Aguzzi, Gianluca
(2024)
A language-based software engineering approach for cyber-physical swarms, [Dissertation thesis], Alma Mater Studiorum Università di Bologna.
Dottorato di ricerca in
Computer science and engineering, 36 Ciclo. DOI 10.48676/unibo/amsdottorato/11386.
Documenti full-text disponibili:
|
Documento PDF (English)
- Richiede un lettore di PDF come Xpdf o Adobe Acrobat Reader
Disponibile con Licenza: Salvo eventuali più ampie autorizzazioni dell'autore, la tesi può essere liberamente consultata e può essere effettuato il salvataggio e la stampa di una copia per fini strettamente personali di studio, di ricerca e di insegnamento, con espresso divieto di qualunque utilizzo direttamente o indirettamente commerciale. Ogni altro diritto sul materiale è riservato.
Download (9MB)
|
Abstract
IT systems are becoming increasingly ubiquitous and interconnected, driven by the rise of Internet of Things devices and advancements in edge-cloud computing. This evolution is having a growing impact on society and the economy. An advanced perspective on these systems identifies them as Cyber-Physical Swarm, which consist of large networks of interconnected devices embedded in the physical world, exhibiting collective behaviors. This thesis explores the engineering challenges associated with these systems, focusing on managing the complexities arising from their collective intelligence and large-scale nature. To address these challenges, this work introduces a language-based approach centered on aggregate computing, which is a top-down paradigm designed to describe large-scale collective behaviors. This paradigm was chosen because it facilitates the design of self-organizing behaviors, which are crucial for the resilient operation of CPSWs. Adopting a language-based approach has led to significant advancements in both hybrid methods–combining declarative and sub-symbolic solutions–and standard engineering approaches. Specifically, we identified two major facets to focus on engineering aspects: design patterns–i.e., the reusable solutions to common problems within a given context–and the platform aspects–i.e., the underlying infrastructure that supports the aggregate software. Regarding the former, we have developed new algorithms, APIs, and design methodologies. For the latter, we have taken action at the deployment and collective execution levels. Concurrently, we have explored how learning can be integrated into aggregate programming through a roadmap. We have effectively combined machine learning with aggregate computing both at the pattern level, with work based on many-agent reinforcement learning, specifically collective program sketching and a novel algorithm called field-informed reinforcement learning, and at the platform level with an innovative approach that aims to create distributed schedulers for collective computation. Ultimately, the goal of this research is to provide a holistic approach for the effective design and deployment of large-scale, adaptable, and robust CPSWs.
Abstract
IT systems are becoming increasingly ubiquitous and interconnected, driven by the rise of Internet of Things devices and advancements in edge-cloud computing. This evolution is having a growing impact on society and the economy. An advanced perspective on these systems identifies them as Cyber-Physical Swarm, which consist of large networks of interconnected devices embedded in the physical world, exhibiting collective behaviors. This thesis explores the engineering challenges associated with these systems, focusing on managing the complexities arising from their collective intelligence and large-scale nature. To address these challenges, this work introduces a language-based approach centered on aggregate computing, which is a top-down paradigm designed to describe large-scale collective behaviors. This paradigm was chosen because it facilitates the design of self-organizing behaviors, which are crucial for the resilient operation of CPSWs. Adopting a language-based approach has led to significant advancements in both hybrid methods–combining declarative and sub-symbolic solutions–and standard engineering approaches. Specifically, we identified two major facets to focus on engineering aspects: design patterns–i.e., the reusable solutions to common problems within a given context–and the platform aspects–i.e., the underlying infrastructure that supports the aggregate software. Regarding the former, we have developed new algorithms, APIs, and design methodologies. For the latter, we have taken action at the deployment and collective execution levels. Concurrently, we have explored how learning can be integrated into aggregate programming through a roadmap. We have effectively combined machine learning with aggregate computing both at the pattern level, with work based on many-agent reinforcement learning, specifically collective program sketching and a novel algorithm called field-informed reinforcement learning, and at the platform level with an innovative approach that aims to create distributed schedulers for collective computation. Ultimately, the goal of this research is to provide a holistic approach for the effective design and deployment of large-scale, adaptable, and robust CPSWs.
Tipologia del documento
Tesi di dottorato
Autore
Aguzzi, Gianluca
Supervisore
Dottorato di ricerca
Ciclo
36
Coordinatore
Settore disciplinare
Settore concorsuale
Parole chiave
cyber-physical swarms, aggregate programming, language-based software engineering, self-organization, collective intelligence, swarm intelligence, machine learning, many-agent reinforcement learning.
URN:NBN
DOI
10.48676/unibo/amsdottorato/11386
Data di discussione
24 Giugno 2024
URI
Altri metadati
Tipologia del documento
Tesi di dottorato
Autore
Aguzzi, Gianluca
Supervisore
Dottorato di ricerca
Ciclo
36
Coordinatore
Settore disciplinare
Settore concorsuale
Parole chiave
cyber-physical swarms, aggregate programming, language-based software engineering, self-organization, collective intelligence, swarm intelligence, machine learning, many-agent reinforcement learning.
URN:NBN
DOI
10.48676/unibo/amsdottorato/11386
Data di discussione
24 Giugno 2024
URI
Statistica sui download
Gestione del documento: