• Abstract

# Implementation and Operational Use of Ground-Based Augmentation Systems (GBASs)—A Component of the Future Air Traffic Management System

These systems detect and correct aircraft landing position errors by comparing satellite data with data from compact, inexpensive airport based equipment.

This paper discusses a satellite navigation augmentation system designed for use by aviation. The ground-based augmentation system (GBAS) was originally developed as a precision approach and landing aid. This paper describes the GBAS concept, discusses the system architecture, and discusses ground and airborne equipment that compose the system. This paper also describes typical operational use of the system and the experience gained during early implementations. Advantages over the current Instrument Landing System technology are also discussed.

SECTION I

## INTRODUCTION

This paper describes a particular satellite navigation technology known as a ground-based augmentation system (GBAS). Satellite navigation has become a critical component of the emerging worldwide air traffic management (ATM) infrastructure. As congestion grows and rising costs demand ever greater efficiencies, the management of air traffic will rely more and more on the management of airplane trajectories in four dimensions (lateral, vertical, and longitudinal path and time). GBAS promises to become an indispensable tool in the future for the management of airplane trajectories for ATM, particularly near and at airports and landing sites.

The International Civil Aviation Organization (ICAO) committee on Future Air Navigation Systems (FANS) developed a vision for a Global Navigation Satellite System (GNSS) to support aviation navigation needs. The Global Positioning System (GPS) was offered to the world's aviation community by the United States in a letter to the ICAO in October 1994 [1]. ICAO accepted the offer, establishing GPS as an important component of the GNSS. The Russian Federation made a similar offer with respect to use of the Global'naya Navigatsionnaya Sputnikovaya Sistema (GLONASS). Hence GPS and GLONASS became the core constellations in the system of systems defined by ICAO as the GNSS. However, because of certain limitations (real and perceived) in the performance of GPS and GLONASS, additional system components were added to GNSS to augment performance, including:

1. space-based augmentation systems (SBASs);

2. ground-based augmentation systems (GBASs);

3. airplane-based augmentation systems (ABASs);

4. ground-based regional augmentation systems (GRASs).

The GNSS as defined by ICAO includes the core constellations (GPS and GLONASS) as well as these augmentation systems. Formal standards and recommended practices (SARPS) for GNSS were developed and published in 2000 [2]. These SARPS are intended to ensure interoperability between components of the GNSS and to ensure that equipment based on GNSS operates safely and with consistent performance that meets the operational needs of aviation users.

The augmentation systems listed above were developed to provide improved accuracy, integrity, continuity of service, and availability of navigation to support a wide variety of operational needs. These augmentation systems have been the subject of nearly two decades of research and consequently much has been written about them. This paper focuses on only one of these four augmentation types: GBAS [3], [4].

GBAS is significant for air transport users for a variety of reasons. It is the only augmentation system defined at this time that is expected to be capable of meeting the most stringent operational needs of aviation (e.g., takeoff and landing in very low-visibility conditions). Furthermore, the system is relatively inexpensive, physically compact, and self-contained, so that deployment in response to demand anywhere globally is technically feasible. Although GBAS relies on the core constellations, it does not rely on any other large and expensive infrastructure. Lastly, GBAS offers very cost-effective, very precise navigation service that can serve all runway ends at a given airport as well as provide improved navigation performance in the terminal area. It can do so with lower installation, maintenance, and lifecycle costs.

SECTION II

## GBAS SYSTEM DESCRIPTION

### A. GBAS System Architecture

GBAS is fundamentally a local differential satellite navigation system [5]. As such, the basic principle is that pseudorange observations made by ground-based receivers are used to develop differential corrections for each satellite. These corrections are provided to the airborne user's receiver via a data link. The airborne receiver then applies these corrections in order to produce a set of corrected pseudoranges that are then the basis of a position solution. The underlying assumption is that, for relatively short baseline separations between the ground-based receivers and the airborne-based receivers, the most significant error sources will be common to both observers and will therefore be eliminated by the differential processing [6].

Fig. 1 illustrates a typical GBAS, which consists of a ground segment, airborne segment, and space segment. The space segment for GBAS consists of GNSS satellites from the core constellations (GPS [7], [8], [9] and GLONASS [12]) as well as ranging sources that may optionally provided by an SBAS [13]. A GBAS may provide augmentation signals based on GPS alone or, optionally, may include augmentation information for GLONASS and/or SBAS satellites as well.

Fig. 1. GBAS system.

The GBAS ground segment consists of three or four GBAS reference receivers that are sited typically on or near an airport property. These reference receivers track the signals from navigation satellites and pass pseudorange measurements and other information relevant to signal health and system performance monitoring to a central processing facility. The central processing facility uses the multiple, redundant observations of the pseudoranges to compute estimates of the pseudorange corrections for each satellite signal observed by the reference receivers. The central processing facility also monitors the signal integrity and computes parameters for each satellite that the user may use to determine the availability of the signal in space for a desired level of service and a given satellite geometry [14], [15], [16]. The differential corrections and integrity information are broadcast to the user over a very high-frequency (VHF) data broadcast (VDB) signal transmitted in the 108.0–117.975 MHz band.

The GBAS ground facility may provide corrections for SBAS satellite signals. However, if SBAS signals are used, the GBAS only uses relevant information for the signal ranging function. The SBAS augmentation (i.e., corrections and integrity) data are not used.

The GBAS also broadcasts information that is used to define a reference path typically leading to the runway intercept point. A GBAS ground station can uplink reference path information for as many as 49 different reference paths using a single radio frequency. (Even more reference paths could be supported by using additional radio frequencies.) Hence, a single GBAS facility can potentially provide service to all the runway ends at a given airport. By uplinking the reference path information on the data broadcast, the integrity and availability of this information are controlled by the service provider. Should a runway be closed, then the approach can be effectively deactivated by removing the associated reference path data from the VDB transmission stream. Also, if a runway's usable length is altered due to maintenance activities, the approach can be revised as necessary to support continued operations on the runway.

The 108–117.975 MHz band is currently also used by VHF omnidirectional ranging (VOR) systems and instrument landing systems (ILSs), which are well established conventional aviation navigation systems [55]. Consequently, the band is somewhat congested in some regions of the world. The VDB signal structure was designed to provide very high spectral efficiency to help mitigate the potential difficulty a service provider may have in finding unused frequencies in the band. A single ILS frequency assignment provides only a single approach to a single runway end. With GBAS, the same 100 kHz currently required for an ILS assignment can theoretically support up to 192 approaches with the capability for multiple approaches to the same runway end if desired. Clearly, GBAS offers significant flexibility over the existing ILS and VOR uses. In an age where the electromagnetic spectrum continues to become more crowded, the efficient “reengineering” of this valuable band allocated to aviation navigation is an important feature of GBAS. The fact that a single ground station can support all the runway ends at an airport is perhaps a valuable economic boon. However, the fact that precious spectral resources are being used efficiently is probably more important in the long run with respect to viability of this system.

### B. GBAS Signal in Space

The GBAS signal in space is defined as the combination of the satellite signals from the core satellite constellations and the VDB signal [14]. The current standards allow a GBAS to augment signals from either the GPS or GLONASS constellations.

The GPS signals are direct sequence spread spectrum signals. These signals consist of a binary phase-shift keyed (BPSK) modulated carrier with a pseudorandom binary code at a chipping rate of 1.024 Mchips per second. In addition, a 50 bit per second navigation message is combined with the direct sequence signal. The primary civil GPS signal, designated “L1 Coarse/Acquisition” or “L1 C/A,” is centered nominally at 1575.425 MHz. All satellites transmit at the same nominal frequency, and code-division multiple access is used to share the band. The GPS signal structure is well documented in the GPS Interface Control Document (GPS ICD-200 [8]) and the interested reader is referred there for further details. A good description of the signal structure as well as a general description of entire GPS system can be found in [10] and [11].

A GBAS currently uses only the primary GPS civil signal L1 C/A. GPS also includes signals at another frequency, designated “L2.” A GBAS makes no use of those signals, as they are not fully supported for civil applications and are primarily intended for use by Department of Defense authorized users. Modernization plans for GPS include the addition of a third frequency, designated “L5,” which is intended to support civil applications. The current definition of GBAS does not support the use of these planned signals, but it is anticipated that the standards will be expanded to allow for use of these signals once they are available.

The GLONASS signals are similar to GPS signals in that they are BPSK modulated direct-sequence spread-spectrum signals. However, unlike GPS, GLONASS uses frequency-division multiple access as the means for satellites to share the frequency band. GLONASS satellites broadcast on different frequencies. The GLONASS signal structure is well documented in the GLONASS Interface Control Document [12] and, again, the interested reader is referred there for further details.

The VDB signal produced by the GBAS ground station employs a differential 8-phase shift key (D8PSK) waveform. This waveform was chosen because of the relatively good spectral efficiency in terms of the number of bits per second that can be supported within a 25 kHz frequency assignment. Fig. 2 illustrates a typically D8PSK modulator and the resultant 8-phase constellation.

Fig. 2. D8PSK modulator and phase constellation.

Table 1 illustrates the coding of 3 bits to a relative phase change. Every phase transition represents 3 bits. The signal is modulated with 10 500 phase transitions per second, resulting in 31.5 kbps of data. Then rate 1/3 error correction coding is employed, so the information bit rate is 10.5 kbps. However, the system uses relatively low duty cycle transmission bursts so that the effective information bit rate is much smaller (on the order of one-fourth or less) for a given GBAS.

Table 1 D8PSK Data Encoding

The VDB signal is broadcast with either horizontal or elliptical polarization. The standards require a minimum signal strength in the horizontal component throughout the coverage region. Optionally, the ground segment may also provide a vertical component to support certain military aircraft that cannot practically carry a horizontally polarized antenna due to physical constraints. If a vertical signal component is provided, the resultant composite signal must be elliptically polarized and must meet nominal requirements regarding the ratio of power in the horizontal and vertical components as well as maintain a phase relationship between the two within a specified tolerance.

The choice of horizontal polarization in this particular frequency band of 108.0–1117.95 MHz was driven by the fact that most existing commercial airplanes are already equipped with navigation systems that use antennas that cover this band, i.e., ILS and VOR. For most installations, the same antenna will be used for the reception of VDB signals.

The GBAS VDB signal uses a time-division multiple-access (TDMA) scheme to enable multiple ground stations to share the same physical frequency. GBAS VDB transmissions are constrained to occur within a time slot structure that consists of two frames per second with eight slots per frame. The basic update rate for differential corrections from a GBAS is 2 Hz, so one set of corrections is broadcast via the VDB in each frame. The GBAS TDMA frame and slot structure is illustrated in Fig. 3.

Fig. 3. GBAS VDB TDMA frame and slot structure.

A GBAS service provider will assign a subset of the eight available time slots to a particular ground station. The ground station will broadcast a single burst of data within the assigned time slot or slots. Each transmission burst includes a training sequence and forward error correction coding bits. Within the payload portion of the transmission burst, the service provider will put one or more message types, as illustrated in Fig. 4. Four types of messages are currently defined for GBAS.

• Message Type 1—Differential Corrections: Includes differential correction and integrity related data for each satellite tracked by the ground system.

• Message Type 2—Contains important information about the ground system (such as the GBAS reference location).

• Message Type 4—Approach Path Definitions: Includes Final Approach Segment definitions for each runway end or approach served by the ground segment.

• Message Type 5—Predicted Ranging Source Availability: Is an optional message that gives an indication of when ranging source corrections are expected to become available of unavailable in the Type 1 Message). (This optional message is generally not used by the current implementations and will not be discussed further in this paper.)

Fig. 4. VDB transmission burst and message types.

Message type 1) messages include differential corrections for each satellite tracked by the ground station with an elevation angle of greater than 5°. In addition, the message includes information regarding integrity in the form of parameters to be used in the computation of “protection levels.” Finally, the Type 1) message includes a cyclic redundancy check (CRC) intended to ensure the same ephemeris information is being used by the ground segment and the airborne user to compute the position of the satellites. A CRC is computed by the ground station using the ephemeris data currently being used to compute the satellite position in order to develop the pseudorange corrections. This CRC is then uplinked via message type 1) to the user who compares this CRC to one computed with the ephemeris data being used by the airborne equipment for the same satellite. If the CRCs do not match, the associated satellite is excluded from the position solution.

Message type 2) contains information related to the GBAS ground station such as the reference location of the station as well as tropospheric and ionospheric modeling parameters for integrity monitoring. This message type includes a number of optional extension blocks that carry some specific information needed for specific types of service.

Message type 4) contains Final Approach Segment (FAS) definitions. A type 4) message will include one or more FAS definitions, which consist of specific points that together unambiguously define a straight line that is the intersection of two reference surfaces: a plane defining the desired lateral path and a conical surface defining the desired vertical descent path. The apex of the conical surface is typically on the runway at the Glide Path Intercept Point (GPIP). The vertical place defining the lateral path typically also includes the GPIP and bisects the runway at the centerline. However, the FAS definition is general enough that final approach paths that are offset and not aligned with the runway can be defined. In fact, virtually any straight line segment path can be defined by the parameters that make up a FAS definition. The points used to define an FAS are illustrated in Fig. 5. The deviation reference surfaces are further illustrated in Fig. 6.

Fig. 5. GBAS reference path definition for approach services.
Fig. 6. Deviation reference surfaces.

### C. GBAS Performance

GBAS performance is characterized in several ways. One fundamental metric is the “signal in space” (SIS) space performance. The SIS performance is defined in terms accuracy, integrity, continuity, and availability of the service. This SIS performance is referenced to the output of a “fault free” instance of compliant user equipment. For example, integrity is defined at the output of user equipment that is conforming to certain mandatory functional requirements that define how the data from the ground station is combined with measurements made by the airborne equipment. A full discussion of SIS performance for GBAS is beyond the scope of this paper. Fortunately, copious references are available [14], [15], [16], [21], [22], [23], [24], [25], [26], [27], [28].

Underlying the design of GBAS is a philosophy that allocates responsibility for potential error sources and fault modes. This philosophy can be summarized as follows.

• The airborne equipment is responsible for making good pseudorange measurements, monitoring the quality of those measurements, and for following the established protocols for the combination of those measurements with data from the ground station. The airborne equipment is responsible for monitoring, detecting, and mitigating any faults that originate within the airborne equipment.

• The ground segment is responsible for monitoring the satellite signals, computing differential corrections, and for detecting and mitigating faults in the satellite signals that could result in unacceptably large errors in the position solution at the output of a compliant “fault free” receiver. The ground segment is therefore responsible for monitoring for, detecting, and mitigating any fault conditions that originate in the ground station or the satellite constellation.

There are several potential sources of error that could cause unacceptably large position errors in a differential satellite navigation solution, including:

1. hardware/software failures in the ground station or satellites (e.g. clock failures);

2. multipath at the ground segment receiver antennas or airborne receiver antennas;

3. large errors in the ephemeris information broadcast by the satellites;

4. residual ionospheric errors due to relatively small scale structures in the ionosphere;

5. residual errors due to tropospheric effects;

6. deformation of satellite signals at the source.

The detection and mitigation of these error sources has been an active area of research for many years now, and some very sophisticated monitoring techniques have been developed. Among the error sources listed above, the satellite ephemeris failure and the residual ionospheric error effects have proven to be very challenging [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40]. The potential error for both of these sources tends to decorrelate as a function of increasing distance between the user and the ground station. Development of effective monitors and mitigations for these error sources has been challenging. Again, a detailed discussion of integrity monitoring for GBAS is beyond the scope of this paper. However, it should be noted that fault monitoring, detection, and mitigation is an automatic function that is essentially transparent to the operational use of the system. The user is given a positive indication if the system should not be used.

Many other performance metrics for GBAS are also important from a practical perspective. For example, transmit power, out-of-band spurious emissions, and signal polarization are all very important when considering the spectrum management aspects of the operational implementation of GBAS.

#### Typical GBAS Accuracy Performance

Over the last decade and a half, many organizations in many countries have conducted flight trials and experiments with GBAS [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54]. The basic technology has not changed (carrier smoothed code phase) during that time, although certain developments, such as the multipath limiting antenna (discussed below), have improved nominal performance. Not surprisingly, there is very consistent experience with the nominal accuracy of GBAS across these various flight trials.

The instantaneous accuracy of a GBAS depends on the current satellite geometry. However, given the current GPS constellation, horizontal accuracy on the order of 1 m 95% (or better) is typical. Similarly, nominal vertical accuracy on the order of 1.5 m 95% is typical. The 95% position accuracy specified by ICAO to support CAT I operations is 4 m 95% in the vertical dimension and 16 m 95% in the lateral direction. Hence GBAS easily meets the accuracy requirements for CAT I approaches. However, the more difficult problem for GBAS, as noted above, has always been in meeting the integrity requirements.

### D. GBAS Landing System (GLS) Approach Selection

Unlike ILS or the microwave landing system (MLS), GNSS-based approaches are not referenced to a ground-based antenna. Both the desired flight path and the measured or estimated position of the airplane are referenced to an earth fixed coordinate frame. The desired reference path is defined by a set of coordinates supplied via the data broadcast. A mechanism is required that allows the correct set of coordinates defining the approach to be selected. Also, a mechanism to verify that the correct selection has been made is required.

