Research Article: Large-Scale Mobile Sensing Enabled Internet-of-Things Testbed For Smart City Services
Research Article: Large-Scale Mobile Sensing Enabled Internet-of-Things Testbed For Smart City Services
Research Article: Large-Scale Mobile Sensing Enabled Internet-of-Things Testbed For Smart City Services
Research Article
Large-Scale Mobile Sensing Enabled Internet-of-Things
Testbed for Smart City Services
Jorge Lanza, Luis Snchez, Luis Muoz, Jos Antonio Galache, Pablo Sotres,
Juan R. Santana, and Vernica Gutirrez
Network Planning and Mobile Communications Laboratory, University of Cantabria, 39005 Santander, Spain
Correspondence should be addressed to Luis Munoz; [email protected]
Received 30 March 2015; Revised 6 July 2015; Accepted 8 July 2015
Academic Editor: Paolo Bellavista
Copyright 2015 Jorge Lanza et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Smart cities are one of the key application domains for the Internet-of-Things paradigm. Extending the Web into the physical
realm of a city, by means of the widespread deployment of spatially distributed Internet-addressable devices with sensing and/or
actuation capabilities, allows improving efficiency of city services. Vehicles moving around the city become excellent probes when
the objective is to gather information across the city in a cost effective manner. Public transportation fleets, taxis, or vehicles such
as waste collection trucks cover most of the urban areas with a limited number of vehicles. This paper presents the deployment of a
large scale Internet-of-Things testbed that has been carried out in the city of Santander. It extends previous descriptions by providing
a specification of one of the unique features of the testbed, namely, the devices that have been installed on 140 buses, taxis, and vans
that every day drive around the city. Besides the physical characteristics of the devices installed and the lessons learnt during the
deployment, the paper introduces the three mobile sensing network strategies used for distributing the data gathered. Finally, the
paper sketches some of smart city services which might be provided using the information coming from the mobile IoT devices.
1. Introduction
Improving efficiency of city services and facilitating a more
sustainable development of cities are the main drivers of
the smart city concept. The growth and change in cities are
accelerating and make it even harder to provide a sustainable
urban living environment [1]. The use of an Information and
Communication Technologies- (ICT-) based infrastructure
alongside the traditional utilities and services infrastructures
will be the next big step in the development of cities [2, 3].
Information systems will help to optimise infrastructure,
inform citizens, and build a communication network that
spans the city and allows tailoring the utility and services
delivery to the actual needs rather than to overprovision
for peak demands; in turn, the confluence of ICT and city
services will fuel economic growth and prosperity and will
form new city ecosystems. This revolution is still only at the
beginning as suitable infrastructures are being deployed and
significant investments into the city infrastructures are being
made.
2
to realise the most interesting and impact-generation experimentation scenarios. In this sense, part of the SmartSantanders testbed is based on the deployment of 140 IoT devices
on vehicles that are continuously driving around the city.
Such a deployment has, to the best of our knowledge, not
any analogous one in any of the testbeds existing nowadays
in the world. In addition to enabling experimentation on
vehicular networking technologies, IoT devices embarked on
vehicles get, as mobile sensors, a variety of information with
increased capillarity in time and space that allows gathering
data from all over the city in a much more efficient manner.
This capacity for getting information from almost the whole
city area is critical in supporting the provision of smart city
services.
Two main contributions are presented in this paper.
Firstly, as the deployment of distributed multipurpose multistakeholder mobile IoT infrastructure is complexity-fraught
and not risk-averse (often a compromise over platform capabilities, overall usefulness, and cost), we regard the experience
gained and lessons learnt from our physical deployment
process as a valuable contribution. In this respect, the paper
provides detailed insight on the actual physical deployment
of 140 IoT devices embarked on buses, taxis, and other city
services vehicles and how they interoperate with the rest of
devices present at the SmartSantander IoT infrastructure.
It is, however, important to highlight, for the sake of
completeness, and to fully understand the technical relevance
of this first contribution, which are the main functionalities
and related challenges that underlie the deployment and
setup of the experimental infrastructure. In this sense, the
experimentation possibilities enabled by existing testbeds,
which are primarily focused on wireless sensor network
(WSN) research, do not fully fulfill the testbed methodologies needed to perform experimentally driven research
when moving from islands of WSNs to a global networked
infrastructureas envisioned by the Internet of Things. This
vision opens up new challenges that demand new capabilities
and features from suitable testbeds [5, 6]. Key features that
have been observed for the deployment of the SmartSantander platform in general and the mobile IoT devices in
particular to support the experimentation when moving from
WSNs to IoT are (1) scale, (2) heterogeneity, (3) mobility, (4)
experimentation realism, (5) data-centricity, (6) concurrency,
and (7) autonomy.
Second contribution presented in the paper is the description and discussion of the three strategies that are used to
address the mobile crowd sensing challenge. Respectively,
they are based on periodic reporting through mobile broadband network, opportunistic Vehicle-to-Infrastructure (V2I)
networking on top of IEEE 802.15.4 links, and Delay Tolerant
Network (DTN) approach using IEEE 802.11. Moreover,
analytical assessment of the V2I-based strategy using real
data traces as obtained from the mobile IoT devices is also
presented.
Finally, the description of two smart city services that
make use of the data gathered by the embarked IoT devices is
another minor contribution presented in the paper. Presenting these services is meant to showcase the potentiality of the
deployed infrastructure besides the experimentation support,
2. Related Work
This section overviews current trend in terms of application
scenarios and main research path in vehicular networks
(VANETs) and how the smart city applications described
in this paper represent an advance on the applicability of
VANETs. Then it outlines related real-world deployments and
initiatives in which vehicles are used as sensing platforms
comparing them with the sensing infrastructure described
in this paper. Finally, a review of various mobile sensing
proposals in the literature is presented comparing the architecture and data dissemination strategies used with the ones
implemented in the SmartSantander mobile sensing testbed.
Many different services have been proposed in the
literature using VANETs like CarTel [7], TrafficView [8],
or MobEyes [9]. These services are mainly focused on
safety applications like collision warning, up-to-date traffic
information, or active navigation [10]. Thus, much of the
existing literature has focused on challenges around efficient
data dissemination [11] as it is of particular interest for
safety services. Anyway, the common denominator for the
majority of applications is that they are navigation or road
safety related and, thus, concentrate on the vehicles and
circulation domains rather than serving to domains outside
the traffic circumstances. However, vehicular networks have
the potential to become important sensor platforms, for
example, for proactive urban monitoring and for sharing and
disseminating data of general interest. Each vehicle can sense
one or more events (e.g., detecting toxic chemicals), process
sensed data, and report this information to a common sensor
platform.
In this sense, the general philosophy of smart cities [2, 12]
is a paradigm shift combining IoT and M2M infrastructures with a citizen-centric model, all together leveraging
massive data collected by sensors, connected devices, social
applications, and so forth. Most of the existing smart city
initiatives with relevant sensor deployments like Santander
[13], Barcelona [14], Oulu [15], or Cambridge [16] rely on
static environments, with application-specific monitoring
tasks. However, in the very last years, there are a number of deployments and applications in which vehicles are
3
data through short- and medium-range wireless access technologies such as IEEE 802.15.4 and IEEE 802.11 has also been
used. Analysis performed to real data traces obtained from
the embarked devices has demonstrated the suitability, under
certain conditions, of the implemented mechanisms.
4
The proposed architecture is agnostic to the communication technologies between the different elements at the different tiers. In this sense, realizations of the architecture can
be carried out using different communication technologies
between servers, GW nodes, and IoT nodes.
