INTRODUCTIONThe Internet of Things can be considered as aconvergence among a number of heterogeneousdisciplines (wireless communication, identifica-tion, real-time localization, sensor networks, per-vasive computing) that enable the Internet to getinto the real world of physical objects interactingwith web services. Measuring, labeling, and tim-ing of made things and people, and their map-ping into the environment is sure to stimulatenew context-aware services. Pleasant user experi-ences will be planned in the workplace and pub-lic areas as well as in the home environment byembedding computational intelligence into thenearby environment and simplifying humaninteractions with everyday services.

Pervasive interconnection with Things willrequire the deployment of a multitude of elec-tronic tags [1] to attach over objects, and con-ceiving new functions concerning habitat control,disaster relief, mobile healthcare, the monitoringof industrial processes, security surveillance, andthe realization of augmented spaces in general(Fig. 1).

Radio frequency identification (RFID) isone of the key enabling technologies since itnot only permits a digital code to be associatedwith an object in a wireless modality, but also

allows its physical status to be captured. RFIDtags may actually be equipped with a largevariety of sensors according to very differentmodalities of integration, exploiting a broadrange of possible functionalities and costs.Active RFID tags, which make use of indepen-dent batteries, a true microcontroller, and ded-icated electronics, ensure long operating rangesand can support high data rates and the great-est versatility in sensor interconnection. Themain drawbacks of this solution are high cost,limited lifetime, and large weight and size.Conversely, passive RFID tags are completelybattery-free and could be permanently embed-ded into tagged objects for structural, medical,or product monitoring. The major limitation ofthis class of systems is the need for proximityto a reading device and a rather reduced set offunctionalities.

The use of an antenna as part of a passivesensor is not new. In the late 1940s the Russianinventor Leon Theremin developed one of thefirst covert listening devices (or “bugs”) using acapacitor microphone and electromagneticinduction to transmit away, through reflection,the audio signals captured in nearby environ-ments. A modern remarkable example of a true-RFID passive sensing device is given by theWireless Identification Sensing Platform (WISP)project [2]. The device involves two mercuryswitches to mechanically toggle between twocommercial RFID microchips, permitting themovement of the tagged object to be monitoredby modulation of the transmitted chip’s identifi-er (ID modulation).

The unique feature of the asymmetric link ofan RFID system further adds a completely newsensing possibility by taking into considerationthat RFID tags are tiny computers of increasingperformance with tiny low-power radios whichmerge together both digital (the microchip datageneration) and analog (antennas and propaga-tion phenomenology) features. Data transmittedback to the reader during the interrogation pro-tocol are digitally encoded, but the strength ofthe backscattered power is governed in an ana-log manner by the interaction with nearbyobjects, the propagation modality, and even themutual position and orientation among readerand tags. This fact poses the basis for a differentsensing modality wherein the captured data can

IEEE Wireless Communications • December 201010 1536-1284/10/$25.00 © 2010 IEEE

Blood

z x

t2

y

THE IN T E R N E T O F TH I N G S

GAETANO MARROCCO, UNIVERSITY OF ROME TOR VERGATA

ABSTRACTThings equipped with electronic labels having

both identification and sensing capability couldnaturally be turned into digital entities in theframework of the Internet of Things. Radio fre-quency identification (RFID) technology offersthe natural background to achieve such function-alities, provided that the basic physics governingthe sensing and electromagnetic interaction phe-nomena are fully exploited. The sensing ofThings is here reviewed from an electromagneticperspective with the purpose of showing howadvanced performance may be achieved bymeans of low-cost batteryless devices. A possibleclassification of basic sensing modalities is intro-duced, and many ideas, at different stages ofmaturity, are then discussed with the help ofexamples ranging from the sensing of non-livingThings up to the more challenging sensing ofHumans.

PERVASIVE ELECTROMAGNETICS: SENSINGPARADIGMS BY PASSIVE RFID TECHNOLOGY

The author reviewsthe sensing of Things from an electromagnetic perspective with thepurpose of showinghow advanced performance may be achieved bymeans of low-cost batteryless devices.

MARROCCO LAYOUT 12/9/10 11:06 AM Page 10

IEEE Wireless Communications • December 2010 11

be collected by a “sensorless” tag, just by exploit-ing the physics of the RFID response.

The research on the basic physics of data cap-ture together with the low-level data processingcan be considered as a particular edge discipline,Pervasive Electromagnetics, which differs fromthe well assessed remote sensing, generallybased on the raw scattering from objects, sincethe former seemingly addresses both the designof devices and data processing, and may profitby digital intelligence distributed in Things. Per-vasive Electromagnetics merges together electri-cal, radio, material, signal processing, and evenarchitectural issues and promises to stimulate, inthe near future, a fan of new low-cost radiosensing devices, ready to be seamlessly embed-ded into objects and the environment itself.

This article describes the basic sensing mech-anisms and paradigms achievable by low-costpassive RFID devices, with particular attentionto showing how a simple tag can be turned intoa radio sensor, and introduces the new conceptof RFID grids, their analog identifiers and finger-prints, as well as their possible application insmart sensing. The data capture mechanisms areillustrated by some experimental and computersimulated examples concerning the sensing ofThings in manufacturing logistics and qualityassessment, and the sensing of Humans (e.g., theemerging radio technology for the monitoring ofhuman motion and the progression of someinternal biological processes).

THE WAY THINGS ARE SENSED

The concept of sensing, in the framework of theInternet of Things, deserves to be consideredwith the largest scope, ranging from the acquisi-tion of an elementary status of an object (e.g.,presence or absence within a given region —localization) to multidimensional description ofthe chemical-physical and relational parametersof a Thing with respect to the nearby environ-ment and other Things. Here the term Thing iscapitalized to highlight the fact that the usualtagging of an object is augmented with the physi-cal interaction between the objects and theRFID tag itself to produce richer, more informa-tive content. Some authors [3] refer to this con-cept by the more evocative term spime (space +time), emphasizing the idea of material objectscorrelated with evolving information.

The sensing modalities may fall at least intothe classes of stationary and non-stationary sens-ing. Stationary sensing occurs when the measure-ment is performed in controlled conditions, suchas when the mutual position between the readerand the Things remains unchanged during thewhole phenomenon to monitor or, similarly,when a same position can be replicated exactlyin successive readings. This is the case withTheremin’s spy device or a card-like reader. Sta-tionary sensing is the simplest configuration tohandle since the variation of the nearby environ-ment can easily be removed by signal processingand the tag response is unambiguously related tothe physical parameter to be sensed.

Much more complex is the case of non-sta-tionary sensing when interrogation is performedat different times or the Thing is moving. The

eventual change of the reader-tag position andalso the change in the environment can be a fur-ther unknown of the sensing problem, makingdata retrieval more difficult, if it is even still pos-sible, but at the cost of more complicated dataprocessing or more complex tag electronics. Ingeneral, additional independent data or func-tionalities are required to properly manage thesensing, as discussed next.

SPECIES OF SENSING RFIDS

Most passive RFID tags for sensing of Thingscan be grouped (Fig. 2) according whether ornot they make use of integrated specific sensorsto detect variation in the tagged object and theenvironment, and according to analog or digitalmodality to transmit such variation to the read-er. Hence, at least four main classes of sensingRFIDs can be envisaged having different degreesof complexity. Self-sensing tags detect the changeof the Thing through variation of an antenna’simpedance and gain. This variation can be com-municated to the reader by analog modulationof the backscattered power or, if multiplemicrochips are included in the tag (whichbecomes an RFID grid), by means of digitalencoding over the microchips’ IDs. True sensortags include a specific physical receptor, whichmodifies the antenna’s impedance and gain, andthe data is transmitted to the reader by eitheranalog modulation of the backscattering poweror digital modulation if analog-to-digital func-tionalities are provided in the tag. It is, however,

Figure 1. Sensing of Things and humans by fixed and handheld RFID readers.

Environmentdata

Growth of trees Growth of animals

Drugquality

Counterfeitingof goods

Monitoringindustrial processes

Fixed reader

Personalhealthcare

Handheld reader

MARROCCO LAYOUT 12/9/10 11:06 AM Page 11

IEEE Wireless Communications • December 201012

reasonable to imagine that the representation ofreal phenomena in a virtual context will benefitfrom a combination of the above sensing solu-tions, depending on the required miniaturizationand acceptable costs.

SENSORLESS TAG WITH ANALOGDATA COMMUNICATION

Any conventional RFID tag may be consideredas a sensor for the effective permittivity of theobject to which it is attached. Like any antennaplaced onto a real medium, the electrical prop-erties of an RFID tag, such as the inputimpedance and radiation gain, are strongly cor-related with the physics of the nearby environ-ment. A change of the tagged object is globallyseen by the tag’s antenna as a change of equiva-lent permittivity, which will in turn produce achange of the tag’s antenna features and henceof the backscattered signals. Denoting by Pin thepower entering the reader’s antenna, the powerbackscattered by the tag and collected in turn bythe reader itself may be written in the free spaceby the Friis formula as

(1)

where the parameter Ψ generically indicates anyphysical or geometrical feature of the target thatis subjected to change in the phenomenon moni-tored by the RFID platform, d is the reader-tagdistance, GR is the gain of the reader antenna,σT is the radar cross-section of the tag whichdepends on its radiation gain and on theimpedance matching between the antenna andthe microchip, ηP is the polarization efficiency ofthe reader-tag link, and finally, r̂_ is the unitaryvector indicating the mutual orientation betweenthe reader and the tag. By processing PR←T, for

instance, through the received signal strengthindicator (RSSI) or another equivalent quantitycollected by the reader, it is theoretically possi-ble to detect a macroscopic change of the taggedobject. So a self-sensing, completely sensorless,passive device [4] is obtained wherein the sensoris the antenna and the antenna is the sensor.

The tag’s responses need to be unequivocallyrelated to the variation of the Thing’s status bymeans of data inversion curves, or lookup tables,produced by offline experimentation or comput-er simulation according to a training procedure.The antenna design effort is hence oriented toachieve monotonic inversion curves, PR←T ↔ Ψ,so that measurement ambiguities may be avoid-ed. However, since no dedicated sensor is pre-sent, this sensing mechanism is rathermacroscopic and non-specific; it is able to detectjust an overall change of the Thing, even if sucha variation is due to different causes (e.g., incase of multivariable systems). So RFID self-sensing should be applied only to Things with asingle time-variant feature.

The transmitted data signal is of analog type(e.g., without any digital modulation), and henceit appears fully exposed to the interaction withthe environment. The signal collected by theinterrogator is indeed dependent on the particu-lar mutual position between reader and tag sincethe tag’s and reader’s radiation gain are general-ly non-isotropic. The interrogation of the Thingis hence strongly undetermined in case of non-stationary sensing (e.g., when the reader-tagposition is not controlled). This is, for instance,the case of an operator performing a manualsweep around a Thing by means of a handheldreader. A possible way to overcome this physicaluncertainty is the processing of both direct andreverse RFID links, and in particular by alsorecording the turn-on power, the minimum powerPto(r̂_ ,d)[Ψ] to inject into the reader’s antenna toachieve the tag’s microchip activation. In thiscase it is possible to demonstrate [5] that the fol-lowing non-dimensional indicator F[Ψ] is com-pletely independent of the reader-Thing mutualposition and the effect on the environment:

(2)

where ZA = RA + jXA and ZC = RC + jXC arethe input impedance of the tag’s antenna andmicrochip, respectively, and pn is the generallyknown power sensitivity of the tag’s microchip(e.g., the minimum radiofrequency power thatneeds to be harvested to wake up the microchipand perform actions). The metric F[Ψ] gives aunique feature of the tag and may be consideredan analog identifier, complementary to the digitalidentifier; that is, a structural property invariantwith the particular measurement conditions(position and orientation of reader-Thing) andnearby environment. If a scattering randomobject was present in the reader zone, it wouldaffect both the direct and reverse links; the pro-cessing in Eq. 2 will remove this effect. The con-tinuous variation of the analog identifier may

Fp

P r d P r d

R

Z

n

R Tto

C

A

[ ]( ˆ, )[ ] ( ˆ, )[ ]

ΨΨ Ψ

=⋅

=

←2

[[ ]

,Ψ + ZC

P r d

d

rG

R T

TR

← =

⎛

⎝⎜

⎞

⎠⎟

( ˆ, )[ ]

( ˆ)[ ]( ˆ

Ψ

Ψ

λπ

σπ

02

2

4 4rr r PP in) ( ˆ) ,η2

Figure 2. Possible classification of passive sensing RFIDs according to the sens-ing mechanism (self-sensing or by a specific sensor) and data transmission(TX) modality (analog or digital encoding).

Complexity

Antenna + chip Multiple chips Dedicated sensors Microcontroller

Perf

orm

ance

Sens

ing

spec

ifici

tyM

acro

sen

sing

Two-port tag

WISP

Isotropic inertial switch

Conventional tag

Self-sensing TX: analog

Self-sensing grid;TX: analog/digital

Specific sensor;TX: analog

Specific sensor;TX: digital

MARROCCO LAYOUT 12/9/10 11:06 AM Page 12

IEEE Wireless Communications • December 2010 13

unequivocally refer to the variation of the physi-cal parameter of the Thing under test in anynon-stationary interrogation. Physical parame-ters that could be suited to this kind of sensingare the shape deformations of the Thing, andeven the change of its chemical and physical fea-tures, provided that an electrical permittivityvariation of the Thing is accordingly produced.The growth of animals, plants, and living tissuescould also be controlled by this sensing modality.

SENSORLESS TAG WITH DIGITALDATA COMMUNICATION

The tag works, like in the previous case, as aself-sensing device, but multiple microchips arenow included to encode the occurrence of dis-crete events such as discrete values [4] of thechanging physical parameter Ψ∈{φ1, φ2, …, φM}to be transmitted through an ID modulation(two-chip tag in Fig. 2). In particular, the anten-na is provided with a number of ports of inputimpedance {ZA,n[Ψ]}, and a single microchip isattached to each port. Depending on the valueof the sensed parameter, the impedance mis-match between the antenna ports and themicrochips is such to enable (or not) themicrochip to harvest the necessary power andrespond to the reader with its own digital identi-fication code (IDn). Hence, several encodingschemes are possible. The simplest one requireseach discrete event φn to be univocally associat-ed to the IDn of the nth microchip, which is theonly one responding in that condition. A moregeneral approach, reducing the number ofrequired microchips, considers instead that morethan one microchip is responding at the occur-rence of the nth event φn so that such an event isunequivocally associated with a set of digitalidentifiers. In any case, due to digital encoding,this reading mechanism is not affected by uncer-tainty in the mutual reader-tag position and theinteraction with the environment, and is hencewell suited to non-stationary sensing.

Multichip tags may be achieved as a generallycoupled multitude of RFID tags, including sin-gle-microchip tags in close proximity, as well astags with a multiplicity of RFID microchips. Thisinterconnected object, called an RFID grid, hasrecently been theoretically studied in [6] as aunique digital/analog entity having improvedread-distance capabilities and other relevantproperties. RFID grids could be suited to envel-op a body, like a smart skin, and transmit a setof digital and analog identifiers forming the gridfingerprint F[Ψ = {F1[Ψ], …, FM[Ψ], ID1, …IDM}, a multidimensional dataset carrying infor-mation about one or more parameters to besensed independently on the reader-tag orienta-tion.

TAG WITH SPECIFIC SENSOR ANDANALOG COMMUNICATION

The tag is equipped with a real sensor (motion,as in Fig. 2, temperature, pressure, chemicalspecies [7], or other), which could be eitherlumped into a device, connected in some part ofthe tag’s antenna, or distributed all over theantenna surface by chemical receptor painting.

Such a sensing mechanism hence may be consid-ered a lumped or distributed impedance loadingZS(Ψ) placed on the tag’s antenna. The variationof ZS(Ψ), caused by the change of the environ-ment, will produce a change of the tag’s radarcross-section and hence a backscattered powermodulation as in the case of sensor-less tags.Moreover, these kinds of devices are suited toinclude specific and direct sensing mechanisms,rather than macroscopic ones as in the case ofself-sensing RFIDs. The sensor’s equivalentimpedance ZS(Ψ) should be preferably reactive(e.g., the sensing substance should work as vari-able inductors or capacitors) to avoid the intro-duction of additional electric loss, which wouldinstead reduce the reading range capabilities ofthe tag. Also, in this case the RFID gridparadigm may be applied, with the additionalpossibility to collect data from different physicalphenomena.

TAG WITH SPECIFIC SENSOR AND DIGITALDATA COMMUNICATION

This kind of tag works as a true data-logger:data collected by a specific sensor are handledby a microcontroller, and sampled and encodedinto digital information that can be stored in themicrochip’s memory and then recovered by thereader through regular RFID interrogation.

Although these devices are typically providedwith an internal battery, batteryless configura-tions have very recently been successfully tried.In such cases the microcontroller has very lowpower consumption (a few milliwatts), and theenergy required to drive the data acquisitionmay be directly harvested from the interrogationsignal, as conventional passive tags do, or gener-ated by piezo-electric energy scavengers trans-forming micro-oscillations of the Thing intoelectrical energy. These devices are the mostpowerful and versatile ones since they are ableto address specific sensors even if the powerissues may reduce the real data rate. Moreover,since the transmitted information is digitallyencoded, this sensing architecture is practicallyinsensible to the environment interactions andmay work properly for both stationary and non-stationary sensing. However, the size and costare not currently suited to massive distributionin the environment.

Examples of this class of devices are the newversion of the programmable WISP platform(Fig. 2) and low-frequency integrated tempera-ture microchips in [8].

SENSING NON-LIVING THINGS

Some of the most interesting parameters ofmanufactured Things to be sensed over a largescale are definitely geometry deformation, tem-perature, and chemical changes. For instance, inlogistics it is sometimes required to monitor thestatus of containers (bottles, packages, bags) todetect variation of their solid, powder, or liquidcontents. The self-sensing natural capabilities ofRFID tags may be exploited to handle theseevents in both continuous and discrete evolu-tions. An example of continuous sensing is thepowder level control inside a plastic container,

This tag works as atrue data-logger:

data collected by aspecific sensor are

handled by a microcontroller, andsampled and encod-

ed into digitalinformation that can

be stored in themicrochip’s memoryand then recovered

by the readerthrough regular RFID

interrogation.

MARROCCO LAYOUT 12/9/10 11:06 AM Page 13

IEEE Wireless Communications • December 201014

as in Fig. 3. In the reported experiment [5] (per-formed at 870 MHz UHF RFID frequency, asalso were the next examples) the filling powderis sugar whose monitoring is a very challengingproblem for RFIDs due to the low relative per-mittivity ((εr = 2.76), which is very similar tothat of air ((εr = 1). The variation of filling levelis hence expected to produce only a slight modi-fication of the tag’s radar cross-section and inturn of the sensed data. The analog identifier inEq. 2 has been collected when the sugar levelincreases from 0 cm (empty container) up to 14cm (about the full tag’s length) at differentangular orientation between the reader and thetag on the horizontal plane (0° rotation meansthat the tag and reader’s antennas are facing).The analog identifier begins to sense the sugarvariation when the filling level exceeds half thetag size and spans an overall dynamics of about50 percent, up to saturation. The collected dataare very little dependent on the mutual orienta-tion between the reader and the container, evenfor rotation of 180° e.g., when the two antennasare completely separated by the sugar itself. ThisThing may be therefore interrogated in non-sta-tionary modalities, for instance by a robot or ahuman operator handling a reader and movingall around a warehouse or a factory.

A similar concept was applied very recentlyto the remote sensing of beverage glasses [9],while another interesting application of (station-ary) analog self-sensing by regular tags aimed atthe detection of geometrical displacements canbe found in [10].

Temperature monitoring has so many impli-cations in pharmaceutical (blood bags, vaccines)and food (frozen goods) supply chains, as well asin hospitals and industrial processes in general.Passive RFID tags promise to be suited to con-ceive a digital quality label, for example, a kindof “seal” or “fuse” [11] able to certify that thetagged item was exposed (or not) to a tempera-ture higher than a particular threshold. For thispurpose, the RFID device has to store this infor-mation over its work life in the absence of con-tinuous interrogation. Furthermore, theover-temperature flag needs to be unchangedeven if the tag temperature falls below the criti-

cal value. A possible approach to this problem isto embed in the RFID tag a shape-memory-alloy(SMA) conductor having the capability to changeits shape when the local temperature exceeds atransition level. The shape’s change can be engi-neered to permit or prevent an RFID microchipresponding to the reader depending on the ther-mal status through the ID modulations. Forexample, Fig. 4 shows the time-response of atwo-microchip RFID grid wherein the microchipemitting code ID.2 is conditioned by an SMAwire that forces such a microchip to be short-cir-cuited depending on the temperature value.When the temperature is less than a threshold(around 80°C in the experiment), the reader willonly receive code ID.1 of the first microchip(indicated by vertical bars in Fig. 4). This datahas the sense of the code-name of the taggedobject (a drug box in the figure). As the temper-ature exceeds the threshold, the SMA wiredeforms and enables the second microchip tosend back its own ID.2, which now has the mean-ing of “seal status.”

In principle, multiple thresholds may beincluded in the device by using a plurality ofSMAs with different thermal characteristics.

SENSING HUMANS

One of the most fascinating and ambitious con-jugations of RFID sensing is a kind of Internetof the Body (e.g., when persons and even theirinternal organs are tagged). True applicationsare not so distant and could enable the monitor-ing of the proper functionalities of critical organsor detecting the progression of a disease throughhandheld devices, enabling the patient to becomethe primary hub of bio-data acquisition. Figure 5shows some possible human-body districts andorgans where sensing tags could be placed toproduce bio-information.

SENSING HUMAN BEHAVIORA wearable tag provided with passiveaccelerometers may be attached to arms tomonitor human movements in entertainment,healthcare, and diagnostic applications. Forinstance, it was demonstrated in [12] that wear-able tags with inertial switches are able todetect limb motion in some common sleep dis-orders like restless legs syndrome and periodiclimb movements. Tags applied to the chest maybe useful to detect breath. More generally,motion-detecting wireless devices may help toproduce statistics to support diagnosis and dis-creetly monitor the activity of a patient inside astructure, and generate warnings about unusualbehaviors such as when the patient falls downor stays motionless for long times. For example,one of the diagrams within Fig. 5 gives therecord of clusters of arm motion collected bythe RFID reader according to ID modulation.The microchip only responds with its ID whenit is in steady state, while it does not respondwhen in motion. The motion events are indicat-ed in the figure by black bars. The samemotions have been recorded for comparison bythe digital micro-electromechanical system(MEMS) accelerometers of an iPhone (used asan active sensor) attached to the arm, in close

Figure 3. Sensing the level of powder (sugar) material by analog, self-sensingmodality.

Sugar level hs (cm)4 2

0.7

F(h s

)

0.65

0.75

0.8

0.85

0.9

.95

1

6 8 10

Tag

Sugar

20 cm 14 cm

13 c

m

hs

12 14

Rotation: 0Rotation: 45Rotation: 180

MARROCCO LAYOUT 12/9/10 11:06 AM Page 14

IEEE Wireless Communications • December 2010 15

proximity to the RFID tag, which provides trueacceleration. Another diagram in Fig. 5 showsthe breath rhythm when the tag is attached tothe chest of a volunteer. In both cases, RFIDinformation is very poor with respect to a morespecific (active) sensor, but the RFID data arenaturally aggregated without any particular pro-cessing, and hence are useful to generatemacro-indicators (occurrence frequency, dura-tion) of human behavior.

SENSING FROM THE INSIDEImplanted sensing tags are much more challeng-ing to handle due to their intrinsic obtrusivenature and the high power attenuation of thehuman body, which makes it difficult to establish

the RFID two-way link. However, from anotherpoint of view, the high electric permittivity ofhuman tissue helps in the miniaturization oftags’ size, making body implants feasible.Implanting a bio-compatible, thin, self-sensingtag is in fact accepted during surgery. Theimplanted tag will scatter back toward the readerindirect information about variations in localequivalent permittivity of the tissues due to thehealing process and possible complications (e.g.,abnormal cell proliferations, edema, and inflam-matory events).

A possible locus of implant for self-sensingRFID tags is a biological duct with the purposeof monitoring its safety condition. In this casethe tag may be integrated with a stent, a

Figure 4. Detection of thermal threshold crossing of drugs by a two-port RFID grid used in digital mode.

Time (s)

50 0

95

Tem

pera

ture

(°C

)

90

85

80

75

70

65

60

55

50

100

100 150 200 250

ID 2

ID 1

Figure 5. Sensing humans by wearable and implanted passive RFIDs.

Time (s) 20 0

ID

1

40 60 80 100 120 140 160 180 200

Passive RFID

Active accelerometer

20 0

a(t) 2 3 4

40 60 80 100 120 140 160 180 200

MusclePlaque

Restenosis of the duct

PR←T

50% 100% 0

0.2 0

0.4

0.6 0.8

1.0

Wearable RFIDs

limb motion

Wearable RFIDs

Breath detection

Implanted RFIDs

Brain edema monitoring

Implanted RFIDs

Duct obstruction

Obstructed duct Open

duct

Time (s) 0

ID

50 100 150 200

One of the most fascinating and

ambitious conjuga-tions of RFID sensingis a kind of Internet

of the Body, e.g.,when persons andeven their internalorgans are tagged.

True applications arenot so distant and

could enable thepatient to becomethe primary hub of

bio-data acquisition.

MARROCCO LAYOUT 12/9/10 11:06 AM Page 15

IEEE Wireless Communications • December 201016

mechanical device having the form of a cylin-drical mesh of thin wires commonly implantedinto a natural passage/conduit in the body(e.g., coronary arteries, urinary tract, prostate,esophagus, carotid artery, biliar ducts) to pre-vent or counteract localized flow constriction,called stenosis. In some cases a new reductionof the circumference of the duct’s lumen mayoccur, indicated as restenosis, due to a newformation of cholesterol plaques against thevein walls or a new proliferation of cells in themuscle wall of the vessel. In both cases therestenosis produces a local modification of tis-sue (from blood to fat or from blood to mus-cle) and hence a variation of the effectivepermittivity “sensed” by the stent. So a newgeneration of self-sensing stents may be con-ceived by integrating RFID microchips into thestent geometry to turn it into a modulatedbackscattering antenna.

In a first feasibility study in [13] the case ofthe carotid artery was considered, where the tagsimulating the stent has a four-turn spiral geom-etry. One of the diagrams in Fig. 5 shows thecomputer-simulated power backscattered, as inEq. 1, by such a simplified version of a stent withrespect to the in-stent restenosis evolution, froma clear duct up to a duct completely filled (in thecorrespondence of the stent) by fat or muscle.The curves are monotonic, and in the particulartype of restenosis caused by cholesterol plaques,the data dynamic is 10:1, particularly promisingfor early diagnosis

Another feasibility study [13] recently dis-cussed, by means of electromagnetic computersimulations, the possibility of monitoring brainedema evolution after surgical treatment forbrain cancer. The physical rationale for RFIDsensing relies on the modification of the electro-magnetic characteristics of brain tissues that maybe detected by the self-sensing properties of anRFID tag placed in the surgery-treated region.Figure 6 shows the computer simulated variationof the backscattered power in Eq. 1 emerging

from a 1-cm-long dipole tag implanted in thehead, collected by the reader, with respect to thevariation of the edema size. Edema, which hereroughly refers to water-imbued brain tissue, issimulated by a sphere of increasing radiusembedding a blood core; the sensing tag isplaced in the origin of such a sphere. The curvesof power backscattered by the tag are also in thiscase monotonic and hence suited to data inver-sion, with dynamics of 15 percent (870 MHz)and 35 percent (2450 MHz) for increasing edemasize.

CONCLUSIONS

The pervasive sensing of Things is still in itschildhood, but it already offers an unprecedent-ed opportunity to mix together complementaryexpertise and stimulate fast progress in edgedisciplines. The presented examples and ideasshow only that low-cost passive sensing is physi-cally feasible, but there is a need for significantresearch to master the design of “sensitiveantennas” and, more generally, to fully under-stand the many implications of the engineereduse of a multiplicity of radio identificationobjects. When RFID microchips, augmented intheir electronic and software capabilities, arefully perceived as atomic components of a morecomplex distributed system, just like resistors,capacitors, and diodes are in microelectronics,new devices will be conceived, working at thesame time as sensors, actuators, radios, andmedia generators.

ACKNOWLEDGMENTSThe ideas and results described in this articlewould not be feasible without the help of mypresent and past collaborators L. Mattioni, C.Calabrese, C. Occhiuzzi, F. Amato, S. Cippitelli,S. Caizzone, Sabina Manzari, and GiordanoContri. Moreover, I would like to thank G.Manzi of NXP for technical and equipment sup-port.

Figure 6. Monitoring of brain edema by a self-sensing implanted tag, used in analog mode, following cancer removal surgery.

Edema progressiont1t0

0.7

0.65

0.75

0.8

0.85

0.9

0.95

1

t2 t3

870 MHz2.45 MHz

Insulateddipole

Edema Blood Insulatedtag

z x

t2t1

t0

t3

y

PR←T

MARROCCO LAYOUT 12/9/10 11:06 AM Page 16

IEEE Wireless Communications • December 2010 17

REFERENCES[1] D. Preuveneers and Y. Berbers, “Internet of Things: A Con-

text- Awareness Perspective,” The Internet of Things: FromRFID to the Next-Generation Pervasive Networked Systems,L. Yan et al., Eds., Auerbach, 2008.

