Internet of Things: Concept, Building blocks, Applications ... · Internet of Things: Concept, Building blocks, Applications and Challenges Riad Abdmeziem, Djamel Tandjaoui January

Internet of Things: Concept, Building blocks, Applications and

Challenges

Riad Abdmeziem, Djamel Tandjaoui

January 28, 2014

Keywords: Internet Of Things, RFID, WSN,Smart Environments, Pervasive computing.

Abstract

Internet of things (IoT) constitutes one of the mostimportant technology that has the potential to affectdeeply our way of life, after mobile phones and Inter-net. The basic idea is that every objet that is aroundus will be part of the network (Internet), interactingto reach a common goal. In another word, the Inter-net of Things concept aims to link the physical worldto the digital one. Technology advances along withpopular demand will foster the wide spread deploye-ment of IoT’s services, it would radically transformour corporations, communities, and personal spheres.In this survey, we aim to provide the reader with abroad overview of the Internet of things concept, itsbuilding blocks, its applications along with its chal-lenges.

1 Introduction

During the past few years, the acess to the Internethas evolved from static (desktop) access to a moremobile and dynamic one, using several devices - suchas Mobile phones,Tablets, Televisions, etc. In thiscontext a novel paradigm named Internet Of Things(IoT) is rapidly gaining ground. The basic idea isthat every objet that is around us will be part of thenetwork (Internet), interacting to reach a commongoal, in another word, the Internet Of thing conceptaim to link the physical world to the digital one.

The pervasive presence around us of various wire-less technologies - such as Radio-Frequency IDen-tification (RFID) tags, sensors, actuators, mobilephones, etc. in which computing and communicationsystems are seamlessly embedded will form the build-ing block of the IoT concept [9]. The IoT’s full de-ployement will give rise to new opportunities for theInformation and Communication Technologies (ICT)sector, paving the way to new applications, providingnew ways of working; new ways of interacting; newways of entertainment; new ways of living [2].

Technology advances along with popular demandwill foster the wide spread deployement of IoT’s ser-vices, it would radically transform our corporations,communities, and personal spheres. From the per-spective of a private user, IoT’s introduction will playa leading role in several services in both workingand domestic fields -such as Domotics, e-health, e-learning, security and surveillance, etc. In the samemanner, from the business point of view, IoT willbring a deep change in the way automation, indus-trial manufacturing, logistics, business/process man-agement and transportation of goods and people arehandleled.

Implementation of IoT paradigm rely on the inte-gration of RFID systems, Wireless Sensor Networks,intelligent technologies (using knowledge to solve cer-tain problems and mainly covering Artificial Intelli-gence [10]) and nanometer technologie (concentrat-ing on the characteristic and application of materi-als of size between 0.1 and 100 nm). Up to this

1

arX

iv:1

401.

6877

v1 [

cs.C

Y]

2 J

an 2

014

day, since research of IoT is still embryonic thereexist no common IoT architecture. Nowadays theElectronic Product Code (EPC) network architec-ture supported by EPCglobal [1] together with theUnique/Universal/Ubiquitous IDentifier (UID) archi-tecture in Japan [16] are the most representativeamong others. The main idea underlying EPCglobalnetwork is to use RFID and wireless technologies towrap every day’s live objects and connect them to thetraditional Internet, while, UID provides middlewarebased solutions for a global visibility of objects.

Several challenges stand between the conceptualidea of IoT and its full deployement into our dailylife. Main issues are : making complete interoper-ability of heteregenous interconnected devices whichrequire adaptation and autonoumous behaviour whileguaranteeing trust, privacy, and security; networkingaspect is not in rest, low computation and energycapacities that characterized the things of the IoTbring ressource effeciency as a fundamental elementin the proposed solutions. Around the globe, severalindustrial, standardization and research bodies arecurrently involved in the devloppement of solutionsin order to bring answers to the highlighted techno-logical requirements.

In this survey, we aim to provide to the reader anoverview of the IoT concept, the different enablingtechnologies, research challenges and the implicationsof a wide spread diffusion of IoT. The remainder ofthis paper is organized as follows: in Section 2, def-initions of IoT from various perspectives are intro-duced. Section 3 introduces the main IoT enablingtechnologies. The applications of IoT already avail-able are summarized in Section 4. Section 5 statesthe research challenges. Finally Section 6 gives theconclusion.

2 Definition and vision

In Internet of things (IoT), huge number of small de-vices will be connected to the Internet in some way.IoT’s definition is usually studied through variousperspectives. According to [10], the IoT paradigm

Figure 1: Internet of Things paradigm as a result ofthe convergence of different visions

shall be the result of the convergence of three mainvisions: internet-oriented (middleware), things ori-ented (sensors) and semantic-oriented (knowledge) asshown in Figure 1. Perspective of Things: Thisperspective focuses on how to integrate generic ”ob-jects” or ”Things” into a common framework, and the”Things” under investigation are RFID tags. RFIDis considered as a one of the leading technologies [12],mainly due to its maturity and low cost, and conse-quently its strong support from the business com-munity. Nevertheless, IoT is more than a globalEPC system where the only objects are RFID tags.Besides that,United Nations (UN) has also statedthat the perspective of ”Things” of IoT goes beyondRFID. It is stated in a UN report that a new era ofubiquity is coming where the users of the Internetwill be counted in billions, and where humans maybecome the minority as generators and receivers oftraffic. Changes brought about by Internet will bedwarfed by those prompted by the networking of ev-eryday objects.

2

The consortium CASAGRAS 1 also proposes anIoT vision statement that goes well beyond a mere”RFID centric” approach. Its members focus on aworld where things can automatically communicateto computers and each other, providing services tothe benefit of the human kind. It not only proposesIoT would connect both virtual and physical genericobjects as a global infrastructure, but also empha-sizes the importance of incorporating the traditionalInternet related technologies and infrastructures inthe development of IoT. Similarly, other relevant in-stitutions have stressed the concept that IoT has pri-marily focused on the ”Things” and that the road toits full deployment has to start from the augmenta-tion in the Things intelligence. From the thing’s per-spective, the International Telecommunication Union(ITU) has given the following definition of the IoT:from anytime, any place connectivity for anyone, wewill add a connectivity for anything . The same def-inition is given by the European Commission, it re-lates to : things having identities and virtual per-sonalities operating in smart spaces using intelligentinterfaces to connect and communicate within social,environmental, and user contexts [3].

Perspective of Internet: “A world where thingscan automatically communicate to computers andeach other providing services to the benefit of the hu-man kind”, this vision is brought by the CASAGRASconsortium, a vision of IoT as a global infrastructurewhich connects both virtual and physical generic ob-jects and highlights the importance of including exist-ing and evolving Internet and network developmentsin it. In this sense, IoT becomes the natural enablingarchitecture for the deployment of independent fed-erated services and applications, characterized by ahigh degree of autonomous data capture, event trans-fer, network connectivity and interoperability. Thisdefinition leads to the Internet oriented vision of IoT,while the perspective of things focuses on integrat-ing generic objects into a commun framework, theperspective of ”Internet” pushes towards a networkoriented definition.

1Coordination And Support Action for Global RFID-related Activities and Standardisation

Within the latter category falls the IoT vision ofthe IPSO(IP for Smart Objects) Alliance [8], a forumformed in September 2008 by 25 founding compa-nies to promote the Internet Protocol as the networktechnology for connecting Smart Objects around theworld. This vision favors the Internet protocols asthe network technology for connecting smart objectsaround the world. According to IPSO, the IP stackis a lightweight protocol that already connects a hugeamount of communicating devices and runs on tinyand battery operated embedded devices. Reducingthe complexity of the IP stack in order to design aprotocol to route IP over things is the promoted idea.

Furthermore, as the Internet is running out of ad-dresses, in the near future it will be moving to anew protocol, IPv6. The current system, IPv4, hasroughly four billion addresses. The new address spacecan support (about 3.41038) addresses, which means,to take a commonly used analogy, that it providesenough addresses for every grain of sand on everybeach in the world! While it is unlikely that we willbe assigning IP addresses to grains of sand, the idea ofassigning them to each of the more or less 5,000 dailyobjects that surround us, is quite appealing. Withthe right technology in each object (e.g., an RFIDtag) and the right network in the surroundings, itwill become easy to locate and catalogue items in afew seconds and to reap the benefits of the vast ar-ray of new information that communications amongthem will provide. IPv6 is undoubtedly one of thesteps to making the Internet of Things a reality[6].

Reducing the complexity of the IP stack and in-corporating IEEE 802.15.4 into the IP architecture islooked as the wisest way to to move from the Inter-net of Devices to the Internet of Things. According tothe IPSO and 6LoWPAN 2, the IoT will be deployedby means of a sort of simplification of the currentIP to adapt it to any object and make those objectsaddressable and reachable from any location.

Perspective of Semantics: The basic idea behindis that the number of items involved in the futur In-

2IPV6 Low Power Wireless Personnal Area Network

3

ternet Of Things is likely to become very high, thus,issues related to how to represent, store, intercon-nect, search, and organize information generated bythe IoT will become very challenging.

Such development will also necessarily create de-mand for a much wider integration with various ex-ternal resources, such as data storages, informationservices, and algorithms, which can be found in otherunits of the same organization, in other organiza-tions, or on the Internet. Therefore, issues of rep-resenting, storing, searching, and interconnecting in-formation generated in IoT will become very chal-lenging.

In this context, semantic technologies could play akey role. In fact, these can exploit appropriate mod-eling solutions for things description, reasoning overdata generated by IoT, semantic execution environ-ments and architectures that accommodate IoT re-quirements [9].

3 IoT elements

The Internet of things is unlikely to rise as a brandnew class of systems. An incremental develope-ment path, along which IoT technologies will beprogressivly employed to extend existing ICT sys-tems/applications, providing additionnal functionali-ties related to the ability of interacting with the phys-ical realm. This section focuses on the enabling tech-nologies that are expected to form the building blocksof the IoT, each technology is briefly presented withits supposed impact on the IoT’s devloppement. Fig-ure 2 gives an overview of the main technologies thatwill be involved in the future IoT.

3.1 Sensing, computing and identifi-cation technologies

The ability of sensing the environement and to self-organize into ad hoc networks represent an impor-tant feature from the IoT perspective. Nevertheless,some limiting factors stands in front of a widespreadadoption: Energy management of the futur embed-ded devices is a crucial issue, in order to get minimal

Figure 2: IoT Elements

computationnal capablilties sensor nodes will have tobe equipped with a battery. While a number of solu-tions for increasing energy efficiency at various layersof the OSI model has been devised, the need to re-place batteries from time to time represents a hugebarrier to the widespread development of IoT. Be-sides, nodes in a classical wireless sensor network areexpected to possess a set of common characteristics,and to share a number of common features includinga full protocol stack. While advances in embeddedelectronics and software are making such a require-ment less and less stringent, solutions able to accom-modate heterogeneity in terms of supported featuresshould be introduced to ease incremental deployment[2].

Wireless technologies will play a key role in thefutur IoT, in a way where the major part of datatraffic between objects will be carried in a wirelessway. Otherwise the reduction in terms of size, weight,energy consumption, and cost of the radio can leadus to a new era where radios could be integrated inalmost all objects and thus, add the world anything

4

Figure 3: RFID tag and reader

to the Anytime, anywhere, anymedia vision.

Wireless Sensor Networks (WSN) and radio-frequency identification (RFID) are considered as thetwo building blocks of the sensing and communicationtechnologies in the futur IoT:

3.1.1 RFID

RFID technology is a major breakthrough in the em-bedded communication paradigm which enables de-sign of microchips for wireless data communication.RFID tags are expected to play a key role as enablingidentification technology in IoT. They help in au-tomatic identification of anything they are attachedto acting as an electronic barcode. From a physicalpoint of view, as shown in Figure 3 ,a RFID tag is asmall microchip attached to an antenna (that is usedfor both receiving the reader signal and transmittingthe tag ID) in a package which usually is similar toan adhesive sticker. Dimensions can be very low: Hi-tachi has developed a tag with dimensions 0.4 mm x0.4 mm x 0.15 mm.

