DEVELOPMENT AND DEPLOYMENT OF GPS WIRELESS DEVICES FOR E91 1
`AND LOCATION BASED SERVICES
`
`Jock Christie, Richard Fuller, Jonathan Nichols, Aubrey Chen, Roger Hayward,
`Konstantin Gromov, Timothy Pfafinan
`
`Axiom Navigation, Costa Mesa California
`
`Position, location and Navigation Symposium, 2002
`
`Dora1 Palm Springs Resort
`Palm Springs, California
`April 15-18,2002
`
`ABSTRACT
`
`The Federal Communications Commission
`recognized as early as 1994 that the rapid
`adoption of wireless technology in the USA was
`destined to put additional strain on emergency
`dispatch and rescue personnel due to the absence
`of any location information associated with a
`wireless handset. Previously, the FCC had gone
`to considerable effort to ensure that all landline
`phones were associated with an unambiguous
`street address, so that rescue personnel could be
`dispatched efficiently during a 91 1 call. The
`FCC was compelled by their duties to establish a
`mandate for the development of "Automatic
`Location Identification" (ALI) technology by the
`wireless industry. Th~s paper will outline a
`system that requires no changes to the wireless
`network and can provide a wide range of location
`capabilities, including precise location solutions
`with GPS. This system was designed for
`flexibility, and can be used for both commercial
`location-based services, as well as E91 1 / PSAP
`solutions.
`
`requirements, and could be deployed
`economically be the carriers.
`
`Thus a great technological race began to
`produce a location determination system that
`wireless carriers would deploy.
`
`Unfortunately, over 5 years have passed, and
`the deployment of ALI technology is currently
`very limited to small geographies. There are a
`number of technological factors for the slow
`rollout, but cultural factors have also played an
`issue. Privacy advocates are understandably
`concerned about the security of such personal
`information. Certainly security should be the
`primary concem during commercial transactions.
`However, when a userk life is dependent on the
`timely dispatch of an ambulance, the system must
`operate in a different fashion.
`
`Tremendous technological progress during
`the last five years has finally led to a hybrid
`system architecture that is both flexible and
`accurate and can be used for applications ranging
`from navigation services to E91 1 dispatch.
`
`INTRODUCTION
`
`TECHNOLOGY OVERVIEW
`
`The FCC's "E91 1 mandate" is regarded as an
`"unfunded federal mandate" by the wireless
`carriers. The carriers are businesses, and thus
`interested in deploying services that generate
`additional revenues and additional profits. In the
`mid 1990s there was no commercially available
`location technology that would fulfill the E91 1
`
`The original E91 1 mandate has provided an
`impetus for significant development of
`miniaturized location determination systems in
`the last few years. Despite private sector
`investments totaling over $100 million in the last
`few years, no location determination solution is
`
`0-7803-7251 -4/02/$10.00 0 2002 IEEE
`
`60
`
`Apple Inc. 1015
`U.S. Patent No. 9,445,251
`
`

