Standardization of Mobile Phone Positioning for 3G Systems

IEEE Communications Magazine • July 2002108

Standardization of Mobile PhonePositioning for 3G Systems

0163-6804/02/$17.00 © 2002 IEEE

ABSTRACT

Finding the location of the mobile phone isone of the important features of the 3G mobilecommunication system. Many valuable location-based services can be enabled by this new fea-ture. Telecommunication managers andengineers are often puzzled by location termi-nologies and techniques as well as how to imple-ment them, since location systems are notnatural evolution from past generations oftelecommunication systems. In this paper, wediscuss briefly why locating mobile phonebecomes a hot topic and what technologies arebeing studied. We then describe and clarify thelatest standards issues surrounding the position-ing methods specified for 3G systems. Theseinclude cell-ID-based, assisted GPS, and TDOA-based methods, such as OTDOA, E-OTD, andA-FLT.

INTRODUCTIONThe U.S. Federal Communications Commission(FCC) has made Emergency 911 (E911) amandatory requirement for wireless communica-tions services such as cellular telephone, wide-band (broadband) personal communicationsservices (PCS), and geographic area specializedmobile radio (SMR). This ruling and upcomingservice is called wireless E911. For Phase IIimplementation, the FCC required that publicsafety answering point (PSAP) attendants ofwireless communications networks must be ableto know a 911 caller’s phone number for returncalls and the location of the caller so that callscan be routed to an appropriate PSAP andrelated emergency assistance attendants. In1999 the FCC decided to tighten the Phase IIlocation accuracy requirement from 125 m in 67percent of all cases to new numbers: for hand-set-based solutions, 50 m in 67 percent of callsand 150 m in 95 percent of calls; for network-based solutions, 100 m in 67 percent of calls and300 m in 95 percent of calls. Handset-basedsolutions are for nonlegacy phones. Network-based solutions are for legacy phones. In 2000the FCC required wireless communicationsoperators to offer operational location-capable

phones by October 1, 2001. On September 8,2000, the FCC granted a limited waiver to Voic-eStream with relaxed accuracy for an extendedperiod. Right after the October deadline of2001, waivers were granted to Alltel, AT&TWireless, Cingular, NEXTEL, Sprint PCS, andVerizon, permitting them to postpone sellingand activating location-capable phones that sat-isfy the new numbers until 2002 or later.

The executive body of the European Union(EU), the European Commission (EC), has simi-lar initiatives for their wireless emergency calls,E112. Coordination groups within EC have beenorganizing meetings to specify similar require-ments as their counterpart in the United States.On March 26, 2002, EU transport ministersapproved $482.4 million in funding for the devel-opment of the European satellite system, calledGalileo. This new system will rival the U.S.-developed global positioning system (GPS),which has already been adopted as one of thetechnologies for positioning, as discussed later.

In the United States, a recent survey by Har-ris Interactive indicated that consumers weremore interested in E911 than other new mobilephone features. Out of 1006 adults, 59 percentselected E911 service. At a distant second, 7 per-cent selected e-mail service.

Besides emergency assistance, wireless E911and its positioning capability will certainly trig-ger and enhance many location-based services.For instance, the telematics system [1] currentlyoffered by automotive manufacturers, such asGM’s OnStar and Mercedes-Benz’s TeleAid, canbe improved significantly. Future systems mayno longer need to have separated location andcommunication devices attached permanently tothe vehicle. Therefore, it is not difficult to under-stand why telecommunications manufacturersand operators have been actively pursuing thetechnologies to locate the mobile phone.

In this article we discuss location technologiesspecified by the 3rd Generation Partnership Pro-ject (3GPP) and 3GPP2, respectively. 3GPP hasbeen concentrating on wideband code-divisionmultiple access (W-CDMA) and Global Systemfor Mobile Communications (GSM) systemswhile 3GPP2 has been focusing on cdma2000and cdmaOne systems.

Yilin Zhao, Motorola, Inc.

STANDARDS REPORT

IEEE Communications Magazine • July 2002 109

BASIC LOCATION TECHNOLOGIES

There are three most commonly used locationtechnologies: standalone, satellite-based, andterrestrial-radio-based [2]. As examples, a typicalstandalone technology is dead reckoning; a typi-cal satellite-based technology is GPS; and a typi-cal terrestrial radio-based technology is the “C”configuration of the Long Range Navigation(LORAN-C) system. For wireless E911, E112,and many other applications, radio-based (satel-lite and terrestrial) technologies are most popu-lar. Cellular networks are terrestrial-basedcommunications systems. It is natural to utilizethe signals of the network to determine themobile phone location or assist in location deter-mination. Research in this area has been veryactive recently as evidenced by the latest roundof publications and conferences. The principlesbehind them are discussed below.

Radio-based technology typically uses basestations, satellites, or other devices emittingradio signals to the mobile receiver to determinethe position of its user. Signals can also be emit-ted from the mobile device to the base. Com-monly studied techniques are angle of arrival(AOA) positioning, time of arrival (TOA) posi-tioning, and time difference of arrival (TDOA)positioning. All these methods require radiotransmitters, receivers, or transceivers. In otherwords, they depend on emitting and receivingradio signals to determine the location of anobject on which a radio receiver or transceiver isattached. To make the position determination,these methods generally work under the assump-tion that one end of the positioning system isfixed or known and the other movable (e.g., amobile phone). However, the location determi-nation capability can be at either the fixed or themobile end. Generally, it is up to the systemdesigner to decide where the final location deter-mination capability should reside. For perfor-mance improvement, hybrid methods (variouscombinations of the techniques discussed or withadditional techniques) are possible. To simplifyour discussion, in the following we use two-dimensional (2D) cases as application examples.Readers should be able to expand the principlespresented to 3D cases.

The AOA system determines the mobilephone position based on triangulation, as shownin Fig. 1a. It is also called direction of arrival insome literature. The intersection of two direc-tional lines of bearing defines a unique position,each formed by a radial from a base station tothe mobile phone in a 2D space. This techniquerequires a minimum of two stations (or one pair)to determine a position. If available, more thanone pair can be used in practice. Because direc-tional antennas or antenna arrays are required,it is difficult to realize AOA at the mobile phone.

The TOA system determines the mobilephone position based on the intersection of thedistance (or range) circles (Fig. 1b shows a 2Dexample). Since the propagation time of theradio wave is directly proportional to its tra-versed range, multiplying the speed of light tothe time obtains the range from the mobilephone to the communicating base station. Tworange measurements provide an ambiguous fix,

and three measurements determine a uniqueposition. The same principle is used by GPS,where the circle becomes the sphere in spaceand the fourth measurement is required to solvethe receiver-clock bias for a 3D solution. Thebias is caused by the unsynchronized clocksbetween the receiver and the satellite.

The TDOA system determines the mobilephone position based on trilateration, as shownin Fig. 1c. This system uses time difference mea-surements rather than absolute time measure-ments as TOA does. It is often referred to as thehyperbolic system because the time difference isconverted to a constant distance difference totwo base stations (as foci) to define a hyperbolic

� Figure 1. Position determination methods: a) angle of arrival; b) time ofarrival; c) time difference of arrival.

h2

h1

d2d3d1

Site 2

Site 3

Site 1

Possiblelocation

Possiblelocation

Site

SiteSite

(a)

(b)

(c)

Site 1 Site 2

Possiblelocation

α1 α2


curve. The intersection of two hyperbolas deter-mines the position. Therefore, it utilizes twopairs of base stations (at least three for the 2Dcase shown in Fig. 1c) for positioning. The accu-racy of the system is a function of the relativebase station geometric locations. For terrestrial-based systems, it also requires either preciselysynchronized clocks for all transmitters andreceivers or a means to measure these time dif-ferences. Otherwise, a 1 µs timing error couldlead to a 300 m position error.

Other location methods can also be used.One simple method for mobile phone position-ing is to use the cell area (or cell ID) of thecaller, assisted by other coarse estimates, asthe approximate location of the mobile phone.We will discuss this method further in the nextsection. Another method is to use short-rangebeacons (or signposts in some literature)installed in the coverage area to provide loca-tion-specific information to the mobile receiver[2]. Due to its limited communication zones,discontinuous communication, and high systeminstallation and maintenance costs, it has notbeen considered for mobile phone positioning.Another alternative is to use other radio signals,such as TV or AM/FM radio broadcast signals,in place of cellular or satellite signals. In mostareas there are plenty of these radio signals.The coverage is generally better than that pro-vided by cellular signals. Other methods arebased on measuring the signal strength or signalcharacteristic patterns and multipath character-istics of radio signals arriving at a cell site froma caller. For measuring the signal strength, itemploys multiple cell sites to find the location.For measuring the signal characteristic pat-terns, it identifies the unique radio frequencypattern or signature of the call and matches itto a similar pattern stored in its centraldatabase.

LOCATION TECHNOLOGIESSPECIFIED BY 3GPPLOCATION TECHNOLOGIESSPECIFIED FOR UTRAN

In the Universal Terrestrial Radio Access Net-work (UTRAN), a handset is called user equip-ment (UE) and a base station is called node B.There are two operational modes for UTRAN:frequency-division duplex (FDD) and time-divi-sion duplex (TDD). The original standards spec-ifications were developed based on FDD mode.

Three location techniques have been specifiedfor UTRAN [3–8]: the cell-ID-based, observedTDOA (OTDOA), and assisted GPS (A-GPS)methods. When the mobile phone position is cal-culated at the network, we call it a UE-assistedsolution. When the position is calculated at thehandset, we call it a UE-based solution. Note thatexcept for the UE-assisted OTDOA method, therest of the methods are optional in the UE.