As new operational capabilities are introduced, consistency with existing operations is highly desirable. Consistency reduces the cost of integrating the new functionality, the cost of training pilots, and safety hazards that can occur if inconsistent system interfaces are used. Consistency with existing ILS and MLS operations is also important to enable air traffic management service providers to handle mixed-mode operations at airports. The transition from ILS to GBAS is likely to take decades, and some airports will need to support operations with both systems during the transition.

The data broadcast on a single frequency may contain FAS datablocks for several different runway ends, multiple approaches to a single runway end, or even multiple runway ends at multiple airports. Due to the TDMA frequency sharing, a receiver tuned to a given frequency could see FAS definitions provided by multiple different ground stations for airports separated by large distances. Therefore, unlike ILS, the simple selection of a frequency does not uniquely identify a specific approach.

Fig. 7 illustrates the data required to uniquely identify a GBAS approach. A selection scheme that required entry of all the unique data would require entry of 12 alphanumeric characters. Such a requirement would be burdensome in situations that entail high pilot workload. If an onboard database were used, the amount of data that is required to be entered could be reduced. However, GBAS approaches, particularly for CAT II/III operations, should not be dependent on the existence of an onboard database. The option for a simple manual entry is required because the onboard database equipment may fail. (Availability of a CAT III approach today does not depend on having a functional flight management system (FMS) or database.) Furthermore, some airplanes are not FMS equipped, some airlines prefer not to carry FMS equipment, and some airlines prefer to not include the FMS on their minimum equipment list for dispatch. Even on some FMS equipped airplanes, a simple control head type interface for the approach selection is preferable, particularly in a retrofit situation. Therefore, at a minimum, the approach selection methodology is required to be simple enough that the pilot can manually enter all the required data, even in a high workload situation.

Fig. 7. Data required to uniquely identify a GBAS approach.

The solution for GBAS approach selection is based on the assignment of five digit channel numbers in the range of 20 000 to 39 999. Each approach is assigned a channel number that allows the user to unambiguously select that approach.

Use of a five-digit channel assignment for GBAS allows GBAS to be consistent with ILS. Flight deck integration and crew operations can be consistent between ILS and GLS. If a certain airplane model uses a simple control head for tuning ILS, then the control head interface can be expanded to include GBAS approach selections. If an airplane has an “autotune” interface based on an FMS and navigation database, then the GBAS tuning can be realized in a manner consistent with that interface. The five-digit channel could be stored in a database in the same manner that the ILS frequency is stored in the database today. In many airplanes, the pilot never deals with the ILS frequency (unless there is an FMS failure) and the same sort of interface is supported by the GBAS five-digit tuning. It is important to understand the distinction between the signal-in-space/equipment interface and the pilot interface. For ILS, the signal-in-space/equipment interface is the frequency. The pilot interface could be a selection of airport and runway. Similarly, the signal-in-space/equipment interface for GBAS is the five-digit channel number. The pilot interface can be something more sophisticated (with the aid of a database). In both cases, the lowest common denominator for the pilot interface is the signal-in-space/equipment interface.

The GBAS channel range is restricted to 20 000 to 39 999 so that the channel space is segregated from other approach selections. Table 2 lists the channel or frequency range for ILS, MLS, GBAS, and SBAS. The ranges are segregated to help reduce the possibility of entry error. Also, segregating the ranges allows the channel assignment to identify the system being tuned, thereby eliminating another pilot entry.

Table 2 Channel/Frequency Range for Precision Approach Systems

#### Approach Selection Verification for GBAS

Positive and unambiguous feedback to the pilot is required for verification that the correct approach has been selected. The channel number allows a specific FAS datablock to be identified. Once the desired datablock has been found, any information contained in the datablock could be presented to the pilot as an independent verification that the correct block has been found. Here again, similarity to the current ILS system is advantageous. Hence each unique FAS datablock is assigned a four-character approach identifier or “ident.” For example, an approach to Boeing field could be assigned the identifier “GBFI,” which is analogous to the ILS audio ident assigned to the ILS approach at the same runway “IBFI.” Optionally, the Airport ID, Runway ID, and Route Indicator from the selected datablock could additionally be used as feedback that the correct approach has been selected. However, the four-character approach ident is the preferred feedback method and should be supported as a minimum for the following reasons.

• The four-character approach ident is similar to the ILS ident and will enable cockpit integrations to be consistent across ILS and GLS.

• The four-character ident supports audio feedback. Some airplanes may need an audio feedback mechanism to be used in the event of equipment failures, etc. The ILS ident is presented to the pilot as an audio (Morse code) signal. For consistency, the GBAS approach ident could be provided the same way. Audio feedback is advantageous because it is consistent with past precedent, but modern display systems make it unnecessary. The four characters could also be displayed on the primary display or on the control head.

• The pilot selects the approach by entering a five-digit number, and verification comes back in the form of a four-character code. The feedback is in a dissimilar form and could only have come from the desired datablock (i.e., it is not entered or stored in the database, etc.).

• The approach idents can be selected to minimize the potential for confusion. For example, if there were two approaches to the same runway, one with a 3° path and one with a 4° path, they could be assigned dissimilar idents such as “GABC” and “GXYZ.” If the Airport ID, Runway ID, and Route Indicator were used as feedback, they would be seven characters long and only differ by one character.

Both the channel number and the approach ident will be shown together on the approach plate just as the ILS frequency and ILS ident are shown on the approach plate today.

A strong mechanism to avoid selecting the wrong approach is needed for GLS because multiple approaches to the same runway end are supported. Consider the example given in Table 3, where a single runway end has three different approaches. Each of the three approaches uses a different Final Approach Segment path. Two of the three defined paths are intended to support special operations. One has a steeper glide path, which may be useful for noise abatement or for wake vortex mitigation. The other is an offset approach that could be useful for converging operations to closely spaced parallel runways. In this example, it is conceivable that if a pilot accidentally selected the wrong approach, it would not be immediately obvious since he would be getting what appears to be normal guidance and his location relative to the airport is correct, etc. The approach identifier is the feedback mechanism that tells the pilot which approach is actually selected.

Table 3 Hypothetical Example of Three Approaches to the Same Runway End

Table 3 illustrates that a pilot interface based on Airport ID, Runway ID, and Route Indicator is somewhat prone to accidental selection of an unintended approach since the identifiers only differ by one letter. The GBAS channel assignments are such that no two channels that map to the same frequency have more two digits in common. Similarly, the approach identifiers were chosen such that there are virtually no letters in common. This should aid in preventing input errors if the approach ident is displayed and checked against the approach plate.

Selection of an ILS approach does not require a route indicator because each ILS provides only a single approach. The potential for multiple approaches to a single runway with GBAS is one of the main advantages of GBAS. This advantage comes at a price of needing an extra piece of information to distinguish between the approaches. The five-digit channel method encodes the differences and does it in a manner that reduces the potential for confusion.

### E. Specifics of the GBAS Channel Mapping

The channel number has encoded within it both the frequency selection and the selection of the specific approach. In other words, all 12 characters worth of information are contained in the channel number. Two pieces of information are derived from the channel number using the following two simple formulas.

The data broadcast physical frequency is given by TeX Source $$F=108.0+\left((N-20\,000)\;{\rm mod}\ 411\right)\cdot 0.025\ {\hbox{MHz}}.\eqno{\hbox{(1)}}$$

The reference path data selector (RPDS) is given by TeX Source $${\hbox{RPDS}}=(N-20{\thinspace}000)\;{\rm div}\;411\eqno{\hbox{(2)}}$$where TeX Source \eqalign{{\rm x}\ {\rm div}\ {\rm y}=&\,{\hbox{the integer part of the quotient}}\ {\rm x/y}\cr{\rm x}\ {\rm mod}\ {\rm y}=&\,{\rm x}-\left(({\rm x}\ {\rm div}\ {\rm y})\ast{\rm y}\right).}

The number 411 used in the mapping is larger than the number of physical frequencies available in the band. This might seem wasteful in that some of the availability channel space (20 000–39 999) will actually be mapped to frequencies outside the allocated range. The reason for using the value 411 in the mapping is that this ensures that, for a given physical frequency, all channel numbers that map to that physical frequency will differ from all other channel numbers by at least three of the five digits. This means an erroneous pilot entry of one digit will not result in selection of a different approach supported by the same ground station. Hence, the probability that an erroneous pilot entry would go undetected is reduced.

The following steps describe a typical GBAS tuning scenario.

1. The pilot selects the approach. This can be done either by selecting the GBAS channel number directly or through some other interface that allows the pilot to select the approach using the airport identifier, runway identifier, and route identifier.

2. The channel number N corresponding to the desired approach is provided to the GBAS equipment.

3. The frequency F of the data broadcast channel is computed using (1).

4. The RPDS is computed using (2).

5. The data broadcast receiver tunes the frequency F and demodulates all the data found on that frequency.

6. All the type 4) messages received are searched until a FAS datablock that contains a value in the RPDS field that matches the RPDS computed in step 3) above is found. When the matching RPDS is found, the desired FAS datablock has been found.

