Mobile Positioning Techniques in GSM Cellular Networks: A ...

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012This paper makes a comparative and analytic study of thevarious techniques for mobile positioning and presentsthe limitations of each and the performance of eachmethod based on some performance indices that will beused in the analysis.II. PERFORMANCE INDICES FOR MOBILE LOCATIONTECHNIQUESThe comparative analysis of the techniques will be doneunder the following matrices:Reliability: simply put is the mean time between failuresand the mean time to repair.Generally a uniform measurement for reliability isdifficult to define due to the variations in technologies. InGPS systems, reliability is defined as a measure of howconsistent a GPS horizontal error levels can bemaintained below a specified reliability threshold. Inmobile positioning methods, we can simply definereliability as the ratio of successful positioning attemptsout of all attempts made. Systems used in personalpositioning should uncompromisingly give extremelyreliable performance especially in emergency cases sincefailure could be very detrimental. This is an importantindicator of performance because a positioning devicethat frequently fails or gives false positions will not beseen as trustworthy by the users [4].Latency:Latency measure is basically the time from power-up tothe instant when the first location measurement isobtained. In GPS, this is known as the time-to-first-fix(TTFF). The high demand for low or short latency is notonly realized in the emergency cases but also from theLBS point of view. Example, the QoS and usability ofguidance and tracking applications decreases if onecannot guarantee a real-time operation. Short latency alsosaves power. It is measured in seconds. Latency is furtherbroken down into three main components by defining thecall set-up delay, network delay and the processing delay.Briefly, call set-up delay is the time elapsed between callinitiation at the MS and receiving the first response fromthe network. Network delay is defined as the time neededto transmit all messages excluding call set-up delay.Processing delay as the name implies is the time neededby the positioning device to measure and calculate theposition [5].Applicability:Basically applicability measures the physical limitationsand requirements associated with the implementation anduse of certain technology in terms of technical andfinancial issues]. The key issues affecting theapplicability of mobile positioning methods are powerconsumption, hardware size, software size, processingload, supported positioning modes, network dependency,signal load, cost and standardization. [6].22Accuracy and Precision: Location accuracy of ageolocation method is defined to be the distance betweenthe estimated location and the true location of the mobilestation. It defines how close the location measurementsare to the actual location of the mobile station beinglocated. Accuracy is expressed in meters.A location system should report locations accurately andconsistently from measurement to measurement. Inevaluating the accuracy of a positioning device, one hasto consider how many location measurements are made,how location measurements are scored and in what formthe results are presented. There are a number ofperformance metrics for location accuracy in theliterature. Root Mean Square Error (RMSE), MeanSquared Error (MSE), Circular Error Probability (CEP),and Cramer-Rao Lower Bound (CRLB). A widely usedscore function in accuracy evaluation of multiple locationmeasurements is the square root of the mean of squarederrors (RMSE), where RMSE is given by:AndWhere (X est , Y est ) is the estimated coordinates, and(X true ,Y true ) is the true location coordinates of the mobilestation. Where d is are distance errors and n denotes thenumber of location estimates. RMSE is the mostapplicable location accuracy performance metric in theliterature. It is the most used and a simple metric.Table 1.Grading of accuracy levelsAccuracyMargin of errorLowGreater than 150mMedium50 to 150mHigh Less than 50m.Cost: We can assess the cost of a location-sensing systemin several ways. Time costs include factors such as theinstallation process's length and the system administrationneeds. Space costs involve the amount of installedinfrastructure and the hardware's size and form factor.Limitations: Some systems will not function in certainenvironments. This limitation has implications for thekind of applications we can build.Coverage: coverage in telecommunications systems isrelated to the service area where, at a bare minimum,access to the wireless network is possible. For the mobilelocation techniques under consideration in this paper,coverage corresponds to the availability of a sufficientnumber of measurements to perform a locationcomputation

