Energy Management in

waralligatorΚινητά – Ασύρματες Τεχνολογίες

21 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

70 εμφανίσεις

Energy
M
anagement in
Wireless Sensor Networks

Mohamed Hauter

CMPE257

University of California, Santa Cruz

1


Wireless
S
ensor Networks


Energy and Wireless Sensor Networks


Paper1


Paper2


Paper3


Conclusion



Outline

2

3


C
onsists
of spatially
distributed autonomous
sensors
.



Monitors physical or
environmental conditions (i.e.
temperature, pressure, etc.)



Cooperates to pass data
through network to main
location


Wireless Sensor Network


Usually deployed in remote
regions



Energy consumption vs.
battery life



Energy harvesting

4

Energy and Wireless Sensor
Networks

Energy aware efficient geographic routing in
lossy wireless sensor

networks with environmental energy
supply



5

BY:


Kai Zeng

Kui
Ren

Wenjing Lou


Patrick J.
Moran

Combine the efficiency of Geo
-
Aware routing and energy
harvesting techniques.


6

Basic Idea!


Geographic Routing with Environmental Energy Supply
(GREES)


Packets are delivered through low cost links


Balances residual energy on nodes using environmental
energy supply



Two protocols are proposed:


GREES
-
L


GREES
-
M

7

Proposal


Battery technology has been unchanged for many years


Former energy aware routing protocols:


Batteries have limited/fixed capacity


Decisions are made based on energy consumption


Energy scavengers:


Harvests small amounts of energy from ambient sources


Solar
-
aware routing protocols:


Must have a global knowledge of the whole network

8

Related Work


Maintain one
-
hop neighbor’s information:


Location


Residual energy


Energy harvesting rate


Energy consumption rate


Wireless link quality

9

Protocol Description


To balance the geographical advance efficiency per
packet transmission and the energy availability on
receiving nodes:


GREES
-
L
-

uses linear combination




GREES
-
M


uses multiplication

10

Protocol Description (Cont.)

11

GREES

12

GREES (Cont.)

13

GREES (Cont.)

14

Simulation Results

15

Simulation Results


Strengths:


Maintains a higher mean residual energy on nodes


Achieves better load balancing


Small standard deviation of residual energy on nodes


Does not compromise the end
-
to
-
end throughput
performance


Weaknesses:


Exhibits graceful degradation on end
-
to
-
end delay


What happens when energy harvesting fails?

16

Conclusions

Minimum
-
Energy Asynchronous Dissemination
to
Mobile Sinks
in Wireless Sensor Networks

17

BY:


Hyung

Seok

Kim

Tarek

F.
Abdelzaher

Wook

Hyun Kwon


Achieve energy savings in wireless sensor networks by:


Optimizing communications between sensor nodes and
sinks



Tradeoff?


Increase in path delay.



Is the tradeoff a good one? We’ll see…

18

Basic Idea


Overlay Multicasting


Uses sinks as intermediate nodes in the tree


Uses flooding to disseminate information


Flooding is energy
-
intensive

19

Related Work


SEAD


Scalable Energy
-
efficient Asynchronous
Dissemination protocol


Stationary sensor node takes the mobile sink’s place


Build an optimal dissemination tree (d
-
tree)


Select dissemination paths to stationary sensor nodes


Stationary sensor nodes forward data


Minimize energy cost


As sink moves, forward delay increases (tradeoff)


Reconfigure d
-
tree when needed


20

Proposal

21

SEAD Tree Model in Wireless
Sensor Networks

22

SEAD Sink Search

23

SEAD Sink Search

24

SEAD Sink Search

25

SEAD Sink Search

26

Results

27

Results

28

Results

29

Results

30

Results


Strengths:


SEAD saves energy


Strikes a balance between end
-
to
-
end delay and power
consumption


Power savings are favored over delay minimization


Weaknesses:


Affects the lifetime of the access node


Not robust in high density networks


31

Conclusion

Meeting Lifetime Goals with
Energy
Levels


BY:


Andreas
Lachenmann


Pedro Jos
´
e Marr
´
on


Daniel
Minder


Kurt
Rothermel

32


Levels :
an abstraction for energy
-
aware programming
of wireless sensor networks.


Goal is to meet the user
-
defined lifetime goals while
maximizing application quality


Applied in applications with:

1.

Known lifetime

2.
No redundant nodes


33

Basic idea

1.
Define energy levels

2.
Measure energy consumption of each level (using an
energy profiler)

3.
Decide level of functionality to meet lifetime goal

4.
Maximize performance within allowed energy level

5.
Maintain network connectivity

6.
Maintain optimal application quality

34

How does it work?


ZebraNet monitoring system


Gathers GPS traces



If a node fails due to energy drought, what happens?


Lost track of at least one animal


Possible network disconnection



Solution ???

35

Example


A node can:

1.
Stop forwarding data from other nodes

2.
Decrease energy
-
intensive radio communications

3.
Stop storing other nodes’ data (avoid flash memory
access)

4.
Decrease queries of GPS position

5.


36

Solution

1.
Eliminates low energy
-
levels issues

2.
Ensures reaching targeted lifetime

3.
Low overhead

37

Benefits to developer


Single application running on each sensor
node


Periodic behavior


It is possible to simulate output behavior, thus
acquire energy consumption statistics


Use voltage sensors


Investing time to define energy levels

38

Design Considerations


Provide a programming abstraction and
runtime support that helps to meet the
user’s lifetime goals by deactivating
parts of the application if necessary

39

Design Goals


Divide into sub goals:

1.
Follow definition of optional functionality

2.
Make it easy to use

3.
Minimum overhead

4.
Provide good application quality

5.
Low runtime

6.
Robust with inaccurate energy estimates


40

How to achieve
goals?

Levels approach follows the
well
-
known model predictive
control (MPC) schemes

41

Notice

42

Combining Energy Levels

43

Code Example for Energy Levels

44

Computing the Energy
Consumption of a
C
ode
Block


Energy consumed by lower level
energy_level(1) = total_energy_consumed





energy_estimated_all_other_levels



Energy consumption that depends on some state of the
hardware of software
Example: attempting to turn on an active device. No energy
consumed, thus adjust estimates.

45

Special Cases

46

Battery
D
ischarge
C
haracteristics
(from three experiments)

47

Results

48

Runtime Overhead


Helps meet user
-
defined lifetime goals


Requires small code modifications


Low overhead


Maximize performance within allowed
energy level


Maintain network connectivity


Maintain optimal application quality



49

Conclusion

50

Questions ????