7. The selected ground station is determined from the header of the type 4) message, which contained the FAS datablock with the matching RPDS.

8. The approach ident from the FAS datablock with the matching RPDS is returned from the GBAS equipment as verification of the correct selection. (Optionally, the airport identifier, runway identifier, and route indicator for the selected FAS datablock could be used for selection verification.)

The approach selection and verification scheme is illustrated in Fig. 8.

Fig. 8. GBAS approach selection and verification.

### F. Management of GBAS Channel Assignments

In order for this tuning scheme to identify one and only one approach, the assignments of frequencies and RPDS values must be managed so that an airplane will receive a given RPDS from only one ground station on a given frequency. Each RPDS must be associated with one and only one FAS datablock within radio range for a given frequency. This means that the RPDS assignments for GBAS installations that share the same frequency must be carefully coordinated. The coverage for each station for an airplane at altitude must be considered, as well as the regions where coverage for two stations overlap.

GBAS channels can be reassigned for stations that are out of radio range. Consider the simple example given in Fig. 9, where four airports share a single GBAS VDB frequency assignment. Because of the TDMA nature of the datalink, those airports may be relatively close to each other, or they may be very far from each other such that their coverage areas overlap near the edge of coverage. Coverage here should be understood to mean the actual volume within which the VDB signal can be received and decoded by an airborne user. In many cases for a user at altitude, VDB coverage is likely to extend to the line of sight radio range.

Fig. 9. Simple example: four airports share a single data broadcast frequency 115.525 MHz.

From Fig. 9, it can be seen that there are areas where an airborne user can receive two or three of the GBAS stations simultaneously. However, there is no area where an airborne user can receive the signals from airports B and C simultaneously. Therefore, the RPDS assignments and associated channel assignments used at airport B can be reused at airport C. (The TDMA slots assigned to airports B can also be reused by airport C.)

### G. GBAS Services

A GBAS may support two basic types of service: approach services and the GBAS positioning service (GBAS/PS).

• The GBAS/PS enables the user to compute an accurate differential position solution with integrity. Only one type of positioning service is defined.

• A GBAS approach service enables the user to compute an accurate differentially corrected position solution but also includes the definition of a reference path so that the airborne equipment can compute guidance information (deviations) relative to the reference path. Multiple types of GBAS approach service are defined that provide different levels of performance. Additional approach services are currently being developed to allow GBAS to support additional types of operations.

A given ground station may or may not support either or both of the services described above. In this case, the airborne equipment is required to output deviation guidance relative to the selected reference path based on differentially corrected position. At the same time, the receiver outputs position, velocity, and time (PVT) information for use by other airplane system based on unaugmented GPS. If the ground station does support the GBAS positioning service, the airborne equipment will output PVT based on the GBAS corrections. The ground station indicates in the type 2) message whether or not the GBAS/PS is supported via information conveyed in message type 2). It is also possible for a GBAS ground station to support the GBAS/PS and not provide any approach service.

The type of service used by the airborne equipment in all cases is determined by the approach selection as described above. The fundamental interface is the five-digit channel number. If the RPDS derived from the five-digit channel number [per (2)] matches the Reference Station Data Selector (RSDS) value uplinked in message type 2), this indicates that the receiver has been instructed to select the GBAS/PS only. The receiver then outputs differentially corrected PVT and does not compute or output valid deviations. If the RPDS derived from the channel number matches the RPDS in an FAS datablock within message type 4), then the deviations are computed and output. If message type 2) indicates the GBAS/PS is supported, the receiver will output differentially corrected PVT. The airborne equipment will automatically support both types of service simultaneously if they are available.

Assignments of RSDS must be managed along with RPDS assignments as described in Section II-E. RPDS and RSDS assignments must be unique within radio range on a given frequency.

### H. GBAS Ground Segment

As mentioned earlier, the GBAS ground segment consists of a set of GBAS reference receivers, a processing facility, and a VHF data broadcast facility. This section will describe some typical hardware implementations of those functions and discuss issues associated with siting.

Fig. 10 shows a picture of a GBAS reference receiver installation. This particular reference receiver configuration was implemented by the University of Oklahoma [56], [57] and has been used in GBAS-related research in support of the Federal Aviation Administration (FAA)'s program to develop the technology [58], [59], [60]. The FAA's GBAS project has gone by the name of the Local Area Augmentation System (LAAS), so throughout the literature one will see references to LAAS where the specific FAA GBAS program is being discussed. One can think of “LAAS” as the FAA's brand name for their GBAS product. GBAS is the internationally accepted term for this type of augmentation system.

Fig. 10. GBAS reference receiver with multipath limiting antenna system.

A GBAS reference station such as the one shown in Fig. 10 would typically be sited on or near an airport property. Proper siting of the reference receivers is very important in that a major source of potential error in a differential GPS system is multipath at either the reference or the user receiver [61], [62], [64]. Errors induced by signal energy that arrives at the receiver via reflection, diffraction, or any means other than the direct path will be unique to one side of the differential system and therefore will not be compensated for by the differential corrections. One source of potential multipath is the ground. To combat this source of error, multipath limiting antennas (MLAs) such as the one shown in Fig. 10 can be employed. The MLA antenna technology was originally developed by Braasch at Ohio University [62]. That development work led directly to the development of the commercial antenna product depicted in Fig. 9 [63]. The MLA shown consists of a stacked array of elements approximately 2 m tall. The elements of the array are phase-combined in a manner such that the resultant pattern has a very sharp cutoff at the horizon, and several tens of decibels of rejection is provided against signals that would enter the antenna from arrival angles below the horizon (i.e., 0° elevation). Multipath resulting from signals reflecting off the ground would arrive from elevation angles below zero degrees, and in no cases would valid direct path satellite signals arrive from such angles. Hence the antenna greatly reduces the energy in the multipath signals.

Much research has been done on the effects of GPS multipath. The interested reader is referred to [62], [64], and [66], [67], [68], [69], [70], [71] for more information.

The reference station shown in Fig. 10 employs two antenna systems: the MLA array and a zenith coverage antenna. The MLA provides coverage from near the horizon to between 30° and 40°. The zenith coverage antenna has complementary coverage such that satellites between 30° and 90° elevation can be tracked. Recent antenna designs for MLAs provide full coverage of the upper hemisphere with a single MLA array [72].

Multipath limiting antennas can be very effective at mitigation of ground bounce multipath. However it is still important to carefully site the reference station antennas to minimize the effects of multipath from other sources. Hence a site survey that evaluates potential multipath sources is typically conducted [73], [74], [75]. Post installation evaluation of multipath errors can be accomplished by code-minus-carrier analysis. However, the results of such observations are limited to evaluation of multipath that results from where the satellites happen to occupy the sky over the duration of the test. A limited observation period is never guaranteed to produce the worst possible multipath error that could be generated by an environment. Therefore, siting evaluations typically involve analysis coupled with observations. Finally, performance monitoring over time during operation can be used to identify, isolate, and mitigate multipath effects.

Another factor influencing the siting of reference stations is that fact that the GBAS facility uses multiple reference observations as a means to monitor for and limit the impact of noncommon mode error sources such as multipath. Therefore, reference stations must be spatially distributed in a manner that ensures the effects of environmental multipath will be uncorrelated.

Fig. 11 shows a typical central processing facility for a GBAS. The facility shown in the figure is the prototype GBAS ground station installed at SeaTac airport. This ground station was manufactured by Honeywell [77], [78]. The racks shown contain the computers that process the pseudorange correction and that support integrity monitoring. These racks also include the VHF transmitter equipment and power management equipment. In total, about one and a half racks of equipment are required for this installation. This compares very favorably to ILS since this GBAS equipment can support all the runway ends at the airport, whereas ILS would require multiple shacks of equipment such as the one shown to provide similar approach services. As a case in point, the GBAS shown in Fig. 11 broadcasts FAS definitions for ten different approaches: four runway ends at SeaTac; two runway ends at Boeing Field; two runway ends at Renton Field; and two runway ends at Paine Field. In addition, the GBAS provides the positioning service that can support Aera Navigation (RNAV) operations in the terminal area and perhaps will support surface applications in the future. RNAV is a method of air navigation that allows an aircraft to fly desired paths point to point, rather than being constrained by paths that lead directly to and from navigation aids. This can conserve flight distance, reduce congestion, and allow instrument flight procedures into and around airports without conventional navigation aids like distance measuring equipment (DME) or VOR. Surface applications may include things like:

• enhanced situation awareness on the airport surface through use of digital airport maps;

• guidance on the airport surface during very low visibility conditions;

• surveillance of airplane positions by the air traffic service providers (through Automatic Dependent Surveillance-Broadcast);

• runway incursion detection systems;

• reduction of runway occupancy time through optimum runway exiting applications.

