From Decentralized Data Collection to Decentralized Data Management
Speaker:
Andrea Vitaletti
speaker DIAG:
Data dell'evento:
Friday, 20 September, 2019 - 12:00
Luogo:
Aula A3 - Via Ariosto
Contatto:
Alberto Marchetti Spaccamela
Nell'ambito della procedura ai fini della chiamata nel ruolo di Professore di II fascia, SSD ING-INF/05 – SC 09/H1
Andrea Vitaletti terrà un seminario pubblico venerdì 20 Settembre 2019, ore 12.00, presso aula A3, via Ariosto 25
From Decentralized Data Collection to Decentralized Data Management
Abstract:
Data Collection by sensors is decentralized in nature: a number of wireless sensors organized in a network and distributed in an area of interest, collect and deliver data to a gateway that makes it available for further elaboration. In this domain, a fundamental problem is the design and implementation of solutions capable to maximize the lifetime of the data collection process while delivering high data quality. While data collection is decentralized, most of data processing, and in particular the distillation of knowledge from the raw data, is usually not.
Most of the current platforms for data management rely on cloud infrastructures based on the dominant centralized client/server architecture. These platforms have proven their effectiveness, but require that data owners trust them. Nowadays, also thanks to new technologies such as the blockchain, it is possible to envision a future in which also data management will be decentralized, thus reducing (or even eliminating) the need of trusted third parties and possibly supporting new forms of P2P decentralized data economy. It is now possible to envision a future in which instead of being the data that reach the algorithms in the server, algorithms will reach the data running on the data owner premises.
In this talk I will discuss such vision in some details and I will present some on my research work in this direction.
Data Collection by sensors is decentralized in nature: a number of wireless sensors organized in a network and distributed in an area of interest, collect and deliver data to a gateway that makes it available for further elaboration. In this domain, a fundamental problem is the design and implementation of solutions capable to maximize the lifetime of the data collection process while delivering high data quality. While data collection is decentralized, most of data processing, and in particular the distillation of knowledge from the raw data, is usually not.
Most of the current platforms for data management rely on cloud infrastructures based on the dominant centralized client/server architecture. These platforms have proven their effectiveness, but require that data owners trust them. Nowadays, also thanks to new technologies such as the blockchain, it is possible to envision a future in which also data management will be decentralized, thus reducing (or even eliminating) the need of trusted third parties and possibly supporting new forms of P2P decentralized data economy. It is now possible to envision a future in which instead of being the data that reach the algorithms in the server, algorithms will reach the data running on the data owner premises.
In this talk I will discuss such vision in some details and I will present some on my research work in this direction.
gruppo di ricerca: