Multi-Channel Wireless Networks: Capacity, Protocols, and Experimentation - PowerPoint PPT Presentation

Loading...

PPT – Multi-Channel Wireless Networks: Capacity, Protocols, and Experimentation PowerPoint presentation | free to download - id: 24a990-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Multi-Channel Wireless Networks: Capacity, Protocols, and Experimentation

Description:

An interface can only use one channel at a time. Channel 1. Channel c. 13. Interface Model ... design. Experimentation. 45. Testbed. Channel abstraction module ... – PowerPoint PPT presentation

Number of Views:40
Avg rating:3.0/5.0
Slides: 80
Provided by: crhcIl
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Multi-Channel Wireless Networks: Capacity, Protocols, and Experimentation


1
Multi-Channel Wireless NetworksCapacity,
Protocols, and Experimentation
  • Nitin Vaidya
  • University of Illinois at Urbana-Champaign
  • www.crhc.uiuc.edu/wireless
  • Joint work with Pradeep Kyasanur,
  • Chandrakanth Chereddi

Keynote talk at WASA conference, Xian,
China August 2006
2
Multi-hop Wireless Networks
  • Two wireless paradigms
  • Single hop versus Multi-hop
  • Multi-hop networks
  • Mesh networks, ad hoc networks, sensor networks

3
Wireless Capacity
  • Wireless capacity limited
  • In dense environments, performance suffers
  • How to improve performance ?

4
Improving Wireless Capacity
  • Exploit physical resources
  • Exploit diversity/multiplicity of resources
  • Examples

5
Exploit Infrastructure
  • Infrastructure provides a tunnel to forward
    packets

infrastructure
BS1
BS2
B
C
D
E
A
Z
Ad hoc connectivity
X
6
Exploit Antennas
  • Steered/switched directional antennas

B
C
A
D
7
Improve Spatial ReusePower/Rate/Carrier-Sense
Control
Transmit Spatial Power Rate
reuse High High Low Low
Low High
A
B
C
D
A
B
C
D
8
Exploiting Diversity
  • Exploiting diversity
  • requires suitable protocols

9
This Talk
  • Utilizing multiple channels in wireless networks
  • Capacity bounds
  • Protocol design
  • Experimentation

10
Multiple Channels
  • Typically, available frequency spectrum is split
    into multiple channels

Large number of channels available
11
Channel Model
  • c channels available
  • Bandwidth per channel W

12
Radio Interfaces
  • An interface can only use one channel at a time
  • Switching between channels may incur delay

13
Interface Model
  • Reducing hardware cost allows formultiple
    interfaces
  • m interfaces per node Typical values of m small

14
Channel-Interface Scenarios
  • Scenario 1 m c One interface per
    channel

Common case today
With sufficient hardware
15
Channel-Interface Scenarios
  • Scenario 2 m lt c A host can only be on
  • a subset of channels

This is the likely scenario
16
Need for New Protocols
  • When m lt c , new protocols are needed
  • How to assign m interfaces to c channels?
  • When to switch an interface among channels?
  • How to select good routes?

17
Outline
  • Utilizing multiple channels in wireless networks
  • Capacity bounds
  • Protocol design
  • Implementation issues

18
Capacity Analysis
  • How does capacity improve if more channels are
    added ?
  • How many interfaces needed to best use c channels
    ?
  • Clearly, c interfaces sufficient
  • Not always possible to have c interfaces

19
Worst Case
  • Worst case capacity is m/c fraction of
    thebest-case

Channel data rate W c interfaces cW
throughput m interfaces mW throughput
B
A
  • What is the average scenario ?

