23rd Telecommunications forum TELFOR 2015 Serbia, Belgrade, November 24-26, 2015.

Setting Up a Multi-Purpose Internet of Things

System
Marko Simeunoviü, Member, IEEE, Andrej Mihailovic, Member, IEEE, Milica Pejanoviü-Djurišiü,
Member, IEEE

standardization and harmonization of the features of IoT

Abstract —Internet of Things deployments and systems systems [4], it remains that the defining property of such
that accompany their functioning, data provisions and systems is in the engagement of small devices that achieve
actuations are undoubtedly marking the new era of connectivity in the Internet subject to their purpose, e.g.
opportunities present in the communication networks and
data provisioning or actuations. In addition, it is generally
the Internet. The technologies of Internet of Things are
extremely diverse. At the moment, there are numerous types considered that this connectivity will play a significant
of small devices, communication technologies, operating part in formation of the Future Internet visions with the
systems, protocols, platforms for data collections and use and availability of IPv6 addresses (i.e. projections are
processing…many with proven and expanding commercial that there will be billions of “things” at the end of this
exploitations. This paper outlines a practical approach for decade [5]). Reality of the current momentum behind the
understanding the challenges and design issues encountered
growth of the IoT systems can be generalized into two
when a new Internet of Tings system is built by taking all the
existing technology as experience and options in the design. global drivers: 1) expanding the use and presence of the
Internet of Things system design considered in the paper is existing IoT systems, 2) constant emergences of new
multi-purpose meaning it is unbounded in terms of the connected IoT devices and systems. The first driver can be
technology and applications. In the process of building such a marked by the pursuit for the consumer base for the data
system, there are many generic and converging functional and the actuations services of IoT systems that have
and architecture points. Hence, many of the available
systemized the processes of connecting devices, their data
technology can be understood and integrated not exclusively
but coexisting with their alternatives. and operational functionalities. The second drive can see
Keywords — Internet of Things, Future Internet. designers planning the technology and appropriate data
handling components of their systems subject to design
I. INTRODUCTION objectives and evaluation of the existing solutions and

D EVELOPMENT and proliferations of the Internet of

Things (IoT) solutions has gained a significant
momentum in the recent years. The early efforts on
experience. Such an approach can be appropriately
focused with a use of dedicated technology (e.g. a sensor
system suited to a particular application) or it can be
transforming Wireless Sensor Networks (WSN) [1] into instigated by the availability and opportunities in engaging
IP-connected devices (i.e. “things” of IoT) [1][2][3] have sensors for various purposes in everyday life.
progressed into a wide ranging diversity of topics, This paper addresses some practicalities of a holistic
expansion of research areas and practical objectives that design approach for a multi-purpose IoT systems based on
are being addressed under the themes of IoT. These have a defined strategy [6] for a centralized, university-driven
occurred as a consequence of deploying IoT solutions in regional deployment of IoT devices. The purpose of such
ranges of applications, from e-health, smart agriculture, deployment strategy is to foster growth of diverse IoT
smart cities, smart homes, transportations, environmental applications, advance research capacities, enable
monitoring, automated processes etc. In fact, one can innovations and commercialization of IoT services and
observe that IoT as the term describing families of establish data collections and handling repositories that
technology has inconsistent points of commonalities when would serve all of the previous purposes. As the timing of
plethora of technologies and applications are analyzed this activity sits at the moment of significant available
considering functionalities of devices/”things”, operating expertise and tools for IoT system development, available
systems, layers, communication protocols, applications, throughout the world, the paper addresses some of the
centralized data processing units (e.g. clouds), means of practical design issues that are encountered in the
representing data (e.g. semantics, ontologies) etc. synthesis (in Section II) and proposes a composition of
While there are global initiatives towards technologies of IoT (in Section III).
II. DESIGN QUESTIONS
A design often starts with the planning dictated by the
All authors are with the Research Centre for ICT, Faculty of Electrical capabilities of hardware, or in simple terms, what can be
Engineering, University of Montenegro, Bulevar Dzordza Vasingtona bb,
81000 Podgorica, Montenegro (e-mails: [email protected], done with specific sensors or actuation devices. The
[email protected] & [email protected], scale and applicability of a design determine the forms of
[email protected]) data being delivered and/or the storage and processing
capabilities that are needed. These elementary steps often

