1 cooperation in multi-domain sensor networks márk félegyházi levente buttyán jean-pierre hubaux...

Post on 12-Jan-2016

212 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

1

Cooperation in Multi-Domain Sensor Networks

Márk Félegyházi Levente ButtyánJean-Pierre Hubaux

{mark.felegyhazi, jean-pierre.hubaux}@epfl.ch

EPFL, Switzerland

TERMINODES Project (NCCR-MICS)http://www.terminodes.org

buttyan@hit.bme.hu

Budapest University of Technology and Economics, Hungary

PerSeNS 2005.03.08

2

Multi-domain sensor networks

co-located sensor networks sensors are compatible sinks can be either separate or common

3

Non-cooperative game assumption: paths to the sink exist in both own and

common network the game unfolds in discrete time steps t lifetime of sensor networks: until the first sensor dies finite routing game: ends when one subnetwork dies

4

Benefits of cooperation operators as players two decisions:

ask (or not) the other player to cooperate cooperate (or not) if asked

reduce complexity: strategy is pre-defined in the sensors

5

Strategies success of data gathering: if

moves:

strategy:

i it SR

i twhere:

is the gathered number of measurements

is the requirement for successiSR

pre-programmed strategy in the sensors

feedback in one bit from the sink (successful or not)

• DD - don't ask/drop• DF - don't ask/forward• AD - ask/drop• AF - ask/forward

success XX

next move XX XX DD, DF, AD, AF

6

Utilities

gain, gi(t): if the step was successful,

then gi(t) = Gi

otherwise gi(t) = 0 cost, ci(t): sum of the transmission cost of all

sensors (cunit ~ dα) payoff, utility:

0

T

i it

U t

where: T is the lifetime of the sensor network

i it SR

i i it t c t g

7

Simulation parameters

Number of sensors per domain 10 - 50 (25)

Distribution of the sensors uniformly random

Area size 40 x 20 m

Reception energy (R) 100 units

Transmission energy (Tr) Tr ~ dα units

Path loss exponent 2 – 5 (4)

Success requirement (SRi) 1.0 (all sensors have to report)

Position of the sinks (separate sinks) [10,10] and [30,10]

Position of the sinks (common sinks) [20,10]

Route selection Minimum energy path

8

Best strategies

Three types: Cooperative: (AF, AF) Defective: (DD, DD) Other: for example (AF, DD)

9

Simulations: Separate sinks

10

Simulations: Common sinks

11

Simulations: Path loss exponent

12

Conclusion

Cooperation is beneficial, because it can increase the lifetime of sensor networks.

For separate sinks, operators can use the sinks of each other

For common sinks, cooperation is beneficial: if sensor networks are sparse – overhearing of

packets is less significant if path loss is high - transmission is expensive

top related