20
Capacity ?Gupta-Kumar
  • Random source-destination pairs among randomly
    positioned n hosts in unit area, with n ? 8

21
Capacity ?
  • l minimum flow throughput
  • Capacity n l

22
Capacity Constraints
  • Capacity constrained by available spectrum
    bandwidth
  • Other factors further constrainwireless network
    capacity

23
Connectivity ConstraintGupta-Kumar
  • Need routes between source-destination pairs
  • Places a lower bound on transmit range

A
A
D
D
Not connected
Connected
24
Interference Constraint Gupta-Kumar
  • Interference among simultaneous transmissions
  • Limits spatial reuse

d
D
C
(1D)d
A
B
? is a Guard parameter
25
Capacity of Wireless NetworksGupta-Kumar
  • Bit rate for each transmission W
  • Capacity increases with n and c as

Capacity increases linearly with channels
26
Capacity of Wireless NetworksGupta-Kumar
  • Result holds when m c

1
W
1
1
W
1
W
m c
c
27
Capacity Problem
  • What if fewer interfaces ?How does the network
    capacity scale with channels?

Additional constraints on capacity become relevant
28
Interface Constraint
  • Throughput is limited by number of interfaces in
    a neighborhood
  • Interfaces, a resource

k nodes in the neighborhood ? total throughput
k m W
29
Destination Bottleneck Constraint
  • Per-flow throughput limited bytotal number of
    flows at a host

If node throughput T Per-flow throughput T /
f
f incoming flows
D
30
Mutlti-Channel Network Capacity MobiCom05
Network Capacity
Channels (m1)
31
Outline
  • Utilizing multiple channels in wireless networks
  • Capacity bounds
  • Protocol design
  • Experimentation

32
Protocol Design Goals
  • Utilize all the available channels
  • Without degrading network connectivity

33
Insights from Capacity Analysis (1)
  • Static channel allocation does not yield optimal
    performance in general in all topologies
  • Must dynamically switch channels
  • Need protocol mechanisms for channel switching

Channel 1
B
A
C
2
D
3
34
Insights from Capacity Analysis (2)
  • Optimal transmission range function ofdensity
    of nodes andnumber of channels
  • Goal of interfering nodes of channels

35
Insights from Capacity Analysis (3)
  • Routes must be distributed within a neighborhood
  • This is not necessary in single channel networks

D
D
F
F
B
B
E
A
A
E
C
C
Multi-Channel (mltc)Optimal strategy
Single Channel (mc1)Optimal strategy
36
Insights from Capacity Analysis (4)
  • Channel switching delay potentially detrimental,
    but may be hidden with
  • careful scheduling, and/or
  • additional interfaces

37
Protocol Design IssueWhich Layers to be
Multi-Channel Aware?
  • Practical decision
  • Above MAC layer
  • Allows use of unmodified 802.11

38
Separation of Responsibility
  • Interface management Shorter timescales
  • Dynamic channel assignmentto interfaces
  • Interface switching
  • Routing Longer timescales
  • Multi-channel aware route selection metrics

802.11
39
Interface Switching Wcnc2005
  • Interfaces may be switched or kept fixed
  • Classification
  • Static strategy All interfaces of a node fixed
  • Dynamic strategy All interfaces of a node can
    switch
  • Hybrid strategy Some interfaces fixed, others
    switch
  • We use a hybrid strategy requiring at least two
    interfaces per node

40
Channel Assignment Our Approach
  • 2 interfaces much better than 1
  • Hybrid channel assignment Static Dynamic

Channel assignment locally balanced
41
Routing Approach
  • Existing routing protocols can be operated over
    interface management protocol
  • May not select good routes
  • Does not consider cost of switching interfaces
  • Our solution
  • Develop a new channel-aware metric
  • Metric can be incorporated into any on-demand
    routing protocol

42
Selecting Channel Diverse Routes
  • Most routing protocols use shortest-hop metric
  • Not sufficient in most multi-channel networks

1
3
4
4
3
B
C
A
A needs route to C Route A-B-C better
4
4
D
E
F
2
4
2
Prefer channel diverse routes
43
Impact of Switching Cost
  • Interface switching cost has to be considered
  • A node may be on multiple routes, requiring
    switching

