Host Computer System (HCS)

The FAA Host Computer System (HCS) processes the flight information and radar data for both en- route and terminal area air traffic control in the US (i.e. the Air Route Traffic Control Centers (ARTCCs) and Terminal Radar Approach Controls (TRACONs) respectively) as illustrated in Figure 2.3. Note that some additional processing is undertaken in the TRACON environment by the Automated Radar Terminal System (ARTS), some of which falls into the category of conformance monitoring, for example determining when to trigger the Minimum Safe Altitude Warning (MSAW) if an aircraft descends too fast near the ground. However, details of the ARTS processing was not readily available and so will not be considered further here.

Flight Plan Data Radar Data Flight Data Processor (FDP) Radar Data Processor (RDP) HOST COMPUTER SYSTEM (HCS) Controller Display ARTS Processing Controller Display Display Processing

EN-ROUTE CENTER (ARTCC)

TERMINAL AREA (TRACON) Flight Plan Data Radar Data Flight Data Processor (FDP) Radar Data Processor (RDP) HOST COMPUTER SYSTEM (HCS) Controller Display ARTS Processing Controller Display Display Processing

EN-ROUTE CENTER (ARTCC)

TERMINAL AREA (TRACON)

Figure 2.3: HCS Processing Elements [adapted from NRC(1998)]

The Flight Data Processor (FDP) component in the HCS takes flight plan inputs prior to departure and amendments from the controllers while the aircraft are en-route. From this it calculates a 4D route and time of flight for the flight plan identifying fixes along the route and estimated arrival times at those fixes. This information is transmitted at the appropriate time to the controlling en-route sectors and via the Automated Radar Terminal System (ARTS) to TRACONs and towers to be printed out on Flight Progress Strips (FPSs) for use by the controllers in these facilities. A sample en-route flight progress strip showing the FDP-derived route of flight and estimated time over a given fix on the route is presented in Figure 2.4.

Fixes on active Flight Plan route Estimated time over given fix

Figure 2.4: Sample En-Route Center Flight Progress Strip

The Radar Data Processor (RDP) element of the HCS receives data from the radars within the system, transforms it into a format suitable for display and uses filtering techniques to infer the heading and speed of the aircraft being tracked. In order to improve the accuracy of these filtering processes, the RDP attempts to “associate” a given radar return with previous returns and flight plan routes output from the FDP [Lincoln (1998)]. This is achieved through “association checking”algorithms in the lateral, longitudinal and vertical domains relative to the 4D prediction of the aircraft locations from the FDP

[FAA (1992), FAA (1993)]. Thus, association checking can be considered a simple form of conformance monitoring. The processes employed in these HCS algorithms are described below.

A rectangular “association area” is defined for lateral and longitudinal checking that is centered on the extrapolated flight plan position and aligned with the flight plan velocity vector as shown in Figure 2.5. The dimensions of the association area are defined as the thresholds on allowable lateral and longitudinal deviations from the expected position. The threshold values are functions of the assigned altitude and whether the current position is in a flight plan leg region or a transition region (defined as being within 15 nm of a lateral transition point greater than 30°). The pre-determined threshold values used by the HCS for lateral association checking are presented in Table 2.1. Note that there is no longitudinal adherence checking in the turn region.

>30° 2DO 2DL Expected position Observed position 15 n m Flight plan transition region Flight plan leg region Association area HCS Flight Plan route

Figure 2.5: HCS Lateral Association Checking [adapted from FAA (1992)]

Table 2.1: HCS Lateral Association Checking Thresholds [adapted from FAA (1995)]

Flight Plan Leg Region Flight Plan Transition Region Assigned Altitude Lateral

Threshold (DL) Longitudinal Threshold (DO) Lateral Threshold (DL) Longitudinal Threshold (DO) ≤ 10,000 ft 4 nm 4 nm 8 nm N/A 10,001 – 18,000 ft 6 nm 6 nm 10 nm N/A 18,001 – 33,000 ft 8 nm 8 nm 12 nm N/A > 33,000 ft 10 nm 10 nm 14 nm N/A