978-1-5090-0055-5/15/$31.00 ©2015 IEEE 273

Authorized licensed use limited to: UNIVERSITY OF SOUTHAMPTON. Downloaded on February 27,2024 at 14:58:09 UTC from IEEE Xplore. Restrictions apply.
 constitute the most important building blocks of a system the purpose of the scenario. An exceptions is if the design
to which more features and sophistication can be added as requirement is to have each “thing”, either a gateway or a
the system and its use expand (e.g. guides on setting up device inside the cluster/mesh, act as a web server (e.g.
elementary IoT communications are widely available to embedding mini-IP suite using 6LowPAN [8] and then
beginners: http://developer.sense-os.nl/CommonSense/). using CoAP [9] as the application protocol) where IP
The hardware and PHY/MAC layer communication connectivity then becomes necessary throughout the
protocols of a single device or clusters of devices then cluster/mesh (IPv6 reachability or IPv4 translations still
constitute the manner in which data is packetized and apply).
delivered to the points at which it can be presented on the To resolve some of the above mentioned options, a
web and/or delivered to users. The choice of hardware that design is often primarily concerned from how and where
hosts sensors is nowadays available via microcontroller data are made available in an IoT system. The answers
processing components such as Arduino, Raspberry PI, to these questions usually describe the purpose and
Galileo…In a multi-purpose systems we do not pose any operations of an IoT system. In a dedicated purpose
restrictions on the type of hardware nor communication scenario (e.g. smart home or a system run by a single
protocols. Regarding the communication protocols, i.e. smart city company) flow of data follows the structure and
layer 2 technologies, we see two elementary decisions, logic of the system’s purpose (e.g. it can delivered to and
subject to scenarios: a) communication medium between processed by either a local gateway or an autonomous data
the “things” if there are numbers of them in proximity (i.e. cloud/platform installation). In a multi-purpose setup
cluster), and, b) communication medium with the discussed in this paper, the requirements are more generic
Internet/fixed network over which data is delivered. and encompassing, data is still gathered or requested from
Setting up a communication link with the a centralized database but it is also made available at
Internet/fixed network is a decision that equally applies to repository locations such as gateways or stand-alone nodes
stand-alone “things” or clusters’ gateways, where in the (requiring address/URL resolving). In this manner, IoT
latter case the gateway bears the communication for all devices and clusters are deployed as autonomous entities
“things” within a cluster/mesh. There are many with abilities to supply basic data gathered and provide
PHY/MAC communication protocols available: WiFi, data to a processing platform for further handling. In
standard cellular SIM-card based, WiMax, dedicated practical terms, URL/addresses of Application
cellular 3GPP Rel-12/13, 3GPP GERAN Rel-13, then, Programming Interface – API (e.g. REST API) can be
IoT/WSN suitable short range technologies between resolved to point to centralized database/platform or at
“things” such as IEEE 802.15.4, ZigBee, Bluetooth if the individual locations in networks of “things”.1
access point has the appropriate interfaces, long range The above text contains basic design questions that help in
such as ETSI LTU…The communication hop between the constructing and justifying the main operational elements
gateway/stand-alone “thing” and the Internet induces the of the IoT system architecture components discussed in the
question of addressing and network layer connectivity. As next section. We note a further set of pertinent issues that
IPv6 addresses are available to IoT it is straightforward to complete the design considerations. The range of IoT
assign an IPv6 address per “thing”. However, in practice, devices expands with the use of technology such RFID
many of the access networks still use IPv4, Network tags, QR codes, NFIC and reading or tagging technologies;
Address Translations (NATs) or adapted IP transport as in “things” effectively become physical objects. This can be
cellular networks, hence, hampering the IPv6 address coupled with data profiles available from conventional
distributions from making a significant impact in things (e.g. sensors or actuators connected to a
liberalizing distributions of IoT devices in the Internet. In communicating/processing small device) to formulate
addition, there are plenty of open issues: large scale identity as the “higher derivative” of the simple data that
topological relevance of the IPv6 addresses for numerous can be associated with the sensing or actuation state. As
gateways or stand-alone devices; dynamic DNS for IoT; such, it is decoupled for the network level identity, i.e. IP
mobility; question of the need to identify each “thing” at addresses. In such cases, “things” become embodied at the
the network level since the IoT communication follows a instance of the system where sufficient processing, data
different communication model than the traditional linking, tracking and semantics are applied, this being at
Internet human-to-human/server model… the centralized cloud/platforms. API generation then
Establishing a cluster of “things”, i.e. a mesh, is often happens from the platform rather than being linked to the
dictated by the available protocols such as RPL [7] routing network location of the “things”. An example of a
for IP-addressed “things”. In such a case, network layer is dedicated platform for commercial exploitation of various
extended to the whole mesh. The issues with IP addresses tagged commercial products is http://evythng.com that
remain as discussed in the previous paragraph, especially runs based on a digital identity maintained by the
as the actual address using in the RPL protocol is a cloud/platform for all the (physical) “things” it manages.
truncated-hostID part of the IPv6 address assigned at the
gateway. An alternative is to use ZigBee, that handles
routing and network layer connectivity to and from
gateways towards the mesh, in that case bridging to IP 1
Note: the stand-alone devices/gateways need support from the
from the gateway. A designer often finds that addressing is centralized location for address resolving if data is to be “pulled” from
not that important and that some codes can be adapted to them as web hosts.