In order to realize this architecture, we propose a reference model for IoT experimentation testbeds that encompasses both testbed observation/management and IoT experimentation planes. We contend that such facilities require,
as illustrated in Figure 1, the provision of testbed features
by four main subsystems: (1) Authentication, Authorisation,
and Accounting (AAA), (2) Testbed Management, (3) Experimental Support, and (4) Application Support. The AAA
subsystem controls the access to the testbed by authenticating
users, authorising the invocation of particular testbed services based on user privileges, and monitoring the level of
platform-use by users. The Testbed Management subsystem
encapsulates the functionalities concerning the automatic
management of the facility. The Experimentation Support
Subsystem (ESS) provides operations to assist the user during
the entire experimentation life-cycle [27]. The Application
Support Subsystem (ASS) offers, via its Application Support
Interface (ASI), a wide range of data management functions
that can operate on information retrieved from the devices at
the IoT node tier.
In our reference model, each subsystem comprises several
functional blocks that implement the functionality expected
from the subsystem. Subsystems may span across the three
tiers requiring different components or logic to be deployed
at each tier. Subsystems export a number of interfaces. Interfaces in our reference model architecture are notional entities
that expose the functionality of the different subsystems
through a collection of Application Programming Interfaces
(APIs). In concrete instantiations of the reference model,
these interfaces may be realised through technologies such as
Web Services, RESTful APIs, messaging protocols, or event
handling to name but a few.
3.2. Fixed IoT Infrastructure. As reported in previous work
[6], existing IoT experimentation facilities have several limitations that make them fail to provide adequate support
for the emerging requirements of experimental IoT research.
The SmartSantander facility [13] offers a variety of properties
and features to overcome many of these shortcomings and
integrates them into a holistic experimentation environment.
The IoT experimentation facility deployed in Santander
was settled on a cyclic approach. The objective of the first
cycle of deployment was to create a meshed Wireless Sensor
Network (WSN) on fixed locations that would serve as a testing environment for the experimental validation of advanced
WSN-related mechanisms. The deployment, also influenced
by the city of Santander smart city service requirements
and strategy, intentionally provided a concentration of IoT
devices in the city centre (a 1 Km2 area) in order to achieve
the maximum possible impact to the citizens. Nonetheless,
other city areas are also covered.
Figure 2 shows an excerpt view of the Santander city
centre deployment. The different markers represent the
deployed nodes (i.e., carbon monoxide (CO), light intensity,
Experimental
user client
ESI
Service
provider client
ASI
Scheduling
Session mgt.
Resource reservation
Deployment
Scheduler
Monitoring
Configuration
Synthesis
Experiment spec.
File upload
Sanity check
Result analysis
Control
Resource
pub/sub/notify
Resource
DB
Experiment
DB
Testbed
MIB
O&M
DB
Participatory
sensing
IoT resource
Data
O&M
lookup
Data
pub/sub/notify
Resource lookup
PSens server
Resource mgt.
PSens
DB
Monitoring and fault mgt.
Resource discovery
Resource monitoring
Resource registration
(Re)configuration
Resource configuration
Systems reconfiguration
Application support
User
accounts
DB
ACI
Testbed
admin client
Testbed
management
Experiment support
AAA
MSI
AAA
Data
Subscription/
notify
IoT resource
Resource
mgt.
Configuration
VTD config.
Session mgt.
Deployment
Resource monitoring
Monitoring
Control
Resource discovery
Resource registration
(Re)configuration
Resource configuration
Systems reconfiguration
Testbed
management
Application
support
Gateway tier
Data
Subscription/
notify
IoT resource
Resource
mgt.
Participatory
sensing
PSens client
Session mgt.
Deployment
Resource monitoring
Monitoring
Control
Resource discovery
Resource registration
(Re)configuration
Resource configuration
Testbed
management
Application
support
Platform servers
Internet
University
intranet
GPRS
Cluster 1
City council
intranet
Cluster 3
IoT node
Gateway
IEEE 802.15.4
Cluster 2
Local
persistence
storage
Environmental data
collection module
IEEE 802.15.4
or
IEEE 802.11
(Optional) local
communication
module
7
The Environmental Data Collection Module contains a
bunch of sensors for detecting air pollutants such as nitrogen
dioxide (NO2 ), carbon monoxide (CO), and ozone (O3 ) as
well as detection of particles in suspension, temperature,
and air humidity. Installed on the rooftop, the sensors are
protected within an ABS box that is shown in Figure 5(b). In
this box, the air inlet traverse a labyrinth designed to allow air
flow while preventing water spills in the electronics.
