Methods for Requirements Engineering, Verification, Security, Safety, and Robustness in AIoT Systems
Kapitel i bok, 2024
security, safety, and robustness with a special focus on AIoT systems. It
covers an architectural framework dealing with requirements engineering
aspects of distributed AIoT systems, covering several clusters of concern
dealing with the context description of the system, learning environment of
the deep-learning components, communication concerns, and a set of quality
concerns, such as ethical aspects, safety, power, security, and privacy aspects.
Each cluster contains a set of architectural views sorted into different levels
of abstraction. In addition, it introduces WebAssembly as an interoperable
environment that would run seamlessly across hardware devices and software
stacks while achieving good performance and a high level of security as a
critical requirement when processing data off-premises. To address security
aspects in AIoT systems, remote attestation and certification mechanisms are
introduced to provide a TOCTOU (time-of-check to time-of-use) secure way
of ensuring the system’s integrity.
machine learning
verification
robustness
safety
AIoT
security
WebAssembly
TOC- TOU
IoT
requirements engineering
Författare
Marcelo Pasin
University of Applied Sciences and Arts Western Switzerland
Jämes Ménétrey
Université de Neuchâtel
Pascal Felber
Université de Neuchâtel
Valerio Schiavoni
Université de Neuchâtel
Hans-Martin Heyn
Chalmers, Data- och informationsteknik, Interaktionsdesign och Software Engineering
Eric Knauss
Chalmers, Data- och informationsteknik, Interaktionsdesign och Software Engineering
Anum Khurshid
RISE Research Institutes of Sweden
Shahid Raza
RISE Research Institutes of Sweden
Shaping the Future of IoT with Edge Intelligence
197-228
978-87-7004-027-3 (ISBN)
Very Efficient Deep Learning in IOT (VEDLIoT)
Europeiska kommissionen (EU) (EC/H2020/957197), 2020-11-01 -- 2023-10-31.
Styrkeområden
Informations- och kommunikationsteknik
Ämneskategorier (SSIF 2011)
Data- och informationsvetenskap
DOI
10.1201/9781032632407-12