1. Ching-Ju Lin Institute of Networking and Multimedia NTU
HCDD : Hierarchical Cluster-based Data Dissemination in Wireless Sensor Networks with Mobile Sink
Ching-Ju Lin
Institute of Networking and Multimedia NTU
Po-Lin Chou
Dept. of Computer Science and Information Engineering NTU
Cheng-Fu Chou
Dept. of Computer Science and Information Engineering NTU
ACM International Wireless Communication and Mobile Computing (IWCMC’06)

2. Outline Introduction Related Work
Hierarchical Cluster-based Data Dissemination (HCDD)
Cluster Construction
Sink Location Registration
Data Delivery and Path Routing
Performance Evaluation
Conclusions

3. Introduction
WSN can be applied to various applications, such as military, surveillance, temperature monitoring.
Sensors are randomly scattered to organize a large scale sensor network.
Due to the random and inaccessible locations of sensor nodes, the power-consumption of the sensor is the most important consideration.

4. Introduction
Finding a routing path for disseminating data to mobile sinks is a challenging problem.
Limited resources, such as energy, storage capacity, and computing capability.
Flooding the location information of mobile sink seems to be a naïve method.

5. Introduction
Proposing a Hierarchical Cluster-based Data Dissemination (HCDD) scheme.
Sensor nodes are self-organized to find the route without location information.
Without any expensive and power-consuming GPS device and with light control overhead.

6. Introduction
Much research has endeavored to develop the routing algorithms, but, however, there are some challenges
Globally flooding the sink location cause serious overhead and rapidly exhausted the energy of sensors.
Estimating the sensor’s location by GPS device also has heavily power-consumed and impractical.
To adjust transmission power to keep connections with the mobile sinks when the sinks are moving out of coverage range is also dissipate much energy.

7. Introduction
This mechanism can reduce the control overhead.

8. Introduction
Part of sensor nodes in the highest level, called Routing Agent.
Updating and Recording the Dynamic Global Information for the mobile sink and data source.
Other nodes record Static Local Information used to find the routing path to their Routing Agent.
Sink information only flooded to the Routing Agent instead of all sensor nodes.

9. Introduction
HCDD can reduce overall control overhead but Routing Agent with heavier traffic load will consume much of their energy.
The load of each Routing Agent will be determined by two factors :
Number of hierarchical levels.
The size of each cluster.

10. Related Works
TTDD : A Two-Tier Data Dissemination Model for Large-scale Wireless Sensor Networks
Mobicom 2002

11. Related Works Record a upstream dissemination node towards the source
mobile sink
source
dissemination node

12. Immediate dissemination node
Related Works
Immediate dissemination node
Primary Agent
mobile sink
source
dissemination node

13. Immediate dissemination node
Related Works
Immediate dissemination node
Primary Agent
mobile sink
source
dissemination node

14. Related Works Advantage Disadvantages
Source can effectively deliver data to mobile sinks.
Disadvantages
Each node should acquire location information by GPS.
The grid construction and maintenance per each source leads to a considerable overhead.
Some dissemination nodes have high overhead.
Some dissemination nodes does not use.

15. Hierarchical Cluster-based Data Dissemination (HCDD)
There are three procedures in HCDD
Cluster Construction
Sink Location Registration
Data Delivery and Path Routing

16. HCDD – Cluster Construction
Group the sensors by Max-Min D-cluster Formation. (INFOCOM 2000)
Providing load-balance clustering for extending the life time of sensor network.
Each cluster is composed of a CH, several gateways, and ordinary sensors.
Each CH maintains the Neighboring Cluster Table, records its neighboring clusters and the set of gateways next to it.
Each Routing Agent has Global Information used to communicate with other Routing Agent and records the routing paths for all sensor nodes.

17. Hierarchical Cluster-based Data Dissemination (HCDD)
This mechanism can reduce the control overhead.

18. HCDD – Sink Location Registration
Routing Agent acquire and maintain the information of mobile sink through the sink registration
Sink registration is divided into two phases
Intra-cluster phase
Mobile sink first forwards registration messages and requirements to its corresponding Routing Agent.
Inter-cluster phase
Routing Agents receives the registration messages in the first phase, will broadcast the information of the mobile sink to all Routing Agent.

19. HCDD – Sink Location Registration
Registration : The dotted line means the intra-cluster registration, and the solid line means the inter-cluster registration

20. HCDD – Data Delivery and Path Routing
After the cluster construction, each sensor node can communicate with Routing Agent along the intra-cluster path
A source node delivers data to the sink just has to forward the data to its Routing Agent.
Receiving the data from the source node, the Routing Agent can retrieve the sink information and relay data to the sink along the inter-cluster path.

21. HCDD – Data Delivery and Path Routing
Data Delivery : The data source forwards data along the inter-path (dotted lines) and inter-path (solid lines) learned from sink registration

22. Performance Evaluation and Discussion
Number of nodes : 1000
Sensing field : 500m * 500m
Buffer size : 10 packets
Initial energy : 2.5 Joule
Number of source nodes : 20
Data rate of source nodes : 10 packets/sec
Transmission range : 28m
Moving speed of sink : 10m/sec

23. Performance Evaluation and Discussion
Performance metrics
Control overhead
Power consumption used for the structure formation and the route maintenance
Network lifetime
The duration from the beginning of the simulation to the time once one sensor runs out its energy
Total number of packets received by the mobile sink
Data delivery rate
The ratio of packets successfully received by the sink to total packets sent by source nodes.

24. Performance Evaluation and Discussion
Number of sources vs. Overhead

25. Performance Evaluation and Discussion
Number of sources vs. Network lifetime

26. Performance Evaluation and Discussion
Comparison of different schemes

27. Performance Evaluation and Discussion
Cluster size vs. Overhead

28. Performance Evaluation and Discussion
Cluster size vs. Lifetime

29. Conclusions
Proposing an HCDD approach to deal with data dissemination from a number of source nodes to a mobile sink.
HCDD exploits a hierarchical cluster-based method to reduce the registration overhead.
Simulation shows that the cluster-based method outperforms the grid-based method is not only the longer network lifetime but also lower control overhead.
Working efficiently without any location estimation devices, like GPS.