/ 23
Decentralized Boundary
Detection without Location
Information in Wireless
Sensor Networks
Wei-Cheng Chu Kuo-Feng Ssu
Institute of Computer and Communication Engineering Department of Electrical Engineering
National Cheng Kung University Tainan, Taiwan
Outline
Introduction Related work
Network assumptions
Decentralized Boundary Detection (DBD)
Simulation Conclusions
Introduction
Wireless sensor networks have a wide range of
applications
The environment may contain physical obstacles
and communication holes
A hole or a physical obstacle can be regarded as
Motivation
Obstacle detection (or boundary detection) is a
major concern in most WSN applications
Previous detection mechanisms are
topology-based, which are not suitable for the dynamic hole detection
Goals
Distributed Location-free
Dynamic hole detection Correctness guaranteed
Related Work (THD)
S. Funke and C. Klein, “Hole Detection or: “How
much Geometry hides in Connectivity?””
Breakpoint Broken contour 1 2 3 1 2 3 s s
Related Work (TTG)
D. Dong, Y. Liu, and X. Liao, “Fine-grained
Boundary Recognition in Wireless Ad Hoc and
Sensor Networks by Topological Methods”
Related Work (BR)
O. Saukh, R. Sauter, M. Gauger, and P. J.
Marron, “On Boundary Recognition without Location Information in Wireless Sensor
Networks”
Flower structure Fail to construct the flower structure
Assumptions and System Model
Location-free
Non-unit disk graph
d-Quasi Unit Disk Graph (d-QUDG)
Each edge is a two-way communication link
Each node is aware of its one, two, and three-hop
Overview
s
Broken contour Contour
s
A boundary node should be nearby a broken contour
Definition of Hole
Hole cycle: a cycle with length larger than three
and it cannot be decomposed into several cycles with smaller length
Hole: the face formed by hole cycle
Hole
Hole
Rule 1
For the hole cycle with length larger than five
Determines whether the contour in two-hop neighbor graph is broken or not
Broken contour
Node s is a boundary node
s s
s
The contour in one-hop neighbor graph is difficult to determine whether it is broken or not
s
Exception of Rule 1
Not able to identify the hole cycle with length
s
Rule 2
For the hole cycle with length smaller than six
Identifies the hole cycle in one, two-hop neighbor graph
s
Correctness Proof
Step 1: A hole cycle has continuity and
consistency properties
Step 2: All nodes of hole cycles can be identified
Step 3: All boundary nodes near or belonging to
hole cycles
Step 4: The identified boundaries of DBD contain
Simulation
Network field: 600m × 600m
Communication range: 20m
Embeddings: UDG and 0.7-QUDG
Number of nodes: 700 to 2200 Number of runs: 100
Control Overhead
The control overhead is calculated as the average
Detection Results (1/4)
UDG with degree 12
Detection Results (2/4)
UDG with degree 12
Detection Results (3/4)
0.7-QUDG with degree 12
Detection Results (4/4)
0.7-QUDG with degree 12
Conclusions
No UDG constraint
Any length of hole cycle can be detected
Requires only 3-hop neighbors’ information
Can be applied in the environment with mobile
nodes
The accuracy of hole shape is better than previous