Page 133 - ITU Kaleidoscope 2016

A STACK4THINGS-BASED PLATFORM
FOR MOBILE CROWDSENSING SERVICES

S. Distefano A. Puliafito, G. Merlino, F. Longo, D. Bruneo ∗

Kazan Federal University Universit` a di Messina
Higher Institute for Department of Engineering
Information Technology and Information Systems Messina, Italy
Kazan, Russia Email: {apuliafito,gmerlino,
Email: sdistefano@kpfu.ru flongo,dbruneo}@unime.it

MCS Application
ABSTRACT Frontend Server
As mobiles grow pervasive in people’s lives and expand their Service
Provider
reach, Mobile CrowdSensing (MCS) and similar paradigms Backend Aggregation
Server Backend Aggregation
are going to play an ever more prominent role. There is a Aggregate Aggregate Server
Analytics Analytics
pressing need then to ease developers and service providers
in embracing the opportunity, and that means offering a Data
Data
Provider Provider
platform for such efforts. This in turn means providing a
solid foundational architecture with abstractions and sound
layering for MCS application designs to be mapped over
it. This should base on a flexible infrastructure able to
provide resources to MCS applications according to their re-
quirements, hopefully on-demand. A service-oriented/Cloud Node1 Nodei Nodej Nodek
Node2 Noden
model can perfectly fill this gap. This paper is a first step
MCS Sensing Network/Infrastructure
in this direction, proposing to adopt Stack4Things (S4T), Owners/
Contributors
an OpenStack-based platform for managing sensing and
IoT nodes, for runtime customization of resources and their Figure 1. The MCS reference scenario.
functions to support MCS services and applications. This
implies developing and extending the S4T platform further
to the specific requirements coming from off-the-shelf, e.g., corresponding experiments in advance, just leveraging nat-
Android-based, mobiles, as well as describing an example ural daily life patterns arising from human activities as they
S4T-powered MCS application, Pothole Detection Mapping, happen and leave behind breadcrumbs in form of samplings
to highlight the role of the platform. ready to be collected.
Typical MCS applications mainly implement a client-server
Keywords— Mobile crowdsensing, Cloud, IoT, OpenStack,
interaction pattern where a service provider offers MCS-
Android.
based services to end users, leveraging contributors willing-
ness to provide their physical (sensing) resources. Data are
1. INTRODUCTION AND MOTIVATIONS therefore collected and processed by (backend and frontend)
servers to carry out aggregate analytics and feed back rel-
Mobile CrowdSensing (MCS) comprises by definition a evant results to end users. Starting from the lowest level,
category of applications where individuals carrying sensor- through heuristics and algorithm design, local analytics may
hosting embedded computers (e.g. smartphones) get collec- provide a category of functions, among which simple ones
tively engaged in information gathering and sharing efforts are interpolation, extrapolation and outlier filtering, which
to analyze and georeference events which may be inter- may enhance a standard MCS application as shown in Fig-
esting for individuals and communities alike. One of the ure 1. A few drawbacks of such siloed pattern limiting
main advantages of MCS is the possibility to conduct sam- the potential of the underlying paradigm are: i) unopti-
ple collection, data mining, etc., without accounting for the mized runtime, as multiple applications would execute on
the same nodes without taking into account such configu-
∗ This research effort was supported by the EU 7th Framework Pro- ration, possibly duplicating sensing or processing activities
gramme under Grant Agreement n. 610802 for the “CloudWave” project on resource-constrained devices, thus also limiting scala-
and by the EU Horizon 2020 Research and Innovation Program under the
Grant Agreement n. 644048 for the “BEACON” project, and has been car- bility of the platform; ii) necessity of enrolling, collecting
ried out in the framework of the CINI Smart Cities National Lab. and managing resources, i.e., lack of a coordinated sens-

978-92-61-20431-0/CFP1668P-ART © 2016 ITU – 115 – Kaleidoscope