An aircraft collision avoidance algorithm

An Aircraft Collision Avoidance Algorithm

1090 MHz 56 bit single packet mode S random squitter

single downlink format packet containing position and velocity

Loran-C sensor for best accuracy

B. Keith Peshak

http://www.gtwn.net/~keith.peshak/taillight.htm

The United States Federal Aviation Administration (FAA) has always admitted that it is unable to provide collision avoidance between all equipped aircraft through the Air Traffic Control (ATC) system of Secondary Surveillance Radar (SSR) operating on 1030/1090 MHz utilizing Air Traffic Control Remote Beacon System (ATCRBS) {pronounced "at-crabs"} transponders.The FAA operators of the ATC system have long confirmed problems ("Real Targets-Unreal Displays - The Inadvertent Suppression Of Critical Radar Data" by Thomas G. Lusch, FAA Air Traffic Control Specialist, Journal of ATC, January-March 1992, pp29-33; Proceedings of the Sixth International Symposium on Aviation Psychology, Ohio State University, pages 460-465), accompanying as the first companion paper.Also the FAA technical center has confirmed technical problems (http://www.gtwn.net/~keith.peshak/tn97_7.pdf).Additional more recent problems have led to the FAA endorsement and support of a radical new "replacement" system from II Morrow {pronounced "tomorrow"} named Automatic Dependent Surveillance-Broadcast (ADS-B) (http://www.ads-b.com/Content/index.htm).However, the FAA seems to have admitted serious deficiencies in the GPS navigation system upon which ADS-B is dependent ( http://www.aerotraining.com/html_gif/gpsdead.htm).

The problems with ATCRBS, the only existing aviation collision avoidance system, are composed of two factors.1).Equipment design that was never intended for the aircraft density (responder) and radar density (interrogator) of today's aviation environment.2).Procedures that only guarantee separation of IFR from IFR aircraft (participating positive control).The three other collision permutations involving VFR aircraft (IFR from VFR, VFR from IFR, VFR from VFR) are not provided, except on an "as time permits" basis (and time never permits sufficiently), and then only to a subset of aircraft, and then further only when all are flying above the radar floor.

The old FAA "fix" was to insist that many aircraft carry ATCRBS transponder interrogators.This Traffic Collision Avoidance System (TCAS) is advertised as equipment that will allow the pilot of the aircraft to see and avoid all other aircraft.That was an augmentation system, utilizing the ATCRBS transponder.Unfortunately, there were problems.

The new FAA "fix" is a "replacement" system - requiring all new equipment in each aircraft.ADS-B is presently about $300,000 per aircraft, although the FAA and II Morrow have stated that they intend to attempt to reduce that to $12,000 per aircraft.Unfortunately, the system requires a large installation of ground equipment to be backwards compatible to see ATCRBS only equipped aircraft.There is no discussion of the cost of the Traffic Information System (TIS) ground equipment, which may require an additional investment of billions of dollars.Yet, and this is perhaps the most serious shortcoming of ADS-B, there is a hard limitation on capacity.ADS-B can handle only about 300 aircraft within hearing range, at that level causing a catastrophic failure of the ATCRBS data link.To solve other ADS-B design deficiencies, it is proposed to increase ADS-B transmission power levels, which would further decrease the capacity.A catastrophic failure of the ATCRBS data link would mean that the entire present and future collision avoidance systems (ATCRBS, TCAS, ADS-B) would fail, and that number is at about 0.015 aircraft per square nautical mile for the equipment shown at AirVenture99 (http://www.airventure.org/), less with the proposed fix.

Fortunately, there is an alternative proposal, which will support an "average density" of traffic at about 1.5 aircraft per square nautical mile, for a total cost per aircraft of approximately $500 (keith.peshak@gtwn.net).Automatic Independent Surveillance Positioning (AIS-P) is an "augmentation" system - present equipment in the aircraft is upgraded, instead of being replaced, no aircraft modifications.At that price, there is no need for an additional TIS ground system.And, while we were at it, this augmentation to the ATCRBS transponder also fixes the ATCRBS P4 problem (http://www.gtwn.net/~keith.peshak/P4Problem.htm), created by the FAA, in order to allow TCAS technology.

The AIS-P system can utilize many types of external positioning sensor apparatus.Though a custom integrated circuit was built for GPS and altimeter interface; this paper also explains a Loran-C interface, for greater positional accuracy, utilizing a NMEA data stream converter.GPS requires dynamic real time correction (dGPS, LAAS, WAAS), Loran-C requires semi-static occasional correction (ASF), to yield satisfactory absolute positional accuracy.For collision avoidance purposes, these errors are common mode, and are unimportant.

