Appendix H:    Common Vehicle Components
and Architecture

H.1  Introduction

H.1.1     Common Vehicle Components

All demonstration vehicles employ the same architecture in terms of map databases, generation and delivery methods of map information and reception of GPS data. Two main configurations, near-term and mid-term, have been established as needed by the map databases and applications. This report provides a detailed description of the common vehicle architecture in the following sections.

H.1.2     Common Vehicle Architecture

The Figure H-1 illustrates the common vehicle architecture.

Figure H-1:  EDMap common vehicle architecture

For near-term applications, the vehicle system utilizes the NTBox dead reckoning module to obtain vehicle position data while for mid-term applications the vehicle system utilizes the Honeywell Prototype Automotive Positioning System (PAPS). Table H-1 describes the GPS receiver employed with the near- and mid-term applications.

Table H-1:  GPS receiver employed by application type

Table H-2 describes the ADASRP functions that are common to all demonstration vehicles.

Table H-2:  Common ADASRP functions

H.2  Near-term Vehicle Positioning

The NTBox was designed by NAVTEQ to represent hardware comparable to navigation systems in production today. Given that near-term needs span the period of present day to approximately 2006, there was concern that the NTBox technology was possibly not representative of hardware toward the end of the near-term. A short study was conducted at NAVTEQ. Briefly, here are the results for three of the latest aftermarket DVD based navigation systems:

• Alpine NVE-N851A
  • 8-channel GPS receiver
  • Unable to determine make/model of gyroscope
• Pioneer AVIC-9DVD
  • 8-channel GPS receiver
  • Unable to determine make/model of gyroscope
• Panasonic CN-DV2000
  • 8-channel GPS receiver
  • Matsushita piezoelectric gyro

The Trimble Lassen GPS receiver is comparable in specifications to the receivers in the three aftermarket navigation products. In addition, as far as multipath mitigation is concerned, Trimble claimed that because the SK II was targeted for the automotive market, a significant amount of work to minimize the effects of multipath on the computed position has already been done. The Trimble representative was not aware of any additional multipath mitigation techniques that they could take advantage of based on the fact that Selective Availability (SA) has been turned off. He also pointed out that they are currently supplying BMW, Mercedes, and Fiat with Lassen SK II receivers for their in-vehicle applications. The main concern is the potential for improved accuracy using Wide Area Augmentation System (WAAS) or some other form of code based differential corrections. To cover this contingency, the NTBox was modified to allow RTCM SC­104 corrections to be input to the Trimble receiver.

Although the gyro type being used in the Alpine and Pioneer systems was not determined, odds are that it is also a piezoelectric gyro similar to the one in the Panasonic system. Navigation Technologies did build a limited number of NTBoxes with the Matsushita gyro and found its behavior to be very comparable to the Murata.

Based on what we have observed while test driving these three systems, the overall positional accuracy of the Vehicle Position (VP) tool using the NTBox is as good or better than all of these systems. Therefore, based on these three systems, it appears that the hardware in the NTBox is comparable to the hardware found in today's high-end aftermarket systems.

The GPS receiver outputs the following data:

• Position, velocity, and Dilution of Precision (DOP) information once per second
• Satellite information once every three seconds
• GPS time information once every five seconds

This data is received by the A-Engine and passed through to the host using the Trimble Standard Interface Protocol (TSIP). The A-Engine also samples the analog data from the gyroscope and the temperature sensor ten times per second. This data along with the accumulated speed pulse count and reverse state are transmitted to the host in a separate TSIP packet every 100 milliseconds. The following is a description of the front panel controls from left to right.

