Geography-informed Energy Conservation for Ad Hoc Routing

Ya Xu, John Heidemann, Deborah Estrin

ISI & UCLA

Presented by: Cristian Borcea

Motivation

reduce the energy consumption in ad hoc wireless networks

increase the network lifetime

Solution

identifies equivalent nodes for routing

based on location information

turns off unnecessary nodes

Assumptions

dense node deployment

many nodes can hear each other

each node knows its location

GPS ... but better other methods

Energy Model

listen:receive:transmit energy consumption 1:1.05:1.4 or 1:1.2:1.7

recall from last week listen:receive:transmit times are 1:3:40

duty cycle > 22% ==> more than 50% of energy spent in listening

energy dissipation in idle state cannot be ignored

Effects of turning radio off in the idle state

Determining Node Equivalence

the physical space is divided into equal size squares based on nominal radio range

any two nodes in adjacent squares can communicate with each other

the nodes within a square are equivalent

Geographical Adaptive Fidelity ( GAF ) Routing

nodes in the same grid coordinate each

other

who will sleep and for how long

runs over any ad hoc routing protocol

load balancing energy usage

all nodes remain up for us long as possible

GAF state transitions

Node Ranking

node(active) > node(discovery)

enat1>enat2 ==> node(enat1)>

node(enat2)

enat = estimated node active time

node ids break the ties

Adapting to Mobility

each node estimates the time when it expects the leave the grid: engt

includes this estimation in the discovery message other nodes sleep for min(enat, engt)

GAF-ma ( mobility adaptation ), GAF-b ( basic scheme )

Simulation

ns2 + cmu's extension for 802.11 AODV vs GAF/AODV DSR vs GAF/DSR 50 transit nodes ( "routers" ) 10 traffic nodes ( sources & sinks ) Traffic: CBR Nominal radio range: 250

Energy model - values used in simulation

WaveLAN (pre-802.11, 1995) 2Mb/s listen:receive:transmit 1:1.2:1.6W

0.025 when sleeping 802.11 wireless LAN

0.75:1.5:1.9W 802.11 cards

0.83:1:1.4W

Network Lifetime

GAF energy savings

mean energy consumption per node (E0-

Et)/(n*t)

E0 initial total energy for n nodes

Et total energy after time t

results: GAF+AODV is 40% better than AODV

for both GAF-b, GAF-ma

GAF-b vs GAF-ma

Data Delivery Ratio

Average Delay

Network lifetime: GAF vs AODV

Network Lifetime vs Node Density

Conclusions

GAF increases the network lifetime

does not decrease the performance

substantially