CN102035888A - Method for scheduling data based on scheduling period and bandwidth awareness - Google Patents
Method for scheduling data based on scheduling period and bandwidth awareness Download PDFInfo
- Publication number
- CN102035888A CN102035888A CN 201010589063 CN201010589063A CN102035888A CN 102035888 A CN102035888 A CN 102035888A CN 201010589063 CN201010589063 CN 201010589063 CN 201010589063 A CN201010589063 A CN 201010589063A CN 102035888 A CN102035888 A CN 102035888A
- Authority
- CN
- China
- Prior art keywords
- node
- data block
- scheduling
- target data
- time limit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention provides a method for scheduling data based on a scheduling period and bandwidth awareness. The method comprises the following steps of: 1, selecting a target data block, namely calculating an initial value of the scheduling period of a new data block and transmitting the initial value to a scheduling node by a source node; storing the received data block in a cache region by the scheduling node, comparing a relative value of a data block which does not exceed the playing period in the cache region and selecting the largest data block to serve as a target data block; and calculating and modifying a scheduling period value of the target data block by the scheduling node; 2, selecting a target node, namely calculating the probability that a node does not contain the target data block in neighboring nodes by the scheduling node, and selecting any one node of which the probability is greater than the set value to serve as the target node by the scheduling node; and 3, transmitting the target data block to the target node, receiving data blocks forwarded by the neighboring nodes or the source node, and circularly executing to the end of a stream distribution session by the scheduling node. By the method, the system obtains better distribution delay performance, the data packet loss rate is reduced, and the node distribution redundancy is reduced.
Description
Technical field
The invention belongs to P2P stream media technology field, relate in particular to a kind of data dispatching method.
Background technology
The P2P Streaming Media comprises overlay network topology constructing and two parts of data dispatch.Thereby the overlay network topology constructing is meant that seeking neighbor node is organized an application layer network; And data dispatch is meant that carrying out Streaming Media on the overlay network that has built transmits, and comprises which data decision from which neighbor node obtains, which neighbor node to transmit which data to.Structure P2P stream media system is arranged because of having definite scheduling strategy based on what the tree type covered topology, thus be easy to obtain optimum distribution delay performance, but because to the poor robustness of node dynamic and be difficult to management, seldom use in the reality.(abbreviate as: netted non-structure P2P Streaming Media) system is owing to be with good expansibility and robustness makes it be more suitable for being widely adopted in practice in isomery, dynamic internet environment to cover the non-structure P2P Streaming Media of topology based on netted (mesh).But the performance of netted non-structure P2P stream media system is difficult to prediction, and owing to the dynamic of neighborhood between the node and the real-time of data distribution, make the data dispatch problem become a major issue in netted non-structure P2P Streaming Media field, and rationalization's data and scheduling strategy are designed to the key of data dispatch.
At present, in netted non-structure P2P Streaming Media, the data dispatch strategy is the burst mechanism that adopts similar Bittorrent (BT) file-sharing software, and source node is transmitted between the peer node that participates in the stream distribution session after Media Stream being divided into the small fragment that is called as data block (segment, chunk or block).The data dispatch strategy is exactly the scheduling strategy which data block institute foundation the node that has certain set of data blocks sends to which neighbor node, therefore the key of data dispatch strategy design is the node/piece selection algorithm that carries out transfer of data between any two nodes, i.e. data dispatch algorithm.Data dispatching method is according to being to be divided into the method for mixing (hybridpull/push) based on the method that pushes away (push), based on the method for drawing (pull) and push-and-pull by transmit leg or by recipient's decision that makes a choice.
Mainly initiatively select data block according to the data cached state of neighbor node based on the method that pushes away by transmit leg, and to the destination node push method, this method has and postpones little, the little advantage of control expense, but owing to lack cooperation between the node, therefore may exist has a plurality of nodes to send the situation of identical data to same neighbor node simultaneously, causes the redundancy of data distribution and has wasted network bandwidth resources.Document is arranged at present
[1-2]Proof LU (Latest Useful Chunk Scheduling, nearest useful blocks of data scheduling) method in based on the data dispatching method that pushes away is optimum.People such as T.Bonald
[1]The scheduling strategy of LU/RP (useful/random node recently) is used in the fully connected network shape topological structure, and can obtains excellent data block dispense rate and distribution delay performance.Document
[2]Consider that the high node of upstream bandwidth ability contributes more in the piece distribution procedure, proposed to utilize the node of nearest useful/bandwidth perception of the isomerism of node upstream bandwidth to select (LU/BA) method, the LU/BA method can obtain the lifting of 80% piece distribution delay performance than the LU/RP method in the full-mesh topology structure, under the situation of node bandwidth isomery.But above-mentioned data dispatching method all be in the hypothesis overlay network between the node topology be full connection, be node with overlay network in other all nodes all be connected, and can know that other all nodes have the information of data block by the exchange of periodicity cache information.And in real network, it particularly under the P2P stream media system often has 1,000,000 grades of users' situation the full connection that impossible realize topology between node, this will have influence on the transmission speed of data block in whole system, when playout-delay certain, node only with system in part of nodes set up under the situation of neighborhood, the data packet loss of LU/BA method is higher.