Figure H-2:  NTBox front panel                                      Figure H-3:  NTBox back panel


Table H-3:  NTBox front panel description

Differential GPS serial port	This is a standard RS-232 port that operates at 4,800 baud and permits a differential beacon receiver to be connected to the Trimble GPS receiver for inputting differential correction data in RTCM SC-104 format.
GPS antenna RF Jack	This is the RF antenna jack directly from the Trimble Lassen SK-8 board.
Auxiliary RS-232 serial port	This is a standard RS-232 port that operates at a software configurable baud rate. This port is used for communication between the NTBox and an external sensor box that transmits data serially.
Host RS-232 serial port	This is a standard RS-232 port that operates at 19,200 baud. This port is used for communication between the NTBox and a host computer.
Programming jack	A conventional 3/32 (2.5 mm) phono jack is employed in order to select ‘run mode’ or ‘programming mode’. In ‘run mode’ the NTBox executes the program stored in the A-Engine’s SRAM. In ‘programming mode’ the NTBox software can be updated by downloading a new executable via the Host serial port. If a phono plug is not installed, then the unit is in ‘run mode’. If a phono plug is installed, then the unit is in ‘programming mode’.
LED power indicator	Indicates that the unit is powered on.
Vehicle interface plug	This interface plug is used to connect the vehicle inputs to the NTBox. Pin 1 (+12Vdc)         +12V DC connection. Connected to the vehicle's ignition switch accessory wire. Pin 2 (GND)             Ground connection. Connected to the vehicle's chassis or other suitable ground. Pin 3 (Speed 1)       +5V speed pulse connection. Connected to the vehicle's speed sensor. Pin 4 (Video Sync)  +5V video synchronization pulse (future use). Pin 5 (Reverse)       +12V reverse connection. Connected to the vehicle's backup lights.
Power switch	This switches the +12 volts main power.

An external analog sensor interface plug is located on the back of the NTBox. This interface plug is used to connect additional analog sensors to the NTBox. These sensors might include additional gyroscopes, accelerometers, or inclinometers.

H.2.1     NTBox Specifications

H.2.1.1    Trimble Lassen SK-8 GPS Receiver

General	L1 frequency (1574.42 MHz), C/A code (Standard Positioning Service), 8 channel, continuous tracking receiver, 32 correlator.
Accuracy	Position 25 meters CEP (50%) without SA (Selective Availability) Velocity  0.1 m/sec (1 Sigma) steady state conditions (without SA) Time       ±500 nanosecond (nominal) Datum    WGS-84 (standard DMA datum set)
Dynamics	Altitude   -1000 meters to +18,000 meters Velocity  515 m/sec (maximum) Acceleration          4g (39.2 m/sec2) Jerk                        20 m/sec3 Operating Temperature     -10°C to +60°C

H.2.1.2    Murata ENV-05D Gyroscope

Maximum angular velocity	80 deg/sec
Operating temperature	-30°C to +80°C
Output (angular velocity=0)	2.5 V
Scale factor	22.2 mV/deg/sec
Asymmetry CW & CCW	3 deg/sec
Linearity	0.5 %FS




H.2.2     Coast Guard DGPS Receiver

The Coast Guard DGPS receiver is a CSI Wireless^TM model MBX-3 beacon receiver and is used to provide the NTBox with DGPS corrections.

Figure H-4:  CSI Wireless^TM model MBX-3 beacon receiver

The MBX-3 may be operated in either automatic beacon search (ABS) or manual tune mode. When operated in automatic beacon search mode, the receiver will identify and tune to the station providing the strongest DGPS signal without user intervention. When operated in manual mode, the user specifies the frequency to which the receiver will be tuned.

The default mode of operation for the MBX-3 receiver is ABS mode. The receiver uses two independent channels to identify and lock on to DGPS beacons without interrupting the continuous flow of RTCM data to the receiver. Because the MBX-3 receiver automatically tunes to the strongest DGPS signal, this mode allows for navigation over wide areas without the need for user intervention.

When the MBX-3 is powered up for the first time in ABS mode, it initiates a global search and examines each available DGPS beacon frequency. It records the signal strength of every beacon identified in a Global Search Table. The receiver uses these measured values to compute average signal strength, noise floor and to sort the frequencies in descending order of signal strength. The two channels cooperatively examine the frequencies with the highest signal strength measurements above the computed noise floor to determine which station provides the strongest RTCM signal. The receiver's primary channel locks to the first identified DGPS broadcast while the second channel continues searching in the background for beacon signals of higher quality. If no signal is available, the MBX-3 will initiate a fresh global search and will continue this cycle until it finds a valid beacon.