Figure 2 illustrates the system architecture ofUE positioning (UP). The UTRAN interfaces(Uu, Iub, Iur, and Iupc) are used to communicateamong all relevant entities. In this figure SRNCstands for serving radio network controller, LMUfor location measurement unit, SAS for stan-dalone serving mobile location center (SMLC),and CN for core network. LMU type A is a stand-alone LMU; type B is integrated with a base sta-tion.

For Release 99 (standards release frozen inDecember 1999) and Release 4 (frozen in March2001), the SAS supports A-GPS only. In Release5 (frozen in March 2002), the SAS will supporttwo other location methods as well (Cell ID andOTDOA). Deployment of the SAS is optional. Ifan SAS is not included in the system, the SMLC(location server) will be an integral part of theRNC or SRNC. Note that later releases havemore features (functionalities) than early ones.“Frozen” means that the contents (or features)of a specific release have been decided, but thedetailed protocols and functionalities may not bestable or finalized. These releases are updatedquarterly.

The Cell-ID-Based Positioning Method —The cell-ID-based method determines the UEposition at the network. In other words, theposition of a UE is estimated based on the cov-erage information of its serving node B. Thisknowledge can be obtained by paging, locatingarea update, cell update, UTRAN registrationarea (URA) update, or routing area update.This method is optional for the network. Despitenot being mandatory, it is the author’s sugges-tion that we should implement it as a defaultlocation method. Whenever OTDOA or A-GPSfails to locate the UE, we can always use thismethod to provide approximate information onmobile phone position to the network.

Depending on the operational status of theUE, additional operations may be needed inorder for the SRNC to determine the cell ID.When the location service (LCS) request isreceived from the CN, the SRNC checks thestate of the target UE. If the UE is in a statewhere the cell ID is not available, the UE is

� Figure 2. System architecture of UE positioning.

IupcUu

Uu Iub Iu

Iur

IubOptional

Mandatory

LMUtype A

UTRAN

UE

Node B(LMU

type B)

SAS

Node B(LMU

type B)

RNC

CNSRNC


paged so that the SRNC can establish the cellwith which the target UE is associated. In stateswhere the cell ID is available, the target cell IDis chosen as the basis for the UE positioning. Insoft handover, the UE may have several signalbranches connected to different cells whilereporting different cell IDs. The SRNC needsto combine the information about all cells asso-ciated with the UE to determine a proper cellID. The SRNC should also map the cell ID togeographical coordinates or a correspondingservice area identity (SAI) before sending itfrom the UTRAN to the CN. This can easilymatch the service coverage information avail-able in the CN.

In order to improve the accuracy of the LCSresponse, the SRNC may also request additionalmeasurements from node B or the LMU. Thesemeasurements are originally specified for softhandover. For FDD mode, round-trip time(RTT) can be used as a radius of a cell to fur-ther confine the cell coverage. RTT is the timedifference between the transmission of the begin-ning of a downlink dedicated physical channel(DPCH) frame to a UE and the reception of thebeginning of the corresponding uplink from theUE. For TDD mode, received (RX) timing devi-ation can be used. RX timing deviation is thetime difference between the reception in node Bof the first detected uplink path and the begin-ning of the respective slot according to the inter-nal timing of node B. The measurements arereported to higher layers, where timing advancevalues are calculated and signaled to the UE.For better accuracy, for instance in FDD mode,a mandatory UE Rx-Tx time difference type 1 oran optional type 2, if available, can be combinedwith RTT to determine the distance from a nodeB to a UE. UE Rx-Tx time difference is the dif-ference between the UE uplink DPCCH/DPDCH frame transmission and the first detect-ed path (in time) of the downlink DPCH framefrom the measured radio link, where DPCCHstands for dedicated physical control channeland DPDCH stands for dedicated physical datachannel. The main differences in type 1 and type2 are the measurement resolution and referenceRx path (or first detected path). For type 2, theresolution is better and the reference path mustbe the path among all paths detected by the UE.In contrast, for type 1, the reference path is theone used in the demodulation process.

The cell-ID-based method should determinethe position of the UE regardless of the UE oper-ational mode (i.e., connected or idle). Thismethod results in a position error as large as thecell area if no additional measurements are used.For instance, a picocell could be 150 m in radius,while a large cell could be more than 30,000 m inradius. Therefore, this method has not demon-strated that it can achieve 100 m accuracy reliablyeven under the best of conditions.

The OTDOA Positioning Method — OTDOAis a TDOA-based positioning method. It deter-mines the position of the mobile phone based ontrilateration as shown in Fig. 3. Two methodsare specified for OTDOA: UE-assisted OTDOAand UE-based OTDOA.

As discussed above for a TDOA-based

method, the accuracy of the system is a functionof the relative node B geometric locations. Addi-tional environmental constraints dictate that themore measurements we can obtain, the betterposition fixes we will receive. Since these mea-surements are based on the signals from nodeBs, the locations of these base stations are nec-essary for the network or UE to calculate thehandset positions. If the transmitters of node Bin UTRAN are unsynchronized, the relative timedifference (RTD) must be provided for systemframe numbers (SFN). It is named SFN-SFNobserved time difference. One way to obtainthese measurements is to deploy LMUs, whichperform timing measurements of all the localtransmitters in fixed locations of the network.These measurements can be converted to RTDsand transmitted to the UE or RNC for position-al calculations. In addition, the UE also mea-sures the SFN-SFN observed time difference,which identifies the time difference between twocells as TDOA. Two types are defined. Asexplained later, type 1 is used for soft handoverand type 2 is used for positioning. The main dif-ference of these two types is that type 2 is appli-cable for both idle and connected modes, whiletype 1 supports intrafrequency measurementsand cannot do interfrequency measurements forthe connected mode. Since node Bs in TDDmode are generally synchronized, the RTDs aretypically constant. Similarly, in FDD if the rele-vant cells are synchronized, measurements ofRTDs would not be necessary. For FDD mode,RTT and Rx-Tx time difference can be obtainedto improve the performance of the OTDOA.For TDD mode, RX timing deviation can beobtained to improve performance. Since adap-tive or smart antennas have been specified as afeature for 1.28 Mchips/s TDD networks, AOAcan be used to further improve the OTDOA andcell ID performances.

The OTDOA location method in UTRANhas its problems, such as hearability, unsynchro-nized base stations for FDD mode, geometriclocation of the base stations, and capacity loss.For the hearability problem, this can occur whenthe UE is very close to its serving node B, which

� Figure 3. OTDOA positioning method.

EstimatedUE position

region

OTDOAhyperboloid

RTT circle


could block the reception of other base stationsignals in the same frequency. From Fig. 3 weknow the UE must be able to hear at least threebase stations in order to perform its locationduty. For the unsynchronized base station prob-lem, it causes significant uncertainty to theTDOA measurements. For the geometric loca-tion of the base station problem, the locations ofthe contributing base stations could affect theavailability and quality of the position fixes. Forinstance, on a long straight highway in a ruralarea, OTDOA may fail to produce requiredsolutions because node Bs may simply lie alongthe same highway. For the capacity loss problem,the system could provide less capacity to cus-tomers since the signal and traffic channels aswell as some of the computing power of the UEor network entity could be occupied by the loca-tion services.

In order to improve the hearability of neighbor-ing node Bs, one option specified is the idle perioddownlink (IPDL). In this method, each node Bceases its transmission for short periods of time(idle periods). During an idle period of a node B,UEs within the cell can measure other node Bs, sothe hearability problem is reduced. By using signal-ing over the Uu interface, UEs are made aware ofthe occurrences of IPDLs, so they can arrange thetime difference measurements accordingly. Sincethe IPDL method is based on downlink, the loca-tion service can be provided efficiently to a largenumber of UEs simultaneously.

For UE-assisted OTDOA, essential informationelements (IEs) or assistance data from UTRAN toUE are reference and neighbor cell information.For UE-based OTDOA, they are reference andneighbor cell information as well as node B posi-tions of these cells. UE-assisted OTDOA is manda-tory for the UE and optional for the UTRAN. TheUE-based OTDOA is optional for both the UE

and UTRAN. Notice that for UE-based OTDOA,the length of the downlink IE is longer than theUE-assisted OTDOA. In contrast, the length of theuplink IE is shorter since one reports 2D/3D UEposition and another reports measured TDOAresults. In addition, UTRAN-to-UE informationtransfer is specified for both point-to-point andbroadcast transmissions. For broadcast, UE-assist-ed OTDOA IEs are defined as system informa-tion block type 15.4 and UE-based OTDOA IEsare defined in as system information block type15.5 [8].

The Assisted GPS Positioning Method —GPS provides an affordable means to deter-mine position, velocity, and time around theglobe. The satellite constellation is developedand maintained by the U.S. Department ofDefense. Civilian access is guaranteed throughan agreement with the Department of Trans-portation. GPS satellites transmit two carrierfrequencies. Typically, only one is used by civil-ian receivers. From the perspective of thesecivilian receivers, GPS satellites transmit at1575.42 MHz using CDMA, which uses a direct-sequence spread-spectrum (DS-SS) signal at1.023 MHz (Mchips/s) with a code period of 1ms. Each satellite’s DS-SS signal is modulatedby a 50 b/s navigation message that includesaccurate time and coefficients (ephemeris) toan equation that describes the satellite’s posi-tion as a function of time. The receiver (moreprecisely, its antenna) position determination isbased on TOA.

