Bringing Dynamics to IoT Services with Cloud and Semantic

Technologies

An Innovative Approach for Enhancing IoT based Services

Sébastien Dupont, Amel Achour, Fabrice Estiévenart, Laurent Deru and Nikolaos Matskanis

Centre of Excellence in Information & Communication Technologies (CETIC),

Avenue Jean Mermoz 28, Charleroi, Belgium

Keywords: IoT, Border Routers, Semantic Technologies, Dynamic Deployment, Cloud Services.

Abstract: In this paper, we present an innovative software architecture that brings dynamics to the world of

interconnected small devices and sensors by mixing cloud services, semantics and border router

technologies. Dynamic aspects can be enabled both in the way that the devices are deployed or managed as

well as in the manner in which the data can be combined or interpreted to form additional services. We got

inspired from the architecture of mediators and wrappers in databases and services systems and adapt them

to the IoT world. We illustrate our purposes with a use case scenario that involves different actors from the

energy and smart cities domains.

1 INTRODUCTION

Smart interconnected devices and sensors –

commonly described by the term “Internet of

Things” (IoT) - are increasingly becoming part of

our everyday lives. Environmental sensors, domestic

appliances, cars, wearables and infrastructure control

devices are more and more interconnected,

autonomous and able to communicate with other

peers or services over the Internet. The automatic

deployment, uniform access and availability of

services and information from these IoT devices

introduces significant applications that help people

to achieve their goals, factories to improve their

processes and products, and provide new ways for a

society to improve its quality of life. In this paper we

present a novel architecture that provides i) flexible

and dynamic services, ii) open and easily available

resources, iii) dynamic behaviour of IoT devices and

iv) semantically rich data and services.

The distributed dynamic services, which this

architecture proposes, take into account data from

the observed environment, open public data and

processed data offered by available services. All this

information is composed to produce results relevant

to the use case, for example offer guidance to users

in a smart cities scenario, optimise manufacturing

processes at chemical plants, assist and improve the

quality of the production of an agricultural farm. In

order to consider a cross domain application

scenario that includes domain interoperability issues

we will examine the use case of the smart energy

management of a city. In this use case the actors that

are involved are the energy providers, the energy

producers, and the smart city citizens - the energy

consumers. The data sources in this scenario include

the energy plant sensor data, the power distribution

facilities sensors with data on the power

consumption at local level and the user appliance

devices that provide usage and scheduled

consumption data. Finally environmental data

sensors such as weather stations together with socio-

economic data analysis from social media and news

feeds bring together all the environmental variables

that can influence and improve the power

distribution routing in the grid.

The architecture that we present in this paper a)

uses IoT Border routers, which provide seamless

connectivity to IoT devices; b) includes cloud

services that provide IoT management and data

services; c) adds semantics to the data produced by

the devices or to the metadata of the devices

themselves.

The border router was designed to act as a

gateway that connects the wireless sensors network

to the internet. This interconnexion allows a

seamless exchange of data. On one hand, sensors

send data when it is available and, on the other hand,

Dupont, S., Achour, A., Estiévenart, F., Deru, L. and Matskanis, N.

Bringing Dynamics to IoT Services with Cloud and Semantic Technologies - An Innovative Approach for Enhancing IoT based Services.

DOI: 10.5220/0005933001850190

In Proceedings of the International Conference on Internet of Things and Big Data (IoTBD 2016), pages 185-190

ISBN: 978-989-758-183-0

185

they can be managed remotely, by receiving updates

and commands.

Connected devices produce lots of data that

needs to be stored and analysed, cloud technologies

enable scalability of both storage and computing

power. The devices and services are monitored and

managed in a centralized way, and benefit from

automatic deployment and continuous integration.

