;Continue to repeat aforesaid operations to feasible flow x, until there is no maximums until augmenting path, obtained in network
Stream is the maximum number of user for meeting qos requirement;
S4.3: it optimizes and revises:If being less than given threshold value according to the ratio that its user requests total amount to account for capacity is distributed
Tr, i.e. ∑J:f (cj)=siRcj< Tr, then certain user of adjustment greater than Tr distributes to the service node, until there is no less than Tr's
Service node or until adjustment cannot be continued according to QoS requirement.
The utility model has the advantages that compared with prior art, the present invention have it is following the utility model has the advantages that
1) internet common service dispatch strategy generallys use polling method etc. and carries out load balancing at present, pursuit it is whole
The promotion of body or average service quality does not account for the different application level of user with different priority, and the present invention is more
The quality of service requirement for paying close attention to each individual can be guaranteed;
2) for dispatching method commercial at present mainly by the way of load balancer/reverse proxy+cluster, cluster uses phase
Identical service is disposed in same configuration, so the difference of service node load capacity and the difference of network delay are not accounted for, and
The present invention has fully considered these constraint conditions;
3) present invention not only allows user can for how accurately and efficiently to schedule users to most suitable service node
Not congestion, do not obtain required service with not conflicting, and the problem of can guarantee the QoS demands such as its response time, it is all to provide satisfaction
The scheduling strategy of user's QoS requirement, to meet the service dispatch demand of Military Application.
Detailed description of the invention
Fig. 1 is the service dispatch problem schematic diagram that QoS of the invention is perceived;
Fig. 2 is maximum flow problem schematic diagram of the invention;
Fig. 3 is the exemplary diagram of the algorithm in step S4 of the present invention, in which:
Fig. 3 (a) is the schematic diagram of first feasible flow;
Fig. 3 (b) is the schematic diagram for indicating first feasible flow of weight;
Fig. 3 (c) is the schematic diagram of the reverse flow of first feasible flow;
Fig. 3 (d) is the schematic diagram of first augmenting path;
Fig. 3 (e) is the schematic diagram of the reverse flow of first augmenting path;
Fig. 3 (f) is the schematic diagram of Article 2 augmenting path;
Fig. 3 (g) is the schematic diagram of the reverse flow of Article 2 augmenting path;
Fig. 3 (h) is the schematic diagram of Article 3 augmenting path;
Fig. 3 (i) is the schematic diagram of the reverse flow of Article 3 augmenting path.
Specific embodiment
With reference to the accompanying drawings and detailed description, technical solution of the present invention is described further.
The present invention discloses a kind of QoS perception for how accurately and efficiently to schedule users to most suitable service node
Service dynamic dispatching method, not only allow user can not congestion, do not obtain required service with not conflicting, but also can guarantee its response
The problem of QoS demands such as time, provides the scheduling strategy for meeting all user's QoS requirements, to meet the clothes of Military Application
Business dispatching requirement.