During the background search, the second channel examines all frequencies to identify beacons possessing superior signal quality. If a DGPS broadcast is identified that exhibits a signal strength 2 dB greater than that of the primary station, the receiver will automatically switch to this beacon. No loss of lock occurs on the primary station during the background scan.

The MBX-3 stores the current primary beacon in memory so that it is available upon subsequent power-up. The user may force a new global search at any time using the display and keypad, or by issuing a proprietary Wipe Search command.

In Manual tune mode, the user may select a specific frequency and bit rate for the receiver to tune to. The user also has the option of specifying just the frequency and allowing the MBX-3 to identify the correct bit rate. The MBX-3 also provides the capability to select a beacon by name from the World Beacon Table stored in the receiver's memory         .




H.3  Mid-term Vehicle Positioning

The Honeywell PAPS GPS Receiver represents this advance and is the source of GPS vehicle positioning data for all mid-term applications. The PAPS unit provides the following position accuracy needed for mid-term applications. Figure H-5 illustrates the overall system architecture for the PAPS unit.



Figure H-5:  PAPS system functional block diagram




The eight external interfaces identified in Figure H-5 are defined in Table H-4.

Table H-4:  PAPS external interfaces

#	Function	Description
1	GPS Satellite RF Interface	The PAPS unit receives vehicle position data from a GPS satellite constellation.
2	Base Station Output Interface	A number of GPS base stations located in the test areas broadcast differential GPS correction data via the cell phone network.
3	Differential Data Link RF Interface	The PAPS unit receives the differential GPS correction data via a GSM modem that the vehicle operator uses to call a specific base station.
4	Differential Data Link Serial Interface	The GSM modem supplies the PAPS fusion processor with the differential GPS correction data that the base station broadcasts.
5	Wheel Sensor Serial Interface	The Fusion Processor in the PAPS unit receives wheel speed sensor data in the form of a serial data message (message 504) from the vehicle.
6	Data Recorder Media (Compact Flash Card)	The data recorder provides data that can be used for analysis purposes.
7	Prototype Automotive Positioning Sensor Serial Output	The Fusion Processor takes data from the GPS receiver, GPS base station, wheel speed sensor and inertial unit to determine vehicle position. The PAPS unit then outputs this data to the internal data recorder, ADASRP platform and other user subsystems.
8	Vehicle Power Interface	The PAPS unit is powered by the vehicle's DC power system.

A number of base stations located in the test areas broadcast differential GPS-corrections data. The PAPS unit receives this data through an RS-232 serial input bus via a GSM modem. Although the PAPS is capable of receiving data at 115,200 bps, the GSM modem is only capable of providing differential GPS corrections at 9,600 bps therefore the Port setup for the PAPS must be set to the lower speed or the unit will not operate properly. The PAPS also receives wheel speed data through an RS-232 serial input bus (message 504) at 115,200 bps. Each OEM is responsible for providing wheel sensor data in the proper format.

The PAPS provides data through an RS-232 serial output bus at a baud rate of 115,200 bps. The unit provides a 140-byte INS/GPS Output Message (message 502) at a rate of 50 Hz and a 64‑byte INS/GPS Detailed Status Message (message 503) at a rate of 1 Hz.




H.3.1     Available Applanix POS Systems

H.3.1.1    POSLV 420

Performance

The POSLV 420 employs a Litton IMU. This IMU has a drift of 1 deg/hr.