The passive RFID tags are not battery poweredand they use the power of the readers interrogationsignal to communicate the ID to the RFID reader.This has resulted in many applications particularlyin retail and supply chain management. The appli-cations can be found in transportation (replacementof tickets, registration stickers) and access control ap-plications as well. The passive tags are currently be-ing used in many bank cards and road toll tags whichis among the first global deployments. Active RFID

Figure 4: Wireless Sensor Network

readers have their own battery supply and can in-stantiate the communication. Obviously the radiocoverage is the highest for active tags, though, this isachieved at the expenses of higher production costs. Of the several applications, the main application ofactive RFID tags is in port containers for monitoringcargo [9].

Sensor networks will also play a crucial role in theIoT. In fact, they can cooperate with RFID systemsto better track the status of things, i.e., their loca-tion, temperature, movements, etc. As such, theycan augment the awareness of a certain environmentand, thus, act as a further bridge between physicaland digital world.

3.1.2 WSN

Sensor networks consist of a certain number, whichcan be very high, of sensing nodes communicating ina wireless multi-hop fashion [7] as shown in Figure 4.Usually nodes report the results of their sensing toa small number (in most cases, only one) of specialnodes called sinks. A large scientific literature hasbeen produced on sensor networks in the recent past,addressing several problems at all layers of the proto-col stack. Design objectives of the proposed solutionsare energy efficiency (which is the scarcest resourcein most of the scenarios involving sensor networks),scalability (the number of nodes can be very high),

5

reliability (the network may be used to report ur-gent alarm events), and robustness (sensor nodes arelikely to be subject to failures for several reasons).Today, most of commercial wireless sensor networksolutions are based on the IEEE 802.15.4 standard,which defines the physical and MAC layers for low-power, low bit rate communications in wireless per-sonal area networks (WPAN). IEEE 802.15.4 doesnot include specifications on the higher layers of theprotocol stack, which is necessary for the seamlessintegration of sensor nodes into the Internet.

Integration of sensing technologies into passiveRFID tags would enable a lot of completely new ap-plications into the IoT context. Sensing RFID sys-tems will allow to build RFID sensor networks, whichconsist of small, RFID-based sensing and computingdevices, and RFID readers, which are the sinks of thedata generated by the sensing RFID tags and providethe power for the network operation.

3.2 Middleware

The middleware is a software layer or a set of sub-layers interposed between the technological and theapplication level. Its main feature of hiding the de-tails of different technologies is fundamental to keepthe programmer away from issues that are not di-rectly pertinent to her/his focus, which is the devel-opment of the specific application enabled by the IoTinfrastructures. The middleware is gaining more andmore importance in the last years due to its majorrole in simplifying the development of new servicesand the integration of legacy technologies into newones. This spares the programmer from the exactknowledge of the heterogeneous set of technologiesadopted by the lower layers[10].

As far as frameworks for developing IoT applica-tions are concerned, a major role is expected to beplayed by approaches based on service-oriented com-puting(SOC). SOC proposes a possibly distributedarchitecture, whereby entities are treated in a uni-form way and accessed via standard interfaces pro-viding a common set of services and an environment

Figure 5: IoT middleware architecture

for service composition. Figure 5 addresses the mid-dleware issues with a complete and integrated archi-tectural approach. Each layer is briefly presented:

Applications: Applications are on the top of thearchitecture, exporting all the systems functionalitiesto the final user. the integration between distributedsystems and applications is ensured through theuse of standard web service protocols and servicecomposition technologies.

Service composition: This is a common layer ontop of a SOA-based middleware architecture. Itprovides the functionalities for the composition ofsingle services offered by networked objects to buildspecific applications. On this layer there is no no-tion of devices and the only visible assets are services.

Service management: This layer provides themain functions that are expected to be available foreach object and that allow for their management inthe IoT scenario. This layer might enable the remotedeployment of new services during run-time, in orderto satisfy application needs. A service repositoryis built at this layer so as to know which is thecatalogue of services that are associated to each

6

object in the network.

Object abstraction: The IoT relies on a vastand heterogeneous set of objects, each one providingspecific functions accessible through its own dialect.There is thus the need for an abstraction layer capa-ble of harmonizing the access to the different deviceswith a common language and procedure.

A service-oriented architecture (SOA) is mainly acollection of services, which communicate with eachother via a set of standardized interaction patterns.The communication can involve either simple mes-sage passing or it could involve two or more ser-vices coordinating some activity via appropriate pro-tocols. Currently, many SOC deployments make useof Web-based protocols (e.g., http) for supportinginteroperability across administrative domains andenabling technologies. SOC can be used to man-age web services and make them act like a virtualnetwork, adapting applications to the specific usersneeds. Service-oriented architectures support a givenlevel of heterogeneity and flexibility in the softwaremodules, nevertheless, to be deployed and executed,SOC/SOA in general and Web services in particu-lar, cannot be straightforwardly applied to the con-struction of IoT applications. In particular, such ap-proaches, at least in their current form, may provetoo heavyweight for being deployed on resources-constrained devices. Nonetheless, they represent avery powerful approach in terms of abstracting func-tionality from the specific software implementationas well as for ensuring integration and compatibilityof IoT technologies into the bigger Future Internet-Future Web perspective[2].

3.3 Ambient intelligence and selfmanagement systems

A parralel can be established between IoT and am-bient intelligence, as a matter of fact, IoT shares anumber of characteristics with ambient intelligence.In Ambient Intelligence (AmI), environments richin sensing/computing/actuation capabilities are de-signed so to respond in an intelligent way to the pres-ence of users, thereby supporting them in carryingout specific tasks. Ambient intelligence builds upon

the ubiquitous computing concept, loosely defined asthe embedding of computational devices into the en-vironment [4].