274
Authorized licensed use limited to: UNIVERSITY OF SOUTHAMPTON. Downloaded on February 27,2024 at 14:58:09 UTC from IEEE Xplore. Restrictions apply.
 III. IOT SYSTEM COMPOSITION requiring IP hosting capabilities (e.g. via 6LoWPAN
and using CoAP at the application level), this might
A. IoT Architecture Components
require a mesh formation using RPL routing towards
This section contains deployment practicalities of the the gateway.
previously elaborated design considerations presented as Regarding the gateways, we generalize two main types of
the main architecture elements of an IoT system. The gateways (or stand-alone nodes as described in the
description is indented to reveal the operational previous sections):
convergence points of the architecture where 1. IoT Gateway: primarily establishes a communication
generalization can help in understanding how various IoT with the fixed network/Internet for the purpose of: i)
components and solutions can be integrated in a multi- exchanging data with the centralized database, ii)
purpose system as advocated in this paper. Generic acting as an autonomous web host for collective data
architecture elements are depicted in Figure 1. Starting gathered, i.e. able to support APIs from arbitrary
with sensor and actuator installations, we identify two locations, subject to IP address resolving or direct link
types of sensor and actuator hosting devices (i.e. “things”): establishment (e.g. with users’ smart phones). The
1. Gathering devices: hosts sensors or actuators with gateway might also perform data formatting and/or
small to arbitrary processing capabilities (e.g. a additional processing (e.g. metadata/ontological/
microcontroller processor) and IoT communication semantical formation or processing) mainly for the
links. Defining property is that the gathering devices purpose under ii). Requires medium to arbitrary
do not provide data in a web-presentable form nor they processing capability. Being an autonomous web host,
function as web hosts. They rely on the gateway to as one of the options for IoT gateway, it naturally
structure raw or user defined data formats and provide supports web protocols from the Internet via APIs such
visibility for the data at the web level and/or distribute as REST: HTTP and CoAP (e.g. CoAP nowadays runs
it to the centralized database (Note: although not from Firefox browsers). However, the gateway should
covered in detail in Figure 1 nor in our discussion, this support proxying of HTTP/CoAP protocols to any
type o device can generalize a tagging system that available protocol run inside the cluster/mesh
supplies the data to gateway or databases where it supported by the gateway (e.g. HTTP, CoAP, MQTT,
attains identify and web form). modifier protocols, user defined) and support the logic
2. Processing devices: Similar hardware and of the commands (e.g. pull/push analogy, delayed
communications as with the gathering node. It adds the reporting…). Note: for gateways and IP-capable
functionality of data formatting in pre-presentable devices, managing the hardware resources and mini-IP
form, e.g. using JSON or XML data formats. It can suite features becomes optimal when a dedicated IoT
also extend its functions to act a as a web host, operating system is installed for the purpose of multi-