POSLV 420	GPS Outage Duration (IARTK/PP/DGPS)
		0 sec			15 sec			30 sec			1 min			2 min
	IARTK	PP	DGPS	IARTK	PP	DGPS	lARTK	PP	DGPS	IARTK	PP	DGPS	IARTK	PP	DGPS
X, Y (m)	0.035	0.02	1.0	0.07	0.05	1.13	0.013	0.07	1.25	0.20	0.12	1.5	0.30	0.27	1.63
Z (m)	0.05	0.03	1.5	0.07	0.06	1.63	0.10	0.09	1.75	0.15	0.15	2.0	0.20	0.35	2.13
Roll & Pitch (deg)	0.02	0.005	0.02	0.02	0.005	0.02	0.02	0.005	0.02	0.02	0.005	0.02	0.02	0.005	0.02
True heading (deg)	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.03	0.02	0.03

IARTK:     Inertially-Aided RTK
PP:                           Post-processed
DGPS:                      Real-time DGPS

Physical

	IMU	PCS	DMI (Applanix)	Antenna
Size:	204 x 204 x 168 mm	2.5U, 19" Rack Mount (445 x 111 x 350 mm)	153f x 97 mm (not including collets or restraint rod)	160f x 58 mm
Weight:	3.5 Kg	9.4 Kg	2.7 Kg	0.46 Kg
Power:	N/A	120/220 VAC, 60/50 Hz or 12 VDC, 110 W (peak)	N/A	N/A
Temperature:	-40° to +60° C	0° to +50° C	-40° to +85° C	-40° to +70° C
Humidity:	0 to 100 %	5 to 95% RH non-condensing	0 to 100 %	0 to 100 %
Cables:	8 m	N/A	8 m	10 m

Sensors

GPS:     12 channel dual frequency (1-1/1-2), low noise,  
            12 channel single frequency (1-1), low noise

Ethernet Input/Output (10/100 base -T)

Data:                Position, attitude, heading, velocity, track & speed, acceleration, status & performance, raw data. All data has time & distance tags.

UDP Port:        Display Port - low rate (1 Hz data)

TCP/IP Ports:   Data Port - high rate (1 - 200 Hz data)

Data Port 2 (buffered for data logginq)

Control Port - used by LV-POSView^TM (controller software)




Logging Output to Removable PC-Card Disk Drive

Parameters:      Position, attitude, heading, velocity, track & speed, acceleration, status & performance, raw data. All data has time and distance tags.

RS232 NMEA Output (DB 9 male)

Parameters:      Position ($INGGA)

                        Heading ($INHDT)

                        Track and Speed ($INVTG)

                        Statistics ($INGST)

Rate:                1 - 50 Hz (user selectable)

RS232 High Rate Digital Output (DB 9 male)

Parameters:      Roll, pitch, true heading, latitude, longitude and altitude.

Rate:                1 - 200 Hz (user selectable)

RS232 Input (DB 9 male)

Parameters:      Auxiliary GPS input

                        Position ($GPGGA)

                        DOP ($GPGSA)

                        Statistics ($GPGST)

                        Satellites in View ($GPGSV)

Rate:                1 Hz

RS232 Base 1 & Base 2 Input (DB 9 male)

Parameter:        CMR, CMR+, RTCM 18/19, RTCM 1, RTCM 9

Other I/0

PPS:                 1 pulse-per-second Time Sync output. Normally high, active low pulse where the falling edge is the reference.

Event Input:      Two input discretes used to mark external events. Discretes are TTL pulses > 1 msec width where rising or falling edge is time-tagged and logged. (Maximum rate 300 Hz.)

H.3.1.2    POSLV 220

The POSLV 220 employs a Honeywell HG1700 IMU. This IMU has a drift of 3 deg/hr.

