Title: Course Matters
1Course Matters
2Recent Papers
3Good PostersBad Posters
4Improving I/O Performance of Intermediate
Multimedia Storage Node
- Pal Halvorsen
- Thomas Plagemann
- Vera Goebel
- Multimedia Systems 03
5Problem
- Improve the performance of I/O in integrated
multimedia storage node - 3 areas of improvement are identified
- reduce memory copy
- checksum computation
- FEC computation
6Reduce Memory Copy
- File system maintain pointer to an area in memory
- Communication system maintain pointer to same
area in memory - Memory copy avoided!
7Network Level Framing
- Packet payloads are stored with checksum
- When packets are retrieved for sending,
destination address in header is updated - Checksum is updated with new destination
- No need to recompute checksum for payload
8Integrated Error Management
- Data are stored on RAID-4 (single parity checks)
- Use the same error correcting code for RAID and
packets - Avoid multiple computation of error correcting
code
9Experiment Memory Copy
- Read 28662512B file 38 times in a loop
- The time to transmit data through the storage
node is reduced by 45-50 when there is no CPU
load, and by 70-73 when CPU load is high
10Experiment Network Level Framing
- Transmit 255MB file
- Time to calculate checksum is reduced by 95-99
- Time spent in kernel is reduced by 51-61
11Experiment Integrated Error Management
- With encoding FEC, the maximum throughput is
22-24 Mbps. - Without encoding FEC, the maximum throughput is
1Gbps
12Lets try again..
13Improving I/O Performance of Intermediate Media
Storage Node
- Pal Halvorsen
- Thomas Plagemann
- Vera Goebel
14Contributions
- Improve performance by
- reducing memory copy
- reducing checksum computation
- reducing ECC computation
15Reduce Memory Copy
- One shared copy of data for different OS
component
Memory
Network System
File System
data
16Network Level Framing
store payload with payloads checksum
1
3
update header and checksum
2
read payload with checksum
- Reduce time to packetize data and compute
checksum for data
17Integrated Error Management
reuse RAID ECC as ECC packet
RAID
ECC
data
data
data
ECC
- Avoid multiple computation of error correcting
code
18Results Zero Copy
19Results Network Level Framing
Accumulated UDP Protocol Execution Time for
sending 225MB file
20Results Integrated Error Management
With ECC Encoding Without ECC Encoding
22-24 Mbps 1 Gbps
Maximum Throughput with/without Encoding
using Cauchy-based Reed Solomon Erasure Code
21Ad Hoc Networks
22Mobile Ad Hoc Network
Radio
Router
Host
23Animation
- http//www-i4.informatik.rwth-aachen.de/mesut/man
et/manet_en.html
24Mobile Ad Hoc Network
25Examples
- Battlefield
- Highway
- Disaster Zone
26Challenges
- All the difficulties of wireless LAN
- Plus
- Nodes can move
- Connections can go up/down
- No fix route
27Two Papers
- IEEE JSAC
- Special Issues in Wireless Multimedia
- Baochun Li from U. of Toronto
- Shiwen Mao from Polytechnic U.
28NonStop Continuous Streaming Service on MANET
- Baochun Li
- IEEE JSAC 2004
29Streaming over MANET
30Network Partition Problem
31Only Solution ..
- Predict Partition
- Replicate Service
32Partition Prediction
33Network Partition Problem
34How to Predict Partition?
- given velocity of each node
- cluster nodes into mobile groups
- find mean group velocity
35Clustering Algorithm
36Clustering Algorithm
37Clustering Algorithm
38Clustering Algorithm
39Choosing Server
?
40How to Choose Server?
- find stable group
- choose server within stable group with the most
similar velocity
41Stable Group
probability
A
B
distance
mean lt radio range and variance is not too large
42Stable Group
B
A
C
H
D
G
E
F
43Stable Group
BCD are in my group
AGH are in my group
B
A
C
H
D
G
E
F
44Stable Group
BCDGH are in my group
ABDGH are in my group
B
A
C
H
D
G
E
F
45How to Choose Server?
- find stable group
- choose server within stable group with the most
similar velocity
46Summary
- Server construct mobile group by clustering nodes
using velocity - Use mean mobile group velocity to predict network
partition - Replicate service before partition to ensure
continuous service
47Summary
- Node construct stable group by comparing distance
over time - Choose server within stable group with most
similar velocity
48Multipath Transport and Multistream Coding for
MANET
- Shiwen Mao et. al.
- IEEE JSAC
49Single Path Transport
50Multipath Transport
51Multipath Transport
52Video Coding 1 of 3
I
P
P
P
P
P
P
P
Typical Frame Dependency
53Video Coding 1 of 3
I
P
P
P
I
P
P
P
Multistream Coding
54Dynamic Reference Frame
- Choose last received frame as reference
believed to be received
55Predict Network States
GOOD )
NACK
ACK
BAD (
56Example
ACK0
ACK2
ACK4
ACK6
0
2
4
6
8
10
1
3
5
7
9
NACK1
NACK3
NACK5
ACK7
57Video Coding 2 of 3
I
P
P
P
Enhancement Layer
I
P
P
P
Base Layer
58Video Coding 2 of 3
BL
S
D
EL
59Video Coding 2 of 3
NACK
NACK
60Video Coding 2 of 3
S
D
BL
61Video Coding 3 of 3
- Multiple Description Coding
62MDMC
63Example of MDMC
359
363
370
7
Typical Reference Frame
64Example of MDMC
359
363
370
MDMC Reference Frame
65Example of MDMC
359
363
9
66Example of MDMC
359
?
9
67Example of MDMC
361
359
363
370
9
68Example of MDMC
359
?
9,2
69Comparisons
Reference Frame LayeredCoding MDMC
Feedback
Buffer
Decoding Delay
70Improving Multicast
71Organizing Multicast Receivers
- Brian Neil Levine
- Sanjoy Paul
- JJ Garcia-Luna-Aceves
- Multimedia Systems 03
72Retransmission in Mcast
73Idea
- Ask a neighbour for a missing packet
74Helper Tree
75Pick Helper By Hop
76Pick Helper by Hop
- Not entirely accurate
- Need to consider latency, link condition etc.
77Pick Neighbour By Latency
78Pick Helper by Latency
- How to measure latency?
- unicast?
- multicast?
- shared-tree? per-source tree?
79Ideal Location of Helper
- share a common path, and is closer to source
(acceptable)
80Idea
- Suppose node A and B know their path back to the
source, then they can deduce if A is acceptable
to help B
81MTrace
D
F
A
B
C
H
G
I
E
82ERS
my path is ABCDF
D
F
A
B
C
H
G
I
E
83ERS
D
F
A
B
C
H
G
I
E
84Respond if
- acceptable
- not too many helpee
85Picking Helper
10
30
86Maintenance
- Periodic refresh states (soft-states)
- Periodic repeat procedure
87Summary
- Organized receivers based on common path
- Enable peer-to-peer retransmission
88THE END