ADS-B and AOP Performance within a Multi ... - Semantic Scholar

5 downloads 0 Views 2MB Size Report
probability of message collisions as traffic converged upon a metering fix. Several aspects of the AOP's performance under analysis are the ability to effectively.
ADS-B and AOP Performance within a Multi-Aircraft Simulation for Distributed Air-Ground Traffic Management Ghyrn Loveness and Richard Barhydt NASA Langley Research Center Airborne Systems Competency Aeronautics Division

Abstract Demand for use of the National Airspace System (NAS) is projected to triple in the next decade while its actual capacity will only increase by 30%. The Distributed Air Ground – Traffic Management concept (DAG-TM) is designed to improve system capacity and increase flexibility and efficiency for NAS users while maintaining or improving Federal Aviation Administration (FAA) mandated safety standards. To accomplish these goals, a collaborative decision making process has been developed that gives decision making authority to the most appropriate user: the flight crew, the air traffic service provider, or the airline dispatch. This process requires the sharing of information about flight intent, traffic, hazardous weather, and airspace restrictions. Automatic Dependence Surveillance Broadcast (ADS-B) is more accurate than radar and allows more data transfer over a larger coverage area than the Traffic Alert and Collision Avoidance System (TCAS). ADS-B is modeled in the DAG-TM concept as the primary data link between NAS users to facilitate their decision making process. Tools have been designed to aid flight crews and controllers using the information that ADS-B provides. In a joint research effort with NASA Ames, NASA Langley’s expertise is focused on flight deck operations. This includes both evaluation of human factors and the development of tools like the Autonomous Operations Planner (AOP) for the DAG-TM concept. A Joint Human-in-the-Loop Simulation involving teams at Langley and Ames, as well as subjects from various airlines and air traffic control facilities, provided extensive data. Two of our research interests focus on the performance of the ADS-B data link and AOP’s ability to assist the flight crew in successfully meeting separation assurance and traffic flow management requirements. Though ADS-B transmissions occur at constant rates, the number of messages encountered by any given receiver is dependent upon the number of aircraft in the vicinity and the range to the targeted aircraft. Previous studies have set broadcast rates to provide adequate information transfer, but the Joint Simulation provided a unique opportunity to model reception probability degradation with increasing traffic levels. We confirmed that as the traffic level increased, the reception probability decreased. Similar confirmation was found for an increasing probability of message collisions as traffic converged upon a metering fix. Several aspects of the AOP’s performance under analysis are the ability to effectively warn flight crews of conflicts and to provide reasonable and efficient solutions. Both human factors and performance metrics will be evaluated for the AOP. Conflict resolution time, seriousness of conflict level, and adherence to time and flight path constraints can help quantify AOP performance. Initial feedback from simulation participants indicates that the AOP exceeded expectations and can be improved for future implementation. Initials results indicate a high degree of precision for subject pilots using the AOP to meet required time of arrival (RTA) restrictions at traffic metering fixes as well as altitude and speed constraints. Causes for outlier cases were analyzed and In-Conformance trends compared between traffic levels do not show significant change, supporting the scalability characteristic of the DAG-TM concept. This interim analysis from the Joint Simulation will provide the basis for conclusions in the final feasibility report for the Enroute Operations Concept Element (CE-5) of the DAG-TM concept. In addition, the analysis is supporting several NASA technical reports and presentations.