POSLV 220	GPS Outage Duration (IARTK/PP/DGPS)
		0 sec			15 sec			30 sec			1 min			2 min
	IARTK	PP	DGPS	IARTK	PP	DGPS	lARTK	PP	DGPS	IARTK	PP	DGPS	IARTK	PP	DGPS
X, Y (m)	0.035	0.02	1.0	0.20	0.05	1.13	0.35	0.08	1.25	0.70	0.15	1.5	1.0	0.60	1.75
Z (m)	0.05	0.03	1.5	0.20	0.08	1.63	0.30	0.12	1.75	0.50	0.20	2.0	0.90	0.70	2.2
Roll & Pitch (deg)	0.07	0.06	0.07	0.07	0.06	0.07	0.07	0.06	0.07	0.07	0.06	0.07	0.07	0.06	0.07
True heading (deg)	0.07	0.025	0.07	0.07	0.025	0.07	0.07	0.025	0.07	0.07	0.025	0.07	0.07	0 .03	0.08




H.3.1.3    POSLV 320

The POSLV 320 employs a Litton IMU. This IMU has a drift of 3 deg/hr.

POSLV 320	GPS Outage Duration (IARTK/PP/DGPS)
		0 sec			15 sec			30 sec			1 min			2 min
	IARTK	PP	DGPS	IARTK	PP	DGPS	lARTK	PP	DGPS	IARTK	PP	DGPS	IARTK	PP	DGPS
X, Y (m)	0.035	0.02	1.0	0.10	0.05	1.13	0.20	0.08	1.25	0.35	0.15	1.5	0.60	0.40	1.63
Z (m)	0.05	0.03	1.5	0.10	0.07	1.63	0.15	0.10	1.75	0.25	0.20	2.0	0.40	0.50	2.2
Roll & Pitch (deg)	0.05	0.02	0.05	0.05	0.02	0.05	0.05	0.02	0.05	0.05	0.02	0.05	0.05	0.02	0.05
True heading (deg)	0.05	0.025	0.05	0.05	0.025	0.05	0.05	0.025	0.05	0.05	0.025	0.05	0.06	0.03	0.06

Notes:

The positional accuracy listed in the tables are RMS values
The accuracy specifications assume good GPS mission planning (when GPS is available)
IARTK performance values are valid at baselines out to 40 km
Specifications are valid as of March 2003




H.4  Delivery of Differential Corrections

The fundamental idea behind differential correction is to estimate certain errors that exist in a GPS position measurement and then correct an originally derived position by those errors. In order to do so, a reference station with a precisely known antenna location is used. Due to the known location of the reference station antenna, the base station receiver can estimate the correction data. Then, this data is transmitted to a second receiver, e.g. in a moving vehicle. Based on the correction data that receiver can correct its position accordingly.

There are two primary kinds of differential corrections, each allowing different levels of accuracy. Code-based differential corrections like those issued by beacon transmitters operated by the United States Coast Guard can provide a GPS receiver with information allowing the receiver to obtain accuracy in the 1 to 5 m ranges. For the mid-term positioning sensor in EDMap, carrier-phase differential correction information allows the GPS receiver to obtain accuracy in the 0.02 to 0.6 m range.

Unfortunately, no source for carrier-phase differential corrections existed at the time of EDMap positioning definition. OmniSTAR (http://www.omnistar.com/) now offers a service that may be used with certain GPS receivers to consistently obtain sub-meter positioning accuracy. This service was not available when EDMap chose its positioning strategy and is not compatible with the current positioning hardware. Additionally, there is an effort underway within the USDOT to investigate the potential of an extended Nationwide Differential GPS (NDGPS) Coast Guard system to provide carrier phase observable differential corrections. The High Accuracy NDGPS work continues, but was not at a level where it could be used in EDMap. The viability of these and other potential sources of differential corrections will be addressed in the Task8b interim report.

A wireless link for transmitting differential corrections is a requirement to demonstrate map-based applications requiring high accuracy positioning in real-time. As a result, the EDMap team defined base stations at each OEM location. Table H-5 outlines the real-time message requirements from the base station GPS receiver to the vehicle mid-term positioning sensor. Overall, the communication link requires an average transmission of 97 bytes per second from the base station to the vehicle.

