Post on 20-Dec-2015
transcript
Globecom 2004
Energy-Efficient Self-Organization for Wireless Sensor Networks: A Fully
Distributed approach
Liang Zhao, Xiang Hong, Qilian LiangDepartment of Electrical EngineeringUniversity of Texas at Arlington
Globecom 2004
Outline
Introduction to Cluster-based Sensor Networks
LEACH Protocol Review Problem Formulation Expellant Self-Organization Scheme Simulation Conclusion
Globecom 2004
Cluster-based Sensor networks(1/2)
Homogeneous versus Heterogeneous Single hop versus Multi-hop
BS
Advantage: Robust to node failures.Drawbacks: (1) High load on cluster head(2) The hardware has to support
(a) long range transmissions(b) complex data computations(c) co-ordination of MAC(d) routing within a cluster.
Advantage: Sensor nodes only requires simple hardware support.
Drawbacks: (1) Less robust to node failure(2) Sensors far from cluster head have
highest energy expenditure(3) Uniform clustering has to be
considered.
Single hop homogeneous networks Single hop heterogeneous networks
BS
Globecom 2004
Cluster-based Sensor networks(2/2)
BS
Multi-hop homogeneous networks Multi-hop heterogeneous networks
BS
Sensors near cluster head have highest energy expenditure
Distance from cluster head
Energy Single hopMulti hop
[ Infocom04 , Globecom04 ]
[ 2 * Globecom04 ]
Globecom 2004
LEACH Protocol
1. Single hop homogeneous networks
2. Random cluster head rotation
3. Perfect data correlation model– All individual packets from members of the same
cluster can be combined into a single representative packet.
Globecom 2004
LEACH Protocol
In LEACH, each node i elects itself
to be a head at the beginning of
round r+1 with probability Pi(t).
(1)
(a) Ci(t) is the indicator function determining whether or not node i has been a head in the most recent (r mod (N/c)) rounds.
(b) c is the desired number of clusters
Globecom 2004
Problem Formulation
1. N in (1) and Etotal in (2) are global information
2. Too few or too many cluster heads []3. Non-uniform distribution of cluster heads
Globecom 2004
Random election
(1) Suppose
(2) All nodes have equal amount of energy,
if N nodes want to elect c heads, then the self-electing probability for each node is
[back]
Globecom 2004
Expellant Self-Organization Scheme
Contribution:
Turn the question about “how many clusters should the nodes be partitioned? “
into “What is the appropriate cluster size?”
Globecom 2004
Expellant Self-Organization Scheme
ExampleStep 1: Each node broadcasts vital information (energy, location,…etc) at cluster radius Rc (Rc is a predetermined system parameter)
Rc
Step 2: Each node counts its neighbors and broadcasts the number of its neighbors at cluster radius Rc
2
1
2
3
1
1
0
•N is the set of neighbors•B(i) is the number of neighbors of the ith neighbor• r = 0.8 in this paper
Step 3: Each node decides to become cluster head according to its potential and broadcast its claim( Using random back-off to avoid collision)
Potential:
)()1()(max iBmeanriBrBNiNi
Th
Globecom 2004
Expellant Self-Organization Scheme
Step 4: Nodes sends “Joint” message to join the nearest cluster head
2
1
2
3
1
1
0
Step 5: Nodes that are outside the neighborhood of existing cluster head forces itself to be a cluster head
“Joint”
[back]
Globecom 2004
Simulation
100 nodes Each has 2J initial energy Evenly distributed in a circular region with
diameter 100m Base station located at (125m, 0)