Introduction DAG-TM NASA is currently studying ways to improve the National Airspace System to meet current demand and projected increases. The Distributed Air-Ground Traffic Management (DAG-TM) Concept has the goal of significantly increasing NAS capacity while maintaining or improving FAA safety standards. Traditional air traffic management focuses on a ground-based and centralized control system. NASA believes that the DAG-TM concept has the potential to solve capacity problems by shifting to a distributed traffic management system where three major users share information, decision-making, and responsibility [1]. In a centralized system, resources become less available as traffic increases, but in a distributed paradigm, the available resources for solving problems increase in proportion to the demand. In addition, operational efficiency gain under the new concept offers potential users an incentive for implementation. Under DAG-TM, Flight Crews, Air Traffic Service Providers, and Airline Operational Control (Airline Dispatch) will have distributed access to information that allows decision authority to go to the most appropriate user. Pilots operating “autonomous” aircraft under Autonomous Flight Rules (AFR) with the appropriate support tools are free to navigate their own routes subject to maintaining separation from all other aircraft and while maintaining traffic flow management constraints. ATSPs have separation assurance responsibility for managed aircraft flying under current day Instrument Flight Rules (IFR) with an IFR clearance. NASA believes that airspace usage should not be segregated into autonomous and managed regions so that every potential user has access to the NAS. This means that AFR and IFR aircraft will be able to operate in the same airspace without a dependence on equipping all aircraft for autonomous flight. The ATSPs also issue traffic flow management requirements as needed to users of the NAS such as a required time of arrival (RTA) for a metering fix [1]. ADS-B Automatic Dependent Surveillance Broadcast (ADS-B) is a new technology that enables the necessary information to be distributed to DAG-TM participants. It provides surveillance data to each flight crew, ATSP, and AOC about traffic within its coverage, much like radar, but adds new functionality. All aircraft operating in a DAG-TM environment would have to be equipped with ADS-B. The FAA has already been planning implementation of this technology and some carriers like the United Parcel Service have the capability already. Aircraft equipped with ADS-B transmit position, velocity, and intent information to nearby aircraft and ground stations. The ability to send this type of information and over a much larger range than TCAS (~90NM versus 14NM) supports distributed operations such as conflict detection and avoidance as well as traffic flow management. The success of these operations is dependent on ADS-B system performance, and thus feasibility studies of DAG-TM must include a study of ADS-B performance in a realistic operating environment. Joint Experiment: ATOL, ASTOR, and AOP The Air Traffic Operations Simulation (ATOS) hosted by NASA Langley’s Air Traffic Operations Laboratory (ATOL) was created to test critical DAG-TM components to evaluate feasibility. In a recently conducted Joint Experiment between NASA Langley and NASA Ames Research Centers participating in the Advanced Air Transportation Technologies (AATT)

project, the ATOS supported a human-in-the-loop study that looked at many different factors for the CE-5 En-Route Operations Concept Element of DAG-TM. This paper focuses on ADS-B performance as well as the ability of subject pilots to meet traffic flow constraints like RTAs, altitude assignments, and speed assignments at a metering fix. These are only a few of the analyses that will come from the Joint Experiment. The Aircraft Simulation for Traffic Operations Research (ASTOR) was a desktop-based PC simulator that operated with a full physical flight model that performed like a Boeing 757. ADS-B was simulated in the ATOS as the data-link between the ASTOR stations in the ATOL. Flight deck decision support tools were implemented through the Multi-function Control and Display Unit (MCDU) in conjunction with the Flight Management System (FMS). For the Joint Experiment, each ASTOR closely modeled a Boeing 777 glass cockpit. In addition, each station had simulated radio equipment allowing communication between pilots at Langley and controllers at Ames. Researcher stations had the ability to observe any of the subject pilots remotely and took notes on items of interest during the simulation. The Autonomous Operations Planner (AOP) is the decision support tool designed by the team at Langley who has the expertise in flight deck operations. NASA Ames’ expertise and design responsibility is for the controller side of DAG-TM operations. The AOP uses the state and intent information broadcast over ADS-B to detect conflicts and then help the pilot resolve them as well as help to meet traffic flow requirements. Strategic resolutions could upload into the FMS and could then be activated and flown by the autopilot. Tactical resolutions are also displayed when strategic resolutions were unavailable or when the conflict is near term and must be resolved before other constraints were considered. Since separate studies are concurrently being done to look at calculation performance, conflict statistics, and other AOP metrics, we will look at AOP performance in supporting DAG-TM traffic conformance goals.

ADS-B Performance Modeling During the experiment, ADS-B messages were transmitted and received by each ASTOR using a simulated 1090Mhz Mode S transponder. An algorithm developed by Seagull Technology enabled each ASTOR to determine the probability of message reception based on two components: the broadcast range between communicating aircraft and the validity of message reception due to message collisions. Broadcast range was modeled as a function of transmitter and receiver power, antenna gain, and signal loss. The validity of message reception was determined as a function of interference with other messages broadcast over 1090Mhz. The overall message reception probability was calculated from the product of these two components. Range The range profile is a static baseline that defines the message reception probability given that there is no interference by other broadcast messages. Reception probability is close to 100% at close range and is linearly modeled to decrease, first slowly, and then quickly once the maximum range is reached. Rmax is defined as the range where the probability of reception without interference has dropped to 90%. In this particular Joint Experiment, Rmax was 81NM. Equations 1 and 2 below give Rmax as a function of power, antenna gain, cable loss, and the standard free space loss (a). Rmax = a ◊10

Ê LFS ˆ -Á ˜ Ë 20 ¯

(1)

where a = 1.188e-5 mw-NM is the standard free space loss for 1090Mhz and LFS in dB is the maximum allowable signal power transmission loss. LFS is defined in Equation 2 and is a function of minimum triggering received power level (MTL), Mode S transmitter power (PT), transmitter cable loss (LTC), transmitter antenna gain (GT), receiver antenna gain (GR), and receiver cable loss (LRC). Transmitter power is set to 500 watts or 54 dBm and MTL is set to 3.686×10-5 watt or –88.67 dBm, which both conform with recommended high-end system performance [2]. Cable loss (LTC or LRC) is assumed to be 50% or –3dB of original send or receiving power and no antenna gain loss is assumed. LFS = MTL - (PT + LTC + GT + GR + LRC)

(2)

Given Equations 1 and 2, Rmax is set to 81NM and establishes the ADS-B reception probability only due to broadcast range as shown in Figure 1. 100 90 y = -0.1173x + 99.5

Probability (%)

80 70

y = -4x + 4.5

60 50 40 30

y = -5.5556x + 550

20 10

R

0 0

10

20

30

40

50

Range (NM)

max

60

(81, 90%) 70

80

90

100

Figure 1: Probability of Correct Message Reception vs. Range (No-Interference).

Message Reception Validity The validity of message reception is dependent upon any interference from other messages. These include radar interrogation replies and messages over 1090Mhz. A message collision occurs when a receiving ASTOR encounters two or more messages with overlapping broadcast time intervals. These intervals vary for the type of message and type of content a particular message contains. The intervals and types of messages used in the ATOS are: 1) 2)

Mode S short messages [3] (56 bit, 64 μsec), which include ground surveillance, data link, All-Call interrogation, and TCAS, and Mode S long messages (112 bit, 120 μsec), which include ground data link (GDL), ADS-B position message, ADS-B velocity message, ID message, and on-demand squitter (ODS) message.

Short messages only have Mode S address (replies to Mode S radar interrogations) and long messages have state and intent information. An ASTOR’s probability of having correct message reception is determined by Equation 3 and is based on a Poisson probability distribution [4]. pT , Long = p R ◊ p Short ◊ p Long

(3)

where, pT,Long = Probability of correct long message reception between two ADS-B equipped airplanes pR = Probability of correct long message reception based on relative range without message collision (no interference, Figure 1) pShort = Probability of correct long message reception in presence of potential short message collisions pLong = Probability of correct long message reception in presence of potential collisions with other long messages Equations 4 and 5 define the probabilities of reception given the two collision conditions. These are functions of the broadcast interval ( τ) and the frequency ( λ) at which the message type is broadcast. p Short @ e - lShortt Short (4)

p Long @ e

- lLongt Long

(5)

Values of λare the total number of message replies/second for each message type and are defined in Equation 6. λi = Ni × Fij (6) where Ni is the number of aircraft in range emitting a particular type of interference and Fij is the broadcast rate for the corresponding type. The default values that were used in the Joint Experiment are listed in Table 1. Table 1. Message Reply Rates and Intervals Used in Message Collision Probability Parameter

Modeled as:

λShort

Na/c_in_rangeⅹ(FSSR+FAll_Call)+NTCASⅹFTCAS

λLong

Rmax used to determine Na/c_in_range NTCAS is # aircraft within 14 NM TCAS range FSSR = 2 Hz FAll_Call = 1 Hz FTCAS = 1 Hz (FGDL+FPosition+FVelocity +FID+FODS)ⅹNa/c_in_range

λShort

FGDL = 2 Hz FPosition = 2 Hz [2] FVelocity = 2 Hz [2] FID = 0.2 Hz FODS = 0.3 Hz 186 μsec [5]

λLong

240 μsec [5]