Message Description	Message Size	Message Frequency
RTCA Standard Type 1	83 bytes maximum	Every 5 seconds
RTCA Reference-Station Satellite Observations	156 bytes	Every 2 seconds
RTCA Reference Station Position Information Type 7	24 bytes	Every 10 seconds

Table H-5:  Mid-term differential correction messages

H.4.1     Base Station / Vehicle Setup

For the test setup, the base station transmits the carrier-phase correction data via a landline telephone modem to a moving vehicle equipped with a cellular modem. A Novatel OEM3 or OEM4 receiver (participant dependent) at the base station is configured to generate the RTCA differential correction data. Figure H-6 shows the complete system layout.

The generated differential corrections are then transmitted to a rover GPS receiver. In order to do so efficiently, the base station PC hosts a multi-modem PCI adapter and is connected to a pool of conventional land-base telephone lines. A Java software application on the base station PC then handles incoming calls and routes the correction data to the respective rover unit connected over the phone line. The Java software and the modem pool configuration of the phone lines allow the base station to handle multiple rover connections at the same time. Hence, each connected rover unit will receive the same differential corrections.

The rover unit then uses a cell modem to connect to the differential base station. The modem has a roof-mounted antenna to improve reception and software on the rover handles placing the cell call as well as serial port routing of the base station data from the cell modem to the rover GPS receiver. The software also handles certain error conditions, e.g. it automatically redials the connection to the base station to minimize the time of a differential correction data outage in case of a dropped cell modem connection.



Figure H-6: DGPS correction setup

H.4.2     Base Station Survey

The base station provides measurements used for differential corrections at the roving GPS receiver. For EDMap, a roving GPS receiver is in each of the OEM demonstrator vehicles as well as in the NAVTEQ data collection vehicles. The position solution calculated at any given time sample at the roving receiver is relative to the base station position.

The relative distance calculated using differencing is made absolute using a set of reference coordinates at the base unit. With the base unit coordinates known, the relative distances calculated using differencing are simply added to the base coordinates (ECEF) to determine the coordinates of the rover unit. The coordinates of the base unit can be converted to a common map coordinate system such as NAD83, permitting GPS to be used for survey purposes and navigation.

The key element of a survey is to accurately determine the location of a point or set of points with respect to a known position or network of known positions. Typically, the goal is to determine the location of a set of points, e.g., for a housing development, as shown in the network of survey points in Figure H-7. It is important to maintain consistent accuracy from the known point to all the network points. Accuracy is maintained by measuring the distance from the horizontal control point (HCP) to one of the network points to form a baseline HCP-11. Another baseline is then formed using HCP-12. The first survey stage is accomplished by closing the loop between 11-12, and thereby forming the triangle HCP-11-12. The 11-12 baseline provides increased accuracy by adding a geometric constraint to the individual baselines HCP-11 and HCP-12.



Figure H-7:  Network survey representation

The survey proceeds using the 11-12 baseline to create more triangle loops until all network points have been covered. Once all points have been covered, a process called network balancing is performed that uses a least squares adjustment to the errors generated at points where two measurements exist. An example of such a point is location 12 that has baselines 11-12 and HCP-12 pointing to it. A difference between the two measurements for location 12 is mediated by balancing the error through the connected baselines. How much any one point is corrected is determined by statistics (covariance) of the GPS position solution for each point. Using this balancing approach, the network of points can be surveyed to a high level of accuracy from point to point. [Hoffman-Wellenhof, et.al.  1994].

Accordingly, in order to determine accurate positions of the EDMap base stations, HCP sites needed to be identified. There are in-ground HCP locations throughout Michigan as part of the National Geodetic Survey (NGS). However, GPS equipment needs to be set up over the HCP marker to collect data. The alternative used by EDMap was to collect data from Continuously Operating Receiver Stations (CORS). NGS sponsored CORS sites collect GPS data at pre-surveyed locations, and make the data available via the internet [www.ngs.noaa.gov/CORS/cors-data.html]. The data are time delayed, and therefore not useful for real-time corrections.