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012III. PERFORMANCE ANALYSIS OF TECHNIQUESThere are numerous techniques that can be considered foruse in wireless position location systems. However, wewill discuss only those techniques that can be practicallyincorporated in the GSM cellular-type systems. Suchtechniques can be broadly classified into two categories.Either the position location system will require amodification in the existing handsets or the system can bedesigned in such a way that all the modifications takeplace at the base stations or the switching centre with nomodifications in the existing handsets.From a physical localization point of view, all thetechniques discussed in this paper will be examined anddistinguished under the following principal techniques.(1) Signal strength and network parameters(2) Triangulation/Trilateration(3) Proximity(4) GSM finger printing/database correlationmethod.Signal strength and network parameters basedtechniques.(a) The Cell ID :Accuracy is based on coverage radius of serving cell, andtherefore is much better in urban environments andranges from 300 m to 20km. In this method, accuracylevel is dependent on factors such as cell size, cell type(Omni-directional or sectored), among other factor.However, there is no way to determine whether the useris, say 15 kilometres north, east, south or west of theserving base station. In areas where towers are closetogether, users can be located more precisely, as the rangeof each tower is smaller. In rural areas, however, wherethere are fewer towers and the range of each tower isgreater, the degree of accuracy is much less.Its main advantage is that it supports legacy handsets androaming subscribers. The lager the number of antenna’sproviding the sector information in the sectored type, themore accurate the location of the device will be. Here thehandset’s location is calculated using mathematicalalgorithms.(b) Timing Advance and Round Trip Time:Another primitive location estimation method is theTiming Advance (TA) in GSM andRound Trip Time in 3G systems. The TA is a measure ofthe handset range from a BTS (Base Transceiver Station),and has a resolution of 550m. It improves accuracy levelin larger (Omni-directional and sectored) cells, as inGSM 900. With TA, the handset registers with at leastthree base stations which send out a timing signal, forwhich the handset sends back a result to the network for aposition calculation.The Round Trip Time (RTT) of the signal between themobile and the serving base station can be used to get aradial distance measurement. The intersection of thecircle with radius equal to half the RTT and thehyperbolic curve indicates the position of the mobile.(c) Received Signal Strength Indication (RSSI):This is a network‐Based method of locating which usesthe Cell ID of the serving base station of the MS. Thistechnique makes no explicit attempt to resolve theposition of the mobile device beyond indicating the cellwith which the mobile device is (or has been) registered.When applied to 802.11 systems, this technique tracks thecell to which a mobile device associates. The Cell-ID isthen converted to a geographic position using knowledgeof the operator’s network, such as coverage database atthe serving mobile location centre (SMLC). It has a Lowcost of implementation. It is the most primitive method oflocating a wireless user and does not requireenhancements to be made to the network infrastructure orto the individual handset. It is, however, the least accurateof all location-identification technologies and does notmeet the requirements of Phase II of the E911 Act.Received Signal Strength (RSS) is used to calculate thedistance between transmitter and receiver based onknown transmitter power and transmitter location, and anappropriate path loss model. More realistic mobilecommunications channel model such as Okumura‐HataModel, can be used to convert the signal strengthmeasurements into distance.In this method, Radial distance from reference points isused to construct intersecting circles.Triangulation/ Trilateration:In trigonometry and elementary geometry, triangulation isthe process of finding a distance to a point by calculatingthe length of one side of a triangle formed by that pointand two other reference points, given measurements ofsides and angles of the triangle. Such trigonometricmethods used for position determination can bedistinguished as:23

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012- Distance –based (tri)-lateration (example: GlobalPositioning System, GPS). For distance-basedlateration, the position of an object is computedby measuring its distance from multiplereference points.- Angle or direction based (tri)- angulations(example: AOA, TDOA, EOTD)(a) GPSdifficult in heavily shadowed and covered urbanenvironments and impossible for indoor and undergroundcalls. All of these factors make it extremely unlikely thata GPS-based solution will be used to solve the positionlocation challenge in cellular-type systems. However, invehicle based mobile phones, GPS is a more realisticoption as it can be permanently installed in the car andmay be used whenever the phone is being used from thecar.