Home > Article > A Review of Indoor Localization Technologies Towards Navigational Assistance for Topographical Disorientation

A Review of Indoor Localization Technologies Towards Navigational Assistance for Topographical Disorientation

Ambient IntelligeToA time of arrisToF time of flightUWB Ultra-WidebandWLAN Wireless Local Area Networknd navigation, Aguirre and D'Esposito [3] provide a well-accepted taxonomy of TD,cognitive impairments For example,well recognized that TD and spatialdeficits are common sequelae of brain injury [87, 74] Individuals living with post-traumaticskills, or deficitpost-injury [16, 72] It has also beenhat deficitsIt has been suggested that wearablegationSystem(GPS) can be a useful wayfinding tool for individuals with cognitive impairment7HGPS signals have limitedegB36,40,3061,9938])Gid significant periods of time indoors- be it in acute and tertiaryhospitals or, subsequent to rehabilitation, at home, schools, office buildings, shopping malls,ong-term care facilities -identification of potential technologies for indoor navigationalassistance is imperative An initial survey of the literature has suggested that a dcollection of candidate indoor localization technologies existsdisciplines This diversity makes it difficult to grasp the potential of an existing technologyfor the rehabilitation of individuals with topographical disorientationLocalization technologies are critical to emerging locationvision [69], stroke [86]and traumatic brain injury [6]

In particular, regarding indals with topographical disorocalization has oftenbeen human-mediated rather than automatic For instance Liu et al evaluated the benefitsrements [46] However, the location tracking andexperimenters In similar vein, Sohlberg et al [75 found that individuals with wayfindingeech-based auditory directionfrom a wrist-worn PDA navigation system However, like Liu et al, the PDA's navigationalcalization technologies in the emerging fields of cognitive prosthetics andassistive technologies As a consequence, the overarching goal of this reviewstematically orgthe literature on indoor human tracking technologies, andascertain their feasibility for eventual use in the realm of TD rehabilitatintechopen

A review of indoor lotowards navigational assistance for topographical disorientationf personnel within a building Clinical setting are under constant changepersonnel and large metallic structures such as beds and wheelchairs are constaIvIronghlydio map capturedbe used reliablyithout accounting for these dynamInterference and noise are often-mentioned challenges[42, 84] Althoughoperate within a reserved radio band [62, 73, 63], most of the research is conducted on opefor the increased risk ofterference due to other systems sharing the same frequency bands of the radio spectrum 43Finally, the usage of radio transmitting deoften times restricted in criticamost healthcare facilities, according to recommendations made by the Association for theAdvancement of Medical Instrumentation(AAMD)[1] and other standards or regulatoareas, eg, waiting roomLocalization technologies based on rf technoertain infrastructuralnetworks, that mayin the facilities, Care must be taken howerevaluating the imphysical environment on the RF localization technologies, as the solution mayn-operative in certain clinical settings6 Photonic energyLight refers to the phenomena of electromagnetic radiation at wavelengths within the visibleange,which extends approximately between 380 and 750 nanometres, Photonic energyrefers to the energy carried by electromagnetic radiation in this wavelenvisible light, or in its lower or upper vicinity, known as ultraviolet and infrared light,se spectively A photon is the minimum possible discrete amount of light energolutions in this category rely on the photonic energy received from infrared or visible liglemissions or reflections, to estimate the position of an object in space

The articles selected for this section can be distinguished based on the sensor required tcate a mobile subject or device via image processing 58, 5, 94, 24, 21, 93, 97, 89, 39, 70, 3148, 10] In contrast, some other articles present localization solutions basrocessing devices[15, 56, 55, 14, 33, 90)Methods based onall of theThe classicalodel of computer vision-based location detection consists of 4 main stagesImage acquisition, normally through a video catation of the image and extraction of relevant feature3 Selection of the closest match or matches of the detected features against the entriesof a database of features (eg edges or fiducials) Thismatical transformations of the spatial relationships between the featuresunt for the variability in scale, rotationntechopen

Ambient Intelligeof the pose of thnat could have given rise to thetheNe can subsequently distinguish between localization systems that rely on imagethat rely on thefiducial) in the environmenConsequently, the articles included in this section were organized in three main groupd natural feature extraction, 2)image processingfiducialF-Cognition, and 3) non-image processing sensors This organization is Presented in thetural feature extraction refers to the mathematical processing ofcal values thatly represent that image Features oflected from itsgram [21 or from structural edges and theelationships 5] A reduced subset of these features is subsequently defined611 Mobile camera systemsArticles grouped in this section consider solutions where theIn the off-line stage, images of the en, ocess described for wireless localizationcations Each image is processed to extract its unique features