1
4
3
B
C
A
2
Route A-B-C in use D needs route to F Route D-E-F
better
3
D
E
F
2
4
2
4
2
Select routes that do not require frequent
switching
44
Outline
  • Utilizing multiple channels in wireless networks
  • Capacity bounds
  • Protocol design
  • Experimentation

45
Testbed
  • Channel abstraction module implemented in Linux
    2.4
  • Experiments done on testbed nodes having two
    wireless radios
  • Radios are operated in IEEE 802.11a mode

Soekris 4521
46
Testbed architecture
One radio mesh node
Two radio mesh node
One radio unmodifed client
Internet gateway node
47
Need for kernel support
  • Linux used as case study
  • Key requirements
  • User applications must work unmodified
  • Operate with off-the-shelf wireless hardware
  • Kernel support needed to meet requirements
  • Support can be added through a separate module

48
Need for kernel support
  • No support to choose channels based on destination
  • Multi-channel broadcast support is absent
  • Initiating switching from user space has high
    latency - frequent switching not possible

49
Need for kernel support
  • Interface management needs to be hidden from
    data path
  • Buffering packets for different channels
  • Scheduling interface switching

Ch. 2
Interface switches channel
Packet to B
buffer packet
Ch. 1
Packet to C
Packet to C arrives
50
Where to add support?
  • In the device driver
  • Tied in to driver, cannot handle multiple
    interfaces
  • In the network layer
  • Multiple interfaces exposed to network layer
  • Some protocols like ARP need to be modified
  • Between network layer and device driver
  • Easy to add without modifying existing code
  • No changes to ARP, IP stack needed

51
Proposed architecture
Multi-channel protocol
User Applications
  • Channel abstraction module provides kernel
    multi-channel support
  • Module implemented by extending Linux bonding
    driver

IP Stack
ARP
Channel Abstraction Module
Interface Device Driver
Interface Device Driver
52
Channel Abstraction Module
  • Unicast Component
  • Allows choosing channels based on destination
  • Broadcast Component
  • Multi-channel broadcast support
  • Queueing and Scheduling Component
  • Queue packets if interface is not immediately
    available
  • Schedule interface switching

53
Components
No
Yes
Broadcast?

Node 1 ath0 1
Node 2 ath1 2
Node 3 ath1 3
Unicast Table Address
Interface Channel

1 ath0
2 ath1
3 ath1
Broadcast Table Channel Interface
Queue Packets
54
Configuring tables
  • Unicast and broadcast tables can be set/modified
    from userspace through ioctl calls
  • Different multi-channel protocols can use
    different policies for setting the tables
  • Operation analogous to routing
  • Routing table in kernel can be set up by an
    userspace routing daemon

55
Example interaction
Userspace protocol
Channel abstraction module
AddValidChannel( ath0, lt1,2,3gt)
AddBroadcastEntry( ath1, lt2,3gt)
AddUnicastEntry( 192.168.0.2, ath0, 1)
DeleteUnicastEntry( 192.168.0.4, 3)
56
Scheduling algorithm
  • Interface is switched from current channel only
    if another channel has pending packets, and
  • Either rule 1
  • Current channel has no pending packets
  • Time spent on current channel greater than T_min
  • Or rule 2
  • Time spent on current channel greater than T_max
  • T_min , T_max choice affects switching overheads

57
Driver modifications
  • Minor modifications made to madwifi driver to
    improve performance
  • Turned off beaconing to reduce switching delay
  • By default, after channel switch card waits to
    hear a beacon - can be as large as 100 ms
  • Instead, added support to specify default BSSID
  • Added support to measure driver queue length
  • To improve scheduling performance

58
Ongoing work
  • Testbed comprises of 20 nodes
  • Detailed measurements of multi-channel protocols
    is in progress

59
Summary
  • Capacity results hint at significantperformance
    benefits usingmany channels with few interfaces
  • Need suitable protocols to exploit the channels
  • Cross-layer interactions
  • In general, available diversity can be exploited
    using suitable protocol

