Transcript download and video and x

H.264/SVC Video
Transmission Over P2P
Networks
Garry Savulkin & Barak Zacharish
Supervisors:Dr. Ofer Hadar, Dr. Evgeny Kaminsky
COMPANY
LOGO
Contents
1
Motivation
2
Scalable Video Coding
3
Developed Components
4
Simulation
5
Conclusions
Motivation
Video
Adaptation
Device
Characteristics
 CPU capability
 Screen resolution
 Apps requirements
Transmission
conditions
 Network type
 Bandwidth
 Latency
Current approach
Video parameters :
176x144 pixels
7.5 frames/sec
Minimal Quality
Encoding . . .
File
request
Main video
storage server
1920x1080
30 fr/sec
Max Q
352x288
30 fr/sec
Max Q
Local
P2P server
176x144
7.5 fr/sec
Min Q
Scalable Video Coding
Scalable Video Coding
QUALITYVideo File
2
0
Minimal Quality
1
Medium Quality
Maximal Quality
Scalable Video Coding
RESOLUTION
704 x 576
2
352 x 288
1
176 x 144
0
Minimal Quality
Medium Quality
Maximal Quality
Scalable Video Coding
COMBINED
SCALABILITY
FRAME
RATE
704 x 576
352 x 288
176 x 144
Minimal Quality
Medium Quality
2
1
7.5 fr/s
3.75 fr/s
0
1.875 fr/s
2
1
7.5 fr/s
3.75 fr/s
0
1.875 fr/s
2
1
7.5 fr/s
3.75 fr/s
0
1.875 fr/s
Maximal Quality
Scalable Video Coding
• Increasing demand for video content with adaptation to
applications requirements such as display and
processing capabilities of various target devices
• Video content is being transmitted via various network
types which differ from each other in sense of bandwidth,
bit rate, error prone etc.
SVC allows you to find the optimal match between the
client target device capabilities and network transmission
conditions
SVC Encoding and Layers Extraction
320x240 pixels
7.5 frames/sec
Minimal quality
H.264
Encoder&
Bitstream
Extractor
480x320 pixels
15 frames/sec
Maximal Quality
1920x1080 - 30frames/sec - Maximal quality
Original Video
H.264/SVC
H.264/SVC codec implementation has the ability to
BitStream
deprive layers only from top to bottom
Extractor…
Example:
Clients
Broadcast
P2P
Server
Main Data
Storage
Server
Barak
240-by-320 pixels
1.75 frames/sec
Minimal quality
Enhancement
Layer 2
Enhancement
Layer 1
Base Layer
Garry
240-by-320 pixels
3.5 frames/sec
Minimal quality
Using H.264/SVC
352 x 288 pixels
30 frames/sec
Maximal Quality
Extraction . . .
176x144
7.5 fr/sec
Min Q
352x288
30 fr/sec
Max Q
1. Video encoding is a highly CPU consuming process
2. Redundant data transmission
Developed components
Extractor
- Supports the extraction of N-th layer
- Creates instructions frame for future
reconstruction
- Constructs “ready-to-send” packets
Inserter
- Reconstructs layers according to
predefined instructions
- Constructs “ready-to-play” video files
- Sends video stream to client(in future)
P2P Server
Client
- Manages files exchange with multi-client
connections
- Manages server’s Log files
- Manages server’s Data Storage
Topology with our components
Clients
Clients
Inserter
Inserter
P2P
Server/Client
P2P
Server/Client
Inserter
P2P
Server/Client
Clients
Extractor
Inserter
P2P
Server/Client
P2P
Server/Client
Clients
CDN Simulation
Content Distribution Network (CDN) is a system of
computers networked together across the Internet that
cooperate transparently to forward stage content closer
to end users.
Simulation features:
• Cooperative push based content management policy
• LRU cache replacement policy
• TCP / IP networking
http://oswinds.csd.auth.gr/~cdnsim/
CDN Simulation - Results
Mean response time vs. Requests intensity
Mean
Response
Time [Sec]
25.0
20.0
H.264/SVC
implementation
15.0
Our
implementation
10.0
5.0
0.0
0
10
20
http://oswinds.csd.auth.gr/~cdnsim/
30 λ [Requests/Sec]
Proposed approach
Video parameters :
176 x 144 pixels
7.5 frames/sec
Minimal Quality
352x288
30 fr/sec
Max Q
Layers extraction . . .
File
request
Extractor
Enhancement
layers data
176x144
7.5 fr/sec
Min Q
Inserter
Conclusions
Or, why SVC integrated with our tool is better
for video transmission over P2P network ?
COMPANY
LOGO