Subsequently, the extractedfeatures are stored in a databaseIn the on- line stage, thenage features are subsequentlextracted These featuresmpared to the entries in the feature database), or through spatial transformations that yield thebest match between the features in question and those in the database(to obtain mored offlcation estimates) Such transformations reflect the differences in between on-linenage capture locations Once the best matching set of database featuresnding database entry for camera position and orientation is usedestimate the absolute location of theArticles adopting a mobile camera system include [5, 94, 24, 21, 93, 70, 31] It must be noted,ed in such articles are not strictly dependent on properties of the robots anan be easilyuman localization threle indoor environments, Aider, Hopped a localizabile monoculhichline environmental features The maximum localization error reportentechopen

A review of indoor lotowards navigational assistance for topographical disorientationduring their experiments was 20 centimetres, and the maximum angle estimation errorge and their spatial relationships, which were, in turn, usedThrougtation invariant feature transformation, they estimated the most pication of theeated databaseeatures The authors reported inocalization errors less thanmetres, 94 of the timestem foextended to the case of human walking Loptured white between twoints The on-line stage involved autonomous robot navigation between two arbitraryints while capturing images To achieve localization, an algorithm correlated freshlyunprocessed raw imabetween the learned route and the current position Temporary occlretectedin the correlation The maximum position deviation reporexperimental 17 metre longas 09 metres This represents a deviation of 5 3% of the6

12 FixedWhen theras of theted in fixednvironment, the structural features of thocalization purposes Instead, features of the object being tracked must bethis wif the salient features of the object or objects appear in the fieldlocation of the object or person can be calculated with respect to thes fixed positionThe position of the object of interest in the environment is estimated based on its positionwithin the captured image, and the spatial distribution of its salient featurescandidate articles discussed solutions for tracking thInstead, most of the articles prhich the tracking plimitedaTherefore, only articles 58, 97,, 48, 10considered as an indoor tracking solution based on our inclusion and exclusion criteriaIn [58, the authors proposed a system that detected elements of theof interest were obtained, along with a verticalation of the object dthe algorithmlour scheme of the clothes of a person while standing up A camera, located at theentrance of the test environmented to create an initial colour model of each pe thisreported correct user recall values of 87 21% with an availability of 73similar approach was presented in [97]Thers used colour-based featuresvertical diffregions of a human figure while standing up To accountthe creation of a connected graph that represented the areas covered by eachges of the graph denoted physical connections of the areas covered by the field of view ofntechopen

Ambient Intelligeas used in the probabilistic modeling of the movementtracking accuracy89 proposed a distributedr network based ong The robustness of the solution reliedlapping the visual field ofing for localizatiothe elements of the distributed network Featurescomponents analysis of features obtained by differencing consecutive segments of thegeThe authors demonstrated the feasibility of human tracking in a crowded setting Thereportedsimilar vein, [48 proposed a system with 4 cameras withly overlapping coverageur and non-colour features were used to account ft inIgorithm Usingidential filter, particles represented theconducted with multiple users navigatingrrently, without deliberately avoiding occlusions Three hundred particles were usedto represent eachIn some cases the users wore similar clothes Some of these useintermittently

The reportedlocalization accuracy of a singlen, while using aof 30 particles for theinally, the Easyliving project [10] proposedge of image-based localization systemsprovide context awareness within intelligent environments The project was based onmical posture ofor the orientation ofin the environment The identification of the subjectbeing tracked was based on cold structural features, The localization areported62 Image alFiducial image detection diffalgorithmsdesipredefined, synthetically created, patterns in theenvironment These patternsn principle, localization algorithms based on fiducial recognitionf the image tofiducial, a database is still required to determine the location of therelative to aeference frame, fixed with respect to the environment In this case, however, the databasean be created automatically by storing a numeric ID associated with the fiducial along withrdinates in tht 194 The ID of the fiduciabe encoded within thereported advantages of fiducial recognition over natural feature recognitin [94) the authors propoposition detection system for robot navigation in indoorThey conducted a simple experiment using a fiducialThey presmage analysis technique based on homographntechopen

A review of indoor lotowards navigational assistance for topographical disorientationposition of the camerocalizaticapplied to human localizationKim and Jun [39] introduced a wearable indoor localization system composed of a portablethrough image processing, combining fiducial markers and natural visual feature extractionLocalization with fiducials was achieved through an open source library called ARToo[37] The localization algorithm detected a synthetic visual patthe camera, Using affine transformations, the authors estimated the distortion and scaling ofte fiducial due to angle ofd capture distance with this inforthe exactosition (ie distance and orientation) of the camera with respect to the marker could bedetermined The authors modified ARToolkit, adding an adaptive illumination thresholdingrithm and an algorithm fohen there were no fiducial markers in view Naturalieved by analyzing the colhue histogramsoftion was defined by a high dimensional spd64hue histograms A Linear Discriminant Analysis (LDA) was applied to reduce thedimensionality of the features which defined each location the first five lda coefficients ofeach frame were used as desce features of that frame The sequence of features intion phase facilitated thatonof a database of natural features of the building, To estimate the location of the user, theector of features obtained from the last 64 frames, defining the current locationmpared against the features in each entry of the database

The difference between afeature vector captured by the camera in the on-line stage against feature vectors stored inthe databaseuantified by the Euclidean distance If the Euclidean distance was lethan an unspecified threshold, the location of the person was defined as the matchthe fieldof thehe location information obtained by both visual systemsual map, along with instructions to complete a predefined paththrough the HMD The localization accuracy of the system was not reported63 other photonic seArticles included in this section make use of non-image capturing Infrared (IR) sensors [15In [15, the apresented an iR proximity-based localization system, which provideduseum visitors with useful information about exhibits in each hall For this purpose, IRg of the door frames of every room Each emittercarried a Personal Digital Assistant (PDA) with an infrared port The PDA contained adatabasetextual information of the exhibits as well asf a new ID, the PDa automatically presented the map of the correspondinghall While in the hall, a graphical user interface in the PDa helped theto obtainauthors notedntechopen

Ambient Intelligen system, in particular, noise and reflections of IR signals The userity of this system was down to the scale of a roomDalated by the Ier thefrequency used for modulationhanged in a cyclic way, from low to high frequencies As the attenuatiinfrared signal are frequency-dependent, ID's modulated at lower frequesuccessfully detected farther away from the emitter However, the power of the IR signalobtained off-line mments of the received signal in the vicinity of the emitter, inhion to fingerprinting for RF signals The authors measured the Id detectionrate in 10 cm concentric regions, at steps of 5 degrees, repeating this process fordifferent modulating frequencies Instead of creating a database of the signal in each pointhors modeled the decay of the signal with an equation that was dependent of theiver and emittersthe system achieved am localizationof 10timetres, withinetres Althexperimenducted, the usage of a unique ID per transmitter would allow the deploymentlatPetrellis et al [55 presented a localizatienvironment, and two receivers installed on a mobile unit Thefacing each other on the wallshille thethe mobile unit facedfrom each other Each emitter transmitted a series of unique cyclic data patterns,yt of theensors facilitated detection of uentation and, at the same time

discrimbetween signals received via direct path and reflected signals Reflection rejchanced using a predictive model, which took into consideration the imince the system relied on predicind orientation of the mobile device, the estimation rules constrained the meliable given that the emitters transmitteduences,the authors proposed a cellular-like spatial emitter arrangement to coverThe system was sensitive to moving personnel and other objects thareflections although compensation algorithms reduced such effects The localizationaccuracy was not reportedlambuilding A user carried a wearable computer instrhoto-detectorEach fluorescent lamp emitted an ID, associated with theocation of theThe Dncode through pulse freqdulation Thecoded infen did not introduce perceptible illumination effects The photo-detectorrientation information The wearable computer was carried on a vest, worn by the useThe area illuminated by two lamps transmitting different location ID's couldsince this would cause interference, resulting in the inability to detect the encoded signWhenever available, location information was presented to thecomputer Such information was overlaid on the users visual field through a HMD, Thelocalization accuracywithin three to four meters Although this accuracy isntechopen

A review of indoor lotowards navigational assistance for topographical disorientationdon theeights, these numbers were not provided The minimalocalization system for tracking assets, personnel orother mobile elemeestimate the poof a receiver, a pair of eye-safe IR lasertwo differentengths, while a an infrared strobe provides a referencegnal Using signals from 3 or more transmitters, the receiver calculates and transmits itsposition to a central data collecting station In order to estimate the orientation of a solidbody two or morers are attached to it The iGPS system claims to offerb-centimetreScientists from Olivetti Research designed the Active Badge loca

tion system, whichcould then be located when in the vicinity of a receiving station The10 seconds The system offeredsed to redirect pcalls f6Common problems reported for phollocalization are theillumination variability[ 58, 89, 78, 971n image processing solutions, ambient noise is usually overcome by image filteringIR sensing, effects of ambient noise can be mitigated by using acombination of different modulation frequencies [56, 33Another problem commonl89, 10]

