Orchestrating the Life Cycle of Participatory Sensing Campaigns Using Workflows

Scriptie/masterproef: Master's Thesis


In the Information Era where mobile technology is flourishing on daily basis, the potential of
hand-held devices cannot be underestimated. Smart-phones continue to grow in resourceful-
ness thanks to their steep revolutionary inclination in terms of processing power, memory
embedded equipment and overall versatility. This has led to an increasingly high integration of
smart-phones in our life, finding its way into our everyday tasks through the various sensors that
are available by standard nowadays. As a consequence, smart-phones have become able to monitor
their immediate environment, wherever they are. The increasing observational capabilities
in relatively cheap and prevalent devices, drew the attention of the scientific research world to
the domain of citizen science and participatory sensing, where citizens are recruited to partake
in scientific data collection processes, using their own, personal devices. This is contrasted by
the traditional approach, requiring specialised, expensive equipment and domain specialists who
oversee the project.

Participatory sensing as a paradigm offers a fresh perspective on the idea of tackling a communal concern. It allows the community itself to take initiative in doing quantitative data collection
in order to tackle such a concern. They can set up and manage personalised sensing campaigns.
However, in order to efficiently execute such a sensing campaign, the the presence of a do-
main expert is required, who can steer the execution process effectively. It makes sense that it
would be beneficial for the community if this dependency was alleviated. This is possible by automating the coordination of the campaign, making it able to steer campaign execution towards
success. This automatic coordination is what is called orchestration.

In this dissertation, we kept building on this notion of campaign orchestration. More particularly, we have focussed on the distinct concerns that are latently present when dealing with
campaign sustainability. We distinguished campaign coordination, participant recruitment and
data collection and triage as three major issues that require to be dealt with in order to provide a
qualitative orchestration framework capable of taking over the job of the domain expert. More
specifically, we have researched how we can utilize workflows as a paradigm to model the orchestration process and run it using an enactment engine. While we did not focus on presenting
advanced analysis tools which offer a specialised solution for the covered concerns, we equipped
the framework with general-usage tools that can demonstrate the expressiveness and potential of
the framework, as well as its limitations and the weaknesses of the chosen implementation.

In order to analyse the functionality of the framework, we covered several common scenarios
in a participatory sensing campaign and discussed how the framework reacts to these situations.
This gives an insight in to what extent either workflows as a concept, the used technology or the
implementation allow for fully covered orchestration.

The main conclusions we were able to derive is that workflows offer us a very intuitive and
expressive approach in order to model not only the general orchestration framework, but also
gives both expert and layman the opportunity to reason about the campaign coordination processes. However, due to the experimental nature of the underlying technology, we identified
several shortcomings that require attention in order to reach the maximum extent of the potential
of the underlying idea of the framework.
Datum Prijs1 sep 2014
BegeleiderWolfgang De Meuter (Promotor), Jesse Zaman (Advisor), Eline Philips (Advisor) & Ellie D'Hondt (Advisor)

Citeer dit