ConeBased Topology Control in Ad Hoc Networks

1
Cone-Based Topology Control in Ad Hoc Networks

• Li (Erran) Li1, V. Bahl2, J.Y. Halpern1, Y.M.
  Wang2 and R. Wattenhofer3
• Cornell University1
• Microsoft Research2
• ETH, Zurich3
• Presented by Guoliang Xing

2
Outline

• Motivation
• Design goals
• Basic cone-based algorithm
• Optimizations on the basic algorithm
• Performance results
• Summary

3
Motivation

• No topology control large transmission radius

• Lots of interference
• Low throughput
• High energy consumption

4
Motivation (contd)

• No topology control small transmission radius

• Network may partition

5
Motivation (contd)

• With topology control

• Low energy consumption
• Little interference
• High throughput

6
Design Goals

• Reduce transmission power
• Maintain global connectivity
• Distributed algorithm using local information

7
Related Work

• Rodoplu and Meng Jsac 03 Only connect to the
  power efficient immediate neighbors
• Li et al. Infocom 03 Only connect to the
  neighbors on local minimum spanning tree (MST)
• Song et al. mobihoc 04 Bound node degrees
• Burkhart et al. mobihoc 04 Reduce inference

8
Assumptions

• Circular communication range
• Receiver can infer the direction of sender
• Location information not required
• Symmetric links
• power(u?v) power(v?u)

9
Basic Cone-based Topology Control (CBTC) Algorithm

• Each node discovers neighbors by increasing
  transmission power until it finds a node in every
  cone of degree ? or it reaches the max power.

? 120o
10
Maintain Global Connectivity

• Resulted links may be asymmetric
• CBTC guarantees global connectivity if
• ?150o
• Link symmetry enforced if exists a link u?v,
  increase tx power of v to enforce v?u
• 150o is the tight upper bound to guarantee global
  connectivity

11
Guaranteeing global connectivity

• If AB lt Rmax and node A and B are not connected
  after the CBTC, there must exist a multi-hop path
  from A to B.
• Intuition there exist As neighbor X and Bs
  neighbor Y, such that XYltAB

 
r
t
B
d
A
u
s
12
Guaranteeing global connectivity

• If AB lt Rmax and node A and B are not connected
  after the CBTC, there must exist a multi-hop path
  from A to B.
• Intuition there exist As neighbor X and Bs
  neighbor Y, such that XYltAB

 
r
t
B
r2
A
r1
u
s
13
Guaranteeing global connectivity

• If AB lt Rmax and node A and B are not connected
  after the CBTC, there must exist a multi-hop path
  from A to B.
• Intuition there exist As neighbor X and Bs
  neighbor Y, such that XYltAB

 
r
t
h
Z
B
r2
A
r1
k
u
s
14
Guaranteeing global connectivity

• If AB lt Rmax and node A and B are not connected
  after the CBTC, there must exist a multi-hop path
  from A to B.
• Intuition there exist As neighbor X and Bs
  neighbor Y, such that XYltAB

 
r
t
h
Z
B
r2
A
r1
Y
k
u
s
15
Guaranteeing global connectivity

• If AB lt Rmax and node A and B are not connected
  after the CBTC, there must exist a multi-hop path
  from A to B.
• Intuition there exist As neighbor X and Bs
  neighbor Y, such that XYltAB

 
r
t
h
ZB, BY, WA, AX lt AB angle(W,A,X),
angle(Z,B,Y)lt150o angle(Z,B,A),
angle(X,AB)lt75o ? either XY or WZltAB
Z
W
B
r2
A
r1
Y
X
k
u
s
16
Optimization I Shrink-back operation

• A node may still have a gap of ? after reaching
  max power
• Transmission power can be reduced as long as the
  cone coverage is preserved

17
Optimization II Redundant edge removal

• Redundant edges
• An edge (u,v) is redundant if there exists an
  edge (u,w) and ?vuw lt ?/3.
• Redundant edges can be removed by each node
  independently.
• Node degree can be bounded.

18
Optimization III Asymmetric edge removal

• When ?120o, global connectivity can be preserved
  after removing all asymmetric links

19
Simulation Setup

• 200 nodes randomly distributed in 1500m by 1500m
  area. Maximum transmission radius is 500m.
• 60 communication pairs, each has 8Kbps data rate
• Routing layer modifies AODV to use energy metric
  instead of shortest path metric
• MAC layer CSMA
• Performance metrics
• Network connectivity average node degree and
  radius
• Network lifetime number of nodes that are still
  alive over time
• Energy model used is not clear

20
Network Topology
21
Network Topology (Contd.)
22
Dynamic Performance
23
Summary

• simple
• uses only directional information
• guarantees global connectivity
• increases network lifetime
• balances network throughput

24
Critiques

• Unrealistic assumptions
• Circular communication range
• Symmetric links
• Only reduce transmission power
• Max energy saving depends on the amount of
  network traffic
• Do not reduce idle listening power