Loading...
Publication
Interference-free Integration of Commercial Off-The-Shelf Components in Pervasive Computing Systems
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 22.18 MB | Adobe PDF |
Advisor(s)
Abstract(s)
The nineties were marked by futuristic visions of massive proliferation and
interconnection of ubiquitous off-the-shelf systems. This innovation movement
became known as ubicomp (ubiquitous computing). This new computing
age raised new challenges in different areas, such as, ambient intelligence
and applied artificial intelligence, software engineering planning, distributed
systems and services, machine learning and security. These research trends
gained preponderance in ubicomp but particularly the generic use of commercial
off-the-shelf systems became pivotal in the massive integration and
deployment of pervasive computing technology. This gains special importance
when we think about device integration in the same household where
unplanned interactions between users and ubicomp systems may occur (e.g.,
entertainment, communication, and health-related devices). Unplanned interactions
are a serious concern since they may cause unsafe behavior on some
of the systems. Think, for example, behavior interferences between communication
or entertainment applications with health-related devices.
In such ubicomp scenarios the deployed system should be aware of the surrounding
context and adapt to cope with the user needs and nearby deployed
systems. These systems are supposed to support and execute services for
the users but are usually planned and developed independently and by different
vendors. Therefore, the deployment and integration of these systems in
the same environment or house are usually conducted in an unplanned manner,
i.e., without a standard integration procedure that may allow inter-system
connection and reconfiguration. Moreover, the traditional software development
processes do not aim to avoid interference between systems during the
conception stage. Furthermore, it is hard to identify and understand a priori
all possible interactions between every possible system. The integration
task is therefore complex since there are multiple possible systems combinations
and variations. Moreover, since these systems are not usually aware
of other systems in the same household, their behavior may potentially trigger
a chain of unexpected reactions and jeopardize other systems operations.
When these actions result in unexpected behavior, this problem is known as
functional behavior interference or feature interaction. The ubicomp integration of off-the-shelf systems suffers from potential interference issues, and
current deployment strategies do not cope with such problems yet.
This thesis presents a novel graph-based approach for representing the expected
behavior of commercial off-the-shelf devices and applications, their
interactions, and for detecting interference among pervasive computing systems
deployed on the same home context. A set of home care scenarios is
used to assess the applicability of this approach. We consider also two setups
where this approach can be applied: i) in a pre-deployment setup, where simulation
is used to detect possible instances of interference, and ii) at run-time,
collecting observations from devices and applications and detecting interference
as it occurs. For pre-deployment and simulation, we use Opensim to
recreate a household with several deployed systems. For run-time, we use
Simple Network Management Protocol for systems state introspection and a
sliding window mechanism to process the collected data-stream.
The proposed model uses graph-analysis algorithms, which together with
behavior introspection mechanisms, allow us to identify the occurrences of
functional interferences, their potential sources and subjects, and the associated
shared media through which such interferences may occur. The interference
analysis is performed through a chain of working filters, each performing
a specific check for interference. The intra-system filter provides mechanisms
for identifying uncommon individual system behavior. The intersystem
filter analyzes interferences between systems by inspecting if each
medium is affected and shared by more than one system during their operations.
A combined approach, using intra-system and inter-system outputs,
allows us to identify any misbehavior and associated causing interactions and
media dependencies. The detection of functional interference, both in predeployment
and post-deployment scenarios, will allow defining reconfiguration
or adaptation strategies, thus guaranteeing a safe integration of ubicomp
systems.
Description
Tese apresentada à Universidade Fernando Pessoa como parte dos requisitos para obtenção do Grau de Doutor em Ciências da Informação, especialidade em Sistemas e Tecnologias da Informação