Fig. 11. GBAS ground facility—SeaTac airport.

The differential corrections and integrity information are broadcast to the user over a VHF data broadcast signal transmitted in the 108.0–117.975 MHz band. Fig. 12 shows a typical VHF antenna installation to support the VDB function of a GBAS ground subsystem. This antenna is installed at Memphis Airport and is part of the Honeywell prototype GBAS installed there to support the FAA development program done in cooperation with FedEx. The antenna is an array of three elliptically polarized sources, which are themselves constructed of four folded dipole elements. A stacked array of three sources is used to provide increased gain towards the horizon and improved system range, and to reduce the illumination of the ground, thereby reducing the impact of nulls in the pattern that are induced by the ground reflections. As a result, this configuration also mitigates some of the deep nulls that might otherwise result from a horizontally polarized signal source above the ground. As mentioned above, elliptical polarization is optional (although recommended by ICAO), and a horizontally polarized antenna would be somewhat simpler than the one shown in Fig. 12.

Fig. 12. VHF data broadcast antenna.

### I. Airborne Implementation—GBAS Landing System (GLS)

This section will discuss typical airborne implementation of GBAS equipment and a GBAS landing function (GLS).

First, let us review some terminology. The reader may have seen the term GBAS and the term GLS seemingly used interchangeably (even in this very paper). Strictly speaking, this is not correct, as the terms refer to different things. GBAS refers to the total navigation system composed of the ground segment, space segment, and airborne segment. GLS refers to the function as it is integrated into the airplane. Although GLS is based on GBAS and includes the GBAS airborne equipment, it also includes other equipment not specific to the GBAS such as the autopilot, displays, approach selection mechanisms, and other parts of the GLS system on the airplane. However, not everyone (including the author) is always careful about this terminology, and sometimes GLS is used where GBAS might have been more appropriate and vice versa.

Fig. 13 shows a block diagram of the GLS system architecture for the Next Generation 737 (NG 737). The NG 737 implements the GBAS airborne functions in a multimode receiver (MMR) [79], [80], [81]. The MMR has quickly become the preferred package for integration of GBAS functionality (and basic GPS functionality for that matter) on commercial air transport class aircraft. This is due to the fact that the MMR readily allows integration of the GBAS function into an existing landing system architecture based on ILS. The MMR leverages existing interfaces, and this results in relatively modest wiring changes being required for a given airframe. Furthermore, given the choice of VDB frequency and polarization, existing ILS and/or VOR antennas can be utilized for the VDB function, thereby eliminating the need to install an additional antenna.

Fig. 13. NG 737 GLS airborne system architecture.

From Fig. 13, it can be seen that the MMR (or navigation landing sensor) interfaces to many systems on the airplane. This is a consequence of the general nature of modern integrated glass cockpit aircraft. Although this diagram is specific to the NG 737 aircraft, diagrams of similar complexity exist for all other Boeing models in production. In the figure, sources that provide data to the MMR are colored yellow and equipment that uses the output of the MMR are coded green. No fewer than nine systems utilize the output of the MMR. This implies that changes in the interface could require each of the nine downstream systems to be modified. Therefore, an effort has been made to make the interfaces as consistent as possible with ILS in order to minimize the necessary changes.

The MMR supports the landing function as well as the basic GPS positioning in its GPS receiver function. This GPS positioning information is used to support navigation in all phases of flight. The MMR provides PVT outputs to the FMS, which is responsible for management of the multisensor navigation system. FMS systems and the management of multisensor navigation systems is a complex subject that could be the focus of another paper entirely.

When GLS is active, the MMR will simultaneously support both the landing function by providing deviations relative to a selected reference path and basic aircraft navigation through PVT outputs. As discussed above, the GBAS ground segment may support both functions simultaneously, or the airborne receiver may use the GBAS to support the landing function and support the PVT using unaugmented GPS. This means that the receiver has to be able to produce two different solutions at once. The landing system function requires deviations produced relative to a specific point on the aircraft. Therefore, the differentially corrected position solution is translated to a common guidance control point (GCP) on the airplane using the pitch, roll, and heading information from the inertial system and lever arms defining the offset between the GPS antennas and the GCP. (The lever arms are airframe specific and are stored in the MMR.) The basic PVT to support FMS operations is referenced to the GPS antenna. So again, the MMR maintains two different position solutions that are ultimately referenced to two different places on the aircraft.

Fig. 14 shows a picture of the navigation control panel implemented on the NG 737 airplane. The control panel allows the pilot to enter ILS frequencies, VOR frequencies, or GLS channel numbers by using a keypad. The desired selection is keyed into the standby window using the keypad. Then when the pilot wishes to make the control selection active, the double arrow button on the left-hand side of the panel is used to swap the active and standby windows, which changes the tuning commands sent from the control head.

Fig. 14. Control panel for GLS approach selection on NG 737 airplane.

Fig. 15 shows a typical FMS approach selection interface through a page on the multifunction control display unit (MCDU). This interface allows the pilot to select an approach procedure by name (e.g., GLS32LA). Although the menu selection may be an alphanumeric sequence that is meaningful to the pilot, the five-digit GBAS channel number is stored in the FMS database, and it is this channel number that is supplied to the MMR to accomplish the approach selection. In a typical modern integrated glass cockpit aircraft, the FMS navigation database will contain much information about a given approach, including the location of the threshold, heading of runway, and other physical attributes. This information is used to generate display symbology. However, should the FMS fail, the pilot can use a reversionary mode where the five-digit channel number is entered directly, the approach is selected, and raw deviation data is provided through the either the primary display or through the standby instruments.

Fig. 15. Typical FMS approach selection interface through MCDU.

The GBAS airborne equipment is integrated into the airplane in much the same manner that the ILS sensor is. The airplane is designed to support automatic landings and exploits redundancy to ensure that any failures of airborne equipment that might cause a hazardous situation are detected. For this reason, multiple independent threads of equipment are used for fault detection. The autopilot system will perform signal selection and fault detection using the multiple inputs. In some configurations, the receivers themselves are dual redundant internally so that the output of the receiver is always monitored. With such dual receivers, a fail operational capability can be implemented with only two independent sensor threads.

SECTION III

## GBAS OPERATIONAL USES

### A. Operational Use of GBAS

GBAS was originally conceived primarily as an approach and landing aid. Consequently, the principle operational use of GBAS is to support takeoff and landing operations similar to those supported by ILS today. In fact, GBAS is likely to eventually displace ILS particularly for support of very low-visibility landing operations. A typical operation supported by GBAS will look very similar to a typical ILS operation. There will be a transition from some terminal area arrival procedure to the final approach path, typically a straight segment aligned with the runway. Today, typically, at a large airport, Air Traffic Control directs aircraft to the localizer via assigned headings or vectors. Often, several aircraft will be on the approach at the same time, separated by several miles. Once an aircraft has intercepted both the localizer and the glideslope guidance, the aircraft is said to be established on the approach. Typically, an aircraft will be established on the approach by 6 nmi from the runway threshold. The point where the aircraft is established on the final approach segment is at or near a final approach fix.

The deviation of the aircraft from the selected path is indicated to the flight crew by means of a display that includes vertical and horizontal bars referred to as “needles.” The two “needles” are the course deviation indicator (CDI). In a modern glass cockpit, the CDI is typically integrated as one element of a multifunction display with an associated flight director to aid guidance and an autopilot to automatically track the intended path.

The output from the navigation receiver (ILS or GBAS) goes both to the display system and to the flight control computer. An aircraft landing procedure can be either “coupled,” where the autopilot directly controls the aircraft and the flight crew monitors the system; or “manual,” where the flight crew manually controls the aircraft via use of flight director commands to minimize the deviation from flight path to the runway centerline.

#### Decision Altitude (Height)

Once established on an approach, the autoland system or pilot will follow the vertical and lateral deviations and descend along the glideslope until the decision altitude is reached [82], [83], [84]. (For a typical Category I operation, this altitude is 200 ft above the runway.) At this point, the pilot must be able to see the runway or runway environment (e.g., the runway lights) in order to continue the approach. If neither can be seen, the approach will be aborted and a missed approach procedure is performed.

### B. Approach Categories

There are three categories of approach operations.

• Category I: An instrument approach and landing with a decision height not lower than 60 m (200 ft) above touchdown zone elevation and with either a visibility not less than 800 m or a runway visual range not less than 550 m.

• Category II: An instrument approach and landing with a decision height lower than 60 m (200 ft) above touchdown zone elevation but not lower than 30 m (100 ft) and a runway visual range not less than 350 m.

• Category III is subdivided.

1. An instrument approach and landing with a decision height lower than 30 m (100 ft) above touchdown zone elevation or no decision height and a runway visual range not less than 200 m.