Joint Experiment Design The Air Traffic Operations Laboratory at the NASA Langley Research Center hosted 12 subject pilot ASTOR stations, 2 pseudo controllers for “Ghost” and HOWDY arrival sectors, and pseudo pilot stations running TMX to create the necessary traffic levels. The ATOL was connected to the Air Traffic Control research lab at NASA Ames via a high-speed data link, allowing their 5 controllers and 9 subject pilots to integrate with NASA Langley. Both research centers combined to participate in a single traffic research environment. The two most important features of this experiment were the ability to look at Mixed IFR/AFR operations and operations at various traffic levels. These components are central to answering the questions about DAGTM scalability and about the ability to implement DAG-TM in a mixed environment where IFR and AFR aircraft operate in the same airspace. A previous study conducted by Ames determined the traffic levels that would set the experimental conditions. A base traffic level was set (L1) that corresponded to the amount of traffic that current day ATC could handle, but with some degree of difficulty. Managed (all IFR) and Mixed (AFR/IFR) operations were tested at L1 and then more AFR aircraft were added to create the L2 and L3 traffic levels to test scalability. Both the L2 and L3 conditions were set with traffic levels above the threshold determined to be unmanageable with current day ATC capabilities. Aircraft operation type and various traffic levels then established four experimental conditions that the subject pilots and controllers would experience.

Figure 2: Joint Experiment Airspace for BAMBE Arrivals. HOWDY Arrivals from SE.

The Air Traffic Operations Simulation modeled airspace and aircraft around the Dallas Fort-Worth International Airport (DFW). Subject controllers from NASA Ames operated the Ardmore, Amarillo, Wichita Falls High, and Bowie Low airspace sectors (Figure 2). Pseudo Controllers provided hand-off service to the subject controllers for aircraft leaving a “Ghost” sector. No data were recorded in these ghost sectors and aircraft runs were terminated when either the arrival aircraft passed the metering fix (BAMBE or HOWDY) or when overflight aircraft finished crossing their sector. Overflights operated through either the Amarillo or Ardmore sectors and BAMBE arrivals began in these sectors and descended down to the

metering fix. HOWDY arrivals were separate from any Ames subjects and approached the DFW TRACON from the southeast (Figure 2). A NASA Langley controller handled these arrivals. Common to both arrival streams and overflights were scripted conflicts that forced pilots to use the onboard decision support tools to prevent and resolve any conflicts with other aircraft, whether IFR or AFR, where a loss of separation (LOS) was predicted. The pilot’s responsibility at all times was to resolve any conflicts that occurred and not create any new conflicts so that the FAA mandated separations of 5 miles horizontally and 1000 feet vertically were adhered to. The arrival runs were assigned a Required Time of Arrival (RTA) over the metering fix as determined by a ground-based scheduler. For BAMBE arrivals, these RTAs were uploaded via Controller-Pilot Data Link Communications (CPDLC) and were assigned with a significant time delay relative to an aircraft’s projected ETA. The subject pilots then had to use the decision support tools to absorb the RTA error in addition to solving conflicts if there were any. Maneuvering as a result of solving an RTA sometimes prevented scripted conflicts from occurring. HOWDY arrivals received RTAs equivalent to their ETAs and received them via the radio from the HOWDY sector controller. This allowed a unique study that ensured that the scripted conflicts would occur during descent. Not only did arrival aircraft have time constraints for the meter fix, they were also issued altitude and speed restrictions of 11,000 feet and 250 knots. Conformance to these constraints and the performance of ADS-B are two areas of results we will initially focus on. ADS-B Model Performance Results Since the Joint Simulation had sophisticated traffic scenarios, and since ADS-B was implemented as the primary data link between aircraft, valuable information was obtained for how the system performed in varying traffic levels. Difference in ADS-B performance between these various traffic levels was one research interest. For the purposes of the ADS-B study, only traffic sequenced into the TRACON through the BAMBE metering fix was used. Though ADSB transmissions occur at constant rates, the number of messages encountered by any given receiver is dependent upon the number of aircraft in the vicinity and the range to the targeted aircraft. In the Joint Simulation, this data link was modeled so that all transmissions were successful. A filter in the simulated receiver determined the probability that a message would be accepted and made a decision. The product of long and short message collision probabilities, PShort and PLong were compiled and plotted as were the data for overall message reception probability. Source data for the broadcast range reception came from 9 ASTOR stations (6 overflights and 3 arrivals for each run), 12 runs, and a 20Hz sampling rate. Total inputs to the DLS Analysis Parse tool numbered over 7.7 million data points. This tool allowed the data to be sorted into bins corresponding to the range between transmitter and receiver. By comparing this information for each ASTOR during each run we were able to see performance trends over the ADS-B operational range. Further, these trends could be compared between the different traffic levels, allowing us to test the performance of the ADS-B data link for various traffic densities. This analysis separates subject pilots flying overflights from those flying arrivals. In Figure 3, range and probability data were plotted for ASTOR stations flying overflight runs that transitioned them above and through the descending arrival streams. A small decrease in message reception probability as the traffic level increases is apparent and confirms the hypothesis that as more traffic is broadcasting messages within a target’s ADS-B range, the more likely that messages collisions will occur and reception probability decrease.