AmI shares with IoT a number of aspects. Thiscomprises the inclusion in the system of sens-ing/computing capabilities embedded in the envi-ronment. Nevertheless, AmI applications have beenmainly developed for closed environments (e.g., aroom, a building), whereby a number of specific func-tions, known at design time, can be accommodatedand supported. Accordingly, one of the main fo-cus of research in AmI has been the development ofreasoning techniques for inferring activities of usersand proposing appropriate response strategies fromthe embedded devices. IoT expands the AmI con-cepts to integrate open scenarios, whereby new func-tions/capabilities/ services need to be accommodatedat run-time without them having been necessarilyconsidered at design time. This requires IoT solu-tions to be inherently autonomic, i.e., presenting theself-configuration and self-organization, possibly cog-nitive, capabilities needed to provide this additionaldegree of flexibility.

IoT application scenarios require applications toprove adaptable to highly diverse contexts, with dif-ferent resources available and possibly deploymentenvironments changing over time. A number of ap-proaches have been proposed to overcome devicesheterogeneity in related scenarios. All the efforts re-quired in terms of development of IoT architectures,methods for management of resources, distributedcommunication and computation, represent the base-line for the introduction of innovative services thatwill improve user’s experience.

IoT services will be responsive in nature, being ableto anticipate user needs, according to the situationthey are in, by means of dynamic resource manage-ment schemes and on-the-fly composition of differentservice components. This requires applications to beable to understand the context and situation the useris in. Such a theme has been addressed within theambient intelligence, ambient assisted living and per-

7

vasive computing fields, leading to a number of solu-tions able to leverage contextual information comingfrom a number of sources.

Services in IoT are expected to be able to seam-lessly adapt to different situations and contexts. Anumber of research efforts for building self-adaptivesituated services have been undertaken in the last fewyears [11]. However, we are still far from reaching aglobal understanding of how to develop self-adaptiveservices presenting the flexibility level required byIoT scenarios. Most of the approaches proposed havebeen conceived to be applied to a single, well-definedspecific application field. What is needed to fosterthe deployment of IoT applications is instead a setof design patterns that can be used to augment end-user applications with self-adaptive properties. Thisrequires methods for discovering, deploying and com-posing services at run-time in a distributed fashion,supporting autonomicity within all phases of the ser-vice life-cycle. While smart objects may be able torun some limited and lightweight services, one keyaspect of IoT is the integration with the Internet in-frastructure, i.e., the cloud. This may take the formof appropriate Web-based services and applications,able to leverage data and/or atomic services madeavailable by smart things to provide value-added ser-vices to the end user.

4 Applications

Enabling the objects in our everyday working or liv-ing environment to possibly communicate with eachother and elaborate the information collected willmake a lot of applications possible.

As shown in figure 6 ,there are several applicationdomains which will be impacted by the emerging In-ternet of Things. IoT technologies, which are eitherdirectly applicable or closer to our current living habi-tudes, might be classified into the following domains:Personal and social, Business environement, Serviceand utility monitoring and Mobility and transporta-tion based on the type of network availability, cover-age, scale, heterogeneity, repeatability, user involve-ment and impact.

Figure 6: Applications domains and relevant majorscenarios

4.1 Personal and social

In this domain, the sensor information collected isused only by the individuals who directly own thenetwork. Usually WiFi is used as the backbone en-abling higher bandwidth data (video) transfer as wellas higher sampling rates (Sound). Ubiquitous health-care has been envisioned for the past two decades.IoT gives a perfect platform to realize this vision us-ing body area sensors and IoT backend to uploadthe data to servers. For instance, a Smartphone canbe used for communication along with several inter-faces like Bluetooth for interfacing sensors measuringphysiological parameters. So far, there are severalapplications available for Apple iOS, Google Androidand Windows Phone operating system that measurevarious parameters.

An extension of the personal body area networkis creating a home monitoring system for aged-care,which allows the doctor to monitor patients and el-derly in their homes thereby reducing hospitalizationcosts through early intervention and treatment. Con-trol of home equipment such as air conditioners, re-frigerators, washing machines etc., will allow betterhome and energy management. This will see con-sumers become involved in IoT revolution in the samemanner as the Internet revolution itself.

8

Social networking is set to undergo another trans-formation with billions of interconnected objects. Aninteresting development will be using a Twitter likeconcept where Things in the house can periodicallytweet the readings which can be easily followed fromanywhere creating a TweetOT. Although this pro-vides a common framework using cloud for informa-tion access, a new security paradigm will be requiredfor this to be fully realized [9].

4.2 Business environement

We refer to the Network of Things within a work en-vironment as an enterprise based application. Infor-mation collected from such networks are used only bythe owners and the data may be released selectively.Environmental monitoring is the first common appli-cation which is implemented to keep a track of thenumber of occupants and manage the utilities withinthe building.

Sensors have always been an integral part of factorysetup for security, automation, climate control, etc.This will eventually be replaced by wireless systemgiving the flexibility to make changes to the setupwhenever required. This is nothing but an IoT subnetdedicated to factory maintenance.

Real-time information processing technology basedon RFID and NFC (Near Field Communication) willbe widely used in supply chain, due to their lowcost and low requirement. Accordingly, accurate andreal-time information relating to inventory of finishedgoods, work-in-progress, and in-transit stages withreliable due dates would be obtained. As a result,the demand forecast would be more accurate and ex-tra buffers would be unnecessary.

For example, a manufacturer of soft drinks canidentify with the click of a button how many contain-ers of its soda cans are likely to reach their expirationdate in the next few days and where they are locatedat various grocery outlets. Using this information, itmight modify its future production and distributionplans, possibly resulting in significant cost savings.

As a result of applications, the reaction time of tra-ditional enterprises is 120 days from orders of cus-tomers to the supply of commodities while Wal-Martapplying these technologies only needs few days andcan basically work with zero safety stock [3].

4.3 Service and utiliy monitoring

The information from the networks in this applicationdomain are usually for service optimisation ratherthan consumer consumption. It is already being usedby utility companies (smart meter by electricity sup-ply companies) for resource management in order tooptimise cost vs. profit. These are made up of veryextensive networks (usually laid out by large organi-sation on regional and national scale) for monitoringcritical utilities and efficient resource management.The backbone network used can vary between cellu-lar, WiFi and satellite communication.