It is interesting to explore what can be achieved

if we provide the meaning of the data together with

the data itself from an IoT device via the Border

Routers and through the cloud services. The receiver

of this information can interpret its format and

meaning and utilize the data in a straightforward and

general fashion. Going back to our use case, this

data can be environmental information produced by

meteorological stations or power usage data coming

from home appliances. By being accompanied by

semantic descriptions, sensor data can be published

or made available in many ways that would allow

discoverability, inference of relationships with other

data and allow predicting spikes and other events by

applying feedback rules. Additionally, this

semantically enriched sensor data can be analysed

for discoverying trends in power usage. This could

fullfil the specific needs of citizens of smart cities,

especially when combined with data analytic

techniques from various socio-economic feeds.

The benefit of combining the dynamic behaviour

of border routers with the storage and computation

flexibility of cloud services and providing

knowledge about the data is that the receiver of the

data does not need to have knowledge of the node

capabilities or type of data collected, but can query

and discover this information and dynamically use

computation, inference and composition of data

services to receive new information and services.

The rest of this article is organized as follows:

Section 2 presents related technologies in the areas

of IoT, Cloud Services and Semantic technologies.

In Section 3 we present our suggested solution of a

platform architecture that consists of 4

implementation layers based on the aforementioned

technologies and finally we provide a conclusion

and our implementation plans in Section 4.

2 RELATED WORK

We have identified three areas of scientific research

that are relevant to our architecture. These are the

Border router devices at the hardware and

communication protocols level, the cloud services at

infrastructure level and the semantic interoperability

services at the application level.

2.1 IoT Border Routers

A wireless sensor network (WSN) is a network of

spatially distributed autonomous sensor devices to

monitor physical or environmental conditions

(Madhav, B., et al

., 2012). These devices have

wireless communication capabilities and

autonomously form networks through which sensor

data is transported. The sensors are constrained

devices with restricted capabilities, for this reason,

algorithms and protocols need to address the

following issues: increased lifespan, robustness and

self-configuration. Such operating systems for

wireless sensor network nodes increasingly resemble

embedded systems. The reason behind this trend is

the need for low cost and low power. This implies

that most wireless sensor nodes use low-power

microcontrollers, and mechanisms such as virtual

memory are either unnecessary or too expensive to

implement

(Navjot Kaur, J., et al., 2015).

Contiki (Dunkels, A., et al, 2004) is a wireless

sensor network operating system that consists of the

kernel, libraries, the program loader and a set of

processes. It is designed to deal with the sensors

limitations mentioned before. Contiki provide

mechanisms that assist in programming smart

objects (or sensors) applications. It provides libraries

for memory allocation, linked list manipulation and

communication abstractions. Contrary to existing

systems such as TinyOS (Hill, J., et al, 2000),

Contiki provides a dynamic structure allowing the

replacement of programs and drivers during run-

time and without relinking.

The biggest advantage of IP based Wireless

Sensor Networks is their ability to be seamlessly

connected to the Internet. A device, which passes

data from the lowpan to ordinary network is called a

"Border Router". Such a device must support at least

two network interfaces: 802.15.4 for the lowpan, and

Ethernet (IEEE 802.3) or WiFi (IEEE 802.11) for

the uplink. Systems based on proprietary or non

standards stacks (e.g. the ZigBee - see

http://www.zigbee.org

- and the zWave - see

http://www.z-wave.com

) require to have a Gateway

at the edge between the lowpan network and the

Internet network. The gateway can also have a Data

storage mechanism if necessary. Any modification

in the format of the data requires an update of the

Gateway, thus hindering the modification or the

development of the lowpan network with new

external elements. Currently, the trend is to replace

these stacks with IP compliant stacks and open data

IoTBD 2016 - International Conference on Internet of Things and Big Data

186

format, for example ZigBee IP is a complete switch

from the proprietary ZigBee stack towards a

IPv6/RPL stack.

CETIC has developed an open-source IPv6 based

Border Router (Deru, L., et al, 2013), build on top of

Contiki, that allow the seamless integration of an

IPv6/RPL based lowpan to an Internet infrastructure.