100

90

90

80

80

70 60 50 40

No Interference

30

L1

20

L2

10

L3

0

% Probability of Reception

% Probability of Reception

100

70 60 50 40

No Interference

30

L1

20

L2

10

L3

0 0

10 20 30 40 50 60 70 80 90 100 110

Range to Target Aircraft (NM)

Figure 3: Overflight Reception Probabilities.

0

10 20 30 40 50 60 70 80 90 100 110

Range to Target Aircraft (NM)

Figure 4: BAMBE Arrival Reception Probabilities.

An important reality check is between the No-Interference curve (Figure 1) with the actual performance considering both aircraft target separation and message interference. As stated previously, parameters relating to radio performance were set to give a maximum range (Rmax) equal to 81NM. All three curves from the actual simulation fall below this curve (included in Figures 3 and 4) due to message collisions. Figure 4 shows the same broadcast range data but for ASTOR arrival flights descending down from en-route operations towards the BAMBE metering fix. Again we have the expected relation to the No-Interference curve and see the same trend of decreased message reception probability with increasing range and traffic level. A second analysis for the performance of ADS-B in a dense traffic environment is to look at converging aircraft streams where the density of traffic does not necessarily remain constant for the entire run. This occurred in the Joint Simulation as traffic was sequenced through a metering fix like BAMBE. Figure 5 shows that as ASTOR arrival traffic approached the fix, the total collision probability (the product of PShort and PLong) increased. Here message collisions are much more apparent as we again see the influence of traffic level on ADS-B performance. L1 has the lowest collision probability curve and L2 and L3 are each higher than the next lower traffic level.

Figure 5: BAMBE Arrival Message Collision Probabilities.

Figure 6: Number of Aircraft within Rmax.

In the Joint Simulation, the Ardmore sector arrivals were initialized closer to the fix than the Amarillo sector arrivals. The traffic conditions at the initialization of each run regardless of route are lower than the rest of the run. The decrease in collision probability in Figure 5 at 150NM reflects the closer start of the Ardmore sector arrivals with the initially lower traffic levels. These lower traffic levels lower the collision probabilities at this point over all the arrival runs. This is echoed in Figure 6 as well, showing the level off in number of aircraft at approximately 150NM. ADS-B is required to perform within a future high traffic level scenario. The LA Basin 2020 scenario from the Technical Link Assessment Team (TLAT) report defines the industry standard [2]. In addition to the number of aircraft within Rmax (81 NM) for Joint Experiment aircraft, Figure 6 shows the number of aircraft that would have been within Rmax (81 NM) for an aircraft approaching the BAMBE meter fix with LA Basin 2020 traffic levels. Though the Joint Experiment is modeling very high traffic levels for airline operations, a significant source of additional ADS-B message interference comes from Mode A/C communications from all the traffic operating at vicinity airports and General Aviation. The modeling of only Mode S communications in the ATOS is more representative of a mature operating environment for ADS-B. However, these additional sources of ADS-B messaging can account for the significant increase in messaging traffic experienced by aircraft approaching the Los Angeles Basin. Traffic Flow Conformance RTA, altitude, and speed restriction conformance can give insight into how well the pilot—with the help of the decision support tools—could fly autonomously while within various traffic densities and while performing en-route and arrival descent operations. These are initial results of a continuing analysis that will broaden into other performance metrics. Only looking at RTA conformance and crossing restrictions, comparisons between traffic level indicate the effects of having to navigate increasing numbers of aircraft while maintaining or resolving constraints in a particular flight plan using the decision support tools like AOP. 12 Departed FMS path for unknown reason. No conflicts.

