In coming decades, population is set to become slightly smaller in more developed countries, but much older. This increase results in a growing need for supports (human or technological) that enables the older population to perform daily activities. This originated an increasing interest in Ambient Assisted Living (AAL), which encompasses technological solutions supporting elderly people in their daily life at their homes. The structure of an AAL system generally includes a layer in charge of acquiring data from the field, and a layer in charge of realising the application logic. For example, a fall detection system, acquires both accelerometer and acoustic data from the field, and exploits them to detect fall by relying on a machine learning technique. Usually, AAL system are implemented as vertical solutions in which often there is not a clear separation between the two main layers. This rises several issues, which include at least a poor reuse of the system components since their responsibilities overlap, and a scarce liability to software evolution mostly because data is strongly coupled with its source, thus changing the source requires modifying the application logic too. To promote reusability and evolution, an AAL system should keep accurately separated issues related to acquisition from those related to reasoning. It follows that data, once acquired, should be completely decoupled from its source. This allows to change the physical characteristics of the sources of information without affecting the application logic layer. Moreover, the acquisition layer, should be structured so that the basic acquisition mechanisms (triggering sources at specified frequencies, and distributing the acquired data) should be kept separated from the part of the software that interacts whit the specific source (i.e., the software driver). This allows to reuse the basic mechanisms and to program the drivers for the needed sensors only. If a new or different sensor is required, it suffices to add/change the sensor driver and to properly configure the basic mechanisms so that the change can actually implemented. The aim of this work is to propose a novel approach to the design of the acquisition layer that overcomes the limitation of traditional solutions. The approach consists of two different sets of architectural abstractions: Time Driven Sensor Hub (TDSH) is a set of architectural abstraction for developing timed acquisition systems that are easily configurable for what concerns both the type of the sensors needed and their acquisition frequencies. Subjective sPaces Architecture for Contextualising hEterogeneous Sources (SPACES) is a set of architectural abstractions aimed at representing sensors measurements that are independent from the sensors characteristics. Such set can reduce the effort for data fusion and interpretation, moreover it enforces both the reuse of existing infrastructure and the openness of the sensing layer by providing a common framework for representing sensors readings. The final result of this work consists in two concrete designs and implementations that reify the TDSH and SPACES models. A test scenario has been considered to contextualise the usefulness of the proposed approaches and to test the actual correctness of each component. The example scenario is built upon the case of fall detection, an application case studied in order to be aware of peculiarities of the chosen domain. The example system is based on the proposed sets of architectural abstraction and exploits an accelerometer and a linear microphonic array to perform fall detection.

Nei prossimi decenni è previsto che la popolazione subirà una lieve riduzione nei paesi sviluppati, ma sarà in media più anziana. Ciò risulta in un aumento nell’interesse per quelle soluzioni (umane o tecnologiche) in grado di supportare la popolazione anziana nella vita di tutti i giorni. Questa situazione ha portato un aumento nell’interesse riguardo all’Ambient Assisted Living (AAL), che racchiude tutte quelle soluzioni tecnologiche volte al supporto degli anziani nella quotidianità della propria casa. Solitamente i sistemi di AAL includono un livello incaricato di acquisire informazioni dall’ambiente e uno che si occupa di realizzare la logica applicativa. Ad esempio un sistema di rilevazione delle cadute potrebbe acquisire dati accelerometrici ed acustici e sfruttarli per rilevare possibili cadute grazie a tecniche di machine learning. I sistemi di AAL sono solitamente implementati come soluzioni verticali senza una chiara separazione tra i due livelli. Questo prevede diverse problematiche, tra le quali scarse possibilità di riuso dei componenti del sistema, poiché le loro responsabilità e conoscenze sono intrecciate. In aggiunta sono poco flessibili all’evoluzione tecnologica poiché i dati sono fortemente legati alla sorgente fisica e quindi, un cambiamento della sorgente richiederebbe anche la modifica della logica. Per favorire l’evoluzione del sistema e il riuso dei componenti, i sistemi AAL dovrebbero mantenere separate le problematiche legate all’acquisizione dei dati rispetto a quelle sulla logica applicativa. Ne consegue che i dati acquisiti dovrebbero essere slegati dalla loro sorgente fisica. Una tale rappresentazione permette di cambiare le caratteristiche fisiche delle sorgenti di dati, senza andare ad impattare la rappresentazione delle informazioni e quindi lo strato di logica applicativa. Inoltre il livello di acquisizione dovrebbe essere strutturato per far si che i meccanismi di acquisizione siano separati dai componenti software che interagiscono con le specifiche sorgenti (i driver). Questo permette di riutilizzare gli stessi meccanismi e di programmare i driver solo per i sensori necessari. Nel caso nuovi o diversi sensori si rendano necessari, è sufficiente aggiungere o modificare i driver e configurare i meccanismi di base per applicare direttamente le modifiche. L’obiettivo di questo lavoro è quello di proporre un approccio innovativo per il design di un livello di acquisizione in grado di superare le limitazioni delle soluzioni tradizionali. La proposta si compone di due set di astrazioni architetturali: Time Driven Sensor Hub (TDSH) è un set di astrazioni architetturali dedicato allo sviluppo di sistemi di acquisizione temporizzati, di facile configurazione sia per ciò che concerne il tipo di sensori che per quanto riguarda le frequenze di acquisizione. Subjective sPaces Architecture for Contextualising hEterogeneous Sources (SPACES) è un set di astrazioni architetturali volto alla rappresentazione di misurazioni di sensori tali che siano indipendenti dalle caratteristiche fisiche del sensore di acquisizione. Tali astrazioni possono semplificare l’interpretazione dei dati e la data fusion, inoltre favoriscono sia il riuso delle infrastrutture esistenti, sia la trasparenza dei livelli di acquisizione fornendo un framework comune per la rappresentazione dei dati sensoriali. Il risultato del lavoro consiste in due design concreti e relative implementazioni, atte a reificare i modelli definiti da TDSH e SPACES. Uno scenario di test è stato considerato per contestualizzare l’utilità degli approcci proposti e per testare l’effettiva correttezza di ciascun componente. Lo scenario d’esempio è basato sulla rilevazione di cadute e sfrutta un accelerometro e un array microfonico lineare per la rilevazione.