Most significantly, mobile IoT devices are also equipped
with GPS so that all their observations come geolocalized and
they also report speed and course of the vehicle. Additionally,
the nodes deployed on the Santander public buses provide
information from the vehicles CAN-Bus (represented with
dashed line in Figure 4 due to its optionality). The information taken from the vehicles is used to feed the traffic
assessment service.
It is also important to mention that observation generation (i.e., gathering of information from the device sensors)
frequency is higher than the reporting (i.e., sending this
information to the repository at the server tier) one. In this
sense, the IoT nodes have a sampling rate that is deliberately
configured to generate a large number of observations. As
the devices count with a Local Persistence Storage memory,
the Environmental Data Collection Module is continuously
producing new observations.
3.4.2. Deployment Insights and Lessons Learnt. Apart from
presenting the mobile IoT nodes architecture and features, it
is essential to highlight some issues that have been addressed
during the deployment or have arisen as lessons learnt.
Never forget that the deployed infrastructure aim is to
support IoT experimentation, the more information that can
be captured, and the widest experimentation possibilities
that are opened. Having a budget restriction, the possibility
of enlarging the available information by increasing the
number of IoT devices has a limit. The approach taken was to
overrate the sensing frequency. This way, a larger amount of
information could be made available. However, there is also
a restriction that applies to this approach; the reporting of all
the observations gathered poses a challenge to the networking
solution to be implemented. In this sense, the solution
adopted reported one observation every five minutes, on a
real-time manner, while one observation every 80 seconds
is captured. Those observations that are not immediately
reported are kept locally and reported through Delay Tolerant
Networking (DTN) or opportunistic networking strategies
which will be described in Section 4.
The main application for the information produced by
mobile IoT nodes, namely, environmental monitoring, does
not need observations at such high pace; however, other
applications, such as traffic assessment, benefit from this
oversampling. Moreover, as it has been already mentioned,
the whole SmartSantander deployment is meant to support
experimentation on IoT technologies. Massive production of
sensor observations is critical to maximize the research and
experimentation opportunities on top of the deployed facility.
Concerning the powering of the mobile nodes, it is
realized by connecting them to the vehicle electrical batteries.
This way, we guarantee a large, practically endless, energy
(a)
(b)
Figure 5: (a) Detail of sensor nodes installed on public bus; (b) Environmental Data Collection Module.
(a)
(b)
Figure 6: (a) Mobile IoT devices real-time location view; (b) mobile IoT devices aggregated location view.
9
guaranteeing that the resulting solution is suited to a real-life
urban setting.
Data-centricity is another key differentiator between
WSN experimentation and IoT one. While, for the first
one, focus is put on the devices and how they operate
(communicate, internetwork, etc.), IoT builds on top of
devices but provides a higher level of abstraction in which
the focus is put on the services that objects provide (generally
data gathering and reporting). The SmartSantander platform
and the mechanisms deployed for intertier communication
were implemented with this feature as a fundamental design
consideration and the deployed mobile infrastructure has
been successfully integrated in it.
Experimentation as a service model has been adopted by
the SmartSantander platform so that concurrent experimentation can be handled. This way the underlying infrastructure
is decoupled from the experimenters requests guaranteeing
scalability both in terms of enlarging the deployed infrastructure, where as a matter of fact mobile IoT devices have been
seamlessly incorporated into the testbed on a plug and play
manner, and in terms of increasing the number of concurrent
experiments.
With the scale and variety of testbed management
events to track, one cannot assume that human intervention
alone is sufficient to provide timely response to events and
remediation to faults. Testbed management automation has
been incorporated keeping the human in the loop only for
decision-making and policy-specification. Moreover, there
are other aspects, more related to the deployment and
networking, that have been put forward in order to maximize
testbed autonomy. Firstly, the energy demands from the
infrastructure, which are augmented due to experimentation
support necessity, have been tackled during the installation
process guaranteeing large enough energy sources for all the
mobile IoT devices. Secondly, multiple intertier communication networking solutions have been adopted in order to
avoid single point of failure problems and maximize testbed
resiliency.