The service dynamic dispatching method of QoS perception of the present invention, comprising the following steps:
S1: determine role and the problem of scheduling: model character includes user node, service dispatch node and service node,
As shown in Figure 1, the purpose of scheduling is that the request of user is sent to corresponding service node by service dispatch node, it is desirable that can satisfy
The QoS demand of user, service node are unable to excess load, and guarantee that the user met is as more as possible;
S2: case study and formal definitions: giving a non-directed graph G=(V, E), and the time delay of each edge (u, v) ∈ E is
luv, the shortest path length between any pair of node { u, v } is d (u, v), 0.1s to 10 minutes, using shortest path first,
Such as dijkstra's algorithm, shortest path is 4 between obtaining AB, i.e., network distance is 4.To user node setEach of
For user c, user's request data rate is denoted as Rc, such as 4kbps-100Mbps, QoS demand be denoted as Qc, 0.1s to 10 minutes;
To server node setIn each server S for, processing capacity is denoted as Us(10,5000), transmission capacity note
For Bs(5,500), transmission time is denoted as tr, calculate the time be denoted as tc;The target of the service dispatch problem of QoS perception is to find one
A function f:C → S so that the number of users for meeting response time requirement is maximum, and meets following two restrictive conditions: when 1) simultaneously
Prolong limitation: to each user c ∈ C, d (c, f (c))+tr+tc≤Qc;2) capacity limit: to each server s ∈ D, ∑ c ∈ C:f
(c)=s, Rc≤min{Us,Bs};
S3: analysis of complexity and problem reduction: according to model built, the complexity of problem analysis, find can reduction shape
Formula model, specifically includes the following steps:
S3.1: maximum flow problem definition: 1) network has a starting point vsWith a terminal vtIf there is several beginning or ends,
Then a starting point and a terminal can be converted by increasing dummy node;2) network is Capacity-network, i.e. stream has direction
Property.If it is Undirected networks or hybrid network, then Capacity-network should be converted into;3) there are a power, table on each arc of network
Show the maximum stream flow allowed to flow through.If with cijIt indicates by viTo vjArc on the maximum stream flow that allows to flow through, with vtIndicate practical stream
, then there is v in the flow for crossing the arct;4) in network, for any node in addition to beginning and end, the summation of influx is answered
This is equal to the summation of discharge, i.e. ∑ fij=∑ fji,i≠s,t;
S3.2: reduction procedure: one figure G=(V of construction0∪VS∪VC∪V1, E), wherein V0Corresponding to starting point vs, V1It is corresponding
In starting point vs, each server node V ∈ VSCorresponding to an element in S, each user node V ∈ VCIt also corresponds in C
One element (Fig. 2 gives an example).Network is Capacity-network.In V0With i ∈ VSBetween there are a line, allow to flow
The maximum stream flow crossed is min { Ui,Bi};In i ∈ VSWith j ∈ VCBetween there are a lines, i.e. (i, j) ∈ E is (and if only if SiIt can be full
Sufficient CjQoS requirement), the maximum stream flow allowed to flow through be Rj;In j ∈ VCAnd V1Between there are a line, allow
The maximum stream flow flowed through is Rj;It is not difficult to find out that therefore, these elements for meeting maximum flow problem optimal solution constitute QoS perception
Service dispatch problem optimal solution, vice versa;As shown in Figure 2;
S4: it algorithm design: proposes the service dynamic dispatching algorithm of QoS perception, provides the execution sequence of algorithm;According to above
The problem of providing definition and reduction, due to QoS perception service dispatch problem can reduction be maximum flow problem, the base of algorithm
This thought is the algorithm (Max-Flow Based Algorithm, MFBA) based on Network Maximal-flow problem, it can be seen that working as
When node capacity abundance, in order to can be with equally loaded while ensureing all QoS, it is therefore desirable on the basis of maximum-flow algorithm
On be adjusted, not only obtained the guarantee of service quality, but also reach the target of load balancing, improved handling up for running
Amount;
The basic thought of algorithm is that (whether such as zero stream), judge in network G about x from any one feasible flow x
S-t augmenting path;If x is exactly max-flow without such augmenting path;It, then can be by increasing if there is augmenting path