The observed positions of the aircraft from the radar input are separated into lateral and longitudinal deviations and compared to the appropriate threshold value. When the observed deviations are within the prescribed lateral and longitudinal thresholds, the aircraft are considered to be “flat-tracked” and the HCS utilizes the Flight Plan to improve its estimates of heading and future position. These estimates are then used to update the predicted flight progress for use in future longitudinal conformance monitoring algorithms. However, if the observed lateral or longitudinal deviations exceed the threshold, then the aircraft is flagged as being out of adherence in that dimension, the aircraft switches to a “free-tracked” status and the graphical indication for that flight on the radar screen changes from a diamond ( ) in “flat- tracked” mode to a triangle (U) in “free-tracked” mode. This subtle change in the flight indicator on the radar screen is the primary alert to the controller of a non-conforming aircraft in the lateral or longitudinal domain. The controller must then manually enter flight progress information for “free-tracked” aircraft or else the time dimension of the 4D predicted flight plan diverges from the actual flight behavior. If the longitudinal deviation becomes excessive, either the flight plan will “time-out” requiring the entire flight plan to be re-entered manually, or the flight could arrive at a facility boundary before its flight plan has been transmitted to the relevant sector by the FDP [Lincoln (1998)].

A similar approach is taken to vertical association checking when the aircraft is at a level altitude: the reported Mode C transponder altitude is compared to the assigned altitude, and non-adherence is declared when the deviation exceeds the appropriate value shown in Table 2.2. Vertical deviations that exceed the prescribed limits are alerted by flashing the altitude element of the datablock associated with the flight on the controller’s display. Vertical transitions, however, are handled differently than in the lateral domain. No association checking is attempted during a vertical maneuver, presumably due to the larger uncertainty in the vertical trajectory to be flown. All aircraft are assumed to be in adherence during a vertical transition [Paglione et al. (2000)], although it is unclear at what point the switch back to level altitude association checking is resumed after a level-off should have occurred.

Table 2.2: HCS Vertical Association Checking Thresholds [Paglione et al. (2000)]

Assigned Altitude Vertical Non-Transition (Level Altitude)

< FL290 ± 200 ft

This description of the conformance monitoring processes in the HCS highlights several important issues. Firstly, the conformance monitoring in the HCS can be classified as a signal-based fault detection approach where the observed positional deviation from the expected position on the flight plan route is the signal that is compared to a pre-determined threshold appropriate for the axis and flight regime being considered. These thresholds are generally large because of the need for simplicity across many ATC environments, the limited surveillance capability in the en-route domain and the wide range of tracking capabilities of aircraft with different navigational equipages using the airspace. This is illustrated by the fact that at typical cruise altitudes of commercial aircraft, lateral deviations of up to 8 nm either side of the expected position in the lateral and longitudinal dimensions are considered conforming by the HCS algorithms. This is twice the 4 nm width of a typical airway.

Secondly, the notion of requiring wider conformance bounds at flight plan transition points is important and one that is seen carried through many of the other systems to be discussed.

Thirdly, the different threshold values used in the horizontal and vertical axes demonstrates the fact that tracking performance and expectation of the trajectory to be flown is different in the two domains. For example, deviations in excess of 500 ft during level flight are sufficient to trigger a non-conformance alert, as compared to 8 nm (i.e. over 48,000 ft) in the horizontal domain. This is due to the fact that commercial aircraft have much better altitude tracking performance to a specified barometric altitude target than they do to a lateral path target and can thus be held to a much stricter standard. In addition, the altitude is much more observable than the lateral position state. This has the associated impact on operations that lower vertical separation minima of 1000 or 2000 ft are allowable in the vertical axis compared to 3 or 5 nm in the horizontal. Vertical transitions are a very different story, however. Here, the vertical trajectory to be flown is so uncertain that no conformance monitoring is even attempted in the HCS under these conditions.