2. An instrument approach and landing with a decision height lower than 15 m (50 ft) above touchdown zone elevation or no decision height and a runway visual range less than 200 m but not less than 50 m.

3. An instrument approach and landing with no decision height and no runway visual range limitations.

Specific types of aircraft equipment and crew qualifications are required for low-visibility operations. For example, to conduct an approach and landing using Category III weather minimums below runway visual range of 600 ft, a fail-operational autopilot system is required and crew training on this type of procedure must be current. Approach and landing using Category I minimums does not require a fail-operational system. Approaches to Category I minimums rely on baroaltimeter indications to determine when the airplane has reached the decision altitude. Approaches to Category II and Category III minimums typically use radar altimeter to determine when the decision height or alert height has been reached.

During an ILS approach, the position of the airplane along the desired reference path is determined via a marker beacon, DME, or RNAV fix [55]. The marker beacon is a set of transmitters operating at a carrier frequency of 75 MHz. When the transmission from a marker beacon is received, an indication appears on the pilot's instrument panel and the modulating tone of the beacon is audible to the pilot. The approach procedure will include information regarding the correct height at which the aircraft should be when it crosses each of the marker beacons, DME, or RNAV fixes. In the early history of ILS, this was an important feature, as it supported situational awareness of the pilot with respect to their longitudinal distance to the runway. Also, by having the flight crew cross-check the baroaltitude at the marker beacon transitions, it provided some additional assurance of a correct altimeter setting and of capture of the appropriate glidepath. For ILS, this was very important since a glideslope antenna array will naturally form “false” glidepaths that will have a much steeper descent. Use of marker beacon signals for pilot situational awareness (in commercial air transport class aircraft) has all but disappeared and has been replaced by the use of DME [55] measurements and navigation moving maps.

In some early autopilot implementations, the marker beacon transitions were used to activate time-based scheduled gain changes to the control of the aircraft [85]. Such practices were later abandoned, and gain desensitization scheduling in ILS-based autoland systems is typically driven by an estimate of distance to go based on radio altitude changes. GBAS offers a much cleaner and easier interface for autopilots since differential GPS provides precise position in three dimensions. Furthermore, GBAS airborne equipment outputs deviation indications in both angular scaling (for compatibility with existing autopilots designed for use with ILS) and rectilinear scaling. The use of the rectilinear scale deviations makes sensitivity scaling unnecessary and eliminates a source of inaccuracy inherent in ILS-based autoland control systems.

GBAS does not use marker beacons. Instead, the distance to the runway threshold is computed and displayed directly to the pilot and is available for use by other onboard equipment such as the autopilot. This is an important feature of GBAS and a great improvement over ILS. Most ILS systems include a DME installation so that the pilot has an indication of distance to the airport. This DME is less accurate and cannot be physically sited at the threshold, so typically the distance is referenced to another location. GBAS FAS path definitions typically include a point at the threshold (the landing threshold point) and hence GBAS airborne equipment can continuously compute a distance to that point.

Fig. 16 shows a prototype xLS approach plate. The term xLS refers to a generic landing system operation. The xLS procedure can be flown using either ILS or GLS. In the example chart for SeaTac Runway 16R, the approach selection information for both ILS (i.e., 111.7 MHz) and GLS (i.e., channel number 20 250) is shown. Note that the two approaches have different ident strings (i.e., “ISZI” and “GSZI”). Both approaches are referenced to the same desired approach path. The chart is nearly identical to an ILS approach chart today except for the addition of some GLS-specific information such as the approach selection channel number. (Note, this chart is presented as an example only—not to be used for navigation purposes.)

Fig. 16. GLS approach plate.

### C. GBAS Support of Terminal Area Operations

As mentioned earlier, GBAS may also provide a service that is not specific to a particular approach path or runway, i.e., the GBAS/PS. This service will allow airborne equipment to output very high-accuracy positioning with integrity with very high availability. Such a service can support RNAV and required navigation performance (RNP)-based procedures. A full treatment of RNP is beyond the scope of this paper. RNP is a parallel and complementary development that has been focused on using the capabilities of modern, multisensor navigation systems to implement new operations based on precise control and containment of aircraft trajectories. GBAS has been integrated into the aircraft in a manner such that when a ground station does support the GBAS/PS, the navigation system can take full advantage of the improved navigation capabilities. Through RNP and the use of RNAV, an airplane can reliably and repeatably fly predetermined, complex two- and three-dimensional paths such as curves and sequences of straight and curved flight legs. RNP can provide optimum transitions to GLS procedures and optimum missed approach paths from GLS procedures.

SECTION IV

## OPERATIONAL EXPERIENCE WITH GBAS

### A. Operational Experience During Development

The first commercial air transport airplane with GLS capability was certified in May 2005. The NG 737 achieved this milestone after more than a decade of technology development, standardization, and flight demonstrations. In the course of implementing GLS on the NG 737, Boeing demonstrated many advantageous aspects of GBAS capabilities for the first time. These demonstrations occurred during a technology demonstrator program in 2002 and then again during flight testing in support of the certification program. For example, during this testing, Boeing successfully demonstrated the use of multiple approach definitions to the same runway, the use of offset thresholds, and support of approach capabilities to nearby airports.

During the development program, the NG 737 was flown with GLS using six different ground stations manufactured by three different organizations serving 11 different airports including locations in both the northern and southern hemisphere. Approaches flown included typical ILS-like straight in approaches with autoland as well as multisegment curved approaches using RNAV, RNP, and the GBAS positioning service. Other demonstrations included guidance for low-visibility takeoff, guidance on missed approach, and landing performance during simulated and induced ground station failure conditions.

Boeing's early operational experience with GBAS has generally been excellent. In fact, GBAS has already provided operational advantages to Boeing. For example, during the spring and summer of 2005, GBAS supported a temporary displaced threshold to runway 13R at King County International Airport, Boeing Field (KBFI), in Washington State. King County began construction of an extension for runway KBFI 13R in May 2005. This required the ILS to be placed out of service and a temporary displaced threshold to be created. To maintain precision guidance capability to KBFI 13R, the Boeing Flight Test organization utilized the SeaTac Airport (KSEA) GBAS to create an additional GLS approach to support the displaced threshold for its test aircraft. Using the published information for the distance of the displaced threshold from the existing threshold, Boeing Flight Test engineers were able to derive the latitude and longitude of the temporary displaced threshold and mathematically define the new approach parameters within about 15 min. This information was then e-mailed to Honeywell for final coding of the FAS definition. The additional approach was loaded into the KSEA GBAS station and was flown to an autoland (in visual conditions) by the next available GLS-equipped Boeing test airplane. The flight crew reported the guidance brought the airplane across the displaced threshold at the intended height and set the airplane down on the runway center line. This was the first practical demonstration of a displaced threshold using GBAS.

Fig. 17 shows a comparison of the guidance quality provided by the ILS at Boeing Field with the guidance quality obtained using the approaches to Boeing Field provided by the SeaTac GBAS station. The scales in both plots are the same to facilitate the comparison. The ILS guidance includes errors with large and quick excursions with magnitudes of up to 70 ft (at a distance of less than a mile from the threshold). In contrast, the lateral guidance errors in the GBAS signals are apparently bias-like and limited to 6.5 ft maximum. Similarly, the ILS glideslope error characteristics are noisier than the GLS vertical deviation errors. Boeing Field is a particularly challenging environment for ILS since there are many large buildings (and often large airplanes) near the runway. Furthermore, the airport is in a valley. This results in multipath and consequently significant beam noise. In fact, the guidance quality of the ILS at Boeing Field is so poor that autopilot systems will not stay coupled making CAT II or CAT III operations impractical, if not impossible. However, with GLS, the performance is immune to the challenging multipath environment and the signal quality is sufficient to support autoland and potentially both Category II and Category III minimums. This is another boon to Boeing since it makes automatic landing at Boeing Field a reality where it was previously impossible. The approach minimums at Boeing Field are limited by other factors so it never made sense to get the ILS quality improved. It would not have resulted in any lower minimums. However, the ability to autoland, even though the minimums remain unchanged, is valuable.

Fig. 17. Comparison of ILS and GLS signal quality at Boeing Field.

It is important to note that the autoland capability at Boeing Field is being supported by the GBAS station at SeaTac Airport, which is 6 nm away, and the VDB transmitter is over a 400 ft ridge. The VDB reception is certainly not line of sight, but solid datalink reception while taxiing on the ground around Boeing Field has been consistently demonstrated.

### B. Operational Experience in Revenue Service

Operational use of GBAS in revenue service by scheduled air transport operators is now beginning. Several airlines have acquired airplanes with GBAS equipment, and the system is currently being used regularly by airlines on three continents.

#### Qantas Airways

Qantas Airways has taken delivery of its new 737-800 airplanes with GLS functionality and is committed to retrofitting their entire NG 737 fleet with the capability. The total Qantas NG 737 fleet with GLS capability is expected to be nearly 50 airplanes. Qantas has also ordered the Airbus A380 with the optional GLS capability. They are the launch customer for the GLS capability on the A380 and expect to take delivery in August 2008. Qantas believes GLS will play an important role in solving proposed critical area protection requirements. Qantas has also ordered the new Boeing 787 Dreamliner, which will have GLS on entry into service as a basic feature. (Qantas ordered 65 787s with options for an additional 50, making a total of 115 airplanes.) Between these three models alone, Qantas should have a very substantial fleet of aircraft with GLS capability.

During the aircraft acceptance phase of the NG 737 program, approximately 100 GLS approaches were flown demonstrating the various system capabilities. These flights occurred at 5 airports: KBFI (Boeing Field), KMWH (Moses Lake), KSEA (Seatac), KPAE (Paine Field) and YSSY (Euphrata), using 3 different ground stations. These demonstrations included operations to narrow runways (30 m) and displaced threshold operations.

A GBAS ground station has been installed at Sydney International Airport, and Qantas began operational trials of GLS at Sydney in December 2006. At the time of this writing, all of the Qantas B737 flight crew (580 pilots) have been trained in the used of the new system. All of the GLS capable B787-800s that Qantas has are participating in the trials. All six runways have been used in the trials. The initial operations have included parallel runway operations (during visual approaches). The trials have included autoland operations (again in visual conditions). As of May 2008, more than 1000 GLS approaches had been successfully completed in-service. Approximately 20%–25% of those operations were autoland operations. Qantas pilots have reported excellent performance under a variety of conditions.

#### TUIFly.com

TUIfly.com, in conjunction with Boeing and DFS (Deutsche Flugsicherung GmbH), began GBAS trials at Bremen Airport, Germany, in September 2007. This trial phase includes using the GBAS during visual conditions to support landings during normal revenue operations at Bremen. The trial phase will support the eventual certification of GBAS in Germany.

The GBAS ground station at Bremen was manufactured by Honeywell and is operated by DFS. Boeing has fitted the TUIfly.com aircraft with onboard equipment developed by Rockwell-Collins. The project is funded by the 3rd Aviation Research Programme launched by the German Federal Ministry of Economics and Technology.

DFS has announced that they expect to realize considerable cost savings by eventually replacing their current ILS systems with GBAS. While each runway and each landing direction requires a separate ILS, one GBAS ground station will normally suffice. Moreover, the mandatory maintenance and flight inspection activities for instrument landing systems involve substantial costs. Those costs are expected to be significantly reduced with GBAS, as the system should not require periodic flight inspection of signal quality.

#### Continental Airlines

Continental Airlines is working to begin a GLS in-service evaluation using a GBAS installation in Guam, which is scheduled to be completed in 2008. The trials would be conducted by Continental Micronesia, which is a subsidiary of Continental. The airline also has plans for eventual deployment of GBAS and GLS operations in other parts of the world.

SECTION V

## FUTURE POTENTIAL

This paper addresses only GBAS applications to date. There is a significant potential role yet to be defined for other aviation applications of GBAS that could be yet more beneficial. Some potential expanded uses of this tool include:

• use of GBAS to guide helicopter landings;

• use of mobile, temporary GBAS to support disaster relief or rescue operations;

• use of GBAS to support unmanned aerial vehicles;

• networks of GBAS stations to implement extended coverage;

• use of GBAS to support high-precision airport surface applications;

• use of GBAS to support high-precision air-to-air surveillance applications;

• use of GBAS on ships or sea-based platforms (e.g., for helicopter operations).

SECTION VI

## CONCLUSION

As we begin use of GBAS in daily airline operations, it is important for airspace managers, airline operational planners, and navigation service providers to understand the potential new or improved capabilities enabled by GBAS, as well as its potential for cost reduction for existing airspace system infrastructure. GBAS can offer a significant improvement in performance and cost relative to ILS. Furthermore, the additional services offered by GBAS should contribute to greater reliability of advanced operations.

### Acknowledgment

The authors would like to acknowledge their colleagues at Boeing who have toiled many years to make GBAS and GLS a reality. In particular, many thanks are offered to J. Ackland, R. Friedman, M. Harris, J. VandenBrooke, T. Lapp, S. Duenkel, and a host of others without whose efforts GBAS would not exist. The industry owes a debt of gratitude to Capt. A. Passerini of Qantas for his leadership in moving forward to gain operational experience with GLS. The author would also like to thank several key members of the FAA for their support of the GBAS development over the years, including B. Clark, J. Warburton, and V. Wulschleger. There are a host of other people in industry who should be thanked for their extensive contributions to GBAS, such as M. Brenner, A. Stratton, Dr. G. McGraw, C. Shively, Dr. F. VanGraas, Dr. T. Skidmore, Dr. M. Braasch, K. Class, K. McPherson, and D. Jensen, to name a few.

## Footnotes

Manuscript received nulldate; revised July 08, 2008. Current version published nulldate.

The authors are with the Boeing Company, Seattle, WA 98203 USA (e-mail: tim.murphy@boeing.com).

## References

1. ICAO State Letter 4/49.1-94/89

1994-10

2. ICAO

International standards and recommended practices—Aeronautical telecommunications, ICAO Annex 10, Amend. 77, vol. I, Radio Navigation Aids

3. RTCA

Minimum aviation system performance standards (MASPS) for the Local Area Augmentation System (LAAS), DO-245A, 9-12-2004

4. RTCA

Minimum operational performance standards for GPS Local Area Augmentation System airborne equipment, DO-253A, sec. 2.3.6.6

5. Differential GPS

B. Parkinson, P. Enge

Global Positioning System: Theory and Applications, Reston, VA, AIAA Press, 1996, vol. II, 1, pp. 3–50

6. Development of the LAAS accuracy models

G., McGraw, et al.,

Proc. ION GPS 2000

7. USAF GPS JPO

Global position system standard positioning service signal specification, 5-11-1993

8. USAF GPS JPO, ICD GPS-200

10-10-1993

9. USAF GPS JPO

Global Position System Standard Positioning Service performance standard, 2001-10

10. Global Positioning System, Signals, Measurements, and Performance

P. Misra, P. Enge

Lincoln, MA
Ganga-Jamuna, 2001

11. Understanding GPS, Principles and Applications

E. Kaplan, et al.

Artech House, 1996

12. Russian Federation

GLONASS interface control document, 5.0

13. RTCA

Minimum operational performance standards for global positioning system/wide area augmentation system airborne equipment, DO-229C

14. G(x)

T. A. Murphy, et al.

Proc. ION Nat. Tech. Meeting, 1997-01

15. Specifics of the LAAS context model, (G(x))

T. A., Murphy et al.

Proc. ION GPS Conf. 1997

16. LAAS signal-in-space integrity monitoring: Description and verification plan

T. A. Murphy, et al.

Kansas City, MO
Proc. ION GPS-97, Sep. 16–19, 1997

17. Verification of the LAAS signal-in-space integrity monitoring algorithm

F., Lui et al.

Proc. ION GPS 1998

18. Availability requirements for Local Area Augmentation System (LAAS)

T. Urda

Proc. ION GPS 2002

19. Gaussian and mixed Gaussian methods of LAAS overbounding

C. Varner

Proc. ION GPS 2002

20. A comparison of LAAS error bounding concepts

C. Shively

Proc. ION Nat. Tech. Meeting 2001

. GBAS validation methodology and test results from the Stanford LAAS integrity monitor testbed

S. Pullen

Proc. ION GPS 2000

22. Sigma estimation, inflation, and monitoring in the LAAS ground system

Pervan, et al.

Proc. ION GPS 2000

23. Core overbounding and its implications for LAAS integrity

J. Rife

Proc. ION GNSS 2004

24. B-value research for FAA LAAS station integrity and fault detection

H. Wen, et al.

Proc. Nat. Tech. Meeting Inst. Navigat., Jan. 26–28, 2004

25. LAAS ranging error overbound for non-zero mean and non-Gaussian multipath error distributions

I., Sayim et al.