The replacement ADS-B system additionally poses significant risks to aviation security (http://www.gtwn.net/~keith.peshak/ADS-Bterror.htm).The AIS-P augmentation to the present ATCRBS system does not have that significant design deficiency.

Classically, in the Los Angeles basin, the transponder density and interrogator installation density have each always been too great for the ATCRBS system design to have ever operated correctly, since about the 1960s.TCAS airborne interrogator equipment now also extends this interrogator density problem to the northeast triangle (Boston - New York - D.C.), and all major airports during the busiest times of the day.Of necessity, the "radar" display was further augmented with the post radar computer algorithm named "coast mode", which assumes zero acceleration target motion.The targets coast along, when radar discontinues to see them, even when the controllers order a change to the flight path.The interrogator density problem is that there is not enough link capacity to get all information through.The only solution is to limit the need for interrogators in system operation.The FAA elected to increase the number of interrogators as the solution to that problem.

When mode S (mode Selective) transponders were added as an augmentation to this system design (necessary to allow TCAS technology to work at all), some ATCRBS radars were replaced by mode S radars.These (TCAS and mode S radar) interrogators issue a different interrogation, with an added P4 pulse.Changing the interrogation without changing the ATCRBS transponder detection digital logic exacerbated the problem of aircraft not announcing their appearance.This problem was created by unintentionally causing ATCRBS transponders to suppress reply to thought still valid SSR interrogations (which contained the new P4 pulse two microseconds aft of the P3 pulse in the main beam of the radar).When airborne TCAS, and the new mode S radar, interrogators were added as an augmentation to fix the interrogation density problem, "not appearing on the radar display" got worse.The interrogations from TCAS and mode S interrogators unintentionally contained this extra side lobe suppression instruction to many ATCRBS transponders (depending on logic design), and those did that unintended command.

The December 1976 manual for the Narco AT-150 transponder, the design certified by the FAA, does not contain transistor Q415, nor its driving resistor R508.This circuitry, that had to be added to the transponders (as they came in for service problems or alignment), is employed so that the U403 output can short out Q413 (the presence of the P4 pulse from the receiver is shorted out before it can set the side lobe suppression latch, in order to prevent the received P4 pulse causing unintended suppression of the reply).The AT-150 of today must have these components because ATCRBS transponders that do not contain these components will behave in an unintended fashion when there is a P4 pulse present in the interrogation - they do not reply to the interrogation for TCAS and mode S radar.How many, of the total manufactured, have been modified?And, also, it was discovered, the multivibrator tuning is critical.How many are properly tuned?How many have been over two years since last they were checked (VFR aircraft owners generally believe that a transponder biennial is not required by the FAA)?Why is it necessary to have "biennial" to "tune" "digital" circuitry?Why does not the FAA require avionics repair shops to have transponder test equipment which includes the P4 pulse in the test interrogation?These are the P4 problem.

Additionally, every transponder has another circuit that says "if interrogated and reply" or "if interrogated and suppress", do not respond to further interrogations for a while.It is easy to visualize how the additional TCAS interrogators will cause a diminution of responses to interrogations, which would, then, include those from the ground.Diminution of responses to interrogations that happen not only at the same time, but that also happen within this "out of service" "window".This is the "5 o'clock slam dunk" problem, brought by airborne TCAS interrogators to every busy hub airport.

Unfortunately, these problems are not linearly additive, but seem to be multiplicatively related.The cumulative problem magnitude can be characterized as somewhere between N**2 and N**3, where N is the number of interrogators.

Fortunately, few aircraft have these TCAS systems, because of their extreme cost.That "low population" will increase with the advent of the B. F. Goodrich "SkyWatch" TCAS, which reduces the cost from the original ~ $225,000 to ~$25,000 for general aviation (http://www.bfgavionics.com/docs/skywatch.html).Even the Ryan TCAD, classically a non-interrogating collision avoidance device, now offers active interrogation (http://www.ryan-tcad.com/) to increase the problem.Also the FAA requirement for TCAS has changed from "big airliners" to "all airliners", and FAA is proposing "all freight carriers" to be added to that mix.Fortunately, few ATCRBS radars have been replaced with the new mode S radars, because of their extreme cost, but that number is increasing.As these three populations of interrogators grow, the precarious SSR based collision avoidance system we have now will fail worse than it fails now.Regardless, at the present time, "working" is defined, by number of passengers, 90% of the "populated" aircraft, cannot know about, by number of aircraft, 90% of the aluminum in the sky.This problem is further enhanced because of an FAA equipment design requirement that "VFR" traffic which is "non-participatory" shall, when seen by the radar, be culled by the "1200 filter" before display.Most of the aircraft seen by ground radar are removed from the ground radar display used by the controllers.ATC does, additionally to the above mentioned problems, not even try to display most of the traffic, in which it has "no interest", even though that traffic is seen by the radar, and can collide with and bring down the aircraft "of interest".This is the "not want to look" problem.