The process followed in the Michigan EDMap test area was to collect GPS observations data at each of the EDMap base stations, and process, in conjunction with CORS data, accurate base station positions for the four base stations. Waypoint [www.waypoint.com] survey software was used to calculate the survey. The process followed is summarized in the steps below:

• Planning – Good GPS satellite constellation coverage was ensured by using almanac information from [www.navcen.uscg.gov/ftp/gps/archives/yuma]. Waypoint mission planner provides interface to almanac data. The periods of best DOP and number of satellites at a reasonable personnel time were selected as the survey day and time.
• CORS Data Collected – CORS sites for the southeast Michigan area were identified based on proximity to the four EDMap base stations to be surveyed. Three were chosen:  SIBY (South of Dearborn, MI for Ford), BRIG (near Brighton, MI for CAMP and TTC), and METR (near Auburn Hills, MI for GM and CAMP). GPS observations and related information (coordinates) are downloaded a few hours after the actual EDMap collection due to the data availability delay. In general, CORS sites are known to approximately 1cm in latitude/longitude. Datum NAD83 Epoch 2002 coordinates were used.
• Collect EDMap Base Station Data – Using the date and time period established by the almanac, ephemeris and carrier phase measurements were collected for 2 hours (to be used at 15 second intervals).
• Network Processing – Waypoint Grafnet software was used to process the EDMap base station data along with the CORS reference data. All EDMap base station observations were added, using one CORS site as the primary control point, and the other CORS sites as balancing nodes. Processing and balancing was an iterative process that continued until baseline residual errors are all under a maximum of 5 cm based on EDMap application needs.

Figure H‑8 shows the EDMap network. The three CORS sites (SIBY, BRIG and METR) are shown, along with the four EDMap base stations (TTCLo – Toyota TTC, GMR – GM Research, Ford – Ford Scientific Research Labs, and 212 – CAMP Facility). The triangles highlighted form the survey network with regard to baselines that need to be computed in order to form a minimally constrained network solution. The ellipses around each station is the expected residual error. The calculated base station positions and expected accuracies are shown in Error! Reference source not found. and Table H‑7, respectively. The results are quite good, and provide a reference from which all EDMap applications and map database collection, especially midterm requirements, can be based.



Figure H-8:  EDMap base station network for southeast Michigan

Table H‑6:  Calculated base station positions

Station	Latitude DMS.s D.d	Longitude DMS.s D.d	ELLHGT (m)	*Geoid Height MSL (m)
CAMP	42 29 38.78546 42.494107072	-83 25 58.06541 -83.432795947	243.815	277.554
Ford	42 17 36.26688 42.293407467	-83 14 17.21095 -83.238114153	170.281	204.329
GM	42 31 04.65227 42.517958963	-83 02 30.46258 -83.041795161	165.379
TTC	42 17 50.87191 42.297464419	-83 40 37.76589 -83.677157191	243.564	278.204
DCX	37 24 56.01401 37.415559447	-122 08 57.5735 -122.14932598	9.204




Table H‑7:  Expected base station position errors (90% of errors contained)

Station	North Error (cm)	East Error (cm)	Hgt Error (cm)
CAMP (212)	1.1	1.3	2.2
Ford	1.1	1.2	2.2
GMR	1.1	1.3	2.1
TTC	1.1	2.1	2.2

The base station survey for the Palo Alto region was conducted previously by DCX, but the approach shown here would be readily applied to the DCX base station using appropriate CORS data.

It is interesting to note that standard averaging of GPS (non-differential) data collected at the base stations, typically collected over several days and then averaged, was shown to be insufficient for decimeter level positioning requirements.

H.4.2.1    Investigating Different Means of Differential Correction Delivery

To satisfy the requirement in the EDMap project, the following three wireless technologies for obtaining differential GPS corrections were evaluated under similar testing conditions.