For instance, in the experimentsrformed in [56], the introduction of new objects or humans was specifically avoidedduringr to reduce the risk of visual occlusions by humans andHowever, clinical settings and publoftentimesdensely populated Consequently, occlusion conditions can emergeceiling mounted sensors Therefore, if a photonic-based systemsidered fortracking system to assist in the localization process during periods of optical occlusionIn thelaser based-solutinly class 1 laser devices should be used, whichlassified as"eye-safe"by the IEC 60825-1 standard [32] In clinical settings, however, specialcare must be taken with even class 1 laser devices, to ensure that no harm will be caused byytion Healthfacilities operate under thenon-disclosure to protect thel,pnts Thisleration when a localization system is designed to capture images of thereveal important information about the person wearing the systemabout the patients and health care personnel in the vicinity Frequently, image processingdesigned to send the captured image tontechopen

A review of indoor lotowards navigational assistance for topographical disorientation4 Literature selectionNe combed the literature for candidate localization technologies thaer-reviewed journal articles published in English between 2003 andsought from three different academic databaseds " indoor location indoor localization""indoor trackind indoorpositioning" After removing duplicate records, we arrived at 214 articlesclessubsequentlyed according to the following inclusiohe article must focus on thent and experimasting of a localizationr navigation system: ie, articles focusing on mathematical processing oflocalization data, or localization experiments in simulated environmentsdiscarded2

The reported technol(a) be usable indoors, within a building or a larger space, ie, technologies usedtion accuracy of a mobile target within a 10 meter radius with aelay of 5 second(e) be applicablesystems designed for vehicles, large objed) trackFifty three articles mal criteria Such articles were subsequently scanned foralternate localizationreferenced threetimes and thatnot selected in the initial search Eleven additional articles were included in this mannetotaling sixty four articles for consideration in the present revof localization teThe location of an object in space is determined by measuring a physical quantity thathanges proportionally with the position of the object ofintoty()sonic waves, (4)mechanical energy(inertialtact),(5)(2)phoand (6 atmeEahysicalan be further subdivided according to the underlying hardware technology Figure 1his two-tiered taxononte that the latter two phatmospheric) have been collapsed into one category, named, Other"due to low articlets in these arentechopen

Ambient IntelligePersonal and local area networksBroadcast and wide area networksRFID tacsInfraredInertialAtmospheric pressureWhere appropriate, articles are further differentiated byalization technique Forthis thirdprovided, expanding on those proposed by Hightower and Borriello [29]Triangulation is a family of methods that include lateration, angulation andhereof lateration refers to the calculation of thedistances are commonly obtained indirecthatroportional to distance Time of flight and power attenuation of a radio signalmon indirect distance metrics [20] Angulation refers to the calculation of theosition of the subject using the angles of arrival of signals emitted from fixed pointspace88, 80Proximity refers to a class of methods which establish the presence of the human subject inity of a sensor, which alone has limited sensing range and analyscapabilities

The proximity of the subject can be detected through physical contact,f a detic band tothe monitoring of a physical quantity in the vicinity of the sensor, for instarmagnetic fieldScene analysis involves the monitoring of a wideround the subject of interest from ant The commonly deployed sensors have broad coverage areaand range Examples include ceiling-mounted video cameras or passive infraredntechopen

A review of indoor lotowards navigational assistance for topographical disorientationDead reckoning refers to the usage oflocatiedates, calculatedg infor-estimated location, Position estimatiocommonly based on accelerometry and gyroscopyThe ensuing review of literature will adhere closely to the taxonomy depicted in Figure 1each physical phenomenon,briefly present the generaeview articles in the relevant subcategories and comment on their relative merits andconclude the arwith recommendations of indoor localiztechnologies suitable for addressing the development of assistive devices for individuawith tD5 RadAn electromagnetic wave is the energy generated by anan oscillating, electrically chargedpace The generSolutions in this category estimate the locatia mobile targetnsmitter andreceived by a mobile station These properties typically depend on theAs gnal and the characteristics ofsurrounding erted in Figure 2, most of the articles in this survey describocalization system