Wide road flow carries out augmentation, obtains the bigger feasible flow x of a new flow;Continue to repeat aforesaid operations to feasible flow x, directly
Augmenting path is not present into network;At this point, feasible flow x is the max-flow of network according to augmenting path theorem;It finds out in network
To get the distribution to a user node set C to service node set S after max-flow, enable the service quality of user
Enough meet demands;Since max-flow does not account for the residual capacity of service node, one is carried out to capacity on this basis
Secondary adjustment, the service node s high to those residual capacities, will distribute to the user of other nodes, if s can satisfy it
Certain customers are then reassigned to s by QoS demand, reach load balancing, with further lifting system performance;
The scheduling strategy is specific as follows:
S4.1: the major parameter of parameter needed for acquisition algorithm, dispatching algorithm of the present invention includes:
1) the far and near D of service node distanceij: this parameter is mainly indicated by the response delay of network topology, is passed through
Elongatedness provides service node network distance when calculating the shortest path of point-to-point transmission on network;Such as service requesting node A and clothes
Be engaged in node B in a network, which is represented by weighted-graph G, G=(V, T), wherein V be node set V=A,
B, C, D }, T is the set T={ (A, B), (A, C), (A, D), (B, C), (B, D) } on side, and the weight (i.e. network delay) on side is table
W={ 4,3,2,4,3 } are shown as, shortest path is 4, i.e. network between using shortest path first (such as dijkstra's algorithm) to obtain AB
Distance is 4;
2) service node dependability parameter Rj: what this parameter indicated is certain service node reliability of operation, that is, is provided
To the ratio that user effectively services, referring to service node historical data.Such as A service node is averaged weekly in over the past several months
Fault-free is 160 hours, then its service reliability is 95.24%;
3) the processing capacity parameter C of service nodej: number of users can be handled, is connected according to the performance maximum of service node
Number subtracts the acquisition of active user's number.Such as the maximum processing capability of service node is concurrently visited in the performance test of beginning for 500
The amount of asking, and existing processing number of users is 200, then the existing processing capacity of service node is 300;
S4.2: from any one feasible flow x (such as zero stream), judge in network G whether about x s-t augmenting path:
If x is exactly max-flow without such augmenting path;If there is augmenting path, then can by flow on augmenting path into
Row augmentation obtains the bigger feasible flow x of a new flow;Continue to repeat aforesaid operations to feasible flow x, until not depositing in network
In augmenting path.Obtained max-flow is the maximum number of user for meeting qos requirement;
Concrete example such as Fig. 3;First look for a feasible flow s → v1→v3→v2→v4→ t, such as Fig. 3 (a), according to path
Upper weight setting, the maximum flow valuve of feasible flow are 4, and network G is recorded as 3 (b).Iteration subtracts current max-flow on existing figure
Amount, and add reverse flow t → v4→v2→v3→v1→ s, routine weight value 4, such as Fig. 3 (c), found on new figure augmenting path s →
v1→v2→v4→v3→ t, this path max-flow are 7, then max-flow is 11 at this time, such as Fig. 3 (d);Continue to add reverse flow and seek
Look for augmenting path s → v2→v1→v3→ t, this path max-flow is 8, such as Fig. 3 (e) and Fig. 3 (f);And so on, it is anti-to continue addition
To flowing and find augmenting path s → v2→v3→ t, this path max-flow are 4, such as Fig. 3 (g) and Fig. 3 (h), are arrived in Fig. 3 (i) network not
There are until augmenting path;Then the max-flow in this figure is 11+12=23, such as Fig. 3 (h);
S4.3:If being less than given threshold value Tr (such as 50% according to the ratio that its user requests total amount to account for capacity is distributed
Arbitrary value between~80%), i.e. ∑J:f (cj)=siRcj< Tr, then certain user of adjustment greater than Tr distributes to the service node,
It cannot continue to adjust until the service node less than Tr is not present, or according to QoS requirement;
S5: it executes algorithm and obtains scheduling result.
