TY - BOOK

T1 - Context-Aware Integrability and Maintainability of Cyber-Physical Ecosystems

T2 - Tactics and Tools

AU - Tomlein, Matus

PY - 2018/4/9

Y1 - 2018/4/9

N2 - Rapid advances within the Internet of Things have increased the capabilities of embedded software on industrial products. Software-intensive industrial products have enabled integration into larger Cyber-Physical Systems, which are hybrid systems providing computation and services to support physical processes. This development provides completely new opportunities for developing industrial ecosystems. A paradigm shift can be made from a situation where development of customized systems requires deep ties between organizations or dedicated hardware components to a situation where customized functionality can be provided in open software ecosystems through software development and integration. However, it requires that product manufacturers transition from providing value solely with mechanical and hardware components to continuous innovation through software and new services. This transition to hybrid industrial products gives rise to new research challenges that need to be addressed.This thesis addresses a number of software challenges resulting from the transition that are faced by companies. It develops methods and tools to support two key quality attributes of cyber-physical ecosystems: integrability and maintainability. The presented research was conducted experimentally in collaboration with industrial partners. By working towards the quality attributes, it provides value for different actors in industrial ecosystems: system installers, system builders, service providers and end-users.This work proposes tactics that utilize context-awareness of industrial systems, domain knowledge codified by domain experts or collected from previous experience and reasoning using rules and machine learning. The tactics provide novel approaches in order to achieve and optimize the two quality attributes. The work further introduces and develops novel tools and methods to realize the tactics.The developed tools use Augmented Reality and Visual Programming to enable experts leverage context and domain knowledge to seamlessly integrate customized software applications in industrial systems. Another tool focuses on common maintenance operations and enables reuse of trained models for activity recognition using “transfer learning” to retain the ability to recognize usage and maintenance activities when the sensed context changes due to the maintenance operations. The tools were developed with feedback from experts at industrial companies and evaluated using real-world applications, using sensor data collected from simulated maintenance operations, and in studies with domain experts.The developed tools and tactics support future software ecosystems for Cyber-Physical Systems by enabling a more seamless experience, as well as reducing required effort and time for common integration and maintenance tasks performed by system installers and end-users of industrial systems.

AB - Rapid advances within the Internet of Things have increased the capabilities of embedded software on industrial products. Software-intensive industrial products have enabled integration into larger Cyber-Physical Systems, which are hybrid systems providing computation and services to support physical processes. This development provides completely new opportunities for developing industrial ecosystems. A paradigm shift can be made from a situation where development of customized systems requires deep ties between organizations or dedicated hardware components to a situation where customized functionality can be provided in open software ecosystems through software development and integration. However, it requires that product manufacturers transition from providing value solely with mechanical and hardware components to continuous innovation through software and new services. This transition to hybrid industrial products gives rise to new research challenges that need to be addressed.This thesis addresses a number of software challenges resulting from the transition that are faced by companies. It develops methods and tools to support two key quality attributes of cyber-physical ecosystems: integrability and maintainability. The presented research was conducted experimentally in collaboration with industrial partners. By working towards the quality attributes, it provides value for different actors in industrial ecosystems: system installers, system builders, service providers and end-users.This work proposes tactics that utilize context-awareness of industrial systems, domain knowledge codified by domain experts or collected from previous experience and reasoning using rules and machine learning. The tactics provide novel approaches in order to achieve and optimize the two quality attributes. The work further introduces and develops novel tools and methods to realize the tactics.The developed tools use Augmented Reality and Visual Programming to enable experts leverage context and domain knowledge to seamlessly integrate customized software applications in industrial systems. Another tool focuses on common maintenance operations and enables reuse of trained models for activity recognition using “transfer learning” to retain the ability to recognize usage and maintenance activities when the sensed context changes due to the maintenance operations. The tools were developed with feedback from experts at industrial companies and evaluated using real-world applications, using sensor data collected from simulated maintenance operations, and in studies with domain experts.The developed tools and tactics support future software ecosystems for Cyber-Physical Systems by enabling a more seamless experience, as well as reducing required effort and time for common integration and maintenance tasks performed by system installers and end-users of industrial systems.

M3 - Ph.D. thesis

BT - Context-Aware Integrability and Maintainability of Cyber-Physical Ecosystems

PB - Aarhus Universitet

CY - Aarhus

ER -