Document 1:T.Bonald, L.Massoulie, F.Mathieu, D.Perino, Andrew Twigg, " Epidemic Live Streaming:Optimal Performance Trade-Offs ", ACM Sigmetrics 08, Annapolis, MD, June 2008.
Document 2:A.Couto da Silva, E.Leonardi, et al., " Chunk Distribution inMesh-Based Large Scale P2P Streaming Systems:a Fluid Approach. " IEEETransactions on Parallel and Distributed Systems (99): 1-1.
Summary of the invention
Technical problem at above-mentioned existence, the present invention propose a kind of be used for netted non-structure P2P Streaming Media, based on the data dispatching method of scheduling time limit and bandwidth perception, this method is mainly taked (push) method that pushes away that made a choice by transmit leg, taken into full account the isomerism of data block broadcast time limit and node bandwidth, preferentially transmit from playing nearest data block of time limit to the high bandwidth node, reduced data block and play the probability that the time limit abandons, promoted utilization rate of network resource owing to surpassing.
For achieving the above object, the present invention adopts following technical scheme:
A kind of data dispatching method based on scheduling time limit and bandwidth perception may further comprise the steps:
The selection of S1, target data block, this step further comprises following substep:
S1-1, source node calculate the initial value d of the scheduling time limit d of the new data block that produces
0=r+2T
sAfter data block is directly sent to scheduling node, wherein, r is the moment that data block produces, T
sBe the used time of distributing data piece;
S1-2, scheduling node deposit the data block that receives in buffer area, compare the relative worth that surpasses the data block of playing the time limit in the buffer area then
And select
Maximum data block wherein, is counted as target data block
S1-3, scheduling node be d by formula
k=d
K-1+ δ * T
sRecomputate the also schedule periods limit value d of modifying target data block
k, wherein, d
K-1Be the current scheduling time limit of target data block, T
sBe the distribution used time of target data block, δ is a delay factor;
The selection of S2 destination node, this step further comprises following substep:
S2-1, scheduling node are concentrated from neighbor node and are selected the neighbor node subclass B that does not contain above-mentioned target data block, calculate the probability of each node among the neighbor node subclass B
Wherein, w
qBe node q upstream bandwidth u
qFunction;
S2-2, scheduling node from neighbor node subclass B, select probability greater than any node of set point as destination node;
S3, scheduling node send target data block to destination node, receive the data block that other neighbor node or source node are transmitted simultaneously, and circulation is carried out S1-2 to the stream distribution conversation end then.
It is the mode that adopts poll that source node described in the above-mentioned steps S1-1 directly sends to scheduling node with data block k.
The obtaining value method of the described delay factor δ of above-mentioned steps S1-3 is: when node neighbor node number less than 10 the time, δ is 5; When node neighbor node number was not less than 10, δ was 2.
The described w of above-mentioned steps S2-1
qWith node q upstream bandwidth u
qFunction be
Or w
q=u
q
Compared with prior art, the present invention has the following advantages and beneficial effect:
1) the inventive method takes into full account the importance of data block coding digital content, time urgency and the local scarcity that data block is play, priority scheduling is important and from playing the data block selection method that the time limit is near, hop count is less, under the less situation of node neighbor node number, can obtain more excellent distribution delay performance, simultaneously, the packet loss of data significantly decreases than the LU dispatching method;
2) adopt the node selecting method that comes the select target node according to the node upstream bandwidth, preferentially target data block is sent to the high node of upstream bandwidth, quickened the speed of data block diffusion, simultaneously, the node selection course incorporates the probability random device, reduced in the push-model certainty node and selected to exist a plurality of nodes to select simultaneously and cause the data distribution redundant, improved utilization rate of network resource to same high upstream bandwidth node propelling data.