Smart grid and smart metering is another po-tential IoT application which is being implementedaround the world. Efficient energy consumption canbe achieved by continuously monitoring every elec-tricity point within a house and using this informa-tion to modify the way electricity is consumed. Thisinformation at the city scale is used for maintainingthe load balance within the grid ensuring high qualityof service. Video based IoT which integrates imageprocessing, computer vision and networking frame-works will help develop a new challenging scientificresearch area at the intersection of video, infrared,microphone and network technologies. Surveillance,the most widely used camera network applications,helps track targets, identify suspicious activities, de-tect left luggages and monitor unauthorized access.Automatic behavior analysis and event detection (aspart of sophisticated video analytics) is in its infancyand breakthroughs are expected in the next decade[9].

Disaster alerting and recovery systems could be sig-nificantly enhanced. Natural disasters (flood, land-slide, forest fire, etc.) and accidental disasters (coalmine accident, etc.) are taking place more and more

9

frequently. Technologies in IoT, such like RFID andWSN could play a crucial role in disaster alerting be-fore it happens, and disaster recovery after it ends. Inorder to lessen the effects of natural disasters such likeflood, landslide or forest fire, it is necessary to antic-ipate its occurrence and to alert in time. The timelyaccess to relevant information on hazardous environ-mental conditions gives residents in the nearing areatime to apply preparedness procedures, alleviatingthe damage and reducing the number of casualtiesderived from the event. WSN enables the acquisi-tion, processing and transmission of environmentaldata from the location where disasters originate topotentially threatened cities. Then this informationcould be used for authorities to rapidly assess criti-cal situations and to organize resources. As to acci-dent disaster recovery, for example, after a coal mineaccident occurs, instant tracking and positioning oftrapped workers using RFID technologies could pro-vide timely rescue and lessen casualties and economicloss to the largest extent. Knowing trapped workersgeographic distribution and comparatively accurateposition, the rescue action would be more targetingthus is time-efficient [3].

Water network monitoring and quality assurance ofdrinking water is another critical application that isbeing addressed using IoT. Sensors measuring criticalwater parameters are installed at important locationsin order to ensure high supply quality. This avoidsaccidental contamination among storm water drains,drinking water and sewage disposal. The same net-work can be extended to monitor irrigation in agri-cultural land. The network is also extended for moni-toring soil parameters which allows informed decisionmaking about agriculture.

4.4 Mobility and transportation

Smart transportation and smart logistics are placedin a separate domain due to the nature of data shar-ing and backbone implementation required. Urbantraffic is the main contributor to traffic noise pol-lution and a major contributor to urban air qualitydegradation and greenhouse gas emissions. Traffic

congestion directly imposes significant costs on eco-nomic and social activities in most cities. Supplychain efficiencies and productivity, including just-in-time operations, are severely impacted by this con-gestion causing freight delays and delivery sched-ule failures. Dynamic traffic information will af-fect freight movement, allow better planning and im-proved scheduling. [9]

Cars, buses and taxis as well as roads intersectionsare becoming more instrumented with sensors, actu-ators, and processing power. Important informationcould be collected to realize traffic control and guid-ance, help in the management of the depots, and pro-vide tourists with appropriate transportation infor-mation. One of the successful applications of IoT intransportation is the Traffic Information Grid (TIG)implemented on ShanghaiGrid.

TIG shields all the complexities in information col-lection, storage, aggregation and analysis. It uti-lizes Grid technology to ingrate traffic informationcollected by sensors and actuators, share traffic dataand traffic resources, provide better traffic services totraffic participators, and help to remove traffic bot-tlenecks and resolve traffic problems. The TIG por-tal provides users with various information servicesand can be accessed by Web browsers, mobile phones,PDAs and other public infrastructure. Services pro-vided in TIG included road status information, least-time travel scheme selection, leastcost travel schemeselection, map operation and information query [3].

Apart from the above applications, many otherscould be described as futuristic since they rely onsome (communications, sensing, material and indus-trial processes) technologies that are still to come orwhose implementation is still too complex. The mostappealing futuristic applications included robot taxi,city information model and enhanced game room.

5 Research challenges

In spite of the partial feasability of the IoT conceptthanks to the advances realized in the enabled tech-

10

Figure 7: Security research challenges in IoT

nologies (seen in section 3), a large effort is still re-quired from the research community in order to guar-antee a full, functional and safe deployement of IoTin our every day life. In this section we propose abrief description of the main open issues.

5.1 Security management and privacy

In the Internet of Things, everything real becomesvirtual, which means that each person ans thing has alocalizable, addressable, and readable countrepart onthe internet. The IoT promises to extend (anywhere,anyhow, anytime) computing to (anything, anyone,any service). IoT is unlikely to fullfill a widespreaddiffusion until it provides stong security foundationswhich will prevent the growth of malicious models orat least, mitigate their impact. As shown in Figure7, there are a number of top level security researchchallenges, some of them are described in this section:

5.1.1 Data confidentiality

Data confidentiality represents a major issue in IoTvisions, making sure that only authorized entities canacces and modify data. This is particularly relevantin the business context, whereby, data may representa valuable asset that has to be protected. Two im-portant aspects have to be taken in consideration,first the definition of an access control mechansimand second the definition of an object authenticationprocess.

As data in IoT applications will be related tothe physical realm, ensuring data confidentiality is amain constraint for many uses cases. For instance: asmart community application where a group of homes(located in a close geographic area) are connectedand exchange data might rise privacy concerns [17].In such a context, data should be accessible only bythe appropriate users (home owners for instance), theleakage of private data into the public sphere com-prise in a serious way the user’s privacy.

Usual solutions for ensuring data confidentialitymay not be straightforwardly applied to IoT con-text due to scalability issues generated by the sheeramount of data in the IoT network. Optimal cryp-tography algorithms and adequat key managementsystems, as well as security protocols that connectall the devices through the Internet form the cor-nerstone data confidentiality. Although it is possibleto implement existing standards (AES for instance),some IoT devices such as passive RFID tags, mightbe extremly constrained. Cryptography mechanismsmust be smaller and faster but with little or no re-duction in security level. Mechanisms could includesymmetric algorihthms, hash functions and randomnumber generators [15].