10
information can fall rapidly, smart city services need to be
capable of capturing relevant information in real-time and
distributing it in a format that citizens and organisations can
act on immediately.
WAN-based observations reporting strategy make use
of the GPRS interface with which all the IoT devices
installed on the vehicles of the SmartSantanders facility are
equipped. These IoT devices are programmed to periodically
send an observation to the so-called GW for mobile nodes
(GW4MN). This element, directly connected to the server
tier, forwards these observations to the appropriate repository.
The periodicity with which observations are reported
is a tunable parameter depending on time, distance, or a
combination of both. Thus, an observation can be reported
every certain number of seconds, when the vehicle has
travelled a specific number of meters or when the first of these
two conditions occur. Currently, this reporting frequency is
fixed to five minutes.
4.1.2. Strategy Discussion. This reporting strategy is used as
the baseline for the SmartSantander deployment. It guarantees sufficiently fresh information since GPRS coverage
of the city is almost 100% assured. Moreover, it supports
real-time event-based asynchronous reporting. Additionally,
it is the only one that can be implemented by all the IoT
devices installed on the SmartSantanders vehicles as it does
not depend on the optional Local Communication Module.
The main drawback for this strategy is that frequent
connection through the cellular network is costly in terms of
energy and should be restricted in terms of quantity of data
due to economic costs.
While the power consumption is not a major issue due
to the fact that nodes installed are connected to the vehicle
battery in order to guarantee permanent and almost endless
access to energy, the economic cost associated with this
reporting strategy should not be neglected. The solution
adopted to minimize the impact and keep these costs to
a minimum while maintaining the experimental facility
grade of service was to negotiate the commercial contract
for the mobile nodes deployment in conjunction with the
Santander municipality mobile phone lines service. This
way it was possible to get competitive fees as economies
of scale apply. However, it is clear that although it was
possible to minimize the communication costs, reaching
a trade-off between observations reporting freshness and
cost, it is important to develop other observations reporting
strategies that reduce the costs while increasing the amount
of information retrieved.
4.2. V2I-Based Opportunistic Observations Reporting
4.2.1. Strategy Description. As the vehicles circulate around
the city, they might come close to some of the fixed IoT
devices. As it has been introduced in Section 3, IoT devices
installed on the Santanders public transportation buses were
equipped with a Local Communication Module which in the
majority of cases included an IEEE 802.15.4. This module
was installed on the roofs of the buses in order to enable
Weekday
Weekend
Std. deviation
117.43
153.32
11
(a)
3,795 W
3,8 W
3,805 W
3,81 W
3,815 W
3,82 W
3,825 W
3,83 W
3,795 W
3,8 W
3,805 W
3,81 W
3,815 W
3,82 W
43,463 N
43,462 N
43,461 N
43,46 N
43,459 N
43,458 N
3,825 W
3,83 W
43,463 N
43,462 N
43,461 N
43,46 N
43,459 N
43,458 N
(b)
Figure 7: (a) Fixed IoT devices location at city centre; (b) mobile IoT devices opportunistic reporting positions.
0.12
Probability density
0.1
0.08
0.06
0.04
0.02
0
12
10
8
6
4
2
0
3:00:00
4:00:00
5:00:00
6:00:00
7:00:00
8:00:00
9:00:00
10:00:00
11:00:00
12:00:00
13:00:00
14:00:00
15:00:00
16:00:00
17:00:00
18:00:00
19:00:00
20:00:00
21:00:00
22:00:00
23:00:00
Number of observations
reporting opportunities
Id = 3010
Id = 3044
Id = 3048
Id = 3078
12
WAN-based
V2I-based
Batched
Capacity
Restricted
Unbounded
Unbounded
Delay
Negligible
Restricted
High
Cost
High
Zero
Zero
Power consumption
N/A
N/A
N/A
Real-time support
Yes
No
No
Security
High
High
High
13
(a)
(b)
(c)
(d)
Figure 10: Real-time environmental monitoring Santander area heatmaps. (a) Temperature ( C); (b) relative humidity (%); (c) dust particles
(mg/m3 ); and (d) carbon monoxide (mg/m2 ).
