young-sam kim / m.d course school of information technology dept. of computer engineering korea...
TRANSCRIPT
Young-sam Kim / M.D Course School of Information Technology
Dept. of Computer Engineering Korea University of Technology and Education
Location Tracking in a WSNby Mobile Agents and
Its Data Fusion StrategiesYu-Chee Tseng, Sheng-Po Kuo, Hung-Wei Lee, and Chi-Fu
Huang
Contents
Abstract
Introduction
Network Model & Problem Statement
The Location Tracking Protocol
Fusion & Delivery of Tracking Results
Prototyping Experiences and Simulation Results
Conclusion
Q/A
Abstract (1)
Providing ubiquitous sensing, computing and
communication capability -> greatly facilitate human life
1. WSN (Wireless Sensor Network)
2. Location Tracking (1)
One of the central issues in sensor networks.
To monitor the roaming path of a moving object.
Abstract (2)
Similar to the location-update problem in PCS networks.
Location Tracking is more challenging in two senses.
1) Central control mechanism & backbone network. (X)
2) Wireless communication bandwidth is very limited.
3. Location Tracking (2)
4. Proposed paradigm
Novel protocol based on the mobile agent paradigm.
The communication & sensing overheads are greatly
reduced.
Based on IEEE 802.11b, NICs
WSN (1)
Rapid progress
Wireless communication
MEMS technologies
Wireless Sensor Network
Inexpensive node
collecting
processing
storing
Environmental Info
WSN (2)
1. Sensor are connected
Past : wire lines.
Today : wire lines + wireless ad hoc network
1) The flexibility of installation & configuration is greatly improved.
2) Research activities have recently been commenced in SN
WSN (3)
2. Issues remain to be resolved
Scalability
- coverage and exposure problems
Stability
- stable and fault-tolerant
Power-saving
- since no plug-in power is available
- energy consumption of communications might be a major factor
- mobile agent-based solutions are sometimes more power-
efficient
Proposed paradigm
3. Novel protocol based on the mobile agent paradigm To solve the location tracking problem
Advantage
1) sensing, computing & communication overheads can be greatly
reduced. -> delivery & fusion of the tracking results
2) on-site or follow-me services may be provided
Network Model (2) & Problem Statement
Sensing scope is r Working Area is A0, Backup Area is A1, A2, A3
Practice errors may exist, and thus more sensors will be needed to improve the accuracy.
Protocol Details (1)
Assume that … 1) Sensors can distinguish one object from the other 2) Each object periodically send a unique ID 3) focus on only one particular object
Protocol Details (2)
Master agent : closets to the object - go to master state & protocol
Slave agents : nearby sensors - go to slave state & protocol
Once the object enters the backup areas, the roles of master & slave may be changed
Protocol Details (3)
Master
For track1, the master discovers two slaves losing the target simultaneously.
For track2, only the slave agent in S1 will be revoked, and new one will be invited.
For track3, the master discovers one slave as well as itself losing the target
Protocol Details (4)
Each sensor will keep an object list (OL). unique identity, denote by ID, two sub-field
Two sub-field : status & time-stamp 1) ID.state : Master, Slave, Standby, Inhibited
2) time-stamp is the time when the record is last updated.
Protocol Details (5)
Seven types of control messages 1) bid-master(ID, sig) 2) assign_slave(ID, si, t) 3) revoke_slave(si) 4) inhibit(ID) 5) release(ID) 6) move_master(ID, si, hist) 7) data(ID, sig, ts)
Extension to Irregular Network Topologies
The election process does not need to be changed.
The rules to migrate masters/slaves need to be modified.
Sensors need to know the locations of at least their two-hop neighbors.
Still use one master and two slaves to track an object.
Extension to Irregular Network Topologies
How to define the master & slaves ? The problem can be solved by a divide-and-
conquer solution in time complexity
Fusion & Delivery of Tracking Results (1)
Assume that…
one of the sensors in the network serves as the
gateway connecting to a location server in the
wireline network.
Fusion & Delivery of Tracking Results (2)
Propose three data delivery solutions. 1) Non-Agent-Based (NAB) strategy. Each sensor works independently and forwards its
sensing results back to the gateway
2) Threshold-Based (TB) strategy. 3) Distance-Based (DB) strategy. The delivery action may be taken only when the
master agent moves.
Conclusion
A mobile-agent approach, A data fusion
model and several data delivery strategies
are significantly reducing the
communication & sensing overheads.