Claims (1)
1. a kind of service dynamic dispatching method of QoS perception, it is characterised in that: the following steps are included:
S1: determine role and the problem of scheduling: model character includes user node, service dispatch node and service node, scheduling
Purpose be that the request of user is sent to corresponding service node by service dispatch node, it is desirable that can satisfy the QoS demand of user,
Service node is unable to excess load, and guarantees that the user met is as more as possible;
S2: case study and formal definitions: object, target, relevant parameter and the constraint condition of service dispatch are considered, to service
Dynamic scheduling problem carries out Formal Modeling;
S3: analysis of complexity and problem reduction: according to model built, the complexity of problem analysis, find can reduction formalization
Model;
S4: it algorithm design: proposes the service dynamic dispatching algorithm of QoS perception, provides the execution sequence of algorithm;
S5: it executes algorithm and obtains scheduling result;
The step S4 the following steps are included:
S4.1: parameter needed for acquisition algorithm: the major parameter of the dispatching algorithm includes: the distance of service node distance, service
The processing capacity parameter of node dependability parameter and service node;
S4.2: algorithm is executed: one figure G=(V of construction0∪VS∪VC∪V1, E), wherein V0Corresponding to starting point vs, V1Corresponding to rise
Point vs, VSIn each server node correspond to S in an element, VCIn each user node also correspond in C one
A element;Network is Capacity-network;In V0With i ∈ VSBetween there are a line, the maximum stream flow that allows to flow through is min { Ui,
Bi};In i ∈ VSWith j ∈ VCBetween there are a lines, i.e. (i, j) ∈ E, and if only if SiIt is able to satisfy CjQoS requirement,
The maximum stream flow allowed to flow through is Rj;In j ∈ VCAnd V1Between there are a line, the maximum stream flow that allows to flow through is Rj;From appoint
The feasible flow x that anticipates sets out, judge in network G whether about x s-t augmenting path: if without such augmenting path, x
It is exactly max-flow;If there is augmenting path, then by carrying out augmentation to flow on augmenting path, it is bigger to obtain a new flow
Feasible flow x;Continue to repeat aforesaid operations to feasible flow x, until augmenting path is not present in network, obtained max-flow is full
The maximum number of user of sufficient qos requirement;VSThe node set in figure G, V are mapped to for server node set SCFor user node collection
It closes C and is mapped to the node set in figure G, S is service node device set, and C is user node set, i VSMiddle any node, j
For VCMiddle any node, UiFor the processing capacity of i-node, BiFor the transmission capacity of i-node, siFor VSClothes corresponding to middle i-node
Business device node, cjFor VCUser node corresponding to middle j node, f (cj)=siIt is user node cjIt is mapped as server node si
Situation;
S4.3: it optimizes and revises:If being less than given threshold value Tr according to the ratio that its user requests total amount to account for capacity is distributed,
That is ∑J:f (cj)=siRcj< Tr, then certain user of adjustment greater than Tr distributes to the service node, until there is no the clothes less than Tr
Business node or until adjustment cannot be continued according to QoS requirement.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610920313.8A CN106302226B (en) | 2016-10-21 | 2016-10-21 | A kind of service dynamic dispatching method of QoS perception |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610920313.8A CN106302226B (en) | 2016-10-21 | 2016-10-21 | A kind of service dynamic dispatching method of QoS perception |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106302226A CN106302226A (en) | 2017-01-04 |
CN106302226B true CN106302226B (en) | 2019-08-30 |
Family
ID=57719650
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610920313.8A Active CN106302226B (en) | 2016-10-21 | 2016-10-21 | A kind of service dynamic dispatching method of QoS perception |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106302226B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111262749B (en) | 2018-11-30 | 2023-05-23 | 华为技术有限公司 | Method and equipment for detecting network reliability |
CN110727518B (en) * | 2019-10-14 | 2022-05-27 | 北京奇艺世纪科技有限公司 | Data processing method and related equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1725750A (en) * | 2004-07-22 | 2006-01-25 | 华为技术有限公司 | Downlink shared channel (DSCH) resource scheduling method |
CN102970763A (en) * | 2012-11-01 | 2013-03-13 | 大唐移动通信设备有限公司 | Resource scheduling method and device based on service qualities |
CN103905336A (en) * | 2014-04-24 | 2014-07-02 | 河北工程大学 | Method for scheduling QoS sensing in SIEPON based on network coding technology |
US9135072B2 (en) * | 2010-09-16 | 2015-09-15 | Apple Inc. | QoS-aware scheduling |
US9621469B2 (en) * | 2012-12-06 | 2017-04-11 | Electronics And Telecommunications Research Institute | Apparatus and method for managing flow in server virtualization environment, and method for applying QOS |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6778498B2 (en) * | 2001-03-20 | 2004-08-17 | Mci, Inc. | Virtual private network (VPN)-aware customer premises equipment (CPE) edge router |
-
2016
- 2016-10-21 CN CN201610920313.8A patent/CN106302226B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1725750A (en) * | 2004-07-22 | 2006-01-25 | 华为技术有限公司 | Downlink shared channel (DSCH) resource scheduling method |
US9135072B2 (en) * | 2010-09-16 | 2015-09-15 | Apple Inc. | QoS-aware scheduling |
CN102970763A (en) * | 2012-11-01 | 2013-03-13 | 大唐移动通信设备有限公司 | Resource scheduling method and device based on service qualities |
US9621469B2 (en) * | 2012-12-06 | 2017-04-11 | Electronics And Telecommunications Research Institute | Apparatus and method for managing flow in server virtualization environment, and method for applying QOS |
CN103905336A (en) * | 2014-04-24 | 2014-07-02 | 河北工程大学 | Method for scheduling QoS sensing in SIEPON based on network coding technology |
Non-Patent Citations (4)
Title |
---|
QoS-aware placement of stream processing service;Kun You;Bin Tang;Zhuzhong Qian;Sanglu Lu;Daoxu Chen;《The Journal of Supercomputing》;20110119;919-941 |
QoS-Aware Service Redeployment in Cloud;Kun You;Zhuzhong Qian;Song Guo;Sanglu Lu;DaoXu Chen;《2011 IEEE International Conference on Communications (ICC)》;20110729;全文 |
QoS感知的服务部署问题研究;游坤;《中国博士学位论文全文数据库》;20120115;全文 |
关于网络最大流的两个算法;毛晓亮;《中国优秀硕士学位论文全文数据库》;20131215;全文 |
Also Published As
Publication number | Publication date |
---|---|
CN106302226A (en) | 2017-01-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ordónez et al. | Robust capacity expansion of network flows | |
Yu et al. | Deadline scheduling as restless bandits | |
Senouci et al. | TOPSIS-based dynamic approach for mobile network interface selection | |
CN109800072A (en) | Task schedule optimization method and device based on edge calculations | |
Stolyar et al. | Control of systems with flexible multi-server pools: a shadow routing approach | |
CN110096362A (en) | A kind of multitask discharging method based on Edge Server cooperation | |
Zhou et al. | Learning from peers: Deep transfer reinforcement learning for joint radio and cache resource allocation in 5G RAN slicing | |
CN108551489A (en) | A kind of application server load balancing method, system, device and storage medium | |
CN106302226B (en) | A kind of service dynamic dispatching method of QoS perception | |
CN109787915A (en) | Flow control methods, device, electronic equipment and the storage medium of network access | |
WO2024094104A1 (en) | Dynamic feedback weighted cloud storage resource scheduling method, apparatus and device | |
CN107979545A (en) | A kind of mapping method of virtual network based on nodal community | |
CN105491085A (en) | Method and device for on-line requesting for queuing | |
Verloop et al. | Monotonicity properties for multi-class queueing systems | |
CN104811467B (en) | The data processing method of aggreggate utility | |
CN106303112A (en) | A kind of method for equalizing traffic volume and device | |
CN106101010A (en) | VBRAS flow load balance method and device based on end office's cloud | |
Joe-Wong et al. | Offering supplementary network technologies: Adoption behavior and offloading benefits | |
Liebeherr et al. | Rate allocation and buffer management for differentiated services | |
Farzaneh et al. | An adaptive competitive resource control protocol for alleviating congestion in wireless sensor networks: an evolutionary game theory approach | |
Jaskóła et al. | Simultaneous routing and flow rate optimization in energy-aware computer networks | |
Karpowicz | Nash equilibrium design and coordination in hierarchical systems | |
Korcak et al. | Competition and regulation in a wireless operator market: An evolutionary game perspective | |
Wellons et al. | Augmenting predictive with oblivious routing for wireless mesh networks under traffic uncertainty | |
Lan et al. | Stability and benefits of suboptimal utility maximization |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |