Research Article A Measurement Study of BLE …nsl.cau.ac.kr/~jpaek/docs/ibeacon-mis-published.pdfGoogle s recent release of their own open beacon format called Eddystone (Eddystone

Research ArticleA Measurement Study of BLE iBeacon andGeometric Adjustment Scheme for Indoor Location-BasedMobile Applications

Jeongyeup Paek1 JeongGil Ko2 and Hyungsik Shin3

1School of Computer Science and Engineering Chung-Ang University Seoul Republic of Korea2Department of Software and Computer Engineering Ajou University Suwon Republic of Korea3School of Electronic and Electrical Engineering Hongik University Seoul Republic of Korea

Correspondence should be addressed to Hyungsik Shin hyungsikshinhongikackr

Received 12 August 2016 Accepted 3 October 2016

Academic Editor Hyun-Ho Choi

Copyright copy 2016 Jeongyeup Paek et alThis is an open access article distributed under the Creative CommonsAttribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Bluetooth Low Energy (BLE) and the iBeacons have recently gained large interest for enabling various proximity-based applicationservices Given the ubiquitously deployed nature of Bluetooth devices including mobile smartphones using BLE and iBeacontechnologies seemed to be a promising future to comeThiswork started offwith the belief that this was true iBeacons could provideus with the accuracy in proximity and distance estimation to enable and simplify the development of many previously difficultapplicationsHowever our empirical studieswith three different iBeacon devices fromvarious vendors and two types of smartphoneplatforms prove that this is not the case Signal strength readings vary significantly over different iBeacon vendorsmobile platformsenvironmental or deployment factors and usage scenarios This variability in signal strength naturally complicates the process ofextracting an accurate locationproximity estimation in real environments Our lessons on the limitations of iBeacon technique leadus to design a simple class attendance checking application by performing a simple form of geometric adjustments to compensatefor the natural variations in beacon signal strength readings We believe that the negative observations made in this work canprovide future researchers with a reference on how well of a performance to expect from iBeacon devices as they enter their systemdesign phases

1 Introduction

Bluetooth is ubiquitous in todayrsquos everyday life used inlaptops mobile devices keyboards headsets andmany otherconsumer electronics for wire replacement Bluetooth LowEnergy (BLE) is an extension of theBluetooth standard in ver-sion 40 that enables low-power low-cost short-range wirelesscommunication [1 2] It has significantly reduced its powerconsumption among other extended features compared toclassical Bluetooth versions and is now possible to run BLEdevices for several months to even a couple years on a coincell battery Such aspects make BLE ideal for applicationsrequiring infrequent or periodic transfers of small amountof data thus it can be applied in a wide range of medicalindustrial and consumer applications

iBeacon is a BLE-based proximity sensing frameworkproposed by Apple [3] which allows amobile device to detect

its proximity to an iBeacon station (and possibly its location)by knowing how close it is to (usually wall-mounted) low-complexity low-cost BLE transmitters the iBeacons EachiBeacon transmits periodically short identification framesthat are received by a mobile BLE device to estimate thedistance between the mobile device and the iBeacon usingreceived signal strength indicator (RSSI) Based on this detec-tion of proximity iBeacons provide automatic and locationspecific triggering of services on the mobile device suchas advertising coupons or route guidance Although thespecifications of the iBeacon were initially proposed by Apple(iOS7 or later) it is just one way of utilizing BLErsquos proximityfeatures and thus it is (and can be made) compatible withother devices (eg Android 43 or later) that use BLE Moregenerally any BLE device can use the same concept toprovide similar proximity functionality Technically iBeacon

Hindawi Publishing CorporationMobile Information SystemsVolume 2016 Article ID 8367638 13 pageshttpdxdoiorg10115520168367638

2 Mobile Information Systems

technology is a subset of BLE beacons but iBeacon and BLEproximity beacons became nearly synonyms today despiteGooglersquos recent release of their own open beacon formatcalled Eddystone (Eddystone is a protocol specification byGoogle that defines a Bluetooth Low Energy message formatfor proximity beacon messages) [4]

Note here that the original goal of BLE beacons wasto provide proximity-based application services possibly toextend the features to coarse-grained location-based applica-tions Given that BLE beacons solely focus on RSSI measure-ments combined with lessons learned from the long historyof research in RF-based localization it is somewhat expectedto be able to achieve only a limited accuracy in indoorlocalization services It is well known that the variability of RFsignals with respect to the indoor environment complicatesthe accurate estimation of the mobile devicesrsquo location or itsrelative distance to the BLE beaconing transmitter deviceHowever most prior work on BLE-based indoor localizationmerely states the possibility of error due to signal variationbut does not quantify them and their reported results showonly the positive side of the picture For example the workin [5] reports a positioning accuracy of 053 meters but thisworkwas bounded by a 9times 10m2 areaThework in [6] reportsa 12-meter accuracy with fingerprinting but only within a474 times 159m2 area Recently work in [7] shows that distanceestimation error can increase significantly with longer dis-tance when themeasured signal RSSI vary over time but doesnot show other dimensions of possible error

In this work we perform an extensive set of experimentsto quantify the impact of various indoor obstacles on the BLEsignal variance Specifically we show that different iBeacondevices from different vendors along with the paired mobiledevice platform (eg iOS or Android) can give significantimpact on the RSSI measurements to complicate the processof designing a generally acceptable distancelocation esti-mation model Furthermore practical factors such as thedeployment height or environmental factors also introduce asignificant amount of impact on the RSSI values to introducean additional level of complexity in designing highly accuratelocation estimation systems using iBeacons Lessons fromthese empirical studies lead us to design an application that isldquopractically designablerdquo Specifically we design and evaluate aclass attendance monitoring application system by combin-ing distance estimationsmade from locally accessible iBeaconsignals and an estimation adjustment scheme designed fromthe lessons from our empirical signal measurements

This paper is structured as follows Section 2 introducesthe relationship between RSSI readings and the estimateddistances and then motivates our empirical studies on therelationship Section 3 describes our approaches to theempirical studies and experiment settings including variousparameters that we change during our studies Section 4 isthemain section and it describes ourmeasurements data andillustrates our analysis results based on the data Section 5presents an application case study an automatic attendancemonitoring system which is based on our empirical stud-ies This system also proposes a novel geometric approachto overcome the limitations of practical iBeacon systems

Section 6 reviews prior and related work and finally Section 7concludes the paper

2 Limitations of Bluetooth Low Energy-BasedLocalization Systems

As Bluetooth Low Energy (BLE) and Applersquos iBeacon pro-tocols begin to see widespread adoption and deploymentindoor proximity andor location enabled applications arequickly penetrating into our everyday lives and consumerspace iBeacon has been installed in Apple stores in theUS to provide product information as well as customerservices and the National Football League (NFL) has usediBeacons in the MetLife Stadium to provide personalizedadvertisements to football fans during the Super Bowl game[8] Furthermore iBeacon can be used to track luggage atthe airport [9] provide interactive experiences to visitors inmuseums [10] plan evacuation paths in emergency guidingsystems [11] track patients in emergency rooms [12] guideindooroutdoor routes [13 14] and detect the occupancy ofrooms [15 16] The application space to which iBeacons andBLE proximity applications can be applied is extremely largeThese applications all use proximity measurements based onthe received signal strength indicator (RSSI) as the sole sourceof information thus these applications exploit informationon how close a mobile device is to an iBeacon rather thantrying to pinpoint the exact and absolute location of themobile device

The original purpose of iBeacon technology is to pro-vide proximity-based application services and coarse-grainedindoor location positioning and navigation based solely onproximity Note that proximity (being near to some object) isrelated to the location (where you are) but is not necessarilyidentical A location (or position) ismore than just proximityand it is an absolute value usually defined by some coordinatesystem (eg latitude and longitude for GPS) Recentlyhowever there have been several proposals to use BLE forindoor positioning extending several prior works that useother technologies such as ultrasonic sound infrared WiFiGSM RFID IEEE802154 and earlier versions of Bluetoothfor indoor localization [17ndash22] (there are many more relatedprior work on indoor localization that uses other wirelesstechnologies but we list only a subset here for brevity and tokeep the reference list focused on BLE only since there are toomany of them) Indoor positioning and localization are stillan active area of research and BLE is extending the literatureusing its new wireless features

There are several approaches (beyond simple proximity-based ones [12]) that can be applied to BLE-based indoorpositioning In all cases an iBeacon will periodically broad-cast an advertisement packet containing a unique ID anda calibrated RSSI value (1198770) corresponding to one-meterdistance (1198890) This value allows us to determine the distancebetween an iBeacon and a device using a model in (1) where120574 is a calibration parameter for the path loss exponent Basedon the RSSI or the estimated distances between iBeacons anda mobile device a weighted average model [5] trilateraliza-tiontriangulation model [23] or fingerprinting [7 24] can

Mobile Information Systems 3

Estimote GELO Wizturn Pebble

Figure 1 Three different types of iBeacons used for the experiments top row shows the images of the Estimote GELO and Wizturn PebbleiBeacons Bottom row shows the icons of their corresponding smartphone applications

be applied for the localization process The estimations canalso be combined with Pedestrian Dead Reckoning (PDR)or filtering techniques to improve tracking or to reducepositioning errors [6 25] Of course all of these approachesassume that all installed iBeacons possess their predefinedlocation information including their exact coordinates

RSSI = 1198770 minus 10120574 log( 1198890) 997904rArr = 10(1198770minusRSSI)10120574

(1)

If distances are used instead of raw RSSI values forpositioning RF propagation model such as (1) is used toconvert RSSI readings into distances The main challengehere is the sensitivity of RSSI values to environment changessuch as object movements or obstacles which result in signalpropagation or radio map changes It is obvious that theaccuracy and efficiency of this distance estimation (and thusthe location estimation) depend heavily on the accuracy ofthe measured RSSI values and the model used to deriveand calculate the distance and also significantly influencedby the surrounding environment Therefore in this workwe perform a detailed empirical measurement study onthe locationdistance estimation performance based on RSSImeasurements for different types of iBeacon devices andmobile device platforms We consider various environmentalfactors to gather a practical perspective on the performancelimitations of iBeacon technologies

3 Approach and Experiment Setup

Wecarried out experiments in two different environments anopen soccer field of a university campuswithout any obstaclesand anunblocked corridor of an office building to ensure line-of-sight (LOS) signal propagation In these environments weused three different types of iBeacons Estimote [26]WizturnPebble [27] and GELO [28] iBeacons to examine differences

in performance among vendors Furthermore we used twodifferent mobile devices iPhone 5 with iOS 712 and GalaxyRound (SM-G910S) with Android 442 to verify any existingparticularities Mobile applications used for data collectionare the applications provided by the respective iBeacon ven-dors ldquoEstimote apprdquo for Estimote iBeacons ldquoGELO toolkitapprdquo forGELO iBeacons and ldquoWizturn beaconmanagerrdquo appfor Wizturn Pebble iBeacons (Figure 1)

The default configuration parameters used in the experi-ments are as followsThe transmission (TX) power of iBeaconwas set to the maximum of 4 dBm unless stated otherwiseto verify the maximum distance that can be covered byany iBeacon We also experiment with minimum TX power(which is minus19simminus23 dBm depending on the device) to observethe lower-bounds of connectivity The advertising intervalof the iBeacons was set to 950 milliseconds which was thefactory default for all three devices Device placement heightfrom ground was set to 12 meters for both the iBeacons andthe mobile phones unless stated otherwise We selected thisheight of 12 meters to imitate the height at which a user holdstheir mobile device in their hands Finally the orientation ofthe iBeacon was set to face the direction of LOS towards themobile device Using these configurations a mobile applica-tion continuously detected signals from iBeacons and loggedtheir RSSI measurements while the distance between theiBeacon and the mobile device was increased from 1 meter tothemaximumdistance and recordedmanually until reachingthe maximum distance Here we note that the maximumdistance is defined to be the farthest distance where the RSSIvalue can be detected by the device At each location (egrelative distance to the iBeacon) we took twenty RSSI mea-surements for each configuration

4 Measurement Data Analysis

In this section we focus on identifying and quantifyingthe various characteristics related to the performance of


iOS

25 percentile75 percentileMean

Android

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

Figure 2 Distance versus RSSI plot for iOS and Android comparison (Estimote 4 dBm TX power 12m height and LOS)

iBeacons for localization applications by carefully analyzingthe extensive measurement data that we have collected

41 Difference between iOS andAndroid Phones Wefirst startby comparing the differences between iPhone 5 with iOS712 and Galaxy Round (SM-G910S) with Android 442 inreceiving iBeacon signals The experiment was conductedoutdoors in an open soccer field using an Estimote beaconwith 4 dBm TX power at 12-meter height in LOS Figure 2presents our results from this experiment where the dottedline represents the average RSSI reading at each distance andthe boxes represent the 25 and 75 percentile values along withtheir minimum and maximum error bars

From this result we can identify several interestingfacts First compared to the 100 meters for the Androidplatform iOS showed notably shorter maximum distancesof 85 meters Note that the 100 meters observed for theAndroid is not the actual maximum distance given that100m was the maximum length we could achieve for LOSlinks in our environment Therefore the difference betweenthe maximum distances of the two platforms turned outto be very large Second RSSI readings on Android phonedecreased more gradually whereas iOS showed a suddendrop in RSSI after sim10 meters Lastly RSSI readings onthe Android platform had more temporal variation thaniOS While it is difficult to conclude whether the hardware

specification shows any difference or this is an effect ofthe software platform it is known that there is a differencebetween iOS and Android in scanning and sampling iBeaconadvertisements [16] More importantly the RSSI readingsdiffer between different phones (regardless of it being animpact of the operating system or the hardware) and weshould point out that this effect itself will have significantimplications for distance estimation and localization causingdifficulties for the application developers to generalize infor-mation extracted from the collected RSSI values

42 Effect of Device Placement Height In this experimentwe examine the impact of the iBeacon installation heightfrom the ground on RSSI For this we place an Estimotebeacon on the ground (eg 0-meter height) and measurethe RSSI readings using the Android mobile phone Wecompare this value with the case when the same beaconis installed at a 12-meter height Figure 3 shows the resultfrom this experiments We can first notice that the values aresubstantially different for the two experiment casesWhen theiBeacon is placed on the groundwhile all other configurationsare identical the maximum distance reduces from 100+meters to only 12 meters with significant and drastic drop inRSSIThis result implies that the placement of iBeacon is veryimportant when designing and deploying an iBeacon systemDistance estimation and thus the localization accuracy can be


Estimote ground

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 10 20 30 40 50 60 70 80 90 100

Real distance (m)

Estimote 12m

Figure 3 Distance versus RSSI plot that shows the impact of deviceplacement height from ground (Estimote 4 dBm TX power andAndroid)

significantly impacted by the precise placement of iBeaconsand planning the installation carefully as part of the sitesurvey will be very important Based on our experimentaldata the optimistic advertisements that declare ldquoease ofdeploymentrdquo and ldquoplace nodes anywhererdquo may not be validwhen targeting precise accuracy on the iBeacon systems

43 Difference between iBeacons from Different Manufactur-ers Next we compare the RSSI reading differences betweendifferent iBeacon products for three different manufacturersEstimote GELO and Wizturn Pebble iPhone5 was used asthe mobile device and both the beacons and the mobiledevices were installed at 12 metersrsquo height in an outdoorsoccer fieldOne important thing to note is that themaximumtransmission power supported by GELO iBeacon was 0 dBmwhereas Estimote andWizturn Pebble allowed configurationup to 4 dBm We used respective maximum powers for eachdevice to examine the maximum distance coverage Ourinitial intuition was that GELO iBeacon will exhibit 4 dBmlower RSSI values compared to other beacons at identicaldistances and thus it will result in a shorter maximumtransmission distance

Surprisingly Figure 4 shows that unlike our expecta-tion GELO along with Wizturn beacons shows a maxi-mum distance of 100+ meters far exceeding Estimotersquos 85meters Furthermore the RSSI readings were all similarbetween distinct iBeacon types despite the 4 dBm differencein transmission powerThis implies several interesting pointsFirst the configured TX power difference does not directlytranslate into received RSSI reading difference which makesthe calibration process for distance estimation (and thusindoor localization)more challenging when the transmissionpower must be adjusted for the given environment SecondiBeacons from different vendors exhibit different maximumdistances and higher maximum TX power configurationdoes not necessarily lead to longer maximum transmissiondistances Lastly due to the aforementioned reasons it willbe very challenging for application developers and service

0 20 40 60 80 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

Wiztern meanEstimote meanGELO mean

Figure 4 Distance versus RSSI plot for three different types ofiBeacons Estimote GELO and Wizturn Pebble (4 dBm0 dBm TXpower 12m height LOS and iOS)

providers to design an iBeacon-based system with heteroge-neous iBeacons since multiple sets of calibration parameterswill be required for each pair of iBeacon and mobile devicetype Again these results imply that designing a reasonablyreliable iBeacon-based localization system can be extremelychallenging

44 Reducing to Minimum TX Power We now focus on thefact that not all systems fully benefit from the use of maxi-mum transmission power For example if we are designingan indoor localization system based on trilateralization orfingerprinting method we would prefer a higher transmis-sion power to cover a larger area andor achieve denserdeployments while using fewer number of iBeacons (egfor cost reasons) However when building a system basedon only the proximity function reducing the transmissionpower to the minimum just enough to cover the target region(eg in front of a store or a product for advertisement) whilekeeping the energy cost low to extend the lifetime of thebatteries can be a reasonable system design option With thismotivation in mind we conducted an experiment using therespectiveminimumTXpower for the three types of beaconsTheminimum TX power configuration allowed on EstimoteGELO andWizturn Pebble beacons was minus20 dBm minus23 dBmand minus19 dBm respectively

Figure 5 depicts the results from this study First weobserve the drastic drop in the maximum distance (note thedifferent 119909-axis values compared to the previous figures)Estimote with minus20 dBm showed a maximum distance of 8meters and Wizturn Pebble with minus19 dBm gave 8 meterswhile the RSSI from a GELO beacon with minus23 dBm TXpower was detectable within only 03 meters The secondobservation is that while the RSSI reductions for Estimoteand Wizturn Pebble approximately matched their reductionin TX power (24 and 23 dBm resp) this was not true for theGELO beacons GELO beacons showed RSSI reductions of



minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 1 2 3 4 5 6 7 8 9

Real distance (m)

Figure 5 Distance versus RSSI plot when minimum transmissionpower is used by three different types of iBeacons (minimum TXpower 12m height LOS and iOS)

44ndash58 dBmwhile the TX power was reduced only by 23 dBmAgain as our previous results concluded these differencescan cause significant complication when designing a systemwith multiple types of iBeacons different TX powers differ-ent reductions in RSSI and different maximum distances willmake the calibration process practically impossible unlessonly a single TX power configuration and a homogeneousiBeacon from a particular vendor is used for designingthe entire system This may be a plausible approach in theinitial system design phase but with the system scalingovertime for supporting additional regions or new use casesconsiderations for heterogeneous systems are a must

45 Indoors versus Outdoors iBeacon is often referred toand is well known as an ldquoindoor proximitylocation systemrdquobecause amajormotivator for its development was to provideindoor location awareness where the alreadywidely usedGPSis unavailable However physically there is no reason whyiBeacon cannot be used outdoors In fact there are man-ufacturers (eg GELO) that provide durable weather-proofiBeacons for outdoor use Through our next experimentswe wanted to compare the performance of iBeacons in theindoor and outdoor environments For this we performedadditional experiments in an unblocked 3 metersrsquo widecorridor of a university building Here we use the Estimotebeacons and an Android phone again at 12 meters heightand 4 dBmTXpower so that we canmake direct comparisonswith our experiments performed outdoors

Figure 6 presents our results To our surprise after sim25meters in distance the RSSI readings remained almost steadyor even increased slightly with distance The maximumdistance is shown as 50 meters only because our corridorwithin the building was 50 metersrsquo long and the actualmaximum distance is projected to be longer with morephysical space To verify our unexpected results we alsoconducted additional experiments with the Wizturn Pebble

Indoor meanOutdoor mean

0 10 20 30 40 50

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

Figure 6 Distance versus RSSI plot for indoor versus outdoorcomparison (Estimote 4 dBm TX power 12m height LOS andAndroid)

and GELO beacons and found similar results (plots omittedfor brevity)The only plausible explanation for such behavioris the multipath effect caused from the walls and ceiling ofthe corridor What is more important here is the fact thatthe RSSI value did not decay for a wide range of distances(25ndash50m in our case) This means that the distance modelin (1) is no longer valid and any indoor localization methodthat relies on distance estimation and trilateralization can facelarge errors in the order of tens ofmeters In bolderwords it isnot possible to design an accurate indoor localization systemwith iBeacons using simple distance estimation

46 Effect of WiFi on iBeacon Signal Reception Anotherissue when using iBeacons for indoor applications is theinterference effect of WiFi signals which operates on thesame 24GHz ISM frequency band BLE was designed to usechannels 37 38 and 39 for advertisement (out of 40 totalchannels and the remaining 37 channels for data communi-cation) and their allocated frequencies are designed to avoidthe most popular WiFi channels 1 6 and 11 (cf Figure 7)Unfortunately the WiFi AP deployment at our universitycampus used channels 1 5 and 9 andWiFi channel 5 happensto interfere with BLE channel 38 This is not an unusual caseand can occur generally in anyWiFi environmentThereforewe performed an additional experiment to quantify theconsequences of WiFi-oriented collisions

For this experiment we placed an Estimote beacondirectly under a WiFi AP and measured the RSSI readingsfrom an Android phone 5 meters apart (cf Figure 8) Weshow a subset of observations in Figure 9 where the 119909-axisrepresents the sequence of measurements (eg time) andthe 119910-axis is again RSSI The dotted line with squares showsthe readings when WiFi AP was powered off and the solidline with circles shows the readings when the WiFi AP is innormal operation The shaded rectangle regions along withthe disconnection in the solid line represent the cases when


2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z

Wi-Fi Ch 1 Wi-Fi Ch 6 Wi-Fi Ch 11

LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y

Figure 7 Channel frequency allocation for BLE and its relation to WiFi channels 1 6 and 11

12m12m

5m

Figure 8 Setup for WiFi interference experiment An EstimoteiBeacon was placed vertically below anWiFi AP that uses channel 5and the receiving Android mobile smartphone was placed 5 metersaway Both the iBeacon and the mobile device were 12 meters fromthe ground

wo WiFiw WiFi

Sequential measurementsminus75

minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons

Figure 9 Time versus RSSI plot for a sequence of beacon receptionthat shows reception failures (packet losses) and reduction in RSSI(identical configuration as Figure 8)

there was no iBeacon RSSI reading on the mobile phone dueto the lack of packet reception

We make two important observations from this experi-ment First when theWiFi AP is active the beacon receptionratio dropped to around 75 even at a short distance of5 meters a distance where a 100 beacon reception canbe achieved without the WiFi AP Second even when aniBeacon advertisement was successfully received the RSSIreadings showed significantly lower values (eg more than10 dBm reduction) for sim53 of the beacons Note that thepacket loss of low-power BLE beacons under high-powerWiFi interference is somewhat expected but a consistentreduction in RSSI is an unexpected phenomenon that can beregarded as another type of wireless signal ldquogray regionrdquo [29]These findings suggest that with the widely deployed andubiquitously utilized WiFi in todayrsquos indoor environmentsan iBeacon system can be significantly affected with reducedreliability and higher estimation errors

47 Effect of Obstacles In this experiment we examine theeffect of physical obstacles on the iBeacon signal receptionand compare against the LOS case Specifically here weconsidered six different cases three cases with obstacles of aniron door a wooden door and a window each one case forsignal amplification using a sheet of aluminum foil and twocases where the mobile phone is covered by hand or paperIn this experiment we used the Estimote beacon paired withan iPhone5 and the distance between the iBeacon and themobile phonewassim3meters Again the height was 12metersand the TX power was 4 dBm

We present the results from this study in Figure 10 Asexpected the iron door blocked the signal drastically with anRSSI drop of sim20 dBm but the wooden door or window alsohad some (nonnegligible) effect of sim3 to 8 dBm drop whilecovering the phonewith a pile of paper caused asim6 dBmdrop



Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI

Figure 10 Average RSSI plot for different types of obstacles(Estimote 4 dBm TX power 12m height 3m distance and iOS)

More interestingly however covering the mobile phone withthe hands of the user resulted in a noticeable reduction ofsim30 dBmThis suggests that in an indoor localization systemdespite undergoing a careful deployment phase and extensivecalibrations the distance andor position estimation can havesignificant errors just because the user is holding the phonetightly in hisher hands or the environment changes naturally(eg door statuses) This is another challenge in designingan iBeacon-based system since mobile devices will inevitablybe held by the usersrsquo hands and obstacle statuses will changeactively in an indoor environment

As a final note we were able to amplify the beacon signalby using a sheet of aluminum foil behind the iBeacon whichresulted in an RSSI increase of 6 dBm While it is unlikelythat iBeacon deployment will intentionally be wrapped ina sheet of aluminum foil this result suggests that someenvironmental artifact or ornament can also unexpectedlycause such an effect implying that obstacles not only reducethe RSSI levels but can also amplify the signal strength

48 Distance Estimation Using the Curve-Fitted ModelFinally we now take all the measurement data from thefour iBeacon-mobile device pairs combination of Estimoteand Wizturn Pebble beacons with Android and iOS phonesand fit the data on to the model in (1) using the least-meansquare method to calculate the estimated distance based onthe RSSI We note that all data are for the outdoors LOS 12-meter height and 4 dBm TX power experiments We thenuse this derived model to compare the real distance to theestimated distance as in Figure 11 One point to take awayfrom this plot is the fact that distance estimation can showsignificant errors even under LOS conditions due to highsignal strength variations Furthermore the errors increaseas distance increases (eg as RSSI gets closer to the receptionsensitivity) While we omit additional figures for brevity we

75 percentile25 percentile

Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)

Figure 11 Real distance versus estimated distance plot where thepreviously collected RSSI data was used to curve fit the model in (1)and empirically determine the model parameter

note that we can see similar plots even with data from a singlepair of iBeacon-mobile device Quantitatively the errors canincrease from tens of meters to even hundreds of metersThis means that when designing an iBeacon-based systema system designer should either cope with the large distanceestimation errors (and hence positioning errors) or denselydeploy a large number of iBeacons to reduce the errorswhich can threat the ldquolow-costrdquo argument in iBeacon systemsUnfortunately as per our experimental results neither issatisfactory

5 Application Case Study AutomaticAttendance Checker System

As our experimental results show the main challenge inusing iBeacon signals for accurate indoor positioning is thevariability of RSSI readings and its sensitivity to environmentchanges which result in drastic changes in signal propagationOur findings in Section 4 show that the RSSI value (andthe corresponding signal propagation model for estimatingdistances) varies significantly among iBeacons from differentvendors (eg Estimote versusWizturn Pebble versus GELO)mobile device types (iOS versus Android) height of thedevice installation from the ground indoor or outdoorenvironmental factors and physical obstacle types Overallthese experiences suggest that with such iBeacon devices weshould take these limitations and performance characteristicsinto consideration when designing applications and applyimprovement schemes with respect to the intuitions collectedfrom such real-world pilot deployment experiences

In this case study we extend the line of possible iBeaconapplications by proposing an automatic attendance checkersystem that automatically checks the attendance of a collegestudent to her classes in a university However simpleproximity is not enough to accurately determine whethera student is inside the classroom or not since the beacon


HTTPS

WebDB server

Figure 12 Overall system architecture of our iBeacon-based automatic attendance checker system

signal can be received behind the walls of the classrooms aswell Thus we use the trilateration-based position estimationto make a decision on whether the student is attending theclass in the room or not To compensate for RSSI readingerrors due to unexpected obstacles (eg users holding theirmobile phone tightly or their phones being in a bag) wemake simple yet novel geometric adjustments in trilaterationcalculation to improve the detection accuracy On a system-level perspective our system is composed of not only iBea-cons and mobile devices but also a database server whichmaintains and handles information about students classesclassrooms time-table and attendance information As welater show the complete system allows for a fully automatedattendance checking mechanism to take place with 100accuracy under the circumstance that the students haveenabled our application on their smartphones Our systemnot only saves time wasted for manual attendance checkingduring classes but also prevents attendance cheatings since itis very unlikely that a student will give hisher smartphone toa friend just for attendance checking

51 System Architecture Figure 12 shows the overview of oursystem architecture Three iBeacons are deployed in eachclassroom with unique identification numbers (major-minorpair) for each iBeacon and the relative x-y coordinate ofeach classroom is preconfigured We developed an Androidapplication for our system which receives beacon signalsin the background and communicates with the webDBserver over the Internet for attendance checking withoutuser intervention The server maintains all the necessaryinformation to compute and confirm attendance in thedatabaseThe recorded attendance data can be viewed on oursystemrsquos website by the faculty and staff members as well ason the studentsrsquo mobile devices

Specifically we used an Oracle DB to maintain infor-mation regarding students professors classes time tablesand most importantly deployed iBeacons including their x-y coordinates (relative within each classroom) and identifi-cation numbers This information is relatively static in thesense that student information changes only when they joinand login for the first time and classtime-table informationchanges only at the beginning of each semester for which weuse web-crawling to automatically populate from the univer-sity website To store the actual attendance information (theclass date time and x-y coordinate within classroom) foreach student we used theMongoDB to handle the frequentlyupdated data Figure 13 shows our database structure

Using the aforementioned system the automatic atten-dance check process operates as follows When a studententers a classroom the smartphone application will receivebeacon advertisements from three or more iBeacons (theremay be signals from nearby classrooms) At this point theapplication sends the list of identification numbers (UUIDmajorminor values [3]) andRSSI readings from the iBeaconsto the server and queries the server for verification ofattendance Given this information the server will first checkthe classroom information of each iBeacon and validatesit against the classroom at which the student should beattending at the given time instance If the classroom infor-mation indicates that the student is in a wrong classroomor if there are only two or less number of beacons detectedfrom the target classroom then the server returns FALSE forattendance If no beacon is detected attendance is FALSE bydefault For all other cases the server uses the relative x-ycoordinates (within the target classroom) of the iBeacons toestimate a position of the mobile device and checks whetherthe device is within the boundaries of the classroom If so


Figure 13 Databased structure in our iBeacon-based automatic attendance checker system server

the server returns TRUE for attendance and records it inthe database This process is repeated periodically so thatattendance can be checked during the class as well This isto account for students who are a few minutes late as wellas those students who leave early during class However toreduce the unnecessary power consumption from the use ofBLE for the duration when a student does not have a classto attend her time-table can be locally cached on the mobiledevice to enable attendance checking only when needed

Note that the x-y coordinates stored in the database areonly relative and local to the target classroom This is incontrast to other indoor positioning systems where global x-y coordinate within the entire target area (eg whole floorof the building to map) is required and is made possiblegiven that we maintain preknowledge on which classrooma student should be at a given time This greatly simplifiesthe system architecture not only in terms of estimatingthe position of the mobile device but also in terms ofiBeacon deployment replacement and reconfiguration Inother words if systems are mostly concerned for a small geo-graphical region iBeacon placements can be independently(and better) customized and optimized We provide detaileddescriptions on such calibration procedures in the followingsection

52 Geometric Adjustment for Improved Accuracy The mainchallenge in using iBeacon signals for accurate indoorpositioning is the variability of RSSI readings and theirsensitivity to environment changes such as obstacles or userhandling of the mobile device Our findings earlier show

that RSSI values can drop significantly when a beacon signalis received behind an obstacle and the amount of signalattenuation varies depending on the obstacle types (egwindow woodenmetal door walls etc) Furthermore theheight of the mobile device from the ground also affects theRSSI (height of the iBeacon also introduces a high impact butiBeacons are usually wall-mounted and its height is fixed intypical scenarios) More significantly the signal attenuationdue to holding the mobile device tightly in usersrsquo hands(which happens often) can be asmuch as 30 dBm Such a highvariation is a significant challenge when using trilaterationfor position estimation However we see one commonalityfrom the aforementioned cases the signal attenuates (RSSI islower than themodel for a given distance) and it is extremelyrare to see higher RSSI than the (best-case) line-of-sightenvironment Using this intuition we use a simple yet novelgeometric adjustment scheme to improve accuracy ratherthan trying to calibrate the model based on highly varyingRSSI values

For example Figure 14 shows one example of an exceptioncase where the distance estimation from three iBeaconsresults in three RSSI-coverage distance circles (circles drawnby the estimated distance as radii for trilateration) withoutany intersection one circle is completely enclosed by thebiggest circle and a third circle is completely outside of thebiggest one In this case standard trilateration calculationis not possible However our intuition is that the distanceof the larger circle (larger distance from higher RSSI value)is closer to the estimation model and the signals of thetwo smaller circles have been attenuated due to some reason


Figure 14 An example of possible exception scenario where trilat-eration calculation will fail due to high variation in RSSI readingthus causing error

Original distance

circle

Adjusted distance

circle

Figure 15 [Case 1] where two distance circles (circles representingthe estimated distance using RSSI) from two iBeacons are nonin-tersecting because the sum of two distance estimations are smallerthan the distance between the two iBeacons This results in nointersection point for trilateration calculation Thus we graduallyincrease the sizes of the circles until they intersect

such as obstacles Based on this intuition our approach isas follows For each pair of circles (out of three pairs fromthree iBeacons) we first check whether [Case 1] two circleshave no intersection points (upper figure of Figure 15) orwhether [Case 2] one circle is completely enclosed by anothercircle (left figure of Figure 16) If neither is true then there isnothing else to consider for that specific pair of RSSI-coveragedistance circles

If a pair of circles have no intersection points (ie[Case 1]) we first increase the size of the smaller circlewith anincrement of 1 meter until there are two intersection points orup to the point where the two circles become the same sizesIf no intersection points are created even after two circles areof the same sizes then we increase the size of both circles

Original distance circle Adjusted distance circle

Figure 16 [Case 2] where a distance circle (a circle representingthe estimated distance using RSSI) from one iBeacon is completelyenclosed by another distance circle (from another iBeacon) result-ing in no intersection point for trilateration calculation In this casewe gradually increase the size of the inner circle until the two circlesintersect

Estimated position

Figure 17 An example of position estimation from an exceptioncase where the original distance circles (from distances estimatedby RSSI) had two circles completely enclosed in a third circle

with an increment of 1 meter until there are two intersectionpoints (cf Figure 15) Again this adjustment comes fromthe assumption that the indoor RSSI levels are disrupted byobstacles which make the RSSI levels degrade more rapidly

In the second case where for a pair of circles one circleis completely enclosed by another circle (ie [Case 2]) weincrease the size of the smaller circle with an increment of 1meter until there are two intersection points (cf Figure 16)The resulting circles after the adjustments represent theldquoadjusted distancesrdquo from each iBeacon Now since all pairsof ldquoadjusted circlesrdquo each have two intersection points we canuse these six intersection points to apply the trilateration andestimate the approximate position of the mobile device

Figures 17 and 18 show the position estimation resultsfrom the two exception cases where the estimated distancesfrom three iBeacons did not provide sufficient coverage fortrilateration After the geometric adjustment based on ourprior observations the resulting estimations were detectedto be very close to the actual positions As a final note theentire position estimation process including the geometricadjustment and trilateration takes place at the server Themobile device simply detects iBeacon signals during theperiod of classes for the student sends the list of iBeacon


Estimated position

Figure 18 An example of position estimation from the exceptioncase in Figure 14 where the original distance circles (from distancesestimated by RSSI) had one circle enclosed in another and a thirdcircle is nonintersecting with the other two circles

advertisement data to the server and queries for attendancecheck Thus there is practically no computation burden onthe mobile device and this leads to minimizing the energyusage as well

53 Evaluation Wehave evaluated our automatic attendancechecker system with the help of four undergraduate studentsattending classes in three classrooms While this case studywas done in a limited scale due to challenges in getting indi-vidual consent for accessing students personal informationthe system reported no false reports (neither false positivenor false negatives for attendance) for our student volunteersduring the course of 1monthWe plan to scale the experimentto a larger number of students and classes in the future

6 Related Work

There are several pieces of prior work that proposes appli-cations and services using iBeacon devices The work in[9] uses iBeacons for tracking luggage at the airport andthe work in [10] provides interactive experience to visitorsin museums iBeacon has been used for path planning inemergency guiding system[11] and also for tracking patientsin emergency rooms [12] Furthermore the work in [13]proposes an indoor route guidance system and the workin [15 16] uses iBeacons for occupancy detection withinbuildingsThese applications all use proximity-based on RSSIas the source of information rather than trying to pinpointthe exact and absolute location of the mobile device Thereare also a number of prior works that focus on using iBeacondevices for precise indoor positioning [6 7 12 23 24]However none of these efforts provide an in-depth andextensive measurement study of iBeacon RSSI measurementsand its variability with different environmental factors

The work in [30] implements an Android applicationthat collects statistics of RSSI values from nearby iBeaconsand provides some measurement results but only at a fixeddistance The work in [5] provides some measurement

data regarding the transmission power and reception ratioalong with the estimated distance and the work in [31]attempts to calibrate the distance estimation model usingmeasurement data and perform error analysis Howevernone of these works provide extensive and comprehensivemeasurement data in the dimensions of several different typesof iBeacons mobile devices software platforms transmis-sion power configurations indooroutdoor comparisons andWiFi interference impacts and with several different types ofcontrolled yet practical obstacles Furthermore we combinethe experiences from theRSSImeasurements to design awell-suited and practically applicable system and propose a casestudy application for automatic attendance checking whichuses localized trilateration techniques along with geometricadjustments to limit the scope of error due to RSSI variabilityand meet the target accuracy

7 Conclusion

The original goal of this research was to develop an improvedindoor positioning system using iBeacon technologies How-ever after numerous and extensive experiments we realizedthat the signal variation was too high to retrieve accuratedistance estimation for designing a reliable and robust local-ization system Instead we decided to report on this varia-tion and inaccuracy to clarify the misunderstanding causedby numerous sugar-coated news articles on how accurateiBeacon technology can be Our finding shows that iBeaconRSSI values (and the corresponding signal propagationmodelto estimate the distance) vary significantly across iBeaconvendors mobile device platforms deployment height of thedevice indooroutdoor environmental factors and obstaclesIt is obvious that the accuracy and efficiency of locationestimation depend heavily on the accuracy of the measuredRSSI measurements and the model used to estimate thedistance not to mention the surrounding environmentalfactors We believe that our work provides evidence onthe challenges for designing an indoor localization systemusing commercial-off-the-shelf (COTS) iBeacons devicesand dismantles the misunderstanding of its overestimatedaccuracy Furthermore based on the observations made inthis work our future work is to find a way to approach theseerrors differently and develop an iBeacon-based system thatis resilient and robust to such RSSI dynamics

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This research was supported in part by the Basic ScienceResearch Program through the National Research Founda-tion of Korea (NRF) funded by the Ministry of Education(NRF-2014R1A1A2056626) For Hyungsik Shin this workwas supported by the Hongik University new faculty researchsupport fund


References

[1] ldquoBluetooth Smartrdquo httpswwwbluetoothcomwhat-is-blue-tooth-technologybluetooth-technology-basicslow-energy

[2] C Gomez J Oller and J Paradells ldquoOverview and evaluationof bluetooth low energy an emerging low-power wirelesstechnologyrdquo Sensors vol 12 no 9 pp 11734ndash11753 2012

[3] Apple ldquoiBeacon for developersrdquo httpsdeveloperapplecomibeacon

[4] Google ldquoEddystone-open beacon formatrdquo httpsdevelopersgooglecombeacons

[5] P Martin B-J Ho N Grupen S Munoz and M SrivastavaldquoAn ibeacon primer for indoor localization demo abstractrdquo inProceedings of the ACM Conference on Embedded Systems forEnergy-Efficient Buildings (BuildSys rsquo14) 2014

[6] Z Chen Q Zhu H Jiang and Y C Soh ldquoIndoor localizationusing smartphone sensors and iBeaconsrdquo in Proceedings of theIEEE 10th Conference on Industrial Electronics and Applications(ICIEA rsquo15) pp 1723ndash1728 IEEE Auckland New Zealand June2015

[7] H K Fard Y Chen and K K Son ldquoIndoor positioning ofmobile devices with agile iBeacon deploymentrdquo in Proceedingsof the IEEE 28th Canadian Conference on Electrical and Com-puter Engineering (CCECE rsquo15) pp 275ndash279 Halifax CanadaMay 2015

[8] mlbcom ldquoMLBAM completes initial iBeacon installationsrdquohttpmmlbcomnewsarticle67782508mlb-advanced-media-completes-initial-ibeacon-installations

[9] M Kouhne and J Sieck ldquoLocation-based services with ibea-con technologyrdquo in Proceedings of the 2nd IEEE InternationalConference on Artificial Intelligence Modelling and Simulation(AIMS rsquo14) pp 315ndash321 IEEE Madrid Spain November 2014

[10] Z He B Cui W Zhou and S Yokoi ldquoA proposal of interactionsystem between visitor and collection in museum hall byiBeaconrdquo in Proceedings of the 10th International Conferenceon Computer Science and Education (ICCSE rsquo15) pp 427ndash430Cambridge UK July 2015

[11] L-W Chen J-J Chung and J-X Liu ldquoGoFAST a group-basedemergency guiding system with dedicated path planning formobile users using smartphonesrdquo inProceedings of the IEEE 12thInternational Conference on Mobile Ad Hoc and Sensor Systems(MASS rsquo15) pp 467ndash468 IEEE Dallas Tex USA October 2015

[12] X-Y Lin T-W Ho C-C Fang Z-S Yen B-J Yang and FLai ldquoA mobile indoor positioning system based on iBeacontechnologyrdquo in Proceedings of the 37th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety (EMBC rsquo15) pp 4970ndash4973 IEEE Milan Italy August2015

[13] A Fujihara and T Yanagizawa ldquoProposing an extended ibeaconsystem for indoor route guidancerdquo in Proceedings of the Inter-national Conference on Intelligent Networking and CollaborativeSystems (INCOS rsquo15) pp 31ndash37 Taipei Taiwan September 2015

[14] R-S Cheng W-J Hong J-S Wang and K W Lin ldquoSeamlessguidance system combining GPS BLE Beacon and NFCtechnologiesrdquoMobile Information Systems vol 2016 Article ID5032365 12 pages 2016

[15] G Conte M De Marchi A A Nacci V Rana and DSciuto ldquoBluesentinel a first approach using ibeacon for anenergy efficient occupancy detection systemrdquo in Proceedingsof the 1st ACM Conference on Embedded Systems for Energy-Efficient Buildings (BuildSys rsquo14) pp 11ndash19 Memphis TennUSA November 2014

[16] A Corna L Fontana A Nacci and D Sciuto ldquoOccupancydetection via ibeacon on android devices for smart buildingmanagementrdquo in Proceedings of the Design Automation Test inEurope Conference Exhibition (DATE rsquo15) March 2015

[17] A LaMarca Y Chawathe S Consolvo et al ldquoPlace lab devicepositioning using radio beacons in the wildrdquo in Proceedings ofthe 3rd International Conference on Pervasive Computing (PER-VASIVE rsquo05) 2005

[18] P Bahl and V N Padmanabhan ldquoRADAR an in-building RF-based user location and tracking systemrdquo in Proceedings of theINFOCOM pp 775ndash784

[19] Y Chen D Lymberopoulos J Liu and B Priyantha ldquoFM-based indoor localizationrdquo in Proceedings of the 10th Interna-tional Conference on Mobile Systems Applications and Services(MobiSys rsquo12) 2012

[20] J Brassil C Pearson and L Fuller ldquoIndoor positioning withan enterprise radio access networkrdquo Procedia Computer Sciencevol 34 pp 313ndash322 2014

[21] A Varshavsky E de Lara J Hightower A LaMarca and VOtsason ldquoGSM indoor localizationrdquoPervasive andMobile Com-puting vol 3 no 6 pp 698ndash720 2007

[22] J Paek K-H Kim J P Singh and R Govindan ldquoEnergy-efficient positioning for smartphones using cell-ID sequencematchingrdquo in Proceedings of the 9th ACM International Confer-ence on Mobile Systems (MobiSys rsquo11) pp 293ndash306 WashingtonDC USA June 2011

[23] Y Wang X Yang Y Zhao Y Liu and L Cuthbert ldquoBluetoothpositioning using RSSI and triangulation methodsrdquo in Pro-ceedings of the IEEE 10th Consumer Communications and Net-working Conference (CCNC rsquo13) pp 837ndash842 IEEE Las VegasNev USA January 2013

[24] P Kriz F Maly and T Kozel ldquoImproving indoor localizationusing bluetooth low energy beaconsrdquo Mobile Information Sys-tems vol 2016 Article ID 2083094 11 pages 2016

[25] S-H Lee I-K Lim and J-K Lee ldquoMethod for improvingindoor positioning accuracy using extended Kalman filterrdquoMobile Information Systems vol 2016 Article ID 2369103 15pages 2016

[26] Estimote beacon httpwwwestimotecom[27] Wizturn beacon httpwwwlime-icombeaconpebbleen[28] ldquoGELO beaconrdquo httpwwwgetgelocom[29] J Zhao and R Govindan ldquoUnderstanding packet delivery per-

formance in dense wireless sensor networksrdquo in Proceedings ofthe ACM 1st International Conference on Embedded NetworkedSensor Systems (SenSys rsquo03) Los Angeles Calif USA November2003

[30] M Varsamou and T Antonakopoulos ldquoA bluetooth smart ana-lyzer in iBeacon networksrdquo in Proceedings of the 4th IEEE Inter-national Conference on Consumer Electronics (ICCE rsquo14) pp288ndash292 IEEE Berlin Germany September 2014

[31] T M Ng ldquoFrom lsquowhere i amrsquo to lsquohere i amrsquo accuracy study onlocation-based services with iBeacon technologyrdquoHKIE Trans-actions Hong Kong Institution of Engineers vol 22 no 1 pp 23ndash31 2015

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



technology is a subset of BLE beacons but iBeacon and BLEproximity beacons became nearly synonyms today despiteGooglersquos recent release of their own open beacon formatcalled Eddystone (Eddystone is a protocol specification byGoogle that defines a Bluetooth Low Energy message formatfor proximity beacon messages) [4]

Note here that the original goal of BLE beacons wasto provide proximity-based application services possibly toextend the features to coarse-grained location-based applica-tions Given that BLE beacons solely focus on RSSI measure-ments combined with lessons learned from the long historyof research in RF-based localization it is somewhat expectedto be able to achieve only a limited accuracy in indoorlocalization services It is well known that the variability of RFsignals with respect to the indoor environment complicatesthe accurate estimation of the mobile devicesrsquo location or itsrelative distance to the BLE beaconing transmitter deviceHowever most prior work on BLE-based indoor localizationmerely states the possibility of error due to signal variationbut does not quantify them and their reported results showonly the positive side of the picture For example the workin [5] reports a positioning accuracy of 053 meters but thisworkwas bounded by a 9times 10m2 areaThework in [6] reportsa 12-meter accuracy with fingerprinting but only within a474 times 159m2 area Recently work in [7] shows that distanceestimation error can increase significantly with longer dis-tance when themeasured signal RSSI vary over time but doesnot show other dimensions of possible error

In this work we perform an extensive set of experimentsto quantify the impact of various indoor obstacles on the BLEsignal variance Specifically we show that different iBeacondevices from different vendors along with the paired mobiledevice platform (eg iOS or Android) can give significantimpact on the RSSI measurements to complicate the processof designing a generally acceptable distancelocation esti-mation model Furthermore practical factors such as thedeployment height or environmental factors also introduce asignificant amount of impact on the RSSI values to introducean additional level of complexity in designing highly accuratelocation estimation systems using iBeacons Lessons fromthese empirical studies lead us to design an application that isldquopractically designablerdquo Specifically we design and evaluate aclass attendance monitoring application system by combin-ing distance estimationsmade from locally accessible iBeaconsignals and an estimation adjustment scheme designed fromthe lessons from our empirical signal measurements