5.1.2 Privacy protection

According to [2], privacy defines the rules underwhich data referring to individual users may be ac-cessed. Privacy is one of the most sensitive subjectin any discussion of IoT security. The data avail-ability explosion has fostered some entities to pro-

11

file and track users without their consent. The IoT’sanywhere, anything anytime nature could easily turnsuch practices into a dystopia. Facebook accountsalready affect a user’s employability and personal in-teractions, IoT could certainly raise much more con-cerns regarding to the huge amount of personal dataavailable.

Different approches are in development to protectthe personal information of IoT users. The delega-tion mechanism is one privacy preservation proposal.An unauthorized RFID reader will retrieve only arandom value, so it will not be able to track theuser. However, limiting access to the user is not theonly protection scenario. In some cases, users willwant to provide information without revealing toomuch about themselves. Some solutions in this con-text let the user find others who best match his pref-erences, without actually revealing such preferences[15]. For instance, a user can try to locate someone inthe neighborhood who like a particular type of musicwithout providing his own location and music prefer-ences.

5.1.3 Trust and governance

There is no consensual defintion of the concept oftrust, nevertheless, it is a wide used concept in dif-ferent context related to computer science’s security.A widely used definition is the one provided by Blazeand Feigenbaum [14], which refers to security policiesregulating accesses to resources and credentials thatare required to satisfy such policies.

In the IoT context, trust mechansims have to beable to define trust in a dynamic, collaborative en-vironement and provide trust throught an interac-tion. Another vision of trust encompasses how usersfeel while interacting in the IoT. Feelings of help-lessness and being under some unknown control cangreatly hinder the deployement of IoT-based appli-cations and services. Users must be able to controltheir own services, in another word, there must besupport for controlling the state of the virtual world.

Governance helps strengthen trust in the IoT. Acommon framework for security policies will supportinteroperability and reduce liability. If someone canattribute a malicious action transaction to a par-ticular user or agent, it will be possible to punishthat user or the agent’s owner. But governance is adouble-edge sword. On the one hand, it offers stabil-ity, support for political decisions, and a fair enforce-ment mechanism. On the other hand, governance caneasily become excessive, fostering an environment inwhich people are continuously monitored and con-trolled. Addressing the challenges of a governanceframework when countless stakeholders and billionsof objects are connected requires the combined effortsof several research communities[15].

5.1.4 Fault tolerance

Clearly, the IoT will be more susceptible to attackthan the current Internet, since billions more deviceswill be producing and consuming services. Highlyconstrained devices will be the most vulnerable, andmalicious entities will seek to control at least somedevices either directly or indirectly. In this context,fault tolerance is indispensable to assure service re-liability, but any solution must be specialized andlightweight to account for the number of constrainedand easily accessible IoT devices. Achieving faulttolerance in the IoT will require three cooperativeefforts. The first is to make all objects secure bydefault. Researchers must work on designing secureprotocols and mechanisms, since it might be verycostly to provide an update for billions of deployed,heteregenoues and interconnected devices.

The second effort is to give all IoT objects theability to know the state of the network and its ser-vices. This system would need to give feedback tomany other elements; for example, a watchdog sys-tem could acquire data as part of supplying qualita-tive and quantitative security metrics. An importanttask in this second effort is to build an accountabilitysystem that will help monitor state.

Finally, objects should be able to defend them-selves against network failures and attacks. All pro-

12

tocols should incorporate mechanisms that respondto abnormal situations and allow the object to grace-fully degrade its service. Objects should be able touse intrusion-detection systems and other defensivemechanisms to ward off attackers. Once an attack af-fects their services, IoT elements should be able to actquickly to recover from any damage. Such elementscan use feedback from other mechanisms and IoT en-tities to map the location of unsafe zones, where anattack has caused service outages.

The IoT is already more than a concept. By com-plying with security requirements, it can fully bloominto a paradigm that will improve many aspects ofdaily life. Open problems remain in many areas ofthe security view, such as cryptographic mechanisms,network protocols, data and identity management,user privacy, selfmanagement, and trusted architec-tures [15].

5.2 Identity management and com-munication issues

The diversity of identities, types and relationshipsconfigurations requires a judicious identity manage-ment, according to certain object identity principlesshown in the following exemple:

- An objects identity is not the same as the iden-tity of its underlying mechanisms. The x-ray ma-chine in the radiology department might have anIP address, but it should also have its own iden-tity to distinguish it from other machines.

- An object can have one core identity and severaltemporary identities. A hospital can become ameeting place for a health conference or a shelterafter a fire.

- An object can identify itself using its identityor its specific features. A virtual food identifiesitself by its ingredients and quantity.

- Objects know the identity of their owners. Thedevice that controls a user’s glucose level shouldknow how that information fits in that usersoverall health.

Identification: The function of identification is tomap a unique identifier or UID (globally unique orunique within a particular scope), to an entity soas to make it without ambiguity identifiable and re-trievable. According to [2], the identification processcould be done basically in two ways. The fist one is tophysically tag one object by means of RFIDs (or sim-ilar). In such a way, an object can be read by meansof an appropriate device, returning an identifier thatcan be looked up in a database for retrieving the setof features associating to it. The second possibilityis to provide one object with its own description, itwill be directly communicated through wireless con-nection. These two approches can complement eachother. RFID-based identification is cheaper in termof the electronics to be embedded in the objects, butrequires a database access where information aboutan object is stored. The self-description based ap-proach, on the contrary, relaxes the requirements toaccess to a global database, but still requires to em-bed more electronics into everyday objects.

Authentication: Authentication is the prossess ofproving identity, it is an important part of identitymanagement, objects will have to handle different be-haviour according to it. A house could have sev-eral appliances that only certain residents and vis-itors can use at particular times. The refrigeratorcould lock itself after midnight to any resident or vis-iting teenagers, but remain open for the adults. Apromising approach is diverse authentication meth-ods for humans and machines. A user could open anoffice door using bioidentification (such as a finger-print) or an object within a personal network, suchas a passport, identity card, or smartphone. Com-bining authentication methods can prevent any lossof overall system security. Such combinations typi-cally take the form of what I am + what I know orwhat I have + what I know [15]. Because the IoTdeals with multiple contexts, an entity is not likelyto reveal its identity all the time. For example, in avehicular network, a police car can reveal its identityto cars and staff at the police station, but keep itsidentity hidden during undercover work unless it isinteracting with other police cars.