`

`commercially available that has been widely
`adopted by the wireless industry.
`
`Over the past several years multiple
`technologies have been presented to the wireless
`industry for the location of devices in their
`network. The solutions have ranged from crude
`location capabilities that have little or no impact
`on the infrastructure of the wireless network to
`very precise location determination systems
`which require extensive modifications to the
`network.
`
`A multitude of location technology for
`wireless devices have been proposed and
`implemented in the past several years. These
`break down into two general categories: network-
`based and handset-based. The position
`computation for "handset" solutions is computed
`in the handset, whereas the position computation
`for "network" solutions takes place on the
`network. Handset-based solutions frequently
`utilize information from the network to improve
`the solution. One major advantage of the
`network solutions is that they frequently require
`NO MODIFICATION of the handsets - which
`means that the consumer does not need to
`upgrade their handsets.
`
`For a handset based solution (the most
`stringent FCC condition), the FCC requires that
`position accuracy must be greater than 50 meters
`67% of the time, and greater than 150 meters
`95% of the time. The FCC requirement for
`network based solutions allow position errors
`that are twice as large, i.e. 300 meters 95% of the
`time.
`
`ALL positioning technologies rely on some
`sort of "triangulation" either based on the
`distance to satellites or based on the distance to
`cell towers. Furthermore, all of these positioning
`technologies require knowledge of:
`
`. Transmitter location
`. Transmitter timing information
`. As many measurements as possible
`
`Thus we could further subdivide the location
`solutions into terrestrial and satellite based
`solutions.
`
`CURRENT NETWORK TECHNOLOGIES
`
`Network based technologies generally use
`terrestrial measurements from one of more cell
`towers. These network based technologies have
`led to concerns among privacy advocates,
`because the user does not have a switch to
`disable this function. On the other hand, network
`based solutions are semi-automatic and do not
`require any special actions by the user during the
`stressful situation following an accident while
`dialing E9 1 1.
`
`Cell-ID
`
`By far the simplest approach is to estimate
`the user's position based on the location of the
`nearest cell tower. The accuracy is dependent on
`the cell size and varies from 10m (a micro cell in
`a building) to several kilometers (for a typical
`digital cell tower). The coverage radius of
`analog towers is even larger and can exceed 10
`kilometers.
`
`This approach is frequently referred to as
`"Phase I" in the wireless community.
`
`TDOA
`
`For situations where multiple towers can
`"hear" a given handset, it is possible to utilize
`"Time Difference of Arrival" Simply put, the
`signal from the cell phone should arrive at the
`closest tower first. The difference in propagation
`times can be used to derive the location of the
`cellular phone. However, this is more difficult
`than it first appears, since the clocks in each cell
`tower have errors which must be calibrated.
`"Location Measurement Units" are used for this
`calibration process. A minimum of 1 LMU is
`needed for every 3 cell towers - whch
`significantly increases the cost and complexity of
`this system.
`
`AOA
`
`"Angle of Arrival'' systems require that the
`direction to the cell phone can be determined
`accurately. This system requires only two towers
`that can hear the cellular handset. However,
`most digital wireless systems utilize three
`antenna sections at each cell site. These antennas
`can only resolve the general direction of the
`cellular handset to approximately 60 degrees.
`
`61
`
`

`

`New expensive antennas must be installed at a
`majority of the cell sites to implement AOA.
`
`a-GPS
`
`The "Global Positioning System" utilizes a
`constellation of approximately 24 satellites
`orbiting the earth. Assisted GPS or augmented
`GPS may be implemented as a network based
`solution by transmitting the measurements from
`the handset to a server on the network that can
`determine the handset's location.
`
`CURRENT HANDSET TECHNOLOGIES
`
`EOTD
`
`"Enhanced Observed Time Difference" uses
`differences in timing data received from
`surrounding base stations to calculate position on
`the handset. Some people regard EOTD as the
`opposite of TDOA. The accuracy of EOTD is
`expected to be around 125% and unllke GPS it is
`not reliant on a clear sky above. Like TDOA, it
`is necessary to use LMUs for calibrating the cell
`tower timing information to produce accurate
`location fixes.
`
`GPS
`
`The "Global Positioning System" was built
`with primary purpose of enabling autonomous
`navigation by military units. Microelectronics
`have made GPS receivers portable and
`affordable. Nonetheless, autonomous GPS is
`susceptible to interference or jamming, and does
`not perform well in urban canyons.
`
`CARRIER PERSPECTIVE
`
`As mentioned earlier, the carriers are
`businesses and will not invest large sums of
`money for the common good of their subscribers
`without a clear path to recoup their investment in
`a timely manner.
`
`As the pioneers of ALI technology for E9 1 1
`rushed to market in the late 1990s, many claims
`were put forward about performance. This
`caused most of the wireless carriers to become
`rather jaded when they realized the actual state of
`the technology.
`
`Furthermore, many technology distributors
`distributed self-serving propaganda about the
`costs required to install and maintain a
`nationwide system capable of providing E9 1 1
`functionality. With some estimates approachng
`$20 per subscriber, most of the carriers decided
`to delay indefinitely the purchase of Location
`Determination Technology. Essentially all of the
`US Carriers filed E9 1 1 waivers in the fall of
`2001 because they were not planning to deploy
`any sort of E9 1 1 system for the Phase I1 deadline
`of 01 October 2001.
`
`It appears now that wireless carriers are
`willing to consider any system that can provide
`both LBS and E91 1 functionality.
`
`PSAP PERSPECTIVE
`
`There are thousands of Public Safety
`Answering Points (PSAPs) throughout North
`America, ranging from single station PSAPs in
`rural areas to 50 stations PSAPs in metropolitan
`areas. They are typically funded by local
`municipalities and their equipment ranges from
`archaic Plain Old Telephone Service (POTS)
`lines to modem Intelligent Work Stations (IWS).
`
`The typical 91 1 dispatcher spends a large
`portion of their time answering calls about cats
`which are stuck in trees. A small proportion of
`the time, they have to handle life-threatening
`situations. During these times, it is essential that
`they have the best information possible about a
`given caller. For wireline 9 1 1 calls, there is a
`database with information about past calls, and
`the caller's street address.
`
`Approximately 50% of all calls to E91 1 now
`come from wireless phones. Thls is very
`frustrating for the dispatchers, who can not get
`access to the most crucial piece of information -
`the actual location of the caller.
`
`PulW4-w
`
`LBS PERSPECTIVE
`
`Location Based Service providers are
`focused on taking position information that is
`
`62
`
`

