Wireless Sensor Networks

brainybootsMobile - Wireless

Nov 21, 2013 (3 years and 8 months ago)

140 views

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Cross
-
Layer Optimization for
Wireless Sensor Networks

Yuecheng Zhang, Liang Cheng

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Target Applications


Local data exchanging system


Multi
-
hop Local Positioning System (MLPS)


Global data collecting system


Remote lab, environment sensing, and etc.


Sensor network and robot collaborative system


Large constructions’ automatic examination


Robot + Sensor network + Multimedia streaming


Sensor network and human
-
being collaborative
system


Firefighting aids, smart battlefield and etc.


LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Important Issues


Network Scale


Cost vs. Budget


Throughput vs. Bandwidth


System Lifetime


Power consumption vs. Power supply


Existing solutions


Low power hardware design


Transmission power/range control


Power
-
aware protocol stacks


Power saving working modes


Node Versatility


Functionality vs. Resource constraints


Functionality: collecting, storing, processing, fusing, relaying


Constraints: computing power, storage, specific device and etc.

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Optimization Goals


Enable large scale networks


Low cost


High efficiency


Lengthen system lifetime


Improve power efficiency


Prevent system deconstruction


Support node versatility


Multiple functionality


Multiple applications

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Why Cross
-
Layer


Possibility


Optimization goals can be pursued in multiple layers


layered optimizations assume cross
-
layer cooperation


Some applications do not need all layers


Multi
-
hop Local Positioning System (MLPS)


Necessity


Optimizing single layer is not enough


Optimizations may be conflict with each other


One
-
for
-
all consolidate design for feasible
applications


LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Why Cross
-
Layer (Cont.)

Network
Scale

System
Life
-
time

Node
Versatility

Upper Layers

Data fusion
Compression

Power
-
aware
working mode
Control


Load
-
aware

working mode
decision

Transport
Layer

QoS (Bounded Delay)

QoS
-
power tradeoff

Load
-
aware
transportation control

Network Layer

Node naming

Efficient routing

Node discovery

Power
-
aware
routing

Reduce overhead

Load
-
aware routing

Simplified discovery

Distributed storage

MAC

Layer

Contention control

Channel reuse


Synchronized sleep

Transmission range
control

Load
-
aware channel
allocation


Physical
Layer

Ultra Wide Band

Low power design

Powerful battery

Attach specific
accessories (GPS)

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Why Cross
-
layer (Cont.)


Heterogeneity


Intra connection vs. Inter connection


Scalability


Correctness verification


Reusability


Code reuse


Module reuse


General purpose sensor networks

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Current results


Networking layer


Self
-
nominating


Floossiping


Data
-
Link layer


Efficient broadcasting MAC


Time synchronization module


LESSAR (LEvel Synchronization by Sender Adjuster
and Receiver)


Positioning module


PLACE (Power
-
based Location And Coordination
Estimation)

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Self
-
Nominating


Advantages:


Robust


Flexible


Disadvantages:


Artificial random delay

A
B
C
D
E
F
CT(B)=10
Ready->Rcvd
CT(C)=9
Ready->Rcvd
CT(D)=7
Ready->Rcvd
A
B
C
D
E
F
CT(B)=3
Rcvd
CT(C)=2
Rcvd->Ready
CT(D)=0
Rcvd->Sent
CT(E)=12
Ready->Rcvd
CT(F)=9
Ready->Rcvd
A
B
C
D
E
F
Ready
Sent
CT(E)=12
Rcvd
CT(F)=9
Rcvd
LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

Future tasks


Do some theoretical study on the system limitations


System Throughput & Network Scale


Node Synchronization and Positioning accuracy


Channel allocation and Mode decision


Identify and justify feasible application models


Implement the applications


Complete the protocol stacks


Tune the protocols and functional modules


Optimize the applications


Developing the research platform

LONG Lab: http://long.cse.lehigh.edu
CSE Department Lehigh University

The End

Questions?