Description of drawings
Fig. 1 is the meshed network topology schematic diagram that participates in the stream distribution session among the embodiment, (a) be the initial condition schematic diagram in nodal cache district, (b) be the schematic diagram of scheduling node select target data block and destination node, (c) transmit the schematic diagram of target data block for scheduling node;
Fig. 2 is that target data block of the present invention is selected schematic flow sheet;
Fig. 3 is that destination node of the present invention is selected schematic flow sheet;
Fig. 4 is the comparison of the data block distribution delay performance of the present invention and LU/BA method;
Fig. 5 is the comparison of the dropout of data block rate of the present invention and LU/BA method.
Embodiment
The data dispatching method that the present invention proposes based on scheduling time limit and bandwidth perception, made full use of interior each data block of nodal cache at time that source server produces and the number of times that is scheduled, introduce the relative worth that delay factor calculates each data block, and the data block of therefrom selecting the relative worth maximum does not preferentially comprise neighbor node target data block and the upstream bandwidth maximum and transmits as target data block in neighbor node.
Describe technical solution of the present invention in detail below in conjunction with drawings and Examples:
Participate in the meshed network topology schematic diagram of stream distribution session as shown in Figure 1 in the present embodiment, source node S produces data block C with speed λ
1, data block C
1Size be 0.5Mbits, suppose that source node S produces a data block, i.e. λ=1 each dispatching cycle.Node n
1, node n
2, node n
3, node n
4, node n
5For participating in the node of stream distribution session, on the node＜u, v〉sign in, u represents the upstream bandwidth that this node is current, v represents the neighbor node number that this node is current, the data buffer area of present node is represented in the tabulation on each node next door, and the data buffer area memory contents is that the current data block that has received of this node and this data block do not shift out the current data buffer zone.
Source node can be divided into data block with the Media Stream behind the coding before transmitting Media Stream, source node S in the present embodiment adopts the branch block mode principle of one of a frame that data stream is carried out piecemeal and handles, because so the importance difference of Frame in the data reconstruction process is the importance rate difference of the data block of each new generation.The data block importance degree that the I frame is formed is higher than P frame and B frame, and the importance of P frame is higher than the B frame, and the importance rate of setting I frame, P frame, B frame in the present embodiment is respectively 3,2,1.
The concrete steps of present embodiment are as follows:
The selection of S1, target data block:
S1-1, shown in Fig. 1 (a), the new data block C that source node S at first produces
1, data block C
1The relative time r of generation be 1s, data block C is set
1Scheduling time limit initial value d
C0Be d
C0=r
1+ 2T
s, T
sBe distributing data piece C
1The used time, can be by computing formula
Calculate, wherein, size
C1Be data block C
1Size, T as calculated
sBe 0.5, data block C
1Scheduling time limit initial value d
C0Be 2s;
S1-2, source node S adopt poll, at random or other any feasible method with data block C
1Directly be pushed to neighbor node n
1, the data block that source node sends has following attribute information: the time of generation, schedule periods limit value, importance rate etc.; Shown in Fig. 1 (b), node n
1Receive data block C
1After with data block C
1Deposit buffer area in, then comparison node n
1The relative worth that does not surpass the data block of playing the time limit in the buffer area
And select
Maximum data block is as target data block, wherein, and relative worth
Here node n
1Data buffer area in have only data block C
1, therefore, node n
1Directly with data block C
1As target data block;
S1-3, with data block C
1Before the forwarding, node n
1D by formula
k=d
K-1+ δ * T
sRecomputate and modifying target data block C
1Schedule periods limit value d
k, this moment data block C
1Current scheduling time limit d
K-1Be 2s, data block C
1The distribution time T
sBe 0.5, because node n
1The neighbor node number less than 10, so delay factor δ is 5, data block C as calculated
1Scheduling time limit d
kBe 4.5s.
2, the selection of destination node:
The periodic cache exchanging information of meeting between S2-1, the node in netted non-structure P2P stream media system and the neighbor node, therefore, so node n
1With neighbor node n
2, n
3, n
5The data block information of the periodic mutual buffer area of meeting, node n
1The neighbor node collection be { n
2, n
3, n
5, node n
1Do not contain target data block C from concentrated the selecting of neighbor node
1Neighbor node subclass B because node n here
1The buffer area of neighbor node centralized node all do not contain target data block C
1, then above-mentioned neighbor node subclass is B{n
2, n
3, n
5, according to formula
The selected probability P (i) of each node among the difference set of computations B, wherein, w
qBe node q upstream bandwidth u
qFunction, w
qWith u
qFunctional relation can represent
Also can be expressed as w
q=u
q
Below will with
For example is come the selected probability P (i) of each node among the set of computations B respectively: node n
2, n
3, n
5Upstream bandwidth be respectively 2,2,0.5, node n then
2Probability P (n
2) be 4/9, node n
3Probability P (n
3) be 4/9, node n
5Probability P (n
5) be 1/9;
S2-2, node n
1From neighbor node subclass B, select probability greater than any node of set point as destination node, preestablish numerical value 0.3, node n
2And n
3Probability all greater than set point 0.3, node n
1At random from node n
2With node n
3In select arbitrarily node n
3As destination node.