Proc. ION Annu. Meeting 2003

. Experimental and theoreticalresults on the LAAS sigma overbound

I. Sayim

Proc. ION GPS 2002

27. Derivation of acceptable error limits for satellite signal faults in LAAS

C. Shively

Proc. ION GPS 1999

28. Comparison of LAAS B-values to the linear model optimum B-values

R. Kelley

Proc. ION Annu. Meeting 1999

29. LAAS ionosphere spatial gradient threat model and impact of LGF and airborne monitoring

M. Luo, et al.

Proc. Inst. Navig. GPS Conf., Sep. 2003

30. The effects of large ionospheric gradients on single frequency airborne smoothing filters for WAAS and LAAS

T. Walter, et al.

Proc. Inst. Navig., Nat. Tech. Meeting, 2004-01

31. Ionosphere spatial gradient threat for LAAS: Mitigation and tolerable threat space

M. Luo, et al.

Proc. ION Nat. Tech. Meeting, 2004-01

32. LAAS study of slow-moving ionosphere anomalies and their potential impacts

M. Luo, et al.

Proc. Inst. Navig., GNSS Conf., Sep. 2005

33. Position-domain geometry screening to maximize LAAS availability in the presence of ionosphere anomalies

J. Lee

Proc. ION GNSS 2007

34. Assessment of ionospheric impact on LAAS using WAAS supertruth data

M. Luo, et al.

Albuquerque, NM
Proc. ION 58th Annu. Meeting, Jun. 24–26, 2002, 175–186

35. Data-replay analysis of LAAS safety during ionosphere storms

Y., Park et al.

Proc. ION GNSS 2007

36. A study of the ionospheric effect on GBAS (ground-based augmentation system) using the nation-wide GPS network data in Japan

T., Yoshihara et al.

Proc. ION Nat. Tech. Meeting 2004

37. Targeted ephemeris decorrelation parameter inflation for improved LAAS availability during severe ionosphere anomalies

R. Shankararaman

Proc. ION Nat. Tech. Meeting 2008

38. Mitigation of ionospheric gradient threats for GBAS to support CAT II/III

T. Murphy, M. Harris

Proc. Inst. Navig. GNSS Conf., 2006

39. Ephemeris failure rate analysis and its impact on category I LAAS integrity

L., Gratton et al.

Proc. ION GNSS 2007

40. LAAS integrity risk due to satellite ephemeris faults

C. Shively

Proc. ION GPS 2001

41. GLS in-service experience

A. Passerini

Seattle, WA
Brief. Int. GBAS Working Group, 2007-07

42. Early operational experience with new capabilities enabled by GBAS Landing Systems (GLS)

T. Murphy, et al.

Proc. Inst. Navig., Nat. Tech. Meeting, 2006-01

43. Initial GBAS experiences in Europe

A. Lipp

Proc. ION GNSS 2005

44. Experimental GBAS performance at the approach phase

S., Saitohi et al.

Proc. ION Nat. Tech. Meeting 2003

45. Flight experiment of GBAS in Japan

S., Saitohi et al.

Proc. ION GPS 2001

46. Preparation for GBAS at Branschweig research airport—First flight test results

K., Butzmühlen et al.

Proc. ION Nat. Tech. Meeting 2006

47. Making the case forGBAS, experimantal aircraft approaches inGermany

U. Bestmann, et al.

Inside GNSS Mag., 2006-10

48. Approach with precision

Murphy, et al.

GPS World, Sep. 2006

49. Flight testing and data evaluation of ground based augmentation systems

Schachteneck, et al.

Proc. IEEE 25th Digital Avion. Syst. Conf., 15-10-2006

50. Performance of a prototype Local Area Augmentation System (LAAS) ground installation

M., Brenner et al.

Proc. ION GPS 2002

51. FAA flight test results using airport pseudolites with the LAAS Test Prototype (LTP)

J. Warburton

Proc. ION 1997

52. Validation of the FAA LAAS specification using the LAAS Test Prototype (LTP)

J. Warburton

Proc. ION GPS 1998

53. Ohio University/FAA flight test demonstration results of the Local Area Augmentation System (LAAS)

F. Van Graas

Proc. ION GPS 1997

54. Results of the Boeing/industry GPS landing system flight test experiments

T., Murphy et al.

Proc. ION GPS 1996

M. Kayton, W. Fried

2nd, New York, Wiley, 4.4.4 and 13.5

56. Flight test results of a MOPS compliant LAAS system to provide guided straight and curved path departures and missed approaches

P. Yih-Ru, et al.

Proc. ION GNSS, 2005

57. B-value research for FAA LAAS station integrity and fault detection

H. Wen

Proc. ION Nat. Tech. Meeting, 2004

58. LAAS government industry partnership

J. Miller

Proc. ION GPS 99

59. Flight test results of the FAA local area augmentation system test prototype

J. Warburton, et al.

Proc. ION Annu. Meeting, 1997

60. Validation of the FAA LAAS specification using the LAAS Test Prototype (LTP)

J. Warburton, et al.

Proc. ION GPS, 1998

61. FAA/Ohio University United Parcel Service DGPS autoland flight test demonstration

F., Van Graas et al.

Proc. ION GPS 1995

62. Multipath effects

M. Braasch

Global Positioning System: Theory and Applications, B. Parkinson, et al., Washington, D.C., American Institute of Aeronautics and Astronautics, 1996, vol. 1, 14

63. The LAAS Integrated Multipath Limiting Antenna (IMLA)

D. Bryce

Proc. ION GPS 2002

64. Investigation of multipath effects in the vicinity of an aircraft dependent on different flight profiles

Eurocontrol EEC Rep. 357, project GNS-Z-E2

65. Optimum antenna design for DGPS ground reference stations

M. Braasch

Proc. ION GPS Conf., 1994

66. GPS multipath of air transport airframes

T. Murphy, R. Snow, M. Braasch

Navigation, vol. 34, issue (4), p. 397–406, Winter, 1996–1997

67. Validation of the airframe multipath error allocation for local area differential GPS

J. Booth, et al.

Proc. IAIN/ION Meeting, 2000-06

68. A program for the investigation of airborne multipath

T. Murphy, et al.

Proc. ION Nat. Tech. Meeting, 2004-01

69. Multipath modeling for airborne and ground-based receivers utilizing flight test data

T. Murphy, et al.

Proc. ION Nat. Tech. Meeting, 1997-01

70. Results from the program for the investigation of airborne multipath Er

T. Murphy, et al.

Proc. ION Nat. Tech. Meeting, 2005-01

71. More results from the program for the investigation of airborne multipath error

T. Murphy, et al.

Proc. ION GNSS, Sep. 2005

72. LAAS/GBAS ground reference antenna with enhanced mitigation of ground multipath

A. Lopez

Proc. ION Nat. Tech. Meeting, 2008

73. An advanced multipath model for DGPS reference site analysis

J. Kelly, et al.

Proc. ION Annu. Meeting, 2004

74. LAAS Multipath Limiting Antenna (MLA) performance testing and analysis

D. Bryce

Proc. ION Nat. Tech. Meeting 2005

. Development of local area augmentation system siting criteria

D. Lamb

Proc. ION Annu. Meeting 2001

76. LAAS reference antennas… key siting considerations

A. Lopez

Proc. ION GNSS 2004

77. Performance of a prototype Local Area Augmentation System (LAAS) ground installation

M., Brenner et al.

Proc. ION GPS 2002

78. Signal deformation monitoring scheme implemented in a prototype local area augmentation system ground installation

F., Liu et al.

Proc. ION GNSS 2006

79. Multi-mode receivers for verification of ground and space-based augmentations systems

D., Stratton et al.

Proc. ION GNSS 2004

80. MMR-centric multisensor integration architecture for civil aviation applications

S. Y., Ryan et al.

Proc. ION GNSS 2005

81. Architectures for combined standard positioning system/precise positioning system user equipment

D., Stratton et al.

Proc. ION GNSS 2004

82. FAA

Criteria for approval of Category III weather minima for takeoff, landing and rollout, Advisory Circular AC 120-28D

83. PANS-OPS

vol. II, III, Doc. 8168-OPS/611

84. Joint Airworthiness Requirements All Weather Operations (JAR-AWO)

Subpart 2, Airworthiness certification of aeroplanes for operations with decision heights below 60 m (200 ft) and down to 30 m (100 ft)—Category 2 operations, sec. 236, Excess-deviation alerts

85. Flight Safety Foundation

Pitch oscillations, high descent rate, precede B-727 runway undershoot, Accident Prevention, vol. 58, no. 9, Sep. 2001

## Cited By

Worldwide Vertical Guidance of Aircraft Based on Modernized GPS and New Integrity Augmentations

Proceedings of the IEEE, vol. 96, issues (12), p. 1918–1935, 2008

Evolution of the Global Navigation SatelliteSystem (GNSS)

Proceedings of the IEEE, vol. 96, issues (12), p. 1902–1917, 2008

## Keywords

### IEEE Keywords

No Keywords Available

### INSPEC: Controlled Indexing

air traffic, aircraft landing guidance, avionics, satellite navigation

### More Keywords

No Keywords Available

No Corrections

## Media

No Content Available
This paper appears in:
Proceedings of the IEEE
Issue Date:
December 2008
On page(s):
1936 - 1957
ISBN:
0018-9219
Print ISBN:
N/A
INSPEC Accession Number:
10401574
Digital Object Identifier:
10.1109/JPROC.2008.2006101
Date of Current Version:
16 Jan, 2009