This paper is structured as follows Section 2 introducesthe relationship between RSSI readings and the estimateddistances and then motivates our empirical studies on therelationship Section 3 describes our approaches to theempirical studies and experiment settings including variousparameters that we change during our studies Section 4 isthemain section and it describes ourmeasurements data andillustrates our analysis results based on the data Section 5presents an application case study an automatic attendancemonitoring system which is based on our empirical stud-ies This system also proposes a novel geometric approachto overcome the limitations of practical iBeacon systems

Section 6 reviews prior and related work and finally Section 7concludes the paper

2 Limitations of Bluetooth Low Energy-BasedLocalization Systems

As Bluetooth Low Energy (BLE) and Applersquos iBeacon pro-tocols begin to see widespread adoption and deploymentindoor proximity andor location enabled applications arequickly penetrating into our everyday lives and consumerspace iBeacon has been installed in Apple stores in theUS to provide product information as well as customerservices and the National Football League (NFL) has usediBeacons in the MetLife Stadium to provide personalizedadvertisements to football fans during the Super Bowl game[8] Furthermore iBeacon can be used to track luggage atthe airport [9] provide interactive experiences to visitors inmuseums [10] plan evacuation paths in emergency guidingsystems [11] track patients in emergency rooms [12] guideindooroutdoor routes [13 14] and detect the occupancy ofrooms [15 16] The application space to which iBeacons andBLE proximity applications can be applied is extremely largeThese applications all use proximity measurements based onthe received signal strength indicator (RSSI) as the sole sourceof information thus these applications exploit informationon how close a mobile device is to an iBeacon rather thantrying to pinpoint the exact and absolute location of themobile device

The original purpose of iBeacon technology is to pro-vide proximity-based application services and coarse-grainedindoor location positioning and navigation based solely onproximity Note that proximity (being near to some object) isrelated to the location (where you are) but is not necessarilyidentical A location (or position) ismore than just proximityand it is an absolute value usually defined by some coordinatesystem (eg latitude and longitude for GPS) Recentlyhowever there have been several proposals to use BLE forindoor positioning extending several prior works that useother technologies such as ultrasonic sound infrared WiFiGSM RFID IEEE802154 and earlier versions of Bluetoothfor indoor localization [17ndash22] (there are many more relatedprior work on indoor localization that uses other wirelesstechnologies but we list only a subset here for brevity and tokeep the reference list focused on BLE only since there are toomany of them) Indoor positioning and localization are stillan active area of research and BLE is extending the literatureusing its new wireless features

There are several approaches (beyond simple proximity-based ones [12]) that can be applied to BLE-based indoorpositioning In all cases an iBeacon will periodically broad-cast an advertisement packet containing a unique ID anda calibrated RSSI value (1198770) corresponding to one-meterdistance (1198890) This value allows us to determine the distancebetween an iBeacon and a device using a model in (1) where120574 is a calibration parameter for the path loss exponent Basedon the RSSI or the estimated distances between iBeacons anda mobile device a weighted average model [5] trilateraliza-tiontriangulation model [23] or fingerprinting [7 24] can






(1)









iOS


Android

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








Estimote ground

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 10 20 30 40 50 60 70 80 90 100

Real distance (m)

Estimote 12m





0 20 40 60 80 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 1 2 3 4 5 6 7 8 9

Real distance (m)






0 10 20 30 40 50

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI






2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z


LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y


12m12m

5m


wo WiFiw WiFi


minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons








Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in








(1)









iOS


Android

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








Estimote ground

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 10 20 30 40 50 60 70 80 90 100

Real distance (m)

Estimote 12m





0 20 40 60 80 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 1 2 3 4 5 6 7 8 9

Real distance (m)






0 10 20 30 40 50

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI






2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z


LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y


12m12m

5m


wo WiFiw WiFi


minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons








Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




iOS


Android

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

1 10 20 30 40 50 60 70 80 90 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








Estimote ground

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 10 20 30 40 50 60 70 80 90 100

Real distance (m)

Estimote 12m





0 20 40 60 80 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 1 2 3 4 5 6 7 8 9

Real distance (m)






0 10 20 30 40 50

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI






2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z


LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y


12m12m

5m


wo WiFiw WiFi


minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons








Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Estimote ground

minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 10 20 30 40 50 60 70 80 90 100

Real distance (m)

Estimote 12m





0 20 40 60 80 100

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI








minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 1 2 3 4 5 6 7 8 9

Real distance (m)






0 10 20 30 40 50

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI






2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z


LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y


12m12m

5m


wo WiFiw WiFi


minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons








Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





minus100

minus90

minus80

minus70

minus60

minus50

RSSI

0 1 2 3 4 5 6 7 8 9

Real distance (m)






0 10 20 30 40 50

Real distance (m)

minus100

minus90

minus80

minus70

minus60

minus50

RSSI






2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z


LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y


12m12m

5m


wo WiFiw WiFi


minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons








Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




2402

MH

z2404

MH

z2406

MH

z2408

MH

z2410

MH

z2412

MH

z2414

MH

z2416

MH

z2418

MH

z2420

MH

z2422

MH

z2424

MH

z2426

MH

z2428

MH

z2430

MH

z2432

MH

z2434

MH

z2436

MH

z2438

MH

z2440

MH

z2442

MH

z2444

MH

z2446

MH

z2448

MH

z2450

MH

z2452

MH

z2454

MH

z2456

MH

z2458

MH

z2460

MH

z2462

MH

z2464

MH

z2466

MH

z2468

MH

z2470

MH

z2472

MH

z2474

MH

z2476

MH

z2478

MH

z2480

MH

z


LL 37 0 1 2 3 4 5 6 7 8 9 10

38

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

35

34

36

39

Freq

uenc

y


12m12m

5m


wo WiFiw WiFi


minus70

minus65

minus60

minus55

minus50

RSSI

Missingbeacons








Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





Line

-of-s

ight

Iron

doo

r

Woo

den

door

Win

dow

Shee

t of

alum

inum

foil

Phon

e cov

ered

in h

and

Cov

ered

in p

aper

minus5857

minus7718

minus6662

minus6113

minus5200

minus8788

minus6447

Type of obstacle

minus30

minus40

minus50

minus60

minus70

minus80

minus90

minus100

RSSI






Mean1 1

10 20 30 40 50 60 70 80 90 1001

Real distance (m)

0

20

40

60

80

100

120

140

160

180

200

Estim

ated

dist

ance

(m)







HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




HTTPS

WebDB server















Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in












Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





Original distance

circle

Adjusted distance

circle






Estimated position






Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Estimated position




6 Related Work




7 Conclusion


Competing Interests


Acknowledgments



References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




References





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



Documents

Research Article A Measurement Study of BLE …nsl.cau.ac.kr/~jpaek/docs/ibeacon-mis-published.pdfGoogle s recent release of their own open beacon format called Eddystone (Eddystone