The four main conventional GPS receiverfunctions are:• Measuring distance from the satellites to

the receiver by determining the pseudo-ranges (code phases)

• Extracting the TOA of the signal from thecontents of the satellite transmitted mes-sage

• Computing the position of the satellites byevaluating the ephemeris data at the indi-cated TOA

• Calculating the position of the receivingantenna and the clock bias of the receiverby using the above data itemsPosition errors at the receiver are contributed

by the satellite clock, satellite orbit, ephemerisprediction, ionospheric delay, troposphericdelay, and selective availability (SA). SA is anaccuracy degradation scheme to reduce the accu-racy available to civilian users to a level withinthe national security requirements of the UnitedStates. It decreases the accuracy capability ofautonomous GPS to the 100 m (2D-RMS) level,where RMS stands for root mean square. Toreduce these errors, range and range-rate cor-rections can be applied to the raw pseudo-rangemeasurements in order to create a position solu-tion that is accurate to a few meters in openenvironments. The most important correctiontechnique is differential GPS (DGPS). It uses areference receiver at a surveyed position to sendcorrecting information to a mobile receiver overa communications link. Note that since May2000 SA has been turned off, which often resultsin an accuracy of under 20 m in unobstructedenvironments.

� Figure 4. Assisted GPS positioning method.

Node B/RNC GPS


To deal with the following problems facingthe standalone conventional GPS receiver, theA-GPS method was specified to improve theperformance of GPS (Fig. 4):• Its startup time (from turning on to the ini-

tial position fix) is relatively long due to thelong acquisition time of the navigation mes-sage (at least 30 s to a few minutes).

• It is unable to detect weak signals that resultfrom indoor and urban canyon operationsas well as small cellular-sized antennas.

• Its power dissipation is relatively high perfix, primarily due to the long signal acquisi-tion time in an unaided application.The basic idea of assisted GPS is to establish

a GPS reference network (or a wide-area DGPSnetwork) whose receivers have clear views of thesky and that can operate continuously. This ref-erence network is also connected with the cellu-lar infrastructure, continuously monitors thereal-time constellation status, and provides datasuch as approximate handset position (or basestation location), satellite visibility, ephemerisand clock correction, Doppler, and even thepseudorandom noise code phase for each satel-lite at a particular epoch time. At the request ofthe mobile phone or location-based application,the assist data derived from the GPS referencenetwork are transmitted to the mobile phoneGPS receiver (or sensor) to aid fast startup andincrease sensor sensitivity. Acquisition time isreduced because the Doppler vs. code phaseuncertainty space is much smaller than in con-ventional GPS due to the fact that the searchspace has been predicted by the reference receiv-er and network. This allows for rapid searchspeed and a much narrower signal search band-width, which enhances sensitivity and reducesmobile receiver power consumption. Once theembedded GPS receiver acquires the availablesatellite signals, the pseudo-range measurementscan be delivered to RNC or SAS in UTRAN forposition calculation or used internally in the UEto compute position.

Additional assisted data, such as real-timeintegrity, DGPS corrections, satellite almanac,ionospheric delay, and universal time coordinat-ed (UTC) offset can be transmitted to improvethe location accuracy, decrease acquisition time,and allow for different position computationsolutions. Besides adding a GPS reference net-work and additional location determinationunits in the network, the mobile phone mustembed, at a minimum, a GPS antenna and RFdownconverter circuits, as well as make provi-sion for some form of digital signal processingsoftware or dedicated hardware. Despite thefact that A-GPS can improve the performanceof a conventional GPS receiver, it cannot beused for legacy phones already on the market.As in the case of OTDOA, there are two solu-tions for A-GPS. A UE can support either oneor both of them.

The UE-assisted solution shifts the majorityof the traditional GPS receiver functions to thenetwork processor. This method requires anantenna, RF section, and digital processor inthe UE for making measurements by generat-ing replica codes and correlating them with thereceived GPS signals. The network transmits

an assistance message to the mobile station,consisting of time, visible satellite list, satellitesignal Doppler and code phase, as well as theirsearch windows or, alternatively, approximatehandset position and ephemeris. These IEshelp the embedded GPS sensor reduce theGPS acquisition time. The assistance data ofDoppler and code phase are valid for a fewminutes, while ephemeris data last two to fourhours. It returns from the UE the pseudo-range data processed by the GPS sensor. Afterreceiving the pseudo-range data, the locationserver in the SRNC or SAS estimates the posi-tion of the UE. The differential correction(DGPS) can be applied to the pseudo-rangedata or final result at the network side toimprove the position accuracy.

The UE-based solution maintains a fullyfunctional GPS receiver in the handset. Thisrequires the same functionality described forUE-assisted GPS, plus additional means forcomputing the positions of the satellites andultimately the UE’s position. This additionalhandset function generally adds to the hand-set’s total memory (RAM, ROM) requirementsin addition to extra computing capability suchas million instructions per second (MIPS). Inthe initial startup scenario, data in the form ofthe precise satellite orbital elements(ephemeris) must be provided to the UE. Thisdata is valid for two to four hours and can beextended to cover the entire visible period ofthe GPS satellite (i.e., up to 12 hours), as dis-cussed later. Thus, once the handset has thedata, subsequent updates are rare. For betterpositional accuracy or longer ephemeris life,differential correction (DGPS) data should betransmitted to the UE. The final position of theUE is generated at the UE itself. The calculat-ed UE position can then be sent to an applica-tion outside of the UE if required.