(b) Angle of Arrival (AoA):Fig. 2 A GPS satellite systemOne of the most straightforward solutions to meet theposition location requirements is to use the GPS [7]. Itsaccuracy under ideal conditions, with clear open sky is onthe order of tens of meters (30m‐80m).This technique when applied to cellular phones mayinvolve installing a complete or a partial GPS receiver inthe handset and then transmitting the received GPS dataon the reverse link to the base station for furtherprocessing and position determination. However, numberof factors has made this method undesirable. The totalcost of building, staffing, maintaining and monitoring asatellite used for location identification purposes isextremely expensive. The satellites currently being usedhave life spans of 7–15 years, making periodicreplacement inevitable. Apart from increasing the sizeand weight of the handsets, the GPS receiver would resultin additional drain for the batteries in the mobile phones.This will result in reduced talk time which would behighly undesirable. Since GPS operates in L-Band(1575MHz) [8] which is different from either of thecellular or the PCS frequencies, the mobile antennawould need redesigning. All these factors would result inincreased cost for the handset. Another factor weighingagainst this solution is the “warm-up" time of the GPS.After being turned on, a GPS receiver takes at least one toone and a half minutes or even longer depending on thedesign, to start giving readings and, hence, this would behighly undesirable for an emergency 911 situation. Tofind the geographical location, the GPS receiver needs tohave at least four satellites visible at all times, which isFig. 3 AOA techniqueAOA methods are sometimes also referred to as Directionof Arrival (DOA) methods. AOA methods utilize multiarrayantennas and try to estimate the direction of arrivalof the signal of interest. Thus a single AOA measurementrestricts the source location along a line in the estimatedAOA. If at least two such AOA estimates are availablefrom two antennas at two different locations as shown infig 3 above, the position of the signal source can belocated at the intersection of the lines of bearings fromthe two antennas. To estimate the AOA, algorithms areused that exploit the phase differences or other signalcharacteristics between closely spaced antenna elementsof an antenna array and employ phase-alignment methodsfor beam/null steering. AOA works well in situationswith direct line of sight with low likelihood ofinterference such as sparsely populated areas, but suffersfrom decreased accuracy and precision when confrontedwith signal reflections from surrounding objects. In denseurban areas, AOA becomes barely usable because line ofsight to two or more base stations is seldom present. Theaccuracy of signals using AOA, however, is limited byvarious forms of signal interference. For example, cityskyscrapers may cause the signal to bounce. If the signalbounces, it is less likely to reach the target, resulting in aweak signal or none at all.24

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012Another factor is the considerable cost of installingantenna arrays considering that these antennas are atneeded at the base station. Although adaptive antennashold considerable promise for improving the capacity ofcellular systems, they would only be needed in the areaswhere capacity enhancement will be required. Hence, forrural and suburban areas which are sparsely populated,this would be a costly solution to meet the E-911requirements. Even if antenna arrays are in place at somebase stations, the position location system may needregular calibration since a minute change in the physicalarrangement of the array because of winds or storms, mayresult in considerable position location error as theabsolute angular position of the array is used as areference for the AOA estimates.TOA trilateration makes use of three sensors to allow themobile device location to be resolved with improvedaccuracy. A drawback of the TOA approach is therequirement for precise time synchronization of allstations, especially the mobile device (which can be adaunting challenge for some 802.11 client deviceimplementations) . Given the high propagation speeds,very small discrepancies in time measurement error assmall as 100 nanoseconds can result in a localizationerror of 30 meters. TOA-based positioning solutions aretypically challenged in environments where a largeamount of multipath, interference, or noise may exist.TOA techniques are more susceptible to errors frommultiple reflections than the hyperbolic method.(a) Time of Arrival (TOA)(b) TDOA.Fig. 4 TOA techniqueTime of arrival systems are based on the precisemeasurement of the arrival time of a signal transmittedfrom a mobile device to several receiving sensors.Because signals travel with a known velocity(approximately the speed of light (c) or approximately300 meters per microsecond), the distance between themobile device and each receiving sensor can bedetermined from the elapsed propagation time of thesignal travelling between them. The TOA techniquerequires very precise knowledge of the transmission starttime(s), and must ensure that all receiving sensors as wellas the mobile device are accurately synchronized with aprecise time source.From the knowledge of both propagation speed andmeasured time, it is possible to calculate the distance (D)between the mobile device and the receiving station.D=c(t).Where: D= distance (meters)c= propagation speed of approx. 