2.2 IoT and Cloud Services

Cloud platforms provide fine-grained control on the

resources available in the target cloud via the

Infrastructure as a Service (IaaS) layer. The

Openstack

(see http://www.openstack.org/) project

implements most of the backend requirements for

the proposed architecture: object storage and

processing scalability through its Swift and Nova

modules, time series database with Gnocchi, data

processing in Sahara, and others.

The Platform as a Service (PaaS) layer enables

management of the services and devices. Continuous

delivery practices ensure the systems are always up

to date, and the various components are

automatically deployed using deployment (e.g.

Openstack Heat, HashiCorp Terraform, Amazon

CloudFormation) and provisioning (e.g. Ansible,

Chef, Puppet) tools.

The Open Mobile Alliance proposes a

standardized set of machine-to-machine (M2M)

tools

(see http://openmobilealliance.org/about-

oma/work-program/m2m-enablers/) to manage IoT

devices (e.g. border routers), notably the LWM2M

protocol.

The OpenIoT project (see http://www.openiot

.eu)

implements an open source middleware

framework for self-managed environments of IoT

applications, and uses a cloud utility-based model.

The outcomes of this project are being considered in

our solution.

2.3 IoT and Semantic Technologies

With their machine-interpretable representation of

information and unambiguous reasoning principles,

Semantic technologies have a great role to play in

the world of IoT. Many semantic-related projects

and languages have recently emerged in order to

cope with the biggest IoT challenges like service

discovery in a large and dynamic environment,

scalability, security or privacy.

Developed by the W3C Semantic Sensor

Networks Incubator Group (Kolozali S., et al.,

2014), the SSN ontology (see https://www.w3.or

g/2005/Incubator/ssn/ssnx/ssn)

describes sensor

resources, observations and their related concepts

such as the measurements, timestamps and locations.

SSN also includes a lightweight quality model

describing the quality aspects of collected data.

(Xiang, S., et al., 2014) have conducted a survey

on formal knowledge representations in various

formats that can be potential candidates for

representing sensor data. Their aim was to evaluate

the resource usage of different alternative formats in

a sensor system. Their experiments have shown that

the choice of the format influences the packet size

and processing cycles needed for embedding the

semantic information into the sensor message and

often format. Both of these aspects have significant

impact on the sensor’s resource consumption.

The backbone technologies of the Semantic Web

are RDF, OWL and SPARQL. Those standards have

been adapted to the specificity of an IoT

environment. stSPARQL and stRDF (Koubarakis

M., et al., 2012) are built on top of SPARQL and

RDF. They extend the basic concepts with spatial

and temporal dimensions in order to facilitate the

representation and the querying of sensor data.

C-SPARQL (Barbieri, D., et al., 2010) has been

developed to perform real-time and continuous

queries over streaming data.

Many XML-based languages have emerged from

various R&D projects. Product Markup Language

(PML) describes physical devices in terms of

Electronic Product Code Networks (EPCN) but does

not allow reasoning in its basic form. There is a need

to develop languages on top of PML in order to

enable reasoning based on description logic.

SensorML proposes a standard model to describe

sensor systems and processes but also lacks the

reasoning and annotation capabilities needed for

automated service discovery.

Sensor Observations Service (SOS) is a service-

oriented approach that defines an interface for

requesting, filtering, and retrieving data from

sensors but no implementation is currently available.

Recently, a promising standard, named HyperCat

(Rodger, L., et al., 2013), has emerged for services

interoperability and discoverability within IoT

networks. It is based on RESTful Web standards like

HTTPS and JSON and it will allow any developer to

provide applications that work across multiple

servers.