Table H‑8:  Differential correction delivery methods

Method	Source	License Requirements	For More Information
Spread Spectrum Wireless Modem	Freewave Technologies, Inc.	License free	http://www.freewave.com
35 Watt Radio Modem	Pacific Crest Corporation	License required for 35 watt operation	http://www.paccrst.com
GSM Cellular Modem	Wavecom	License free	http://www.wavecom.com

H.4.2.2    Spread Spectrum Wireless Modem

Two spread spectrum wireless modems were used to test the communications link. The base station modem’s antenna was placed on top of the CAMP building located in Farmington Hills, MI. The other modem antenna was mounted on top of a vehicle. A known repeating ASCII string was sent from the base station antenna to the vehicle antenna. Initial testing showed the communications link consistently failed when the vehicle traveled more than 2 miles from the base station location. Since the link was not capable of maintaining a connection over long distances, this technology was deemed not useful for EDMap purposes.

H.4.2.3    35 Watt Radio Modem

Figure H‑9 below illustrates the results of a test that General Motors conducted in approximately a 180 square kilometer area in 2002 to assess the radio modem's capability to deliver differential GPS correction data. A 35 watt Pacific Crest Radio transmitter and antenna was placed approximately 50 feet in the air on top of a GM Research building located in Warren, MI. A vehicle was equipped with a radio modem and a roof-mounted antenna. As the vehicle traveled through the area, GPS locations of the vehicle were recorded along with data received from the base transmitter location. The figure clearly shows that the Pacific Crest Radio modem was extremely susceptible to poor data quality or complete loss of communication as the distance between the base transmitter and receiver grew. Since the loss of differential GPS correction data results in rapid degradation of the vehicle position accuracy, the EDMap team members quickly determined that this method of transmitting differential GPS correction data was wholly unacceptable.



Figure H‑9:  Pacific Crest 35 watt transmitter test

H.4.2.4    GSM Cell Modem

Figure H‑10 below illustrates the results of a test that General Motors conducted in an approximately 4,100 square kilometer area in 2002 to assess the GSM cell modem's capability to deliver differential GPS correction data. A GPS receiver in the vehicle recorded vehicle location as information was transmitted through the cell link from a computer inside a GM Research building. The figure clearly shows a vast improvement in reception of differential GPS correction data over that of the Pacific Crest Radio modem. Instances of bad data receptions were greatly reduced and the area in which differential GPS corrections could be obtained increased by a factor of more than 20 times. The two continuous areas of bad communications are the result of the cell connection being lost requiring a redial. Because of the performance exhibited by the GSM Cell Modem, the EDMap team members adopted it as the method of obtaining differential GPS correction data.



Figure H‑10:  Wavecom cell modem and T-mobile cell provider test

Figure H‑11 and Figure H‑12 show the cell modem that the EDMap team chose as the delivery method of differential GPS correction data.



Figure H‑11:  Wavecom GSM cell modem



Figure H‑12:  Wavecom GSM cell modem

H.4.2.5    MEMS Gyro Technology

The Honeywell MEMS gyro, shown in Figure H‑13, is based on the Coriolis force. The resonant structure is composed of two elements, called proof masses, driven electrostatically in a plane in opposite oscillatory directions. When the device is rotated about the sensitive axis, the elements are driven to oscillate out of their original plane of motion by an amount proportional to the product of the input rotational rate and the oscillatory drive velocity. Measurements of the mass deflection allow the determination of the rotation rate. Commitment and investment in the Honeywell MEMS gyro remain at a high level and interest from customers has extended well beyond the traditional aerospace market.



Figure H‑13:  Honeywell MEMS gyro

The micro-machined mechanisms are fabricated using a dissolved wafer process, depicted in Figure H‑14. The critical dimensions are controlled with boron diffusion and dry etch techniques. The silicon wafer is anodically (thermal electrically) bonded to a glass substrate that has been prepared with a metal deposition step to provide all the electrical contacts for the sensor operation. After the unwanted silicon is dissolved away, the mechanism is free to operate. This allows for wafer level probing to map all the mechanisms on a wafer in a single automated process. After mapping, the wafer is diced and the individual mechanisms are vacuum packaged and ready for integration with the sensor electronics.



Figure H‑14:  Wafer level automotive gyro fabrication

The simplicity of this fabrication process makes it possible to build the mechanisms on high volume, low cost MEMS silicon lines.