300 meters /microsecondt= time in microsecondsWith distance used as a radius, a circular representationof the area around the receiving sensor can be constructedfor which location of mobile device is highly probable.Fig. 5It is obvious that even small errors in the click at thetransmitting or the receiving end causes a major error inthe distance estimate. Usually the mobile unit cannot besynchronized accurately enough to directly obtain thetime used by the signal to travel between a mobile unitand a base station. One solution to get around thesynchronization problem is to use differences in the timedelays of several base stations instead of absolute times.The Time‐Difference‐Of‐Arrival (TDOA) is measuredbetween multiple pairs of reference points with knownlocations. Each TDOA measurement yields a hyperboliccurve on which the mobile may reside. The intersectionof multiple curves produces the location estimate.If only two base stations are available, then only onehyperbolic curve is produced based on the TDOA.TDOA measurements do not require preciselysynchronized clocks at the fixed location receivers [9].This corresponds to the timing standards already providedat cellular base station sites, making TDOA more realisticthan requiring each mobile unit to have an accurate clock.25

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012Atomic clocks, such as a Cesium time source or a GPSreceiver clock, are typically used for timing at basestations. Hence, it is concluded that the TDOA techniqueappears to be the most feasible option for the wireless E-911 solution.Furthermore, this method offers many advantages overother competing techniques. Since all the processingtakes place at the infrastructure level; no modificationsare needed in the existing handsets. In this respect, thissolution would be more cost effective than a GPS-basedsolution. It also does not require knowledge of theabsolute time of the transmission from the handset like amodified handset TOA method needs. Since thistechnique does not require any special type of antennas, itis cheaper to put in place than the DOA finding methods.It can also provide some immunity against timing errorsif the source of major signal reflections is near themobile. If a major reflector affects the signal componentsgoing to all the receivers, the timing error may getcancelled or reduced in the time difference operation.Hence, TDOA methods may work accurately in somesituations where there is no LOS signal component. Inthis respect, it is superior to the DOA method or the TOAmethod.The required changes to incorporate the TDOA methodare only to be in the software of the system.them to the SMLC. E-OTD accuracy is dependent on celldensity, cell plan, multi-path, interference, noise, LMUperformance, and cell site position accuracy. In thistechnique the accuracy does not degrade much indoors,and it performs well in high base stations density areas.Conversely, E-OTD performance is degraded in areaswith low base stations density. It is an Expensive solutionfor operators with accuracy ranging from 50m to 500mand latency around 5s.Proximity:(a) Enhanced Observed Time Difference of ArrivalE-OTD.Cambridge Positioning Systems, a proponent of the E-OTD system, offers a product with positioning accuracywithin 50 meters. This time-based technique requireslocation receivers in the network infrastructure called aLocation Measurement Unit (LMU), and operates usingtime measurements of signal travelling distances betweenthe MS and the LMU as in the TDOA technique. Every30 seconds, each base station broadcasts a signal in anasynchronous manner.When signals from at least three base stations arereceived by the handset and LMU, the time differences ofarrivals of the signal from each base station at the handsetand the LMU are calculated. The differences in timestamps are combined to produce intersecting hyperboliclines and finally the location estimate of the MS isdetermined by the intersection of the hyperbolas.The LMUs help to determine the clock offsets betweenbase stations. One LMU is needed for every 3 to 5 basestations. However, the cell towers initiate thetransmission to the handset and LMU, not vice versa. TheLocation Measurement Units (LMUs) are essentiallymodified mobile devices, optionally with a GPS receiver,placed in a fixed geographical position, with thecapability to perform E -OTD measurements and return26Fig 6In this technique, Mobile position location is determinedby assigning the coordinates of a known location that is“close” to the mobile. This known location could be aBase station serving cell or the closest WiFi access point.It can also use the intersection of multiple overlappingareas to improve accuracy.Finger printing/database correlationThis technique utilizes the distant RF patterns (multipathphase and amplitude characteristics) of the radio signalsarriving at a receiver antenna from a single caller. Theunique characteristics of the signal, including itsmultipath pattern are analyzed and a “fingerprint” isdetermined for a defined area. The fingerprint is thencompared with a database of previously “fingerprinted”locations, and match is made. By matching the fingerprintof the caller’s signal with the database of knowfingerprints, the caller’s geographical location isidentified to one of the surveyed areas”. It has beenshown that an accuracy of 44m is possible in GSMnetworks.

LocationtechnologyTOATDOAAOARSSISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012ObservedinformationPropagation time ofreceived signalTime of arrival ofsignals emitted atthe same timeAngle of incidenceof received signalStrength of thereceived signalTable 3: Tabular presentation of results from indicesSuitableenvironmentUrban andsuburbanUrban andsuburbanRural areas andareas where directLOS is availableUrban and indoorAdvantages Disadvantages OthersHigh accuracy isachievable where thereis coherent detection-in the case wherethere is coherentdetection, a highaccuracy can beachieved-the receiver does notneed the time oftransmission-requires nomodification to thehandset.-requires nosynchronizationbetween MS and BS.-simple computation-independent of radiosystem-requires nosynchronization withBS-involves simplecalculations-low implementationcost-it is independent ofradio systems-no synchronization forbase Station is required-GPS Outdoors Requires no change tonetwork infrastructure,-Synchronizationis required-it is expensive-It involvescomplexcalculations.-expensive toimplement-Offers lowaccuracy inlarge cells.-High cost ofinstalling arrayantennaOffers lowaccuracy in lowcells-High batteryusage,-lacks indoorcoverage, andcoverage indenselypopulated areas-long signalacquisition timeE-OTD Urban/suburban -subject tourbanmultipathproblems-requires LMUin the networkinfrastructurePrivacy isnetworkcontrolledPrivacy isnetworkcontrolledPrivacy isnetworkcontrolled-privacy isusercontrolled.-changes inhandset areinevitable..27

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012IV. CONCLUSIONA brief summary of the results of the analysis is presentedin the table 4 below.A major element is that the resolution of the locationmethod typically depends on the size of the cells used. Inand around the cities, the cell sizes tend to be small, sothat the mobile station can be tracked with a reasonableaccuracy. Unfortunately in a rural area where there is asmall population and a low usage of the network, the typeof cell to be used will be a large macro cell with a largecoverage area, as there are no other cells within range toperform a handover, which gives information onneighbouring cell. Thus the location finding methodsoften give poor accuracies in the rural areas. A-GPSmethods are obviously the best techniques for locating themobile station, but it does suffer when there is poor radioreception, especially within buildings [10]. As has beenseen, A-GPS suffers from poor coverage in urbanenvironments and also within buildings.Table 4: Summary of resultsLocationTechnologyCategory Accuracy Cost Latency Coverage WirelessstandardReliabilityApplicabilityCell IDCell ID + SSAOATOANetworkbasedNetworkbasedNetworkbasedNetworkbased300m-200m200m-11km100m- Moderate Moderate,200m,,≈10sdecreaseswithincreasingdistancebetweenMS and BS50m-200m Low Moderate,

ISSN 2249-6343International Journal of Computer Technology and Electronics Engineering (IJCTEE)Volume 2, Issue 6, December 2012References[1] Wireless Geolocation System Architecturewww.cwins.wpi.edu/publications/pown/chapter_14.pdf[2] “FCC Report and Order and Further Notice of ProposedRule Making", FCC Docket96-264, June 1996.[3] Mika Keski-Heikkilä, “End-user Location in Digital MobileNetworks”. S-72.333 Post-Graduate Course in RadioCommunications2001-2002. 19th February, 2002[4] WJ Buchanan, J.Munoz, R.Manson, , K Raja, ”Analysis andMigration of Location-Finding Methods for GSM and 3GNetworks”.[5] Isaac K Adusei, K. Kyamakya, Klaus Jobmann. “MobilePositioning Technologies in Cellular Networks: AnEvaluation of their Performance Metrics”University ofHannover Institute of Communications Engineering,Hannover – Germany. 2002[6] L. Lopes, E. Viller, B. Ludden, “GSM standards Activity onlocation”. IEE Colloquium on novel methods of Locationand tracking of cellular mobiles and their systemsApplications. 1999.[7] Muhammad Aatique,” Evaluation of TDOA Techniques forPosition Location in CDMA Systems”. Thesis submitted tothe Faculty of the Virginia Polytechnic Institute and StateUniversity, September 1997[8] Jim Costabile,” Wireless Position Location”. Virginia TechWireless Symposium, June 4, 2010[9] Benjamin Woodacre, “TDOA Positioning Algorithms:Evaluation and Implementation”September 23, 2003[10] Hakan Senturk, “Performance Evaluation of HyperbolicPosition Location Technique in Cellular WirelessNetworks”. Thesis Presented to the Faculty of the GraduateSchool of Engineering and Management of the Air ForceInstitute of Technology, March 200229