Figure 1. Generalized Architecture Components of a multi-purpose IoT system

275
Authorized licensed use limited to: UNIVERSITY OF SOUTHAMPTON. Downloaded on February 27,2024 at 14:58:09 UTC from IEEE Xplore. Restrictions apply.
 threading of different operations of protocols structures and links between the data collected and assist
components, as opposed to a machine level software in formation of data stores (e.g. a data store in London,
(most popular choices between Contiki, TiniOS, RIOT, UK, provided by the city’s authorities
Linux, OpenWSN, FreeRTOS) http://data.london.gov.uk/).
2. IoT Decision Gateway: We separate this type of
gateway as a clone of the IoT gateway with extra IV. CONCLUSION
features of data processing and autonomous decision There is a diversity of issues encountered when various
making. This can be a more powerful processing unit IoT devices, technologies of IoT and platforms are
for scenarios where local decisions are performed for analyzed. The expansion of the current IoT applications
actuation processes or data processing. It provides a and services is largely driven by the immediate
distributed unit for cloud unravelling or for novel deployment opportunities rather than by a collective
functional paradigms of Fog computing in IoT convergence towards a single model for all of the diverse
networks. technical options. Multitude of applications and scenarios
makes this task almost utopian. An example is the
B. Discussion on Building a Centralized Platform
expansion of data and data representations tools that are
Building an IoT system with a multi-purpose objective yet to be coherently standardized and widely endorsed in a
or a platform for collection of data stemming from diverse form such as one Semantic Web for IoTs. Achieving this
IoT devices is a task usually performed for large scale would be a significant challenge since the growth of the
deployments by relevant worldwide companies and current Big Data generated by IoT devices and systems,
consortiums. Examples can include European although successful, is fragmented by the different
recommendation for IoT architecture done in the IoT-A providers, their data models and scattered technologies.
project http://www.iot-a.eu/public, FI-WIRE Future This paper provided a deployment-focused review of the
Internet European large-scale initiative including the design concerns and options encountered in constructing a
generic enablers of IoT systems compositions multi-purpose IoT system. This is done with an awareness
https://www.fiware.org/ and large scale data analytics, of the progress and global convergences in the relevant
repositories and device-data-web linkages by cloud- areas. We also highlight the importance of areas of
providers like Cisco, Microsoft, Amazon… security, privacy and trust in IoT systems as the topics that
Following the regional, university-driven strategy [6] would ultimately bolster the deployment and acceptance of
applied in this paper, the system is expected to comprise of IoT installations and provide easier interoperability
stand-alone IoT gateways and devices with ability to between the various existing systems.
almost directly provide API access (address resolving step
precedes this) without the need to fetch data via the ACKNOWLEDGMENT
centralized platform/database. The system will also Work is funded by European Union project Fore-Mont
accommodate the capability to perform data extractions as a part of Seventh Framework Programme (Grant
and APIs to the interfaces in the platform without direct Agreement No. 315970 FP7-REGPOT-CT-2013)
communication with the IoT gateways or devices. It is http://www.foremont.ac.me.
hence a working platform where extraction of data can
serve for various purposes, from research analytics to REFERENCES
feeding various applications. We make no strict [1] D. Giusto, A. Iera, G. Morabito, L. Atzori (Eds.), The Internet of
Things, Springer, ‘10
requirement on the type of platforms to be deployed, [2] L. Mainetti, L. Patrono, A. Vilei, “Evolution of wireless sensor
opening possibilities for evolution and integration in the networks towards the Internet of Things: A survey”, IEEE 19th
existing open cloud architectures. The main concern with International Conference on Software, Telecommunications and
Computer Networks (SoftCOM), Croatia, Sep. ‘11
the maintenance of the large and diverse volume of data is [3] L. Atzori, A. Iera, G. Morabito, “The Internet of Things: A survey”,
that it can be used for various purposes and users. This Computer Networks, Vol. 54, Is. 15, pp. 2787-2805, October 2011.
makes the process of database formations (structured or [4] M.R. Palattella, N. Accettura, X. Vilajosana; T. Watteyne, L.A.
Grieco, G. Boggia, M. Dohler, "Standardized Protocol Stack for the
unstructured, different data formats) rather unspecific as it
Internet of (Important) Things," Communications Surveys &
is often not immediately visible what are the linkage Tutorials, IEEE , vol.15, no.3, pp.1389,1406, Third Quarter ‘13.
between data and to whom or what would the data be [5] "Gartner Says the Internet of Things Installed Base Will Grow to 26
extracted and provided. The task of computationally Billion Units By 2020". Gartner, 2013-12-12,
http://www.gartner.com/newsroom/id/2636073, Retrieved 2.1.2014.
efficient storage and extractions becomes more [6] A. Mihailovic, M. Simeunoviü, N. Lekic, M. Pejanoviü-Djurišiü, "A
challenging as the volume of data expands and also their strategy for deploying diverse sensor-based networks as an
combinations via the use of tools such as ontology and evolution towards integrated Internet of Things and Future
Internet," Telecommunications Forum Telfor (TELFOR), 2014
semantics (i.e. Big Data problem analogy). Solutions such 22nd, Nov. 2014, Belgrade.
as data stores are an interesting optimal solution for [7] P. Thubert, et al,”RPL: IPv6 Routing Protocol for Low-Power and
overcoming this problem as they can collect various Lossy Networks”, RFC 6550, IETF,March 2012
[8] N. Kushalnagar, G. Montenegro, C. Schumacher, “IPv6 Over Low-
statistical and also IoT data and can provide information Power Wireless Personal Area Networks (6LoWPANs): Overview,
and APIs for various types of users. In addition, Assumptions, Problem Statement & Goals”, IETF RFC 4919, 2007
interactions between engineers and data users (e.g. [9] C. Bormann, A. P. Castellani, and Z. Shelby, “CoAP: An
Application Protocol for Billions of Tiny Internet Nodes”, IEEE
researchers using data from agricultural installations of Internet Computing, vol. 16, no. 2, pp. 62–67, 2012.
IoT, smart city traffic analysts…) can understand the

276
Authorized licensed use limited to: UNIVERSITY OF SOUTHAMPTON. Downloaded on February 27,2024 at 14:58:09 UTC from IEEE Xplore. Restrictions apply.