Download notes 2 - Department of Computer Science

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Live Streaming over
Subscription Overlay Networks
CS587x Lecture
Department of Computer Science
Iowa State University
Outline
Subscription Overlay Network (SONet)

Subscription and topology management
Streaming over SONet
Performance Study
Concluding Remarks
Subscription Overlay Network
Two components

One central server
 Streaming source (eg., live TV broadcast, etc.)

A number of subscribing nodes
 Pay monthly fee in return of video services
 A node can be offline/online/idling/playing
One Service

Streaming video data from source to all online nodes
Three goals
Minimize the server workload
Minimize network traffic
Minimize data latency
1
r
S
4
r
6
r
r
3
2
5
Solution I: Unicast
Dedicating one stream for each online
node
Advantages
Simple implementation
Good data freshness (shortest distance)
Disadvantages
Server bottleneck: not scalable
Solution II: IP Multicast
One multicast stream can serve many
clients simultaneously
Advantages
Simple implementation
Achieve all three goals
Disadvantages
IP Multicast is not widely deployed on the
Internet (due to security issues, etc.)
Solution III: Application Layer Multicast
A node receiving data can forward its
incoming stream to serve others
Existing ALM techniques
Chaining/ESM
NICE/ZIPZAG, etc.
c5
c3
c6
s
c1
c4
c2
Solution III: Application Layer Multicast
A node receiving data can forward its
incoming stream to serve others
Existing ALM techniques
Chaining/ESM
NICE/ZIPZAG, etc.
Advantages
each server stream can serve many clients
Leverage the entire network resource
c5
c3
c6
s
c1
c4
c2
Solution III: Application Layer Multicast
A node receiving data can forward its
incoming stream to serve others
Existing ALM techniques
Chaining/ESM
NICE/ZIPZAG, etc.
Advantages
each server stream can serve many clients
Leverage the entire network resource
Disadvantages
Only the playing nodes can contribute
Difficult to maintain topology, etc.
c5
c3
c6
s
c1
c4
c2
Observation and Motivation
A SONet may consist of a large number of
subscribers, but at any one time, only a small
percentage of them are playing

American watch TV 4 hours/day in average
A majority of SONet not playing may be idling

Unlike regular TV sets, a node not playing is likely to be
online
Recruiting appropriate idling nodes for data
forwarding can effectively reduce network traffic
Motivation Examples
S
A
R6
S
R1
R1
R1
R2
R2
R2
R3
R4
S
I
R3
R5
R7
R4
B
A
R6
I
R3
R5
R7
R4
B
Server workload?
Network traffic?
Data freshness?
A
R6
I
R5
R7
B
Subscription/Topology Management
Account database

The server maintains all subscriber information, including IP,
password, the amount of data forwarded (for discount purpose),
etc.
Topology graph

When a new member M joins, the server S detects its path to the
members


When a member M1 is asked to forward data to another member
M2, M1 reports the actual PATH(M1, M2) to the server



A path is denoted as PATH(S, M) = SR1R2M
The connections among the members are detected and updated as
needed
The topology graph becomes more and more accurate to the server
For each stream, the server records its actual streaming path

Given a router, the server can find out the set of streams flowing
through it.
Notations/Definitions
Path(X, Y)

The sequence of routers on the
shortest path from X to Y (as
known to the server)
S
R1
Ring(R, i)

R2
The set of routers that are i-hop
away from R
R3
I
Capacity(N)

The number of streams N can
forward
Local node and local router

A node and a router is local to each
other if they connect directly
R4
A
R6
R5
R7
B
Changing of Node Status
incentive
offline
online
idling
playing
A node can make itself only offline, idling,
or playing

Only the server can decide when a node can
become an incentive node
A node becomes online
When an offline node becomes online, the server may
bundle the streams flowing through N’s local router
X2
X2
X3
X3
X1
X1
R
Y1
N
Y2
R
Y1
Y3
X1=>Y1, X2=>Y2, X3=>Y3
N
Y2
Y3
X1=>N, X2=>Y2, N=>Y3
Bundling Procedure
A node becomes playing
The node is in incentive

S
Simply turn on its player
The node was in idling

X2
 Find the joint router Rj
R11
R10
R2
X1
R9
R5
 Search Ring(Rj, 0)


Prefer playing node
If not, recruit an incentive
node, more than one
candidates may be available
(choosing criteria?)
 Repeat on Ring(Rj, 1), …,
until a parent is found
Rx3
Rx1
R1
Find a parent (should be as
close as possible to N)
X3
Rj
P
R6
R7
R3
R8
R4
Ry1
Y1
Ry2
Y2
Ry3
Y3
N
Find Parent Candidates
Choosing Incentive Node
Cost(XY, I, N) =
X1
X1
Hop(X, I) +
Hop(I, Y) +
Hop(I, N) –
Hop(X, Y)
I
I
N
N
Y1
Y1
If N’s capacity is not 0
It can help serve others


The server finds the parent by searching
Ring(Rj, 0), …, but stops at Ring(Rj, d),
where d is Hop(N, Rj)
Since N can serve at least one child, it can
redirect its incoming stream to its parent’s
current child
A node becomes offline or idling
Find a new parent for each child of this
node

The parent can be either an incentive or
playing node
Performance Study
Performance Metrics

Mean Relative Delay (MRP)
 The MRP of a node X is to defined to HOP(S,
X)/StreamingPath(S, X)
 Measure the data freshness

Link Stress
 The total amount of traffic flowing through each network
links
Our study focus on



Effect of subscription size
Effect of topology size
Effect of active rate
Effect of Subscription Size
Concluding Remarks
SONet: a framework for video streaming over the
Internet



Similar to cable/satellite broadcast networks
Allows effective incentive mechanisms
Centralized subscription and topology maintenance
A new topology-oriented technique for building
application layer multicast


Unique in its ability of incorporating idling nodes to assist in
data forwarding
Simulation confirms its performance advantage
Future work???