13

Authorization: Authorization is also an identitymanagement concern. Authentication and authoriza-tion share open research issues, such as finding abalance between centralized and distributed systemsto answer the question of who is in charge of defin-ing and publishing roles. However, specific topics,such as delegation, fall mainly under the authoriza-tion umbrella. An IoT element can delegate certaintasks to other objects for a limited time. For exam-ple, an object in the users personal network, suchas his phone, can check on his behalf to see if hissuitcase contains all the clothes needed for an up-coming conference. The services, an object provides,might depend on the number of credentials presented.For example, a classroom could provide anyone whoasks with the name of the course being taught, but itwould release the syllabus of that course only to stu-dents with authorization certificates from the dean.

The futur deployement of IoT requires the de-velopement of advanced techniques able to embedcommunication capabilities into every-day objects.In the last years, researches have been led onlow-cost, low-power consumption and micro/nanpo-electronics. Low-power communications is a well-established research field within the sensor network-ing community. The typical approach pursued insuch works relates to the match of the RF front-endactivation patterns (i.e., sleep periods) to the traf-fic pattern. The use of such protocols, however atpresent, does not provide a final answer to the op-timization of energy consumption versus scalabilityissues. These are of paramount importance for IoTscenarios, as battery replacement is a costly processto be avoided as much as possible, especially for largescale deployments. Furthermore, the basic idea ofsuch protocols is to perform active/sleep duty cyclesin order to save the power dispersed in idle listening.More recently, advances in the field of nano-scale ac-cumulators as well as energy harvesting techniquesappear of prominent interest to limit the need forbattery replacements [5].

5.3 Ubiquitous intelligence

The Internet of Things will create a dynamic net-work of billions or trillions of wireless identifiable(things) communicating with one another and inte-grating the developments from concepts like Perva-sive Computing, Ubiquitous Computing and Ambi-ent Intelligence. Internet of Things hosts the visionof ubiquitous computing and ambient intelligence en-hancing them by requiring a full communication anda complete computing capability among things andintegrating the elements of continuous communica-tion, identification and interaction. The Internet ofThings fuses the digital world and the physical worldby bringing different concepts and technical compo-nents together: pervasive networks, miniaturizationof devices, mobile communication, and new modelsfor business processes.

IoT scenarios will be typically characterized byhuge amounts of data made available. A challeng-ing task is to interpret such data and reason aboutit. This underpins the need to have an actionablerepresentation of IoT data and data streams. Thisrepresents a key issue in order to achieve re-usabilityof components and services, together with interoper-abilty among IoT solutions. Advances in data min-ing and knowledge representation/management willalso be required, to satisfactorily address the pecu-liar features of IoT technologies. A related researchfield is that of distributed artificial intelligence, whichaddresses how autonomous software entities, usuallyreferred to as (agents), can be made able to interactwith the environment and among themselves in sucha way to effectively pursue a given global goal[6].

IoT may well inherit concepts and lessons learnedin pervasive computing, ambient intelligence applica-tions and service-oriented computing. Researchersworking in the field of human-computer interfacesand user-centric design methodologies, in particular,addressed already several issues concerning the im-pact of sensorized and pervasive environment on theuser experience. Since IoT will take the reference sce-narios one step further in terms of scale and offeredfeatures, it will also require the development of suit-

14

able, scalable service delivery platforms that permitmultiple services to coexist.

5.4 Standardisation efforts

Although considerable efforts have been made tostandardize the IoT paradigm by scientific commu-nities, European Standards Organizations (ETSI,CEN, CENELEC,etc.), Standardization Institutions(ISO, ITU) and global Interest Groups and Alliances(IETF, EPCglobal, etc.), they are not integrated ina comprehensive framework.

Efforts towards standardization have focused onseveral principal areas: RFID frequency, protocols ofcommunication between readers and tags, and dataformats placed on tags and labels. EPCglobal, Eu-ropean Commission and ISO are major standardiza-tion bodies dealing with RFID systems. EPCglobalmainly aims at supporting the global adoption of aEPC for each tag and related industry driven stan-dards. European Commission has made coordinatedefforts aiming at defining RFID technologies and sup-porting the transition from localized RFID applica-tions to the IoT. Differently from these, ISO dealswith how to modulate, utilize frequencies and pre-vent collision technically.

The European Telecommunications Standards In-stitute (ETSI) has launched the Machine-to-Machine(M2M) Technical Committee to conduct standard-ization activities relevant to M2M systems and de-fine cost-effective solutions for M2M communica-tions. Due to lack of standardization of this leadingparadigm towards IoT, standard Internet, Cellularand Web technologies have been used for the solu-tion of standards. Therefore, the ETSI M2M com-mittee aims to develop and maintain an end-to-endarchitecture for M2M (with end-to-end IP philosophybehind it), and strengthen the standardization effortson M2M. Within the Internet Engineering Task Force(IETF), there are two working groups 6LoWPANand ROLL dealing with integrating sensor nodes intoIPv6 networks.

6LoWPAN is to define a set of protocols to makethe IPv6 protocol compatible with low capacity de-vices. Core protocols have been already speci-fied. While ROLL recently produced the RPL (pro-nounced ”ripple”) draft for routing over low powerand lossy networks including 6LoWPAN. Lots of con-tributions are needed to reach a full solution [3].

5.5 Social and political issues

The Internet has long since changed from being apurely informational system to one that is socio-technological and has a social, creative and polit-ical dimension. But the importance of its non-technological aspects is becoming even more apparentin the development of Internet of Things, since it addsnew challenges to these non-technological aspects.

Several critical questions need to be asked with re-gard to possible consequences of the full IoT’s de-ployement, much of the public debate on whether toaccept or reject the Internet of Things involves theconventional dualisms of security versus freedom andcomfort versus data privacy. In this respect, the dis-cussion is not very different from the notorious al-tercations concerning store cards, video surveillanceand electronic passports. As with RFID, the uneasefocuses mainly on personal data that is automati-cally collected and that could be used by third partieswithout peoples agreement or knowledge for unknownand potentially damaging purposes.

