Transcript Characterizing Wireless Networks in the National Wireless Testbed

Golden Gate Club Connectivity
1
Studies of Wireless Networks with
Realistic Physical Layer Emulation:
The ORBIT Test-Bed Facility
Funded by NSF NRT project #ANI-0335244 and DARPA IPTO
PnP Networks
www.pnphome.com
Contact: Richard E. Howard
[email protected]
Rutgers, The State University of New
Jersey
www.winlab.rutgers.edu
Contact: Prof. D. Raychaudhuri
[email protected]
2
Modeling Wireless Networks:
The Radio Problem

Ethernet Modeling:

All nodes in a subnet receive all packets
 Low error rate
 Emphasis on collision, routing, congestion, ...

Wireless Network Modeling

Packet reception depends on complex, changing RF conditions





Hidden nodes and range of link qualities
Hard to model—non-local, sensitive dependence on environment
 Computationally intractable—”Hall of Mirrors”
Extra control “knobs”—transmit power, channel, packet length, ...
High error rates under the best conditions
Conventional network modeling must be done after getting RF right.
3
Blocked Mission Traffic--Weighted Fraction (BloMiT-WeFra)
Legend
Perfect control
Cognitive control (CogCon)
Static configuration (SOA)
A
Reconfigure network, power/rate
management, delay low priority
data.
B
Reconfigure network,
power/rate management,
send buffered data.
C
BloMiT-WeFra
0. 1%
1%
A1
B
C1
A2
Adjust fragmentation
threshold, manage
power/data rate.
C2
Midway
10%
Waterloo
100%
Mission Time (mtime)
4
ORBIT: Testbed Overview



ORBIT consists of radio grid emulator + field trial network
Emulator used for detailed protocol evaluations in reproducible complex
radio environments
Field trial network for further real-world evaluation & application trials
Research
User of
Testbed
ns-2+ scripts
&
code
downloads
Static radio
node
Emulator
Global Internet
Firewall
Mapping
Mobility
Server
“Open”
API
3G BTS
High
Speed
Net
Wired
routers
Radio link
Dual-mode emulation Mobile node
Radio device
(robotic control)
1. Radio Grid for Lab Emulation
“Open”
API
Access
Point
(802.11b)
3G
access
link
Adhoc
link
End-user devices
2. Field Trial Network
5
ORBIT: Testbed Facilities

Simulation (Cluster)

Compute facility to run simulations (NS)
 Create extensions to ns-2 PHY modules for improved realism and
cross-layer

Emulation Grid

802.11a radio nodes (~20x20 @ 1m spacing)
 Mapping of various “typical” wireless net scenarios
 Open API for complete flexibility of OS/protocol software; Linux libraries

Field Trial System

Outdoor system for greater realism in protocol testing & for application
development, live demos, etc.
 3G base station router with IP interface
 ~50 open API 802.11a AP’s covering RU NB campus, some downtown
areas…
 Mobile AP’s on buses, etc.
6
ORBIT: Physical Facilities
•~12,000 sq ft (Grid + Lab. space + Offices)
•Rt 1 South @ Technology Center of NJ
•“Move in” late 2004
7
ORBIT: Radio Grid Scenarios

Use programmable, controlled interference in a physically small
environment.

An n x m array of identical radios on grid.
 A secondary array of programmable interferers

