Thesis etd-05212010-234133 |
Link copiato negli appunti
Thesis type
Tesi di dottorato di ricerca
Author
ALBANO, MICHELE
email address
michele.albano@gmail.com
URN
etd-05212010-234133
Thesis title
Reliable & Efficient Data Centric Storage for Data Management in Wireless Sensor Networks
Academic discipline
INF/01
Course of study
INFORMATICA
Supervisors
tutor Chessa, Stefano
Keywords
- erasure coding
- middleware
- quality of service
Graduation session start date
24/06/2010
Availability
Full
Summary
Wireless Sensor Networks (WSNs) have become a mature technology aimed at performing environmental monitoring and data collection. Nonetheless, harnessing the power of a WSN presents a number of research challenges. WSN application developers have to deal both with the business logic of the application and with WSN's issues, such as those related to networking (routing), storage, and transport. A middleware can cope with this emerging complexity, and can provide the necessary abstractions for the definition, creation and maintenance of applications.
The final goal of most WSN applications is to gather data from the environment, and to transport such data to the user applications, that usually resides outside the WSN.
Techniques for data collection can be based on external storage, local storage and in-network storage.
External storage sends data to the sink (a centralized data collector that provides data to the users through other networks)
as soon as they are collected.
This paradigm implies the continuous presence of a sink in the WSN, and data can hardly be pre-processed before sent to the sink.
Moreover, these transport mechanisms create an hotspot on the sensors around the sink. Local storage stores data on a set of sensors that depends on the identity of the sensor collecting them, and implies that requests for data must be broadcast to all the sensors, since the sink can hardly know in advance the identity of the sensors that collected the data the sink is interested in.
In-network storage and in particular Data Centric Storage (DCS) stores data on a set of sensors that depend on a meta-datum describing the data.
DCS is a paradigm that is promising for Data Management in WSNs, since it addresses the problem of scalability (DCS employs unicast communications to manage WSNs), allows in-network data preprocessing and can mitigate hot-spots insurgence.
This thesis studies the use of DCS for Data Management
in middleware for WSNs.
Since WSNs can feature different paradigms for data routing (geographical routing and more traditional tree routing), this thesis introduces two different DCS protocols for these two different kinds of WNSs.
Q-NiGHT is based on geographical routing and it can manage the quantity of resources that are assigned to the storage of different meta-data, and implements a load balance for the data storage over the sensors in the WSN.
Z-DaSt is built on top of ZigBee networks, and exploits the standard ZigBee mechanisms to harness the power of ZigBee routing protocol and network formation mechanisms.
Dependability is another issue that was subject to research work. Most current approaches employ replication as the mean to ensure data availability.
A possible enhancement is the use of erasure coding to improve the persistence of data while saving on memory usage on the sensors.
Finally, erasure coding was applied also to gossiping algorithms, to realize an efficient data management. The technique is compared to the state-of-the-art to identify the benefits it can provide to data collection algorithms and to data availability techniques.
The final goal of most WSN applications is to gather data from the environment, and to transport such data to the user applications, that usually resides outside the WSN.
Techniques for data collection can be based on external storage, local storage and in-network storage.
External storage sends data to the sink (a centralized data collector that provides data to the users through other networks)
as soon as they are collected.
This paradigm implies the continuous presence of a sink in the WSN, and data can hardly be pre-processed before sent to the sink.
Moreover, these transport mechanisms create an hotspot on the sensors around the sink. Local storage stores data on a set of sensors that depends on the identity of the sensor collecting them, and implies that requests for data must be broadcast to all the sensors, since the sink can hardly know in advance the identity of the sensors that collected the data the sink is interested in.
In-network storage and in particular Data Centric Storage (DCS) stores data on a set of sensors that depend on a meta-datum describing the data.
DCS is a paradigm that is promising for Data Management in WSNs, since it addresses the problem of scalability (DCS employs unicast communications to manage WSNs), allows in-network data preprocessing and can mitigate hot-spots insurgence.
This thesis studies the use of DCS for Data Management
in middleware for WSNs.
Since WSNs can feature different paradigms for data routing (geographical routing and more traditional tree routing), this thesis introduces two different DCS protocols for these two different kinds of WNSs.
Q-NiGHT is based on geographical routing and it can manage the quantity of resources that are assigned to the storage of different meta-data, and implements a load balance for the data storage over the sensors in the WSN.
Z-DaSt is built on top of ZigBee networks, and exploits the standard ZigBee mechanisms to harness the power of ZigBee routing protocol and network formation mechanisms.
Dependability is another issue that was subject to research work. Most current approaches employ replication as the mean to ensure data availability.
A possible enhancement is the use of erasure coding to improve the persistence of data while saving on memory usage on the sensors.
Finally, erasure coding was applied also to gossiping algorithms, to realize an efficient data management. The technique is compared to the state-of-the-art to identify the benefits it can provide to data collection algorithms and to data availability techniques.
File
Nome file | Dimensione |
---|---|
AlbanoPhDThesis.pdf | 2.20 Mb |
Contatta l’autore |