Distributed Topology Control for Power Efficient Operation in Multihop Wireless Ad Hoc Networks

1
Distributed Topology Control for Power Efficient
Operation in Multihop Wireless Ad Hoc Networks

• Roger Wattenhofer, Li Li, Paramvir Bahl, Yi-Min
  Wang

Presenter Bibudh Lahiri
2
Presentation Overview

• Introduction and Motivation
• Related Work
• Cone-based Algorithm
• Simulation and Results
• Discussion

3
Introduction and Motivation

• Network lifetime limited by battery power
• Two choices
• Increase battery power
• Energy-efficient algorithms

4
Related Work

• Distributed triangulation-based algorithm for
  logical links (Hu)
• Did not take advantage of adaptive transmission
  power control
• Centralized spanning tree algorithm (Ramanathan
  and Rosales-Hain)
• Do not guarantee network connectivity

5
Related Work

• Distributed topology control (Meng)
• Guarantees network connectivity
• Relies on radio propagation model
• Adjust transmission power to improve throughput
  (Hou and Li)
• Adaptive clustering-based routing protocol
  (Heinzelman et al.)
• Rotate local base stations to higher energy nodes

6
Objectives

• Nodes can only use local information for
  determining transmission radius
• Decisions made to guarantee global node
  connectivity
• Minimize power consumption by finding minimum
  power paths
• Find topology with small node degree
• Minimal interference
• Simple and efficient
• Small and mobile (sensor) nodes
• Few assumptions about radio propagation model and
  hardware

7
Solution

• Cone-based topology control algorithm
• Designed for multihop wireless ad hoc networks in
  2-D
• Uses directional information of incoming signals
  from neighboring nodes
• Power consumption of each route can be made
  arbitrarily close to optimal

8
Cone-based Algorithm

• Phase One
• Neighbor discovery process
• Phase two
• Redundant edge removal
• Does not impact connectivity
• Reduces interference and improves throughput

9
Cone-based Algorithm

• Different from previous work
• Guarantees maximum connected set of nodes will
  always be found
• Computationally less demanding
• Do not need to specify deployment region
• Do not need exact location information, only
  direction
• Not tied to radio propagation model

10
Cone-based Algorithm

• Model
• set V of n nodes
• power p (0 lt p lt P), assumed to be unknown
  function of distance for upper-bound proof
• direction ? (0 lt ? lt 2 p)
• local set of neighbors N(u)
• cone is made up of angle a

11
Cone-based Algorithm

• a lt 2p/3
• If G was connected at full power, this will
  ensure a connected graph at lower power
• Proof by contradiction

a 2p/3
12
Cone-based Algorithm Phase 1

• Each node u beacons with growing power p
• If node u discovers neighbor v, v and direction ?
  added to list N(u)
• Increase power until every cone with angle a has
  is at least one neighbor v in the set N(u), or p
  reaches P

13
Cone-based Algorithm Phase 1

• Each neighbor v in N(u) covers a cone

14
Cone-based Algorithm Phase 1

• Symmetric
• If node u is in neighbor set of v, then node v
  is in neighbor set of u

15
Cone-based Algorithm Phase 2

• If node u has 2 neighbor nodes v, w in N(u) such
  that the power needed to send from u to w
  directly is not less than the total power to send
  via v, remove w from N(u)

16
Cone-based Algorithm Phase 2
w
35
u
10
20
v
Which edge should be removed to minimize power
usage?
17
Cone-based Algorithm Phase 2
w
35
u
10
20
v
u transmitting to v 30 lt 35 remove edge u,v
18
Cone-based Algorithm Phase 2

• two nodes v, w
• v, w in N(u) and w in N(v)
• p(u,v) lt p(u,w)
• p(u,v) p(v,w) lt q p(u,w)
• Remove w from N(U) (and u from N(w))

19
Simulation and Results

• 100 nodes
• Placed randomly in 1500 by 1500 rectangle
• Two-ray propagation model for terrestrial
  communications

20
Simulation and Results
21
Simulation and Results
22
Simulation and Results
23
Simulation and Results
24
Simulation and Results
25
Simulation and Results
26
Discussion

• This work focused on static nodes
• How would the algorithm need to change to adapt
  to mobile nodes?

27
Questions?