Mapping algorithm which matches “real-world” SNR vectors to
selected nodes on grid, using some nodes as interferers
8
ORBIT: Field Trial System
9
Interference Measurements
Using ORBIT Testbed
Walls
~1m
~1.5m
1,4
1,3
1,2
1,1
2,4
2,3
2,2
2,1
~4m
~5m
Link Nodes
~3m
Interfering Nodes
10
Packet Loss as a Function of Channel Spacing
For Different Packet Payload Sizes
1
256 B; 1.0 Mb/sec
512 B; 1.9 Mb/sec
768 B; 2.9 Mb/sec
1024 B; 3.9 Mb/sec
1280 B; 4.8 Mb/sec
Fraction of Dropped Packets
0.9
0.8
0.7
Packet Payload; Offered Load
0.6
0.5
0.4
0.3
0.2
0.1
0
1
2
3
4
5
Channel Spacing from Interferer
PnP-20040524 One sender, 1 receiver, 3 interferers
1 microsecond packet spacing set
11
Packet Loss as a Function of Channel Spacing
For Different Packet Sizes at 1/3 Lower Rate
1
256 B;
512 B;
768 B;
1024 B;
1280 B;
Fraction of Dropped Packets
0.9
0.8
0.7
0.67 Mb/sec
1.35 Mb/sec
2.0 Mb/sec
2.7 Mb/sec
3.4 Mb/sec
Packet Payload; Offered Load
0.6
0.5
0.4
0.3
0.2
0.1
0
1
2
3
4
5
Channel Spacing from Interferer
PnP-20040526 One sender, 1 receiver, 3 interferers
100 microsecond packet spacing set, 1 mW
12
Packet Loss as a Function of Throughput
For Different Channel Spacings
Fraction of Dropped Packets
1
Same
1
2
3
4
5
0.9
0.8
0.7
0.6
Channel Spacing
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1
1.5
2
2.5
3
3.5
Net Throughput (Mb/sec)
PnP-20040526 One sender, 1 receiver, 3 interferers
13
High Power Increases Channel Overlap
FN
Near
Sending
Nodes
Receiving
Nodes
Far
FN
Packet Loss % at 4 Mbps
Far
Packet Loss % at 4 Mbps
Near
100
100
10
10
1
1
Packet Loss %
Packet Loss %
0.1
0.1
0.01
0.01
0.001
0.001
01
Channel
Separation
50 mW
23
45
6 7
8 9
10mW
1mW
01
23
4 5
6 7
Channel Separation
50 mW
10mW
8 9
1mW
14
Optimizing Wireless Networks
Net A
Requires
Knowledge of
Application
Behavior
FN
Greatest
Improvement
Video subnet
optimized for
QOS
Ch 1
Ch 2
Ch 5
Ch 10
Video
Data
Adjacent Channel Interference
Both networks have reduced capacity
Net B
Partition Network Based on
Application Requirements
15
Network States (Measured)
State
BloMIt-WeFr Traffic Rate
Rate
(kb/sec)
1 One pair of nodes communicating
0.00
1,350
2 Add 3 pairs of nodes with similar traffic on nearby channel
3 Change new nodes to same channel as original
0.40
3,240
0.08
4,960
4 Three nodes leave and traffic rate increased for single link
Add 3 pairs of nodes with similar traffic on adjacent channel, higher
5
data rate and longer packets than state (2)
6 Change new nodes to same channel as original
0.00
2,700
0.89
1,224
0.40
6,480
0.22
8,380
0.07
8,380
0.00
3,830
0.98
260
0.38
9,400
0.14
9,400
7
Change new nodes to channel 6 as it becomes available (other nodes
leave).
8 Same as state (7) with command traffic switched to channel 1 link
Three nodes leave and traffic rate increased again for single link by
9
shortening time between packets.
10 Add 3 pairs of nodes with similar traffic on adjacent channel
11
Change new nodes to channel 6 as it becomes available (other nodes
leave).
12 Same as state (11) with command traffic moved to channel 1 link
16
Integrated Mission IT Metrics-Static Path Through Mission
Integrated Mission Traffic (GB)
Static Network
Integrated BloMIt-WeFr
Static Network
2000
1500
1000
500
0
0
1000
2000
3000
16
14
12
10
8
6
4
2
0
0
Mission Time (sec)
500
1000
1500
2000
2500
Mission Time (sec)
17
Integrated Mission IT Metrics-Optimized Path Through Mission
Integrated BloMIt-WeFr
Optimized
Integrated Mission Traffic (GB)
Optimized
200
70
180
60
160
140
50
120
40
100
30
80
20
60
40
10
20
0
0
0
0
500
1000
1500
2000
2500
500
1000
1500
2000
2500
Mission Time (sec)
Mission Time (sec)
Improvement potential for this mission profile
BloMiT-WeFr: 1,578 => 182
Mission Traffic: 14 GB => 62 GB
Note: This is wireless link-layer characterization only.
Guaranteed delivery protocol (e.g. TCP) would add
“thrashing” and increase the difference.
18