It is already possible to find in the literature [32, 33] experiences in which Floating Car Data paradigm is employed to
perform traffic estimation. With a good number of vehicles
moving around the city it would be possible to enrich
currently available traffic condition assessment applications
with the information on speed and course gathered by the
mobile IoT devices as they circulate across the city. Figure 11
shows colour-coded (green for fluid, yellow for moderate,
and red for dense) assessment of the Santander streets traffic
conditions. This map is part of a smartphone app which is
currently using the information from the inductive loops but
is being extended to receive also the information from the
mobile IoT devices.
6. Conclusions
Vehicular networking will be inevitably part of the IoT
scenario and smart cities are one of the most evident examples
of the IoT applicability potential. This paper focuses on
the description of a large-scale IoT testbed deployed at
the city of Santander with special emphasis on the devices
14
that have been installed on 140 vehicles that circulate all
over the city daily. It presents the detailed description of
the embarked devices and analyses the experimentation
alternatives that they enable. In this sense, particular stress
is put on the mobile IoT devices features enabling, through
short-range low-power radio technologies, the interaction
not only among them but also with the fixed infrastructure
deployed at Santander streets. We expect that this description
will promote a better knowledge of the testbed features
and experimentation scenarios that it enables and assist IoT
experimenters to prepare such experiments.
Three different mobile sensing data distribution strategies
have been implemented in the testbed and they have been
described in this paper, while the main data dissemination strategy implemented for the SmartSantander mobile
sensing deployment is based on periodic reporting using
cellular access network. However, opportunistic offloading
of sensed data through short- and medium-range wireless
access technologies such as IEEE 802.15.4 and IEEE 802.11
has also been used. Future work in this area will focus
on complementing the analysis with a detailed study of
the delay conditions for V2I-based strategy. Moreover, the
practical implementation of opportunistic batched reporting
at hotspots different from bus depot will also complement
currently implemented solutions.
Finally, a brief description of two smart city applications
for the information gathered through the mobile sensing
infrastructure has been presented as easily understandable
examples of the added value that can be obtained thanks to
real-world roll-outs of IoT (fixed and mobile) technology.
Conflict of Interests
The authors declare that there is no conflict of interests
regarding the publication of this paper.
Acknowledgments
This work has been partially funded by Research Project
SmartSantander, under FP7-ICT-2009-5 of the 7th Framework Programme of the European Community. The authors
would like to acknowledge the collaboration with the rest
of partners within the consortium leading to the results
presented in this paper. The authors would also like to express
their gratitude to the Spanish government for the funding in
the following project: Connectivity as a Service: Access for
the Internet of the Future, COSAIF (TEC2012-38574-C0201).
References
[1] WHO, Urbanization and health, Bulletin of the World Health
Organization, vol. 88, no. 4, pp. 245246, 2010.
[2] H. Schaffers, N. Komninos, M. Pallot, B. Trousse, M. Nilsson,
and A. Oliveira, Smart cities and the future internet: towards
cooperation frameworks for open innovation, in The Future
Internet, vol. 6656 of Lecture Notes in Computer Science, pp. 431
446, Springer, Berlin, Germany, 2011.
15
International Journal of
Rotating
Machinery
The Scientific
World Journal
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Engineering
Journal of
Volume 2014
Volume 2014
Advances in
Mechanical
Engineering
Journal of
Sensors
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
International Journal of
Distributed
Sensor Networks
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Advances in
Civil Engineering
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Volume 2014
Advances in
OptoElectronics
Journal of
Robotics
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Volume 2014
VLSI Design
Modelling &
Simulation
in Engineering
International Journal of
Navigation and
Observation
International Journal of
Chemical Engineering
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Volume 2014
Advances in
Volume 2014
Volume 2014
Journal of
Control Science
and Engineering
Volume 2014
International Journal of
Journal of
Antennas and
Propagation
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Volume 2014
Volume 2014