The ATCRBS system is the only aircraft collision avoidance system we have.It is so poor in quality and accuracy of operation, that the FAA is seeking a total replacement of all ground and airborne equipment.The ADS-B initiative is the surviving contender favored by the FAA for this entirely new and radical departure from the present air traffic control system.

There are several significant problems with this new and radical system design, which will render it ineffective to the standards of even the present ATC system design, which is itself insufficient.

One issue is the extremely high cost, due to the extreme complexity, of the ADS-B equipment; which is necessary for every aircraft to possess in order to be seen by any other.Therefore, limited employment amongst only a small minority of aircraft is the likely result - a subset system covering exposure to only those that have the equipment (which is what TCAS is now, except that TCAS in the aircraft can also occasionally see some of the other ATCRBS transponders).From the perspective of the system users, different technology, more insufficient result.

Another issue is that, since only partial information is provided in each mode S "packet" transmission by the ADS-B transmitter, multiple "packet" transmissions are necessary to convey the "message" (minimum required information of a single report).This results in far diminished system capacity - the other packet "times" could have been used, otherwise, by more aircraft, if full information were transmitted in one packet.From the perspective of the system users, ADS-B is different technology, with even less total system capacity.Mitre estimates that ~300 aircraft within an 80 NMi radius will collapse the ATCRBS data link with the mode S standard compliance.With that collapse would be the loss of all ADS-B, TCAS, TCAD, and ATCRBS "radar" information, resulting in all "ground radar" targets in coast mode, and airplanes blind to traffic.If the transmitter power output is increased, as is proposed by the ADS-B engineers, any engineer can calculate (for the same 1/(R*R)) how this range radius will increase (allowable aircraft density will decrease).

With 300 aircraft allowable within a "hearing range" of 80 nautical miles, a first hand approximation would be that 300 aircraft are allowable within 20,106.2 square nautical miles.Dividing by that 300 yields an "average density" of traffic of one aircraft per every 67 square nautical miles.Taking the reciprocal, 0.015 aircraft per square nautical mile.If power were then increased to cover the desired design goal of a 250 nautical mile effective range, that same 300 aircraft limit would be allowable in, now, 196,349.5 square nautical miles.Dividing by that 300 yields an "average density" of traffic of one aircraft per every 655 square nautical miles.Taking the reciprocal, 0.0015 aircraft per square nautical mile.AIS-P will allow increase of that capacity to 1.5 aircraft per square nautical mile.

Yet another issue is diminished security with ADS-B.To connect these different packets which form a single message containing all necessary information from one aircraft, a means is necessary to tie the different packets together to form the complete message.The mechanism chosen is to add a number field, which further diminishes, therefore, the amount of message data capacity for the packet.In this field is a number algorithmically derived from the tail number of the aircraft.The algorithm is published in the international mode S standard.This is a common problem shared with the mode S transponder equipment.However, mode S equipment must be interrogated, and tells only who he is.ADS-B equipment need not be interrogated, and tells who and where and which way and how fast.

The problem of "I am Airforce One, and I am at latitude longitude altitude speed direction (the missile should come there to meet me, missile need not carry big battery for interrogator transmitter, directional antenna, or the transmitter)" is addressed in the second companion paper (http://www.gtwn.net/~keith.peshak/ADS-Bterror.htm).The reader is encouraged to contact Litton Guidance and Control Systems of Northridge, CA (818-678-7666) and ask for the color sales brochure of the AN/PPX-3B and TPI-10 interrogator sets, inquire as to the intended purpose, and query the missiles now proven to work with this system.

There is a way to save the present ATCRBS transponder system, currently in place, with a simple technical upgrade.This solves the "P4 in the new interrogation" problem, but also provides the ADS-B collision avoidance goals, while all objections to the ADS-B system design are removed, these together solving the entire above listed set of problems.This is available as a modification to the common ATCRBS transponder (by disabling or removing many old integrated circuits, eliminating potentiometer adjustments, adding a small 49x53 millimeter circuit board containing a clock IC, an Actel 42MX09 FPGA IC (http://www.actel.com/products/antifuse/) also (http://www.actel.com/products/antifuse/mx.html), a hysteresis IC (74LC14), one diode, one resistor, and six capacitors).The interface to the old transponder is an eight wire cable.This is a field upgrade, inside the old transponder, which can be accomplished by any avionics shop, utilizing the Monarch-Air modification kit.Another version of this same design is provided for new transponder designs, available to the transponder manufacturer, from keith.peshak@gtwn.net.This contains all of the transponder digital circuitry, and eliminates all need for any adjustments.A transponder that requires no biennial, because there is nothing to adjust.Both solve the P4 problem and both provide the AIS-P augmentation described.