3 THE ARCHITECTURE

A high level view of the architecture is provided in

the following diagram. The architecture of the

Bringing Dynamics to IoT Services with Cloud and Semantic Technologies - An Innovative Approach for Enhancing IoT based Services

187

platform we propose is composed of the following

layers: the border routers layer that interfaces with

the sensors and IoT devices is described in further

detail in section 3.1; the cloud services layer that

provides data storage and device management

capabilities, which is described in section 3.2; the

semantic layer that provides the data model and

formal description of resources, which is detailed in

section 3.3; and last but not least the high-level and

application level services that are described in

section 3.4.

3.1 The Border Routers Layer.

Sensor networks are gradually moving towards full-

IPv6 architecture and play an important role in the

upcoming Internet of Things. These smart objects

will be integrated into existing network

infrastructure using a Low Power WPAN IEEE

802.15.4 (Gutierrez, J., et al., 2001) link layer. This

technology has the advantage of providing efficient

low-bitrate network connectivity at a minimal cost.

It can also be used with both networks stacks,

Zigbee and IPv6. In order to run IPv6 over

IEEE.802.15.4, an adaptation layer is necessary to

provide header compression mechanism, link

specific fragmentation and reassembly. The

6LowPAN (IPv6 over Low power Wireless Personal

Area Networks) (Kushalnagar, N., et al, 2007) is

designed to provide such a service.

Figure 1: Diagram of the proposed Architecture.

The border router (BR) acts as a gateway to

interconnect the IoT network to the Internet. It is

designed to forward packets between IPv6 and

6LoWPAN networks, to configure the IP-nodes and

to provide context sharing and multi-hop routing

(RPL). The BR has two interfaces, the 802.15.4 one

on the WSN side, and Ethernet or WiFi on the other

one. Most of the current deployments have a unique

BR, which represents a critical point of failure. In

our architecture we consider the 6LBR

implementation (Deru, L., et al, 2013). This solution

proposes a redundant BR deployment to enhance

fault tolerance and sensors mobility without

compromising the energy-efficient control

mechanism provided by the routing protocol. This

implementation supports the RPL (Routing Protocol

for Low Power and Lossy Networks) to route

packets using the route over approach (Winter, T., et

al, 2012). Then, packets are routed on the IP level

with a low networking overhead. Going back to our

smart energy use case. Data from both energy

producers/providers and consumers is collected

thanks to sensors and sent to the cloud through the

border routers. With this inter-connexion between

sensor networks and the rest of the IP world, we

enable dynamic access to the Data.

3.2 The Cloud Services Layer

Data Storage

The various sensors produce large amounts of

information in time series format, i.e. arrays of

measurements indexed by timestamp. This data is

stored in a time series database that can be

efficiently queried to obtain aggregated information

on specific geographic regions, periods of time,

measurement type, etc.

The time series database scales horizontally by

allowing the cloud infrastructure to add or remove

computing and storage resources to the database as

required.

To optimize resources usage, the cloud services

layer also provides long term archiving capabilities

for the data that does not need to be accessed on a

regular basis.

Device Management

This component provides the various functionalities

needed to manage the IoT devices connected to the

cloud. Each device of the fleet can be automatically

provisioned, configured and updated. The device

management provides a control panel for the users to

view the status of the device fleet and manage it, e.g.

add or remove a device, update a device, view

device status and query sensor.

IoTBD 2016 - International Conference on Internet of Things and Big Data

188

Service Management

Like the devices, the services described in this

architecture are automatically provisioned with the

help of composite cloud templates; they are

managed through a control interface that exposes

monitoring information and configuration settings.

3.3 The Data Semantics Layer

The Semantic technologies and data semantics level

is composed of the ontology repository, the semantic

mapping services and the ontology that formally

describes the resources used and the data provided

by the border routers.

As demonstrated in (Rodger, L., et al., 2013) and

in (Xiang, S., et al., 2014), adding semantics to thin

and light sensors and IoT devices can be achieved by

choosing more compact representations of the

descriptions together with appropriate encoding of

the messages that are sent by the devices. It

is interesting what can be achieved by formally

describing the IoT devices and by providing context

to the data that the devices are producing.

