Distributed Clustering in Ad-hoc Sensor Networks: A

Distributed Clustering in Ad-hoc Sensor Networks:
A Hybrid, Energy-Efficient Approach
Mulmedia and Mobile communications Laboratory
2004 / 4 / 20
박건우
Reference
 Ossama Younis and Sonia Fahmy,
“Distributed Clustering in Ad-hoc Sensor
Networks:
A Hybrid, Energy-Efficient Approach,”
INFOCOMM 2004
Contents





Introduction
Problem Statement
The HEED Protocol
Performance Evaluation
Conclusion
Introduction(1/3)
 Useful energy consumption
 Transmitting/receiving data
 Processing query requests
 Forwarding queries/data
 Wasteful energy consumption




Idle listening
Retransmitting
Overhearing
Generating/handling control packets
Introduction(2/3)
 Hierarchical clustering
- reduce useful energy consumption
 Scalability
 Efficient data aggregation
 Routing protocols
 The essential operation in clustering
 Selecting a set of cluster heads
Introduction(3/3)
 HEED (Hybrid Energy-Efficient
Distributed clustering)
 Distributing energy consumption
 Terminating the clustering process within a
constant number of iterations/steps
 Reducing control overhead O(n)
 Producing well-distributed cluster heads
and compact clusters
Problem Statement(1/2)
 Network Model
 Quasi-stationary.
 Energy consumption is not uniform for all
nodes.
 Nodes are location-unaware, i.e. no GPS
service.
 All nodes are equal.
 Nodes are left unattended after deployment.
 Fixed number of transmission power levels.
Problem Statement(2/2)
 The Clustering Problem
 Clustering is completely distributed.
 Clustering terminates within a fixed number
of iterations.
 At the end of each Tcp, each node is either
a cluster head, or a non-head node that
belongs to exactly one cluster.
 Clustering should be efficient.
 Cluster heads are well-distributed over the
sensor field.
The HEED Protocol
 A. Clustering Parameters
 Cluster header selection
 Residual energy
 Select an initial set of cluster heads
 Intra-cluster “communication cost” - Break ties
 Cost : function of density and power
 Fixed power (optimal power)
 Node degree
- Load distribution
 1/(node degree)
- Dense clusters
 Variable power

Average minimum reachability power (AMRP)
The HEED Protocol
 B. Protocol Operation
Clustering is triggered every
Tcp+Tno seconds
(Tcp : clustering process time,
Tno : network operation time)
1. Initialize
2. Repeat
3. Finalize
The HEED Protocol
 B. Protocol Operation
->
The HEED Protocol
 B. Protocol Operation
// 1. cluster set이 존재할 경우
// 3. 자신이 cost가 가장 작은 node일 경우
// 9. Sch 없음. 자신이 CH가 됨
// 12. 첫 iteration에서 CHprob를 이용
//
Sch로 속할 것인지 판단
The HEED Protocol
 B. Protocol Operation
// 주변에 CH존재
// 직접 CH가 됨
Phase II 에서 CHprevious = 1 인데도 is_final_CH = false 라는 말은
해당 node가 cluster head가 아니라는 것을 의미한다.
The HEED Protocol
 C. Correctness and Complexity
 Lemma 1: HEED terminates in Niter = O(1) iterations
 Lemma 2: At the end of phase III of the HEED
protocol, a node is either a cluster head or a regular
node that belongs to a cluster
 Lemma 3: HEED has a worst case processing time
complexity of O(N) per node
 Lemma 4: HEED has a worst case message
exchange complexity of O(1) per node, i.e., O(N) in
the network
 Lemma 5: The probability that two nodes within
each other’s cluster range are both cluster heads is
small, i.e., cluster heads are well-distributed
The HEED Protocol
 D. Inter-Cluster Communication
 Lemma 6: Assume that N nodes are uniformly and
independently dispersed at random in an area and
the area is divided into square cells, then each cell
contains at least one node (asymptotically almost
surely) a.a.s. (i.e., the expected number of empty
cells is zero)
 Lemma 7: There exists at least one cluster head in
any square cell area a.a.s.
 Lemma 8: For any two cluster heads v1 and v2 in
two neighboring areas A and B, v1 and v2 can
communicate.
 HEED produces a connected multi-hop cluster
head graph (structure) a.a.s.
Performance Evaluation
Max
AMRP
Min
LEACH
Performance Evaluation
AMRP
Min
Max
LEACH
Performance Evaluation
LEACH
HEED
Conclusion
 Energy-efficient distributed clustering
approach for ad-hoc sensor networks
 HEED (Hybrid, Energy-Efficient Approach)
 Cluster heads are randomly selected based on
their residual energy
 Terminates in a constant number of iterations
 Independent of the network diameter
 HEED
 energy efficiency, scalability, prolonged network
lifetime, and load balancing