6894 workshop in software design lecture 2 september 14, 1998 CTAS

1
6894 workshop in software designlecture 2
september 14, 1998 CTAS
2
the arrival scheduling problem

• a resource allocation problem
• aircraft arriving at an airport need to be
  sequenced for landing
• cant let them land in same order as they arrive
  in terminal area
• runway may not be available when they get there
• faster aircraft would overtake slower ones
• current manual solution
• aircraft pass through meter gates on border of
  terminal area
• arrivals at meter gate are delayed, and aircraft
  may be reordered
• currently, controllers use judgment experience
  to make these decisions
• an opportunity for automation
• better scheduling can increase airport throughput
  and minimize delays
• (but cannot be optimal because of unpredictable
  weather, etc)
• reduce controller workload, and thus improve
  safety

3
what is CTAS?

• basics
• Center/TRACON Automation System
• set of tools to help controllers manage air
  traffic flow at large airports
• developed at NASA by Erzberger and colleagues
• chosen in 1991 by FAA as future automation system
  for terminal area
• technical innovations
• algorithms for trajectory synthesis and runway
  allocation
• user interface human-centered automation
• a suite of tools
• TMA (Traffic Management Advisor)
• provides advisories on sequencing incoming
  aircraft to controllers in Center
• FAST (Final Approach Spacing Tool)
• provides advisories on sequencing for landing to
  controllers in TRACON
• DA (Descent Advisor), EDP (Expedite Departure
  Path), UPR (User Preferred Routing)
• exploit CTASs trajectory info for analyzing
  arrival/departure paths
• CAP (Collaborative Arrival Planning)
• pass scheduling info to airlines to optimize
  planning at hubs

4
what TMA does (roughly)

• from
• rapidly changing info (seconds)
• radar information giving aircraft positions and
  velocities
• flight plan info received from host computers and
  entered by controllers
• availability of runways
• slowly changing info (hours)
• weather data
• very slowly changing info (weeks)
• map of airspace
• static info
• characteristics of different kinds of aircraft
• computes
• suggested arrival times at meter gates for
  incoming aircraft
• so
• CTASs function is to provide advisories
• not safety-critical in a narrow technical sense
• but obviously reliability is of paramount concern
• CTAS especially useful during rushes

5
organization of TMA software
INPUTSOURCEMANAGER
COMMUNICATIONSMANAGER
USERINTERFACE(TGUI, PGUI)
ROUTEANALYZER
DYNAMICPLANNER
6
our project

• what Erzberger told me (June 98)
• algorithmic components (eg, DP, RA/TS)
• well understood
• mostly reliable
• concerned about two areas of software
• TGUI, PGUI code
• CM (communications manager)
• CM a much harder problem because wed need to
  learn domain
• why i chose CM
• a more interesting problem
• large-scale engineering issues
• more about software design, less about
  algorithms, GUIs, etc
• CM is a vital component of TMA and FAST
• CM shapes the entire CTAS system

7
CM (Communication Manager)

• main functions
• maintains aircraft tree (database of aircraft
  states)
• calls algorithmic components periodically to
  process entire tree
• maintains GUIs, handles user requests
• basic stats
• all in C
• connected only by sockets to other processes no
  shared memory
• about 70kloc (including headers)
• but only about 20kloc of essential stuff
• history of development
• algorithmic components are focus of NASA
  research, startingin 1980s
• dynamic planner was designed with OMT, an OO
  method
• route analyzer translated from Fortran
• CM was first written in 1991 most of current
  code from 1997-98

8
basic control loop of CM
cm_motif.cmain_loop(every 300ms)
cm_sock_util.cselect_active_sockets
ism_socket.cism_socket_read
cm_read.cread_ra
update_aircraft.cism_socket.cism_socket_checkha
ndle_weather_updatefind_and_send_fps
(1)request_etas_for_new_fps (2)distribute_ac.c
distribute_aircaft_to_ras (3)update_active_aircra
ft
update_aircraft.cupdate_aircraft
cm_send.c(1) forward_flight_plan_add(2)
send_request_for_flight_plan(3)
handoff_aircraft_to_ra
9
project goals structure

• experiment with new technology
• design notions tools
• evaluate pitfalls successes
• invent refinements, new approaches
• improve CTAS
• help NASA/CSC improve design of CM
• explain existing problems and propose solutions
• uncover problems
• a research project
• be willing to adjust plans as we learn more
• lots of cooperation, between teams too
• can change goals to more exciting ones