`

`similar to that provided to the PSAPs during and
`E9 1 1 call. These services require some
`knowledge of the user’s physical location. Some
`prototype services already exist, but they are very
`low tech. Users punch in the local zip code at a
`prompt. This sort of coarse location information
`is more than adequate for “regional” information,
`like weather information. This regional
`information is also of relatively low value to the
`consumer, as there are multiple ways to obtain
`this information.
`
`However, as more precise location
`information is available, users are willing to pay
`more to get information that is more
`personalized. For example, most cell phone
`users would pay extra for turn by turn driving
`directions, especially when they are late to an
`important client meeting.
`
`BASIC ARCHITECTURE
`
`This system consists of two primary pieces
`of hardware. The “X-Pak” GPS accessory
`module records GPS satellite measurements as
`well as cell timing information from nearby cell
`towers. The X-Pak transmits data via SMS to the
`“X-GPS” Location Server. SMS is used to
`transfer data from and to the phone. SMS
`transmissions utilize the data channel, and thus
`do not interfere with the normal operation of the
`voice channel.
`
`X-GPS ADVANTAGES
`
`Latency for E91 1 applications is one of the
`most crucial attributes of a system. When your
`emergency phone call is truly a matter of life and
`death, 60 seconds is a long, long time. “Time to
`
`First Fix” is the interval between the time you
`turn on a GPS device, and the time you are first
`able to report a position (even if it is 2D).
`
`Autonomous GPS receivers will exhibit
`great TTFF when performing warm starts
`outdoors, with typical times under 30 seconds.
`However, it is well known that when an
`autonomous receiver does not have a clear view
`of the sky, due to buildings or heavy foliage, that
`the TTFF will increase dramatically.
`
`We conducted a series of simple tests using
`one autonomous receiver, and one X-Pak
`communicating with our server. We visited
`several locations to collect data, and to determine
`the differences in acquisition time for the two
`configurations.
`
`The results of this test are shown below. In
`every single test case the network centric solution
`produced a solution as fast or faster than the
`autonomous GPS receiver. In fact there were
`several test cases in which the autonomous GPS
`receiver did not get a position fix, even after
`waiting 6 or 7 minutes.
`
`Examining only the cases for which both
`receivers were able to track satellites, we see that
`TTFF is reduced by a factor of 2 by employing a
`network centric solution.
`
`This is a relatively small sample, speaking
`statistically. But it indicates a strong correlation
`between increased TTFF and operation in
`autonomous mode.
`
`63
`
`