60
Thanks!
  • www.crhc.uiuc.edu/wireless

61
Thanks!
  • www.crhc.uiuc.edu/wireless

62
Static Strategy - Approach 1Draves2004Mobicom
Assume two interfaces per node, 4 channels
1,2
1,2
1,2
B
C
A
All channelsnot utilized
D
E
F
1,2
1,2
1,2
  • All nodes use common set of channels
  • Easiest extension when multiple channels available

63
Static Strategy - Approach 2Raniwala2005Infocom
Assume two interfaces per node, 4 channels
1,2
2,3
1,2
B
C
A
Some routes are longer
D
E
F
2,3
3,4
3,4
  • Different nodes may use different channels

64
Dynamic strategy So2004Mobihoc, Bahl2004Mobicom
1-4
1-4
1-4
B
C
A
Coordination may be needed before each
transmission
D
E
F
1-4
1-4
1-4
  • Interfaces can switch channels as needed
  • Transmissions can occur dynamically on any channel

65
Hybrid strategyNasipuri1999Wcnc, Jain2001Ic3n
  • One common channel used as control channel
  • Remaining channels used as data channels

66
Supporting broadcast
  • Send a copy of packet over all channels
  • Strength All neighbors receive broadcast,
    similar to single channel network
  • Drawback Cost of broadcast goes up
  • Extensions
  • Use a dedicated broadcast channel (with a third
    interface)
  • Send broadcast over subset of channels

67
Protocol Operation
  • Each node has 2 interfaces
  • 1 fixed, 1 switchable

1
3
1
3
B
C
A
  • Timeline
  • A send to B
  • D send to A

1
D
E
F
2
4
2
Connectivity maintained all channels used
68
Example
1) A sends Hello(1)
2) B sends Hello(2, A1)
3) E sends Hello(4, A1, B2)
4) D sends Hello(3, A1, B2, E4)
69
Simulations
  • Qualnet 3.6 simulator
  • IEEE 802.11a channels (varied from 1 to 12)
  • Proposed protocols use two interfaces per node
  • Interface switching delay is 1 ms

70
UDP Chain topology
Throughput (Mbps)
Chain Length (in hops)
71
FTP Chain topology
Throughput (Mbps)
Chain Length (in hops)
72
MCR New Routing Metric
  • Measure switching cost for a channel
  • Measure total link cost of a hop
  • ETT cost Draves2004Mobicom Switching cost
  • Combine individual link costs into path cost

73
Routing Protocol
  • Incorporate metric in on-demand source-routed
    protocol (similar to DSR)
  • RREQ messages modified to include link costs
  • Destination can compute best path
  • Using cost information in RREQ

74
Throughput in Random Networks(50 nodes, 500m x
500m area)
Normalized throughput
Topology Number
75
Throughput with Varying Load
Normalized throughput
Number of flows
76
Implementation details
Hello Protocol Routing Protocol
User Applications
  • Abstraction layer simplifies the use of multiple
    interfaces
  • User-space daemon implements routing and Hello
    protocol
  • Netfilter-based module implements support for
    on-demand routing

On-demand routing support
IP Stack
Interface and Channel Abstraction Layer
Interface
Interface
77
Channel abstraction layer
Address Inte l
192.168.0.2 ath0 1
192.168.0.3 ath1 2
192.168.0.4 ath1 3
Unicast Mapping Table
Add Packet to Queue
Channel Interface
1 ath0
2 ath1
3 ath1
Broadcast Mapping Table
1
2
3
Schedule packet transmissions
78
Testbed measurements
  • We use Netgear cards based on atheros chipset
  • Madwifi driver has been modified to reduce
    switching delay
  • Measured switching delay is approx. 5 milliseconds

79
Throughput
A
B
4 flows A-gtB, B-gtC, C-gtD, D-gtA 8 flows In
addition A-gtD, B-gtA, C-gtB, D-gtC
C
D
Channel data rate is 6 Mbps
Aggregate throughput (Mbps)
Number of channels
About PowerShow.com