For UE-assisted GPS, the essential IEs fromUTRAN to UE are reference UE position, GPSreference time, plus either code phase andDoppler (acquisition assistance) or satellite posi-tions (ephemeris and clock correction). For UE-based GPS, they are reference UE position, GPSreference time, plus satellite positions and failed/failing satellites information (real-time integrity).The UE can also request additional GPS IEs.From UE to UTRAN, UE-assisted GPS reportsmeasured pseudo-range results, while UE-basedGPS reports 2D/3G UE position. For UE-basedGPS, UTRAN-to-UE information transfer isspecified for both point-to-point and broadcasttransmission. For broadcast, the essential IEsare defined as system information block types 15and 15.2 [8].

There are some interesting features definedin the specifications to improve the performanceof the A-GPS. In the DGPS corrections IE [8],three pairs of pseudo-range correction (PRC)and range-rate correction (RRC) related fieldsare specified. Proper usage of these pairs willextend the life of the ephemeris and clock cor-rection IE [8], which in turn will reduce trans-mission bandwidth, handset power consumption,memory, and CPU load. Typically, the ephemerisIE is valid at the UE for 2–4 h. The PRC2/RRC2pair is used to extend the life for 6 h and the

The basic idea of

assisted GPS is

to establish a

GPS reference

network (or a

wide-area DGPS

network) whose

receivers have

clear views of the

sky and which

can operate

continuously.

This reference

network is also

connected with

the cellular

infrastructure.


PRC3/RRC3 pair is used for 8 h. To extend thelife of the ephemeris IE beyond 8 h in broadcastor beyond 2–4 h in point-to-point operations,tailored DGPS can be used: a UE informs theRNC or SAS of its issue of data ephemeris(IODEs) and expected use time of its storedephemeris [8]; then the RNC provides aPRC/RRC pair specific to that UE. The basicidea is to remove the errors contained in theephemeris IE when needed. Since the maximumvisible period of GPS satellites for one path is 12h, this feature avoids transmitting the longephemeris IE more than once during the period.By using this method, a UE wakes up for a par-ticular correction pair only when a location fix isrequired.

The real-time integrity IE [8] is essential forproper operation of A-GPS. What this messagecovers is abnormal situations out of the controlof the GPS control segment (master stations onthe ground). For example, the atomic clock of aGPS satellite can fail suddenly, which in turnproduces wrong satellite signals. In other words,the location accuracy can degrade substantiallywhen undetected GPS satellite failures are pre-sent. Such failures can render the positioninginformation derived by a UE completely unus-able. Although the control segment monitors thehealth of the GPS satellites, this activity is notperformed continuously. It may require morethan 30 min for the control segment to commu-nicate this to GPS users. To provide the real-time integrity IE, one has to deploy an integritymonitor (IM) in the RNC or SAS. Besidesinforming of failed/failing satellites, it also tellsthe UE of measurement quality when satellitesare healthy. This is done through the supplieduser differential range error (UDRE), which isone field in the DGPS correction IE. The UEuses the UDRE as a factor in weighing dataobtained from associated satellites in its positioncalculation.

The above features are also specified in GSMstandards. To reduce infrastructure investment,a shared location server can be implemented tosupport dual-mode phones operating in bothGSM and UMTS/W-CDMA networks. An addi-tional benefit of implementing these features isthat the RNC or SAS has to provide differentialcapability to the system. This capability cangreatly improve positioning performance, such aswith unexpected multipath in the server andpoor geometry of visible satellites for the UE.

Finally, in order to better understand the A-GPS method, we list the pros and cons of UE-based and UE-assisted GPS in Table 1. Readerscan compare them to decide which is suitable fortheir specific application.

To clarify positioning measurements, Table 2lists all UE and UTRAN related measurementelements. For UE, three elements are specifiedfor other purposes. SFN-SFN observed time dif-ference type 1 is used to identify the time differ-ence between two cells. SFN-CFN observedtime difference is used for handover timing pur-pose to identify active cell and neighbor celltime difference, where CFN stands for connec-tion frame number. Note that SFN-CFNobserved time difference is defined as cell syn-chronization information in radio resource con-trol protocol [8]. Both are used for softhandover. The difference is that SFN-SFN is forestablishment of a call directly into soft han-dover, and SFN-CFN is for addition of newradio links into the soft handover for an alreadyexisting call. Rx-Tx timing difference type 1 isused for call setup purposes to compensate forthe propagation delay between uplink and down-link transmissions, and to tell the network thatthe received timing of a cell is moving out ofthe UE’s soft combining window. In addition tosoft handover, it can also be used to improveposition determination performance. ForUTRAN, all the elements listed are optional

� Table 1. Comparison of UE-assisted and UE-based GPS.