`

`Itl
`*\*< -
`
`wovement in Time To First Fix (lTFF)
`--_ .-.-e-
`
`w l l i u u ~ a p s l
`
`I
`
`I
`
`I
`
`No Assist Lock on
`Traxsis Assist lock on
`
`Note: No Bar = No Fix Obtained
`
`Average Tmxsa
`
`-
`-
`
`Burlingam
`
`Broadway Blvd
`
`Burlingam
`
`Burlingame
`
`b s t r o
`107 to S. San Fran
`E
`S. San Fran
`w s
` ar area
`
`Near burlingame
`Burlingame to SB I
`B u I i n g a me
`101 in S San Fran
`
`Mission in SF
`
`Mission in SF
`
`Near burlingame
`
`0
`
`50
`
`100
`
`150
`
`200
`TTFF (sec)
`
`250
`
`300
`
`350
`
`400
`
`Figure 1) TTFF results for autonomous vs. networked 1
`
`E9 1 1 ARCHITECTURE
`
`emergency, rather than asking extensive
`questions about the user’s location.
`
`Standards committees for CDMA, GSM, and
`TDMA met in the late 1990s to create standards
`for how the various components would inter
`communicate.
`
`The objective is very similar in all air
`interfaces - providing the location of the wireless
`user directly to the PSAP. Unfortunately, the
`implementations are all rather different.
`
`The primary challenge with E9 1 1 is that the
`most accurate solution available is needed in the
`minimum time possible. Cell tower (Phase I)
`information can be used for initially routing the
`call, which must happen within the first 6
`seconds after completing the call. This coarse
`position information is good enough to get the
`9 1 1 call into the appropriate call center, but it is
`note adequate for dispatching emergency
`vehicles. Precision location information from a
`GPS receiver fiom the device is nominally
`available at this point, so that the dispatcher can
`focus on understanding the scope of the
`
`E9 1 1 CONSIDERATIONS
`
`The rapid adoption of wireless technology
`by Americans has led to a dramatic increase in
`call volume to emergency dispatch centers. A
`large portion of these calls are accidental or non-
`emergency calls, and are merely a nuisance to the
`dispatchers. Another large portion of the calls
`are legitimate emergencies, where the only
`priority is to dispatch the nearest possible vehcle
`to the scene.
`
`In a tiny fraction of the situations, the caller
`is actually moving, and dynamic Location
`Determination will prove to be essential. In
`these situations, the dispatcher will need to do an
`“ALI REFRESH’ to get updates location
`information for the given phone number. Our
`system is designed to provide updates as
`frequently as once every 10 seconds, enabling
`dispatchers to track an incident in real time from
`their console.
`
`64
`
`

`

`CONCLUSIONS
`
`We have developed a uniquely capable
`system for accurately determining the location of
`a GPS enabled device. The hybrid system can
`utilize measurements from the GPS receiver and
`from cell tower information. The network
`centric approach allows for more sophisticated
`algorithms to be utilized, which is particularly
`important in situations with limited GPS
`visibility. The network centric architecture also
`leads naturally to a centralized database of the
`user’s last known location. This position
`information is stored securely, and is only shared
`with authorized parties. The position can be
`pushed to the appropriate PSAP during an E9 1 1
`call, or requested by an LBS provider.
`
`REFERENCES
`
`[l] J. M. Zagami, S. A. Parl, J. J. Bussgang,
`and K. D. Melillo, “Providing Universal
`
`Location Services Using a Wireless
`E9 1 1 Location Network”, Spread
`Spectrum Scene.
`
`[2] FCC OET-7 1 “Guidelines for Testing
`and Verifying the Accuracy of Wireless
`E9 1 1 Location Systems”, published 12
`April 2000
`
`[3] “Emergency Services Data
`Communications (J-Std-036)” by the
`Telecommunications Industry of
`America, Dec. 2000
`
`[4] R. A. Fuller, J. R. I. Chstie, J. 0.
`Nichois, A. Chen, R. C. Hayward, K.
`Gromov, T. Pfafinan “ A Highly
`Flexible and Scalable System for
`Location Determination of Wireless
`Devices”, IEEE PLANS 2002
`
`65
`
`