Knowledge representations of both the devices and

the data allow logical reasoning that is able to infer

new information from existing assertions. In

addition to this approach, in our solution we can add

semantic information at the border router that is

either inferred or known at deployment time because

of the localization of the devices controlled via the

border router. This knowledge of the type of devices

and interlinking or associations between the data that

they produced can be made available dynamically at

the services one level above, the cloud services

layer, if this knowledge has not been provided

already.

The Ontology Repository (Chondrogiannis, E.,

2011) manages resources and stores information

about their attributes and metadata based on a pre-

defined data model. The data model can be designed

and built using either an XSD schema description or

an OWL-RDF ontology model. The repository

provides a RESTful storage API that acts as a

backend for WebUI components. It also provides an

API for enabling queries - that can be relatively

complex - on the resources in order to assist

discovery and exchange of information about the

stored resources. The discovered resources are

cached at the application level for performance

purposes. The Ontology Repository core and its

standard REST API are domain independent, and

can easily be extended in order to add or customise

its functionalities according to the use case

requirements.

The mapper component is providing mappings

between the raw data schema, which is coming from

the sensors and stored in the cloud, and the resource

ontology in the repository using a mapping file. The

ontology repository contains the formal description

of the device resource and the description of the

data. The streaming data from the device as well as

the details of its schema or format is only available

at the cloud service layer. The mapping enables

querying the data with semantic technologies such as

the SPARQL language using SPARQL endpoints

and ensures seamless integration of raw data with

the devices’ data model in the ontology (Matskanis,

N., et al., 2015).

This three-way approach of inserting semantics

to the IoT devices and device data enhances

discovery and inference of information that can be

used to create more dynamic and flexible services

over the IoT infrastructure.

3.4 Data as a Service

We suggest to tackle the challenge of data

integration in IoT by developing a service-oriented

approach based on a "Data as a Service"

architecture. That solution has many potential

benefits: ease of query (heterogeneous data sources

and sensors expose the same interface), support for

any sensor hardware (the generic service

specification is not bound to specific hardware

features), performance (data can be cached at any

node or any level of the architecture), integration

with any web data source such as web services or

open data repositories.

In this type of architecture each data source (file,

database, sensor data) is wrapped by a standardized

Web Services API (RESTful), which provides an

easy and standard way to access the resource by

using the standard HTTP methods. Data as a Service

consists in making available any data source, i.e.

databases, log files, media collection through a

standardized interface such as REST. This kind of

architecture offers many advantages in an IoT

configuration such as flexibility, the possibility for

dynamic services composition or proven security

methods (e.g. HTTPS).

Figure 1 depicts our Data as a Service

architecture in an IoT context. Each sensor

continuously generates data; that data is stored

inside log files or in structured databases (e.g. file-

based and lightweight database systems like

SQLite). Some types of sensors provide their data

through a Web API but, for any other sensors, a

Bringing Dynamics to IoT Services with Cloud and Semantic Technologies - An Innovative Approach for Enhancing IoT based Services

189

mapper component is needed to wrap the data within

a RESTful access service.

Standard API or languages like SensorML or

PML are used to provide a unique and transparent

way to read or write information from the sensor

network. Finally, a service repository serves as a

central database for any IoT application that wants

to use data coming from sensors as a data source.

IoT applications can use basic services providing

unilateral raw data or high-level and cloud-services

that combine different types of data, working like

mash-up services.

4 CONCLUSIONS

This paper proposed an innovative architecture for

bringing dynamics to IoT networks. Border routers,

cloud services and semantics are the technological

building blocks composing that architecture. We

believe that we can take advantage of those concepts

and build flexible, reliable and secure IoT

applications in various domains such as energy,

agriculture or weather forecast.