The new circuitry knows to ignore the P4 pulse, just as if the FAA had not required its addition to the transmission by TCAS and Mode S radar interrogators.This option is sufficiently economical, so that all presently equipped aircraft can continue to participate in the ATC system with their installed equipments, and also provide AIS-P compliance.New transponders, become much less expensive to manufacture, because this Actel IC replaces many components.No alterations are necessary to any aircraft to allow it to be seen by software modified ADS-B/TCAS/TCAD equipments or by the new AIS-P proximity warning equipments.

There is additional regulatory agency effort required to allow fielding of the AIS-P capability.The packet utilized is a mode S downlink data format, but not one contained in the present international mode S standard.In that standard is a five bit field in the mode S downlink data packet format which specifies a binary data format number (DF# or DFN).This is the identification of one of thirty-two possible different downlink message allowable format types.There are several downlink format numbers not being used.Adding the AIS-P data message format to the mode S specification is as simple as requesting John Mark Loscos at ICAO (514-954-6713) to assign to the AIS-P specification (http://www.gtwn.net/~keith.peshak/peshak22.htm) one of the unused and available downlink format numbers.This will be used for a mode S squitter packet.This would create and add to the mode S system one available packet type which would contain all of position and velocity data, and would not contain ID (tail number) data.We wish for one of the undesignated downlink format numbers to be assigned to the format specification to allow system compatibility with all existing mode S and TCAS equipments.

There would be no need to change or update those equipments.ATCRBS equipments would already just ignore this "noise" {else TCAS and mode S transponders would have already failed within the ATCRBS system}.There would be no need to change any transponder or ground "radar" equipments.Those ADS-B/TCAS/TCAD equipments that would desire to "see" the AIS-P equipped aircraft would require only simple software update.Those aircraft that would desire to "hear" the AIS-P equipped aircraft ("Warning, 9 o'clock, four miles, three hundred, high, thirteen seconds (to impact)") could purchase the $2500 panel mount "transponder sized" Proximity Detector box shown at AirVenture99 (contact narco@netreach.net to request).Those aircraft that would desire to "see" the AIS-P equipped aircraft could purchase the EFIS display from Sierra Flight Systems shown at AirVenture99 (contact ncalvin@sierraflightsystems.com to request).An audio tape is available from EAA (http://www.eaa.org/) for the "TCAS almost free" seminar in Sporty's Pavilion on Saturday night of AirVenture99 - a technical presentation explaining the details and discussing the working equipments.The FAA has been "highly resistive" to the AIS-P concept, as competition to ADS-B, and also highly resistive to the concept of a repair for the P4 problem.This is quizzical, in view of the drastic airline flight delays, occurring as a result of necessary extreme spacing, implemented as a countermeasure to accommodate the failure of transponders.Contact jane.garvey@faa.govto request an explanation, and come to AirVenture00 at Oshkosh, the "Meet The Boss" seminar in the FAA building on Sunday morning, to ask her for an extemporaneous explanation.

We now address the optimal position and velocity sensor to employ for any collision avoidance technique.Loran-C is more repeatably accurate for positioning (http://www.gtwn.net/~keith.peshak/loranpos.gif) than is GPS (http://www.gtwn.net/~keith.peshak/gpspos.gif).The offset from true position (the center) is corrected by the input of extremely recent (last second) dGPS information for that exact location, or the last ATIS broadcast of Additional Secondary Factor (ASF) observation information for that region of the country.Glonass would be a second choice for most repeatably accurate (http://www.gtwn.net/~keith.peshak/glonasspos.gif), if more satellites were to be put back in orbit to replenish the constellation (http://www.gtwn.net/~keith.peshak/visibility.gif).That would remove the red, and congeal the blue and green.GPS is, clearly, the poorer third choice in absolute accuracy!

Satellite position accuracy can be increased by utilizing all satellite navigation constellations, with such as the Ashtech GG-24 receiver (http://www.gtwn.net/~keith.peshak/bothpos.gif), to help solve the problem of the low number of Glonass satellites available.But the ideal solution would favor Glonass only, once that satellite constellation is again filled, and it will, then, be on a par with Loran-C with ASF for the last weather observation, for the area, entered into the sensor.The ASF can be obtained from an ATIS broadcast over the NavCom, then dialed into the Loran-C sensor by the pilot (no data link needed and no equipment alteration required of the aircraft), or that process could be automated if a data link were put in place and an equipment added to the ground and aircraft.