(2017). Software Architectures For Embedded Systems Supporting Assisted Living. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2017).

Software Architectures For Embedded Systems Supporting Assisted Living

MOBILIO, MARCO
2017

Abstract

In coming decades, population is set to become slightly smaller in more developed countries, but much older. This increase results in a growing need for supports (human or technological) that enables the older population to perform daily activities. This originated an increasing interest in Ambient Assisted Living (AAL), which encompasses technological solutions supporting elderly people in their daily life at their homes. The structure of an AAL system generally includes a layer in charge of acquiring data from the field, and a layer in charge of realising the application logic. For example, a fall detection system, acquires both accelerometer and acoustic data from the field, and exploits them to detect fall by relying on a machine learning technique. Usually, AAL system are implemented as vertical solutions in which often there is not a clear separation between the two main layers. This rises several issues, which include at least a poor reuse of the system components since their responsibilities overlap, and a scarce liability to software evolution mostly because data is strongly coupled with its source, thus changing the source requires modifying the application logic too. To promote reusability and evolution, an AAL system should keep accurately separated issues related to acquisition from those related to reasoning. It follows that data, once acquired, should be completely decoupled from its source. This allows to change the physical characteristics of the sources of information without affecting the application logic layer. Moreover, the acquisition layer, should be structured so that the basic acquisition mechanisms (triggering sources at specified frequencies, and distributing the acquired data) should be kept separated from the part of the software that interacts whit the specific source (i.e., the software driver). This allows to reuse the basic mechanisms and to program the drivers for the needed sensors only. If a new or different sensor is required, it suffices to add/change the sensor driver and to properly configure the basic mechanisms so that the change can actually implemented. The aim of this work is to propose a novel approach to the design of the acquisition layer that overcomes the limitation of traditional solutions. The approach consists of two different sets of architectural abstractions: Time Driven Sensor Hub (TDSH) is a set of architectural abstraction for developing timed acquisition systems that are easily configurable for what concerns both the type of the sensors needed and their acquisition frequencies. Subjective sPaces Architecture for Contextualising hEterogeneous Sources (SPACES) is a set of architectural abstractions aimed at representing sensors measurements that are independent from the sensors characteristics. Such set can reduce the effort for data fusion and interpretation, moreover it enforces both the reuse of existing infrastructure and the openness of the sensing layer by providing a common framework for representing sensors readings. The final result of this work consists in two concrete designs and implementations that reify the TDSH and SPACES models. A test scenario has been considered to contextualise the usefulness of the proposed approaches and to test the actual correctness of each component. The example scenario is built upon the case of fall detection, an application case studied in order to be aware of peculiarities of the chosen domain. The example system is based on the proposed sets of architectural abstraction and exploits an accelerometer and a linear microphonic array to perform fall detection.
VIZZARI, GIUSEPPE
MICUCCI, DANIELA
AAL; software; architecture; fall; detection
AAL; software; architecture; fall; detection
INF/01 - INFORMATICA
English
27-mar-2017
INFORMATICA - 87R
29
2015/2016
open
(2017). Software Architectures For Embedded Systems Supporting Assisted Living. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2017).
File in questo prodotto:
File Dimensione Formato  
phd_unimib_701812.pdf

accesso aperto

Descrizione: tesi di dottorato
Tipologia di allegato: Doctoral thesis
Dimensione 8.79 MB
Formato Adobe PDF
8.79 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/151641
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
Social impact