S3, node n
1To destination node n
3Send target data block C
1, receive the data block C that source node S is transmitted simultaneously
2And with the data block C that receives
2Deposit buffer area in, circulation is carried out S1-2 to the stream distribution conversation end then;
Shown in Fig. 1 (c), when whole stream distribution session does not finish, source node constantly produces new data block, and according to poll, at random or other any feasible method the data block that produces directly is pushed to the neighbor node of source node, neighbor node is transmitted data block again between the node of whole stream distribution session.In order to verify the advantage of the inventive method, on the P2PTV-Sim emulation platform, the inventive method is tested.P2PTV-Sim is under the subsidy of European Union's the 7th framework planning NAPA-WINE project, by Turin, Italy Polytechnics take the lead the exploitation P2P Streaming Media analogue system platform.In test, setting the node number that participates in the stream distribution session is 1000, and the overlay network topology is taked ways of connecting at random, and average neighbours' number of each node is 10, source node produces 200 data blocks according to the speed of speed 1Mbps, and the delay time limit of data distribution is set at 30s.Fig. 4 has provided the comparison of the data block distribution delay performance of the inventive method and LU/BA method, and each node is received the maximum delay performance of all data blocks from figure, and the inventive method is than the average maximum delay decline of LU/BA method 3s.Fig. 5 has provided the comparison of the dropout of data block rate of the inventive method and LU/BA method, and as seen from the figure, data loss rate of the present invention is 0 substantially, and the dropout of data block rate of LU/BA method is on average about 2%.
Claims (5)
1. the data dispatching method based on scheduling time limit and bandwidth perception is characterized in that, may further comprise the steps:
The selection of S1, target data block, this step further comprises following substep:
S1-1, source node calculate the initial value d of the scheduling time limit d of the new data block that produces
0=r+2T
sAfter data block is directly sent to scheduling node, wherein, r is the moment that data block produces, T
sBe the used time of distributing data piece;
S1-2, scheduling node deposit the data block that receives in buffer area, compare the relative worth that surpasses the data block of playing the time limit in the buffer area then
And select
Maximum data block wherein, is counted as target data block
S1-3, scheduling node be d by formula
k=d
K-1+ δ * T
sRecomputate the also schedule periods limit value d of modifying target data block
k, wherein, d
K-1Be the current scheduling time limit of target data block, T
sBe the distribution used time of target data block, δ is a delay factor;
The selection of S2 destination node, this step further comprises following substep:
S2-1, scheduling node are concentrated from neighbor node and are selected the neighbor node subclass B that does not contain above-mentioned target data block, calculate the probability of each node among the neighbor node subclass B
Wherein, w
qBe node q upstream bandwidth u
qFunction;
S2-2, scheduling node from neighbor node subclass B, select probability greater than any node of set point as destination node;
S3, scheduling node send target data block to destination node, receive the data block that other neighbor node or source node are transmitted simultaneously, and circulation is carried out S1-2 to the stream distribution conversation end then.
2. the data dispatching method based on scheduling time limit and bandwidth perception according to claim 1 is characterized in that:
It is the mode that adopts poll that source node among the described step S1-1 directly sends to scheduling node with data block.
3. the data dispatching method based on scheduling time limit and bandwidth perception according to claim 1 and 2 is characterized in that:
The obtaining value method of delay factor δ among the described step S1-3 is:
When node neighbor node number less than 10 the time, δ is 5; When node neighbor node number was not less than 10, δ was 2.
