The OpenIoT Approach to Sensor Mobility with Quality-Driven Data Acquisition Management - Interoperability and Open-Source Solutions for the Internet of Things - page 58

6 Related Work
Significant research effort is focused on IoT architectures integrating sensor
data streams with cloud environments which address the issues of energy- and
bandwidth-ecient data collection. In the work reported in [ 4 ] the authors pro-
pose a collaborative mobile sensing framework called Mobile Sensor Data EngiNe
(MOSDEN), designed to operate on smartphones to capture and share sensed
data between multiple distributed applications and users. The engine is designed
so as to be compatible with the GSN (Global Sensor Network) middleware. By
supporting processing and storage on end user smartphone devices, the platform
aims to reduce the necessary data transmission to a centralized server, conse-
quently achieving bandwidth and energy eciency. In their subsequent work [ 7 ],
the authors specifically address sensor discovery and configuration challenges
and issues such as configuring sensor sampling rate to determine an optimal
balance between user (application) requirements and energy consumption.
Specifically focusing on mobility aspects, Mobile Crowdsensing applications
(MCS) take the advantage of a population of individuals to measure large-scale
phenomena that cannot be otherwise measured by individuals [ 4 ]. The challenges
of meeting resource limitations in the context of MCS applications are summa-
rized in [ 3 ]. The authors further discuss resource allocation challenges in the case
of multiple concurrent applications sampling various sensors on a single mobile
device. Potential solutions include prioritizing applications that require sensor
data, hence reducing or increasing the sampling rate of certain sensors while
aiming to achieve e cient energy consumption of the mobile device. A discus-
sion of different mobile crowdsourcing applications and optimizing smartphone
related energy consumption is given in [ 1 ].
In general, model-driven approaches to data acquisition in sensor networks
have demonstrated high-fidelity representation of real phenomena while requir-
ing smaller amounts of live data to be collected [ 2 ]. In [ 6 ], the authors address the
problem of energy eciency in case of redundant sensor readings and present an
approach for model-driven adaptive environmental sensing. Their approach is
complementary to the techniques proposed in this paper since their solution
requires that mobile devices maintain local models of expected sensor read-
ings hence generating predictive readings, and push updates to the back-end
server only in cases when predicted values do not match actual sensor readings.
The authors in [ 10 ] provide an extensive overview of utility-driven data acquisi-
tion techniques for ecient collection of data in participatory sensing, whereby
queries of different types (e.g., one-shot queries, continuous monitoring queries)
may come from different applications. In the context of data acquisition, the
proposed algorithms aim to achieve ecient sharing of sensor data among mul-
tiple queries that may be of different types, and is thus more general than the
problem addressed in this paper.
While a number of aforementioned projects and approaches focus on mobile/
fixed sensing architectures and address the issues of energy- and bandwidth-
ecient data collection, what is missing is a generalized solution for providing
QoS support at different levels (physical level, network level, application level)