[2] M. Philipose et al., “Battery-free Wireless Identificationand Sensing,” IEEE Pervasive Computing, vol. 4, no. 1,2005, pp. 37–45.

[3] B. Sterling, Shaping Things, MIT Press, 2006.[4] G. Marrocco, L. Mattioni, and C. Calabrese, “Multiport

Sensor RFIDs for Wireless Passive Sensing Of Objects —Basis Theory and Early Results,” IEEE Trans. AntennasPropagation, vol. 56, no. 8, Aug. 2008, pp. 2691–702.

[5] F. Amato and G. Marrocco, “Self-Sensing Passive RFID:from Theory to Tag Design an Experimentation,” Euro.Microwave Conf., Roma, Italy, 2009.

[6] G. Marrocco, “RFID Grid: Part I — Electromagnetic The-ory,” to appear, IEEE Trans Antennas Propagation.

[7] R.A. Potyrailo et al., “Selective Quantitation of Vaporsand Their Mixtures Using Individual Passive Multivari-able RFID Sensors,” IEEE Int’l. Conf. RFID 2010, Apr.2010, pp. 22–28.

[8] K. Opasjumruskit et al., “Self-Powered Wireless Temper-ature Sensors Exploit RFID Technology,” IEEE PervasiveComputing, vol. 5, no. 1, Jan.–Mar. 2006, pp. 54–61.

[9] R. Bhattacharyya, C. Floerkemeier, and S. Sarma, “RFIDTag Antenna Based Sensing: Does Your Beverage GlassNeed a Refill?,” IEEE Int’l. Conf. RFID 2010, Apr. 2010,pp. 126–33.

[10] S.E, Sarma, C. Floerkemeier, and R. Bhattacharyya,“Towards Tag Antenna Based Sensing — An RFID Dis-placement Sensor,” IEEE Int’l. Conf. RFID 2009, Orlan-do, FL, 1999, pp. 95-102.

[11] S. Caizzone, C. Occhiuzzi, and G. Marrocco, “RFIDRadio-Sensor Integrating Shape-Memory Alloys,” IEEEAntennas and Propagation Soc. Symp., Toronto, Cana-da, 2010.

[12] C. Occhiuzzi and G. Marrocco, “The RFID Technologyfor Neuroscience: Feasibility of Limbs’ Monitoring inSleep Diseases,” IEEE Trans. Info. Tech. in Biomedicine,vol. 14, no. 1, Jan. 2010, pp. 37–43.

[13] C. Occhiuzzi and G. Marrocco, “Pervasive Body Sens-ing: Implanted RFID Tags for Vascular Monitoring,” IEEEAntennas and Propagation Soc. Symp., Toronto, Cana-da, 2010.

[14] C. Occhiuzzi and G. Marrocco, “Sensing the HumanBody by Implanted RFID Tags,” Euro. Conf. Antennasand Propagation, Barcelona, Spain, 2010.

BIOGRAPHYGAETANO MARROCCO (marrocco@disp.uniroma2.it) receivedhis Laurea degree in electronic engineering and Ph.D.degree in applied electromagnetics from the University ofL’Aquila, Italy, in 1994 and 1998, respectively. Since 1997he has been a researcher at the University of Rome TorVergata, Italy, where he currently teaches antenna designand bioelectromagnetics, manages the Antenna Laborato-ry, and is an advisor in the Geo-Information Ph.D. pro-gram. His research is mainly directed to the modeling anddesign of broadband and ultra-wideband antennas andarrays as well as of sensor-oriented miniaturized antennasfor Biomedicine, Aeronautics and RFID. He holds twopatents in broadband naval antennas and structural arrays,and three patents in sensor RFID systems. He was GeneralChairman of the first Italian multidisciplinary scientificworkshop on RFID: RFIDays-2008: Emerging Technology forRadiofrequency Identification. In 2010 he was co-chair ofthe RFIDays-2010 International Workshop in Finland. Cur-rently he serves as Associate Editor of IEEE Antennas andWireless Propagation Letters.

MARROCCO LAYOUT 12/9/10 11:06 AM Page 17