Personal privacy is indeed coming under pressure.Smart objects can accumulate a massive amount ofdata, simply to serve us in the best possible way.Since this typically takes place unobtrusively in thebackground, we can never be entirely sure whetherwe are being (observed) when transactions take place.Aside of the data protection issues, there is also thequestion of who would own the masses of automati-cally captured and interpreted real-world data, whichcould be of significant commercial or social value, andwho would be entitled to use it and within what eth-ical and legal framework.

15

Another critical aspect is that of dependence ontechnology. In business and also in society, gener-ally, we have already become very dependent on thegeneral availability of electricity; infrequent black-outs have fortunately not yet had any serious conse-quences. But if everyday objects only worked prop-erly with an Internet connection in the future, thiswould lead to an even greater dependence on the un-derlying technology. If the technology infrastructurefailed for whatever reason: design faults, materialdefects, sabotage, overloading, natural disasters orcrises, it could have a disastrous effect on the econ-omy and society. Even a virus programmed by somehigh-spirited teenagers that played global havoc withselected everyday objects and thus provoked a safety-critical, life-threatening or even politically explosivesituation could have catastrophic consequences.

The Internet of Things has now arrived in politics.A study for the (Global Trends 2025) project carriedout by the US National Intelligence Council statesthat foreign manufacturers could become both thesingle source and single point of failure for mission-critical Internet-enabled things, warning not only ofthe nation becoming critically dependent on them,but also highlighting the national security aspect ofextending cyberwars into the real world: U.S. lawenforcement and military organizations could seek tomonitor and control the assets of opponents, whileopponents could seek to exploit the United States.

The European Commission is reflecting vocally butsomewhat vaguely on the problem of governance fora future Internet of Things. The issue here is how tosafeguard the general public interest and how to pre-vent excessively powerful centralized structures com-ing into being or the regulatory power of the Internetof Things falling exclusively into the hands of whatthey describe as a single specific authority [13].

To be truly beneficial, Internet of Things requiresmore than just everyday objects equipped with mi-croelectronics that can cooperate with each other.Just as essential are secure, reliable infrastructures,appropriate economic and legal conditions and a so-

cial consensus on how the new technical opportuni-ties should be used. This represents a substantialresearch issue for the futur.

6 Conclusion

Internet Of Things brings the possibility to connectbillions of every-day’s objects to the Internet, allow-ing them to interact and to share data. This prospectopen new doors toward a futur where the real andvirtual world merge seamlessly through the massivedeployment of embedded devices. This survey hasaimed to provide a brief overview of the IoT’s state-of-art, including the main definitions and visions, en-abling technologies, already or soon available appli-cations and open research issues focusing on the secu-rity perspective. The IoT has the potential to add anew dimension to the ICT sector by enabling commu-nications with and among smart objects, thus leadingto the vision of (anytime, anywhere, anymedia, any-thing) communications. Though a lot still to be donein order to fullfil the IoT vision.

References

[1] The epcglobal architecture framework. page 19,March 2009.

[2] Francesco De Pellegrini Imrich ChlamtacDaniele Miorandi, Sabrina Sicari. Internetof things: Vision, applications and researchchallenges. Ad Hoc Networks, pages 1497–1516,april 2012.

[3] Yi-Duo Liang De-Li Yang, Feng Liu. A surveyof the internet of things. The 2010 InternationalConference on E-Business Intelligence, 2010.

[4] R. Wichert E. Aarts. Ambient intelligence.Technology Guide, Springer, Berlin, Heidelberg,pages 244–249, 2009.

[5] N. Harris-B. Al-Hashimi G. Merrett, N. White.Energy-aware simulation for wireless sensor net-works. Proceedings of IEEE SECON, Rome,Italy, pages 64–71, 2009.

16

[6] Peter Friess-Sylvie Woelffl Harald Sundmaeker,Patrick Guillemin. Vision and challenges forrealising the internet of things. Cluster ofEuropean Research Projects on the Internet ofThings, March 2010.

[7] Y. Sankarasubramaniam-E. Cayirci I.F. Aky-ildiz, W. Su. Wireless sensor networks: a sur-vey. Computer Networks 38 (4), pages 393–422,2002.

[8] D.Culler J. Hui and S. Chakrabarti. Incorporat-ing ieee 802.15.4 into the ip architecture-internetprotocol for smart objects. Ad Hoc Network, jan2009.

[9] Slaven Marusic-Marimuthu PalaniswamiJayavardhana Gubbi, Rajkumar Buyya. In-ternet of things (iot): A vision, architecturalelements, and future directions. journal,page 24, jan 2007.

[10] Giacomo Morabito Luigi Atzori, Antonio Iera.The internet of things: A survey. Computer Net-works, page 19, May 2010.

[11] F. Rosenberg M. Baldauf, S. Dustdar. A sur-vey on context-aware systems. Int. J. Ad HocUbiquitous Comput, pages 263–277, april 2007.

[12] A. Gluhak M. Presser. Eurescom message-themagazine for telecom insiders. Ad Hoc Networks,feb 2009.

[13] Floerkemeier C Mattern, F. Vom internet dercomputer zum internet der dinge. Informatik-Spektrum 33(2), pages 107–121, 2010.

[14] J. Lacy M.Blaze, J. Feigenbaum. Decentralizedtrust management. Proceedings of IEEE nter-national Symposium Security and Privacy, Col-orado Springs, pages 164–173, 1996.

[15] Pablo Najera Rodrigo Roman and Javier Lopez.Securing the internet of things. IEEE Computer,pages 51–58, september 2011.

[16] K. Sakamura. Challenges in the age of ubiqui-tous computing: a case study of t-engine - an

open development platform for embedded sys-tems. Proc.Of the 28th International Conferenceon Software Engineering, pages 713–720, May2006.

[17] Xiaohui Lang Xu Li, Rongxing Lu andXuemin (Sherman) Shen. Smart community: Aninternet of things application. IEEE Communi-catins Magazine, pages 68–75, november 2011.

17