These articles can be further subcategorized according to thePersonal and localnetworks, including technologies such as IEEE 80211a-Wideband (UWB2agB,且四2出g,9a9and broadcast networks not originally intended for localizbroadcast signals [62, 63](RFID) tags [45, 35,Radar [65, 711Each flayf RE localizatio51PeMost articles included in this sectionthe localization of mobile targets using a twostage process98,17,64664253,929243,79,22,%6,18,67,49583, The first stansists of an off line radioanalysisntechopen

Ambient Intelligeile station extracts radio fingerprints, ie, features from one or more metrics of the radsignal measured at predefined points in the erproportional to the distance betweenobile receiver and the emitting station CommonhedonicSanicPhisical magnitude measuredFig 2 Distribution of articles by physical quantity measuredmetrics include the direction or angle of arrival (AoA), Received Signal Strength(RSS),ortime of flight (ToF) of the incoming radio signal [52] A radio map or database ofted, storing signal feature values at each location along with theorresponding spatial coordinates Some authors ernment(79, 34 t such as the spatialautomated or assisted radio mapcreation teThe on-linethe active localization process where the mobile reachieved by proximity techniques, i

e, finding the closest match betwefeatures of the17,64,6,4222,18,67,4,57eral candidate locations (each with a fingerprint bearing some resemblance to that of theed signal) are geometrically combined to provide an estimate of thein space [43, 53, 79, 67,95,83, 9] Algorithms deployed in the selection of the cloor matches from the radio map include: 1)nearest neighbours techniques andthereof [17, 95, 9: 2)Bayesian statistical matching [98, 92, 43, 64, 66, 67, 83, 57: 3)networks [18, 4deve At a given la09m。map is created with a user transporting the mobilea fingerprint is recorded at each possible orientation Since thentechopen

A review of indoor lotowards navigational assistance for topographical disorientationagation of radio signals, the fingerprint generated for each1929%o of the time 42 and within 3 metres 91 6% of the time 53d The lowest number of basestations used to create the feature space of a fingerprint was one 198 In other wordfingerprinting sede reasonable localizationhardware requirements The most pressing challenge however is the non-stationarity of theff-line phases at the same exact location The time-variant nature of the radio map can bettributed to radio signal propagation effects induced by dynamic aspects of therid localizationLAn localization technologies The authors proposedhanding off the localizationbility between GPs and WLan depending on theavailability Fingerprintingthreprocess, indoor WLAN positioningof the most probable locations These locations were predefined by the usafeteria

The histogram of the RSS measurementsbest matched the histogram received If the histogram did not closely match any knownocations, a centroid algorithm estimated the location of the user from the locations ofThe plam levelproDdDe b accuracy, while the WLAN localization algorithm, using the centroid of severalcations, yielded an accuracy of approximately 30 metresRF-based localization can also be achieved without a priori analysis of the radio propertiesment (i e, without development of a) Four of these articles, all ofthem based on UwB radio signals, rely on signal triangulation as the sole localizationIndoor localization based on triangulation of radio waves is a non-trivial problem becauseon-Line(NLOS)conditions emerge In the preserOS conditions, the radiognal can travel to the receiver through a non-direct path,cy-based solutions The properties of ultra-wide band, shortduration pulsestion problems associated with multipropagation The most representative example is theproposed by Venkatesh andsuggested that the statistical parameters describing the distribution of theed rootquare delay spread(RDS)serve as the best discriminant estimatntechopen

Ambient IntelligeOS)and NLOSropagation This means that the statisticalthe rds of theVenkatesh and Buehrer tracked a mobilen though 71 predefined locatbuilding, achieving localization accuracies ranging from 1 centimetre to 2ple of RF triangulation-based schemes, Krejcar and Cernohorsky presented aocalization system [41] that relied on the triangulation of RSs metrics Room-granularityreported but further details of the triangulation or localization schemesFinally, a system basedanalysis and proximity techniques using aBluetooth ad-hoc network was presented in[20] Bluetooth inquiry signals were used feocalization In inquiry mode, a bluetooth device inquires about neighbouring bluetoothtations This inquiry process consists of scanning for dehile high power levels willhat are located farther away, providingestimates in this fashion This approach requires a fixed or anchor noently, the localizedestablish the position of other undetected mobiles nodes in their vicinity, creating an ad-hoclocalization network The reported localization error was 188 meters52 Broadcast and wia network solutions5

21 Solutions based on televiignals and cellular networThe solutions in this section are based on rf infrastructure transmitters that cover a widelocalization services In particular, two such technologiesidentified: television (Tv)broadcast signals [62, 631Rabinowitz and Spilker [62, 63] proposed the use of synchronization signals alrthe Advanced Television Signal Committee(ATSC) standardbeento penetrate indoors, they offer significantly grthan GPs-based solutions, Implementationlocalization solution wouldmodification of the existing broadcast signalthe inherent lack of synchronbetween stations due to clock imperfections, Rabinowitz and Spilker depreference station that transmitted an offset correction signal A mobile station thenAs the positions of the broadre fixed and known, the position of a mobileever could be estimated at least thrtransmitting stapurposes Rabinowitz and Spilker presented experiments in indoorging from 10 to 23 metersHu et al devemethod for the localizationhones inside a cell 73 usingfingerprinting techniques Invoking a methodfingerprint of the cellular signal, Hu et al arguedalization can be achievedindoor environments by statisticallI that granulsecord in the radio map The authors emphasizedlocalizationacy in thisntechopen

A review of indoor lotowards navigational assistance for topographical disorientationon the size of the cell and the characteristics of the environment thisthe improvement achievedover conventional cell-ID localization was not reported53 Solutions based on Rposed of one or more reading devices that can wirelesslybtain the ID of tags present in the environment TIpresent in the environment reflect the signal, modulating itdentification code [45, 59 The tared by a battery, or passive,drawing energy from the incoming radio signal The detection range of passive tagsherefore more limitede located inso situated in a fixed position To locate a mobile tag, the reader scanned through 8different power levels for tags in the vicinity When a mobiwas detected, thecompared the power returned by the reference tags and the mobile tag, determining theighbour algorithm The position of the mobile tagas determined by triangulating the position of the nearest reference tags The authorsreportedalization accuracy of 2 meters, 75% of the time The mm localizationdelay was of 75 secondsJia et al

[35] proposedid radio and vision based system which used rFid tags and astereo camera for robot navigation purposes To estimate the location of the mobile unidetector comprised of a directional antenna, which yielded the general direction of the RFIDgs detected When a tag was in proximity of the robot, it obtained images from the stereoamera to estimate the distance to the obstacle marked by the rFiD tag Subsequently, theid of the tag was compared against entries in a tag location database to determine if itobtained from the rfid taused to estimate the distance and orientation of the robotwith respect to the tag The localization accuracyf 85 centimetres Although thisosture were minimal Therefore, this system might be adapted to hilocalization at theFinally, in [81et al introduced a localization system based on proximity to RFIDgs In this systemtor used a personal digital assistant (PDA), which servedD of thedbsequently transmittedwhich returned infoThis informatios displayed inreen The localization accuracy of this system was not reported However, the accuracyrelated to the density of tags in the environmentntechopen

Ambient Intellige5 4 Solutions based on a radio frequency radarRoehr et al [71] presented an extension to thets were capable of transmitting andquency modulated signal with a 58 GHz carrier The fixed and mobile clocksere synchronized before distance and velocity estimations could be calculated Once theed at the mobile stationSubsequently, the mobile station sent a reply to the fixed station, which was synchronizedsing the signal justed from the fixed station the round trihe signalsed to calculate the distance between the fixed and the mobile stations, while the frequencyused to estimate the velocity of the mobile unit The experimental setupncluded one experiment within an office building, where distances ranging from 5 to 25metresmeasured with the radar system These distance mecompared against theobtained with a laser range finder This experimenonly and yieldeddeviation of less than 3 centimetres when compared to lase fineIn [65], the authors presented a radar indoor human tracking system, which exploited theDoppler effect of moving objects andrticularDoppler featuresred by the radar system

Such features were obtained from a jointmefrequency transform The authors also proposed a scheme for tracking several humanbeings concurrently; the Doppler separation effect between moving humans was exploited,hile target differentiation was realized using an antennaThree antennas and two frequencies were required for multiple target tracking Finally, theauthors presented experiments to determine the range of the target, using frequencydiversity, and studied the effects and errors introduced by the attenuation due to walls ine environment The authors concluded that the properties of humanchallenging within an indoor environment They suggested a few techniques that mightdoor localizah as the placement of the radar as fafrom the wallshe propagation of RF signals in indoor environments poses a central challenge Certainment affect the propnaterials such as wood or concrete attenuate RF signals, while materials such as metals orwater cause reflections, scattmulti-path racbetween the transmitter and the[88, 91, 73, 42, 99, 79, 71, 52 Several authors haveproposed techniques to compensate for these inaccuracies by automatically generating radiomaps which consider the structure of the building 79, 73, 34] However, a comprehensivemodel of allterials in a compThe propagation of radintechopen