11

Howdy L1

(1) Tactically resolving several conflicts, no strategic RTA resolutions available. (2) Unknown, was busy with conflicts. (3) AOP possibly ignored RTA during a descent conflict resolution. Contacted ATC.

10 9

Howdy L2 Howdy L3 Howdy 'L4'

Frequency

8 Strange pitch oscillations, RTA required speed up. Pilot went to redline speed intervention. RTA kept getting later.

7 6

Tactically leaves LNAV route and becomes late when returning to route. Contacts ATC.

5 4

Didn't upload entered RTA before hitting execute.

3

AOP failure message. AOP alert failure, strange RTA readout on ND

2 1 0 0-15

15-30

30-45

45-60

60-75

75-90

90-105 105-120 120-135 135-160

Deviation (Seconds) Figure 7: Frequency of HOWDY Arrival RTA Deviations.

More

The two histograms, Figures 7 and 8, give a high level synopsis of the frequency of RTA deviations and their magnitude for HOWDY and BAMBE arrivals. The deviations are the absolute value of the difference between the assigned RTA and the Actual Time of Arrival (ATA). Researchers observing pilot actions during the Joint Experiment have provided notes on possible causes or circumstances to the cases that are out of conformance. To be out of conformance a flight must have greater than 15 seconds in ATA versus RTA deviation, greater than 200 feet in altitude deviation, and/or be greater than 10 knots in speed deviation, all absolute values. 80.4% of HOWDY arrivals were in RTA conformance and 88.2% of BAMBE arrivals were in RTA conformance. Though this difference is still being looked at, one possible explanation is that the scripted conflicts were much more likely to occur for HOWDY arrivals because they weren’t being asked to absorb a time delay. BAMBE arrivals often avoided scripted conflicts by maneuvering to resolve their RTA delay. 12 Bambe L1

11

Bambe L2

10

Bambe L3

9

Waited to solve RTA resulting in a tactical holding pattern at the fix. No ATC contact.

Frequency

8

Tactical maneuvering over the fix. No ATC contact. (17s late).

7 6 5

Entered wrong RTA and never noticed. ATC issued a direct to BAMBE

4

Initially entered wrong RTA, arrived 2 min early with conflicts. No ATC contact.

3 2 1 0 0-15

15-30

30-45

45-60

60-75

75-90

90-105 105-120 120-135 135-160

More

Deviation (Seconds) Figure 8: Frequency of BAMBE Arrival RTA Deviations.

What is more interesting is that, excluding several of the outliers on the basis that AOP had a failure message (2 occurrences) or that the RTA was not uploaded correctly (3 occurrences), the trend between levels is insignificant. This conformance data supports the scalability experimental hypothesis. Figures 9, 10, and 11 show mean and uncertainty for deviations in RTA, altitude, and speed conformance, respectively. Initial results in further variance studies using an ANOVA show—even for the raw data including all of the outliers—no significance in the variance trends. This analysis—currently being undertaken by Booz Allen Hamilton—will continue in depth for the complete feasibility report, but suggest that no significant loss of conformance occurs as traffic level increases. In addition to the statistical significance analysis, simply a practical look at the precision with which subject pilots were able to conform shows support for the experimental hypothesis. RTA deviations are on the order of 4 to 8 seconds, altitude deviations are on the order of 15 to 25 feet, and speed deviations are on the order of 2 to 4 knots.

Meter Fix RTA Conformance (Seconds)

Meter Fix Speed Conformance (Kts)

Meter Fix Altitude Conformance (Feet)

(Only in conformance data points)

12

(Only in conformance data points)

(Only in conformance data points)

45

5 4.5

40

4

8

6

4

Average Deviation (Kts)

35

Average Deviation (ft)

Average Deviation (Seconds)

10

30 25 20 15

3.5 3 2.5 2 1.5

10

1

5

0.5

2

0

Howdy L1

Bambe L1

Howdy L2

Bambe L2

Howdy L3

0

0

Bambe Howdy L3 'L4'

Figure 9: RTA Deviations

Howdy L1

Howdy L2

Bambe L2

Howdy L3

Bambe Howdy L3 'L4'

Figure 10: Altitude Deviations

Meter Fix RTA Conformance (Seconds)

14

160

8

140

7

Average Deviation (Feet)

6 4 2 0

Bambe L1 Managed

Bambe L1 Mixed

Howdy L1 Mixed

Figure 12: RTA Deviations

Average Deviation (Knots)