Pros/Cons UE-Based GPS UE-Assisted GPS

Advantage Relatively short uplink IE Relatively short downlink assistance IE if codephase and Doppler are used

Assistance IE valid for 2–4 hours or up Network in control of position determinationto 12 hours at the UE if ephemeris lifeextension feature is used (less signaling)

Good for tracking/navigation applications Need less computing power and memory atthe UE

Can be used as a standalone GPS receiver

Do not need LMU

Disadvantage Relatively long downlink assistance IE Relatively long uplink IE

Need more computing power and Assistance IE valid for a few minutes at thememory at the UE UE if code phase and Doppler are used (more

signaling for tracking/navigation applications)

Need accurate timestamps at the UE

Certain event trigger mechanisms will not work

Need LMU for certain assistance data

The basic idea

is to remove the

errors contained

in the ephemeris

IE when needed.

Since the

maximum visible

period of GPS

satellites for one

path is 12 hours,

this feature avoids

transmitting the

long ephemeris

IE more than

once during

the period.


and can be used for position determination.They are also been used for Uu, Iub, Iur, andIupc interfaces, as shown in Fig. 2.

For 3.84 Mchips/s systems, 1 chip is about0.26 µs. This translates to 78.125 m in rangeuncertainty. For 1.28 Mchip/s systems, 1 chip isabout 0.78 µs. This translates to 234.375 m inrange uncertainty.

LOCATION TECHNOLOGIESSPECIFIED FOR GERAN

In the GSM EDGE Radio Access Network(GERAN), where EDGE stands for enhanceddata rates for GSM evolution, three locationmethods are specified: cell ID, enhancedobserved time difference (E-OTD), and A-GPS.They are all inherited and evolved from themethods specified for GSM. E-OTD is a TDOApositioning method based on the OTD featurealready existing in GSM. In principle, it is simi-lar to OTDOA but operates in TDMA-basednetworks. A-GPS method is similar to that speci-fied for UTRAN, but with different IE lengthsand formats. Note that GERAN has not adoptedan uplink TOA method specified for GSM.Readers can refer to [9] for more informationregarding these methods.

LOCATION TECHNOLOGIESSPECIFIED BY 3GPP2

Advanced forward link trilateration (A-FLT)and A-GPS have been standardized by Telecom-munications Industry Association’s TR-45.5 asIS-801 (IS-801-1 is its addendum) [10]. The nextrelease, IS-801-A, is being handled by 3GPP2.Unlike GSM and W-CDMA, cdmaOne andcdma2000 are time-synchronized systems. There-fore, time difference measurement from them iseasier than for GSM and W-CDMA.

The basic idea of the A-FLT method is tomeasure the time difference (phase delay)

between CDMA pilot signal pairs. Each pairconsists of the serving cell pilot and a neigh-boring pilot. The time difference is convertedto range information. Finally, the range data isused to form certain hyperbolic curves atwhich an intersection is defined for handsetlocation. Since the principle of this method isnot much different than TDOA, we will notdiscuss it in detail.

Although the essential A-GPS IEs specifiedin IS-801 are very similar to those defined in3GPP for UE-assisted and UE-based A-GPS, IS-801 offers more options. In other words, it pro-vides more than one way to accomplish the sametask. For instance, two IEs are defined to pro-vide the reference UE position, one based onspherical coordinates and one on Cartesiancoordinates. Despite these additional options, itdoes not support DGPS and broadcast specifiedby 3GPP. The new release, IS-801-A, is likely toaddress these issues in addition to adding otherfeatures and utilizing new channels defined forcdma2000.

CONCLUSIONSIn general, among the three methods specified,the cell-ID-based method has the worst position-al accuracy, while A-GPS has the best positionalaccuracy. For cell-ID-based methods, the accura-cy should be very close to the radius of the cell.For TDOA-based methods, it may achieve anaccuracy of under 100 m in 67 percent of calls.For A-GPS methods, an accuracy of under 20 mis a very reasonable expectation in 67 percent ofcalls when SA is off. Many factors affect the per-formance of these methods, as discussed in theprevious sections, such as cell sizes, hearabilityof other base stations, visibility of GPS satellites,and multipath. An experienced GPS user maynotice that the accuracy of A-GPS in open skyenvironments is similar to an ordinary GPSreceiver in the market. Meanwhile, an A-GPSreceiver may have better coverage in obstructed

� Table 2. Measurement elements.

