Geography Informed Energy Conservation for Ad Hoc Routing

1
Geography Informed Energy Conservation for Ad
Hoc Routing
Vijay Raghunathan EE206A (Spring 2001)
2
Introduction

• Mobile and wireless systems are battery operated
• Extremely stringent energy (battery lifetime)
  budget
• Energy efficiency cuts across all layers of
  abstraction
• Low-power hardware components
• Low-power HW/SW synthesis
• Energy-efficient operating systems
• Smart applications
• Battery friendly network protocols
• Dynamic Power Management (DPM)
• Shutdown based DPM Unused, idle components
  turned off

Energy awareness is the key to long system
lifetime
3
Overview

• Network level, shutdown based DPM technique for
  ad-hoc networks
• Exploits node redundancy to save energy
• On-demand routing protocols (AODV, DSR) are more
  energy efficient than a priori protocols if idle
  mode energy is ignored
• Energy consumed in idle listening mode is
  significant

4
Overview (contd.)

• Ad-hoc network of wireless nodes
• Nodes know their location (GPS)
• Traffic nodes and transit nodes

2
4
1
3
5
Nominal Radio Range

• Nodes 2, 3 and 4 are equivalent for routing
  packets from 1 to 5 and hence, two of them can
  sleep at any given time

5
Determining node equivalence
Virtual grid
2
4
1
3
5
r
r

• Nodes determine their grid Ids using location
  info
• Node 2 should be able to reach node 5.
• r2 (2r)2 ? R2
• r ? R/?5

6
State Transitions
D On hearing discovery message from a higher
ranked node

• Active gt Discovery gt Sleeping
• Only one active node per virtual grid
• Expected Node Active Time (ENAT)

7
Main features

• Eliminates node redundancy
• Balances energy usage between nodes (load
  balancing)
• Set Ta accordingly so that other nodes take over
  in due time
• Adapts to node mobility
• Reduce the sleep time Ts (random between ENAT/2
  ENAT)
• Include mobility information in discovery
  message (Expected Node Grid Time) and choose
  sleep time accordingly
• Independent of the routing protocol. However,
  depends on routing protocol to handle packet loss
  caused by shutdown of a currently routing node.

8
Evaluation

• Data delivery ratio is maintained
• Performs well as network density increases
• Higher density ? More redundancy ? Higher energy
  savings
• Consistent even under a more realistic
  propagation model
• Shadowing model
• Insensitive to small localization error
• Localization error ? Similar to mobility related
  error

9
Results Discussion

• Effect on network lifetime
• GAF-1 results in higher lifetime than GAF-2

10
Conclusions

• Inter-node (network level) DPM technique
• Exploits node redundancy for energy savings
• GAF extends network lifetime
• Mean energy consumption per node is reduced, for
  the same time interval
• GAF maintains high fidelity (i.e same data
  delivery ratio and same mean delay as unmodified
  AODV) at low mobility
• At high mobility
• GAF-1 Higher energy savings, Lower fidelity
• GAF-2 Lower energy savings, Higher fidelity
• Performs well at high node density, under a
  realistic propagation model and is insensitive to
  small location errors

11
Conclusions

• Choice of network parameters
• Total number of nodes 100, No. of neighbors 42
  ?Very high connectivity!
• Even then, it translates to 2.2 nodes/grid on
  average ? Very low redundancy!
• Trade-off Fault-tolerance vs. Energy
  consumption
• Analysis of the effect on FT due to shutdown
• Alternative A node runs to death and wakes up
  another node before it dies, through a paging
  channel (avoids communication overhead)

12
References

• T. Simunic, L. Benini, P. Glynn, and G. De
  Micheli, Dynamic Power Management for Portable
  Systems, Proc. MOBICOM, Aug. 2000.
• M. Zorzi, and R. R. Rao, Energy Management in
  Wireless Communication, Proc. WINLAB Wkshp. On
  3G Wireless Information Networks, pp. 189-201,
  Mar. 1997.
• M. Stemm, and R. H. Katz, Measuring and
  reducing energy consumption of network interfaces
  in hand-held interfaces, IEICE Trans. Comm.,
  E80-B8, pp. 1125-1131, Aug. 1997