4. the data dispatching method based on scheduling time limit and bandwidth perception according to claim 1 and 2 is characterized in that:
W among the described step S2-1
qWith node q upstream bandwidth u
qFunction be
5. the data dispatching method based on scheduling time limit and bandwidth perception according to claim 1 and 2 is characterized in that:
W among the described step S2-1
qWith node q upstream bandwidth u
qFunction be w
q=u
q
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105890637A CN102035888B (en) | 2010-12-15 | 2010-12-15 | Method for scheduling data based on scheduling period and bandwidth awareness |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105890637A CN102035888B (en) | 2010-12-15 | 2010-12-15 | Method for scheduling data based on scheduling period and bandwidth awareness |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102035888A true CN102035888A (en) | 2011-04-27 |
| CN102035888B CN102035888B (en) | 2012-10-31 |
Family
ID=43888204
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010105890637A Expired - Fee Related CN102035888B (en) | 2010-12-15 | 2010-12-15 | Method for scheduling data based on scheduling period and bandwidth awareness |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102035888B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103581055A (en) * | 2012-08-08 | 2014-02-12 | 华为技术有限公司 | Message order preserving method, flow scheduling chip and distribution type storage system |
| CN112751924A (en) * | 2020-12-29 | 2021-05-04 | 北京奇艺世纪科技有限公司 | Data pushing method, system and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1665223A (en) * | 2005-03-04 | 2005-09-07 | 清华大学 | A push and pull based pure distributed data exchange method |
| WO2009002325A1 (en) * | 2007-06-28 | 2008-12-31 | Thomson Licensing | Queue-based adaptive chunk scheduling for peer-to-peer live streaming |
| CN101645921A (en) * | 2009-04-17 | 2010-02-10 | 中国科学院声学研究所 | Updating method of fragment popularity |
-
2010
- 2010-12-15 CN CN2010105890637A patent/CN102035888B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1665223A (en) * | 2005-03-04 | 2005-09-07 | 清华大学 | A push and pull based pure distributed data exchange method |
| WO2009002325A1 (en) * | 2007-06-28 | 2008-12-31 | Thomson Licensing | Queue-based adaptive chunk scheduling for peer-to-peer live streaming |
| CN101645921A (en) * | 2009-04-17 | 2010-02-10 | 中国科学院声学研究所 | Updating method of fragment popularity |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103581055A (en) * | 2012-08-08 | 2014-02-12 | 华为技术有限公司 | Message order preserving method, flow scheduling chip and distribution type storage system |
| CN103581055B (en) * | 2012-08-08 | 2016-12-21 | 华为技术有限公司 | The order-preserving method of message, flow scheduling chip and distributed memory system |
| CN112751924A (en) * | 2020-12-29 | 2021-05-04 | 北京奇艺世纪科技有限公司 | Data pushing method, system and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102035888B (en) | 2012-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | A QoS-oriented distributed routing protocol for hybrid wireless networks | |
| CN100477641C (en) | Data dispatching method of stream medium request broadcast system | |
| CN102868602A (en) | Improved routing method based on social correlation degree in delay tolerant network | |
| JP2019122040A (en) | Network source reuse and routing mechanism defining multi-source by software | |
| CN102811221B (en) | Push mechanism-based scheduling method of peer-to-peer network live streaming media data package | |
| CN101965031A (en) | Maximum probability-based cognitive radio multi-path multicast routing method | |
| CN101895954A (en) | Opportunistic network routing method based on incremental transmission of packet index | |
| CN102035888B (en) | Method for scheduling data based on scheduling period and bandwidth awareness | |
| CN103124292A (en) | Method and device for scheduling data in P2P (peer-to-peer) stream media system | |
| Ren et al. | Fast-mesh: A low-delay high-bandwidth mesh for peer-to-peer live streaming | |
| Atero et al. | A low energy and adaptive architecture for efficient routing and robust mobility management in wireless sensor networks | |
| CN104469874A (en) | Message forwarding method of opportunistic network based on probability centrality | |
| CN103312612A (en) | Optimizing method and system for multi-constrain quality-of-service routing | |
| CN114339941A (en) | Opportunistic network routing method based on cluster division | |
| CN103581329A (en) | Construction method for topological structure based on clustered peer-to-peer network streaming media direct broadcast system | |
| CN107171957B (en) | Self-adaptive DTN routing algorithm based on resource limited condition | |
| CN102571913A (en) | Network-transmission-overhead-based data migration method | |
| Liu et al. | A survey of transport protocol for wireless sensor networks | |
| Logambigai et al. | QEER: QoS aware energy efficient routing protocol for wireless sensor networks | |
| Bradai et al. | On the optimal scheduling in pull-based real-time p2p streaming systems: layered and non-layered streaming | |
| Liang | Energy-efficient, reliable cross-layer optimization routing protocol for wireless sensor network | |
| CN106454937A (en) | Distribution control method for caching space of mobile terminal | |
| CN101977158A (en) | Method for realizing congestion control in multi-source application layer multicast | |
| CN103402262B (en) | Wireless Sensor Networks assemblage method | |
| CN102035894B (en) | Distance-based state synchronization method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121031 Termination date: 20141215 |
|
| EXPY | Termination of patent right or utility model |