Border routers have a real-time knowledge of the

status of the network, the localisation of the

sensors/devices and their deployment status. Thus they

serve as an endpoint to any external component that

will make use of the sensors. We have also shown that,

with their self-expressiveness and their formal

description, Semantic Web languages/standards are

well suited to describe the data provided by IoT

devices. Additionally cloud technologies provide

solutions that can otherwise be challenging to the IoT

networks such as storage and compute scaling, as well

as IoT device and service management.

An architecture that provides semantically rich

data and services, combines the meaning of data of

the IoT devices with the additional context provided

by the Border Router as well as enables dynamic

discovery and management of the devices, we

believe that it can enable dynamic composition of

data and services, dynamic response to conditions

and assist implementation of new client services,

business models for services providers in energy and

in other use cases.

ACKNOWLEDGEMENTS

The work presented in this paper has been partially

funded by the Walloon Region project "Plateforme

BigData" (PIT Hors pôles, grant no. 7481).

REFERENCES

Madhav, B., et al., 2012, Wireless Sensor Network: A

Promising Approach for Distributed Sensing Tasks,

Excel Journal of Engineering Technology and

Management Science.vol.1, no. 1.

Navjot Kaur, J., et al., 2015, Comparative Study of Tree

Based Routing Protocols for WSNs, International

Journal of Advanced Research in Computer Science

and Software Engineering (ijarcsse), vol. 5, issue. 6.

Kolozali S., et al., 2014, A Validation Tool for the W3C

SSN Ontology Based Sensory Semantic Knowledge,

Centre for Communication Systems Research (CCSR),

University of Surrey, Guildford, United Kingdom.

Koubarakis M., et al., 2012, Introduction in stRDF and

stSPARQL,

Barbieri, D., et al., 2010, Querying RDF Streams with C-

SPARQL.

Rodger, L., et al., 2013, HyperCat:an IoT interoperability

specification, IoT ecosystem demonstrator

interoperability working group.

Russomanno, DJ, et al., 2005 Building a sensor ontology:

a practical approach leveraging ISO and OGC

models. Proceedings of the 2005 International

Conference on Artificial Intelligence, Las Vegas,

USA; 637–643.

Dunkels, A., et al, 2004, Contiki - a lightweight and

flexible operating system for tiny networked sensors,

in Local Computer Networks, 2004. 29th Annual

IEEE International Conference on, pp. 455 – 462.

Hill, J., et al, 2000, System architecture directions for n et

etworked sensors, InProc. ASPLOS-IX.

Deru, L., et al, 2013 Redundant Border Routers for

Mission-Critical 6LoWPAN Networks, in Proceedings

of the Seventh Workshop on Real-World Wireless

Sensor Networks.

Kushalnagar, N., et al, 2007 “ IPv6 over Low-Power

Wireless Personal Area Networks (6LoWPANs):

Overview, Assumptions, Problem Statement, and

Goals”, RFC 4919.

Winter, T., et al, 2012, RPL: IPv6 Routing Protocol for

Low-Power and Lossy Networks, Internet Engineering

Task Force, RFC 6550, Available:

http://tools.ietf.org/html/rfc6550.

Xiang, S., et al., 2014, Adding semantics to internet of

things, Wiley Online Library

(wileyonlinelibrary.com). DOI: 10.1002/cpe.3203.

Gutierrez, J., et al., 2001, IEEE 802.15.4: A developing

standard for low-power lowcost wireless personal

area networks, IEEE Network Magazine, vol. 15, no.

5, pp. 12–19.

Chondrogiannis, E., Matskanis, N., et al., 2011, Enabling

semantic interlinking of medical data sources and

EHRs for clinical research purposes, eChallenges

conference.

Matskanis, N., Mouton, S., Ebel, A., Marchiori, F., 2015,

Using Semantic Technologies for more Intelligent

Steel Manufacturing, KEOD2015 conference.

IoTBD 2016 - International Conference on Internet of Things and Big Data

190