9

8

120 100 80 60

4 3 2 1

Bambe L1 Mixed

Bambe Howdy L3 'L4'

5

20 Bambe L1 Managed

Howdy L3

6

40

0

Howdy Bambe L2 L2

(Only in conformance data points)

180

10

Bambe L1

Meter Fix Speed Conformance (Knots)

(Only in conformance data points)

16

12

Howdy L1

Figure 11: Speed Deviations

Meter Fix Altitude Conformance (Feet)

(Only in conformance data points)

Average Deviation (Seconds)

Bambe L1

Howdy L1 Mixed

Figure 13: Altitude Deviations

0

Bambe L1 Managed

Bambe L1 Mixed

Howdy L1 Mixed

Figure 14: Speed Deviations

Another initial result is the comparison of all “Managed” runs with “Mixed” operations at the same traffic level. Figures 12, 13, and 14 show the deviations from conformance for RTAs, altitude restrictions, and speed restrictions for their respective meter fixes. Here we see a very clear decrease in conformance deviation between “Managed” (All IFR) and “Mixed” (IFR/AFR) operations that supports our second experimental hypothesis: “Mixed Operations in very high traffic density sectors ... do not degrade traffic throughput and efficiency compared to operations with all managed aircraft.” We also see the effect of scripted conflicts (apparent in all 8 figures) in the HOWDY runs as compared to BAMBE runs because, as mentioned earlier, BAMBE runs often prevented their scripted conflicts by maneuvering to absorb an RTA delay. More scripted conflicts during descent for the HOWDY runs appeared to increase conformance deviations. Based on an initial ANOVA, the correlation between varied traffic levels appears to be insignificant and we can say with some degree of certainty that the difference between the HOWDY L1 Mixed and BAMBE L1 Mixed runs were due to factors other than traffic level.

Preliminary Conclusions Initial traffic flow management data such as meter fix restrictions appear to support both of the Joint Experiment’s hypotheses about DAG-TM operations. The Scalability hypothesis is supported by small and precise deviations from assigned RTAs, meter fix crossing altitudes, and crossing speeds, while showing no statistical significance for trends with increasing traffic levels. The “Mixed” operations hypothesis is supported by initial data showing significantly lower

conformance deviations for “Mixed” operations at L1 when compared with results of “Managed” aircraft runs, operating with normal day procedures at the same traffic level. ADS-B performance trends were confirmed. Areas of increasing traffic density such as aircraft converging on a meter fix, showed higher probabilities for message collisions. With increasing traffic levels, overall reception probability decreased. When comparing to the industry standard LA Basin 2020 scenario for required traffic density to test ADS-B performance, the Joint Experiment had much fewer targets broadcasting ADS-B, especially nearer to LAX. Joint Experiment traffic levels do not follow the rapidly increasing trend close in to the airport because the significant impact of Mode A/C messaging over 1090Mhz from other vicinity airports and General Aviation traffic was not modeled. The goal of the ATOS is to create a test environment as close to current day and future operating conditions as possible. To continue the in depth feasibility study for key elements of DAG-TM, ATOS will have to include such messaging sources. In the future, additional background traffic will be modeled and all potential ADS-B messaging sources accounted for in order to enhance our analysis of this new technology’s performance.

References [1] Ballin M., D. Wing, M. Hughes, and S. Conway, 1999, Airborne Separation Assurance and Traffic Management: Research of Concepts and Technology, AIAA-99-3989, AIAA. [2] RTCA and Eurocontrol, 2001, ADS-B Technical Link Assessment Team Technical Link Assessment Report, RTCA and Eurocontrol, http://www.eurocontrol.int/ads/ADS_Programme_content.htm [3] Harman, W.H and M.J. Brennan, 1997, Beacon Radar and TCAS Reply Rates: Airborne Measurements in the 1090 MHz Band, Project Report ATC-256, Lincoln Laboratory, MIT. [4] Orlando, V.A. and W.H. Harman, GPS-Squitter Capacity Analysis, DOT/FAA/RD-94/8, Lincoln Laboratory, MIT. [5] Orlando, V.A. and G.H. Knittel, 1993, GPS-Squitter: System Concept and Performance, ATC Quarterly, Vol. 1, No. 4, Air Traffic Control Association, pp. 303-326.

Suggest Documents