Element Accuracy (chips) Choice Comment

UE Rx-Tx time difference type 1 1.5 for FDD Mandatory For soft handover

Rx-Tx time difference type 2 TBD for FDD Optional For position determination

SFN-SFN observed time 1 for FDD, 0.5 for TDD Mandatory For soft handover of establishing a calldifference type 1

SFN-SFN observed time 0.5 for FDD intra-frequency, 1 for Mandatory For position determinationdifference type 2 FDD inter-frequency, 0.5 for TDD

Cell synchronization 1 for FDD, 0.5 for TDD Optional For soft handover on an existing callinformation

GPS timing of cell frames TBD (to be determined) Optional For UE-assisted GPS position determination

UTRAN SFN-SFN observed time 0.5 Optional For position determinationdifference

RTT 0.5 for FDD Optional Can also been used for position determination

RX Timing Deviation 0.5 for TDD Optional Can also been used for position determination

GPS timing of cell frames TBD Optional For position determination


environments. Compared with other radio-basedtechnologies, A-GPS typically has better accura-cy, but worse coverage than TDOA/TOA/AOAin buildings and urban canyon areas. On theother hand, the solution quality of TDOA/TOA/AOA depends heavily on the geometriclocation of the contributing base stations. Tofurther improve these methods, hybrid approach-es can be used [9]. One example is to combineOTDOA and A-GPS. Another is to combineOTDOA and AOA. Current specifications haveleft the door open for such fusions.

In 3GPP, new UE positioning enhancementsolutions have been postponed until existingmethods have been finished and stable. Thesenew solutions include a thin UE GPS method,an uplink TDOA method, and a cumulative vir-tual blanking method to replace IPDL. 3GPPRAN is still working on the impact study ofIPDL on the quality of service and measure-ment performance of the UE, and the openinterface to support all three positioning meth-ods discussed. GERAN continues working onlocation service in Iu mode. The proposed newmethods have to wait until Release 6 (standardsrelease expected to be frozen in June 2003) tobe considered.

In 3GPP2, work for the next standards releasehas already begun. It is very likely they willaddress the issues of supporting broadcast,DGPS, and other new features, as well as how toutilize new channels specified for cdma2000. Itwould be beneficial if 3GPP and 3GPP2, as wellas 3GPP GERAN and 3GPP RAN, can studyhow to harmonize their respective applicationinterfaces and positioning protocols in futurereleases, particularly those independent of multi-ple access techniques.

REFERENCES[1] Y. Zhao, “Telematics: Safe and Fun Driving,” IEEE Intell.

Sys., vol. 17, no. 1, Jan./Feb. 2002, pp. 10–14.[2] Y. Zhao, Vehicle Location and Navigation Systems, Nor-

wood, MA: Artech House, 1997, translated and pub-lished in Chinese by Publishing House of ElectronicIndustry, Beijing, 1999.

[3] 3GPP TS 25.123, Requirements for Support of RadioResource Management (TDD), Mar. 2002.

[4] 3GPP TS 25.133, Requirements for Support of RadioResource Management (FDD), Mar. 2002.

[5] 3GPP TS 25.215, Physical Layer Managements (FDD),Mar. 2002.

[6] 3GPP TS 25.225, Physical Layer Managements (TDD),Mar. 2002.

[7] 3GPP TS 25.305, Stage 2 Functional Specification of UEPositioning, Mar. 2002.

[8] 3GPP TS 25.331, Radio Resource Control (RRC) ProtocolSpecification, Mar. 2002.

[9] Y. Zhao, “Mobile Phone Location Determination and ItsImpact on Intelligent Transportation Systems,” IEEETrans. Intell. Trans. Sys., vol. 1, no. 1, Mar. 2000, pp.55–67.

[10] TIA/EIA/IS-801-1, Position Determination Service Stan-dard for Dual-Mode Spread Spectrum Systems–Adden-dum, Mar. 2001.

BIOGRAPHYYILIN ZHAO [SM] ([email protected]) received his B.E. fromDalian University of Technology (DUT) and his M.S. andPh.D. in electrical engineering: systems from the Universityof Michigan, Ann Arbor. He is a Distinguished Member ofthe Technical Staff at Motorola, adjunct professor at DUT,and senior visiting scholar at Chinese Academy of Sciences.As a delegate to T1, ETSI, and 3GPP, he made importantcontributions in standardizing location technologies forGSM/GPRS and UMTS/W-CDMA networks. His researchinterests include mobile phone architecture and its applica-tions, embedded systems, IP-based networks, location andnavigation systems, and intelligent transportation systems(ITS). He has delivered tutorials and seminars at many uni-versities, IEEE, SAE, and other international conferences. Heis a vice president of the IEEE ITS Council and associateeditor of the council’s Transactions on Intelligent Trans-portation Systems.

It would be

beneficial if

3GPP and 3GPP2,

as well as 3GPP

GERAN and 3GPP

RAN, can study

how to harmonize

their respective

application

interfaces and

positioning

protocols in

future releases.

Business

Standardization of Mobile Phone Positioning for 3G Systems