Loran-C, certainly, could be the most accurate instrument landing and navigation system available, both in repeatability, and absolute accuracy.If no ASF correction, then there will be a relatively constant offset error to Loran-C absolute position, but that error will be common mode to all aircraft (important for instrument landing, unimportant for collision avoidance).Even without ASF correction, Loran-C is a more accurate sensor for the application of collision avoidance.

There are other deficiencies, and solutions, to reliable use Loran-C technology in aviation.

The first issue is that precipitation static renders the technology potentially impotent.This only happens in rain or snow, more the latter than the former.Use of airframe grounding techniques and static wicks reduces the problem, and is a well known solution for the same problem observed by the Airborne Direction Finder (ADF) equipment operating on the 200-400 KHz band.Unfortunately, this "static" equipment is poorly maintained on the average aircraft, rendering it a poor solution.Use of an H field antenna (coil) to replace the E field antenna (long wire) eliminates that deficiency.Static electricity discharges are typically extremely high voltage, but extremely low current.

The second issue is that the sudden loss of one of the three Loran chain stations being used to determine a position causes loss of position.This only happens rarely, and the new updated Loran-C technology of "automated blink" allows the receiver to be made immediately aware of the loss of a station.Like the little red "NAV" flag on the VOR or localizer or glideslope, the pilot can be made aware of the fault, and can change the selected chain or chain stations.The use of a masterless algorithm, like that pioneered on the Ray Jefferson PL-99 handheld Loran-C, of the late 1980s, now used in many experimental class and part 103 class aircraft, minimizes that deficiency by utilizing all stations in the selected chain for position determination.An "all in this chain" position solution convergence algorithm, sometimes called by the name "masterless navigation".

The third issue is multi-chain reliability.GPS requires at least four intersecting "range" spheres to resolve a position.In the early years, GPS used a three channel receiver, opting for a fourth pseudorange from earth centric, by employment of a blind atmospheric pressure altitude sensor.An economic design trade-off.We have since learned the value of the twelve channel "all in the sky" solution convergence algorithm, enabled by 12 channel hardware correlator receivers.Use all chains, each with masterless navigation, and use of geometry weighted position contribution to the final solution, presents an elimination of the minimized deficiency.Locus Incorporated, of Madison Wisconsin, is one company that has introduced an all-chain Loran-C sensor (http://www.locusinc.com/loran-news.htm).

We now detail how the goal of the appropriate AIS-P sentence, containing the positional and motion data gathered from the Loran-C sensor, can be generated for the Actel ATCRBS transponder chip AIS-P input.We here consider the common Ray Jefferson PL-99 Loran-C sensor.

A low cost microcomputer, based on such as the PIC chip, (http://www.radioshack.com/sw/swb/projects/bstampidx.htm) can be placed between the transponder chip and the Loran-C sensor.We must interface to the Actel transponder chip, which was designed to require a specific GPS sentence, from the loran-C position sensor, which speaks a different series of sentences, the combined having the necessary information.The PL-99 does not produce a single output sentence with all of the necessary data, but does provide the necessary information.The PL-99 is not FAA certified equipment, so the interface containing the PIC chip can also remain outside the purview of FAA certification (for experimental class and part 103 class aircraft).This saves monumental DO-178 certification costs (http://www.rtca.org/).

The AIS-P specified packet data input format used by the Actel chip is the standard NMEA-0183 $GPRMC GPS sensor sentence.This micro will produce that, assembled from fragments of the Loran-C NMEA sentences produced by the PL-99 sensor:

$LCGLL,3032.61,N,09754.45,W

Latitude and Longitude we will use

deg-min.xx, N/S, deg-min.xx, E/W

$LCGTD,16339.3,31349.9,41960.2,54801.3,.

Loran time differences

LOP Td for each slave, 5th missing in this example

$LCSTD,2,2,2,2,2,

Td status (A first quality indicator)

0=good

1=low SNR

2=cycle error

4=blink (station not usable)

8=searching

$LCSIU,0,,,,4,5

Stations in use

0=master

4=Y

5=Z

$LCSGR,9610

GRI of chain in use

10s of microseconds

$LCSNT,0,A,9610,V,0,,,3,4,

Status of fix (a quality indicator, rec GRI, rec stations)

position fix quality

0=no fix

1-9 increasing quality (second quality indicator)