Robust  Control  for    Safety  and  Security  

Hamsa  Balakrishnan    

Joint  work  with  H.  Khadilkar,  P.  Park,  V.  Ramanujam  and  C.  Tomlin  

Page  2  

*  Today’s  operations  *  Surveillance  using  ground-­‐based  radar  systems  *  Primarily  “procedural”  approach  to  air  traffic  control  *  Manual  handoffs  between  controllers  with  little  prior  coordination  *  Radio  communications  between  pilots  and  controllers  

*  NextGen  operational  concepts  *  Satellite-­‐based  surveillance  technologies:  ADS-­‐B  *  Increased  potential  for  control  and  optimization  algorithms  *  Increased  availability  of  state  information  (onboard  and  ground)  *  Datalink  capabilities  

New  Technologies  Enable  New  Operational  Concepts    

10/28/13  

Page  3  

*  Increased  potential  for  control  and  optimization  algorithms  *  Enhancing  system  capacity  *  Improving  operational  efficiency  * Maintaining/improving  system  safety  

*  New  challenges  *  Interactions  between  new  and  legacy  infrastructure  *  Information  security  *  GPS  jamming/spoofing  *  Detecting  adversaries  in  the  presence  of  uncertainties  

*  Incentives  for  participation  *  Cost  vs.  potential  benefit  of  collaboration  *  Risks  associated  with  information-­‐sharing  

Opportunities  and  Challenges  

10/28/13  

Page  4  

*  Objectives:  Safety  and  efficiency  *  Conflict  detection  and  resolution  *  Optimize  State  Update  Interval  *  Minimize  flight  times  

*  Decentralized  at  longer  range  *  Low  traffic  density  *  ADS-­‐B  surveillance  *  Max  transmit  power  

*  Handover  zone  *  Decentralized  control    *  Adaptively  adjust  transmit  power  

*  Centralized  close  to  the  airport  *  High  traffic  density  *  Min  transmit  power  *  Ground  radar  surveillance  *  Augmented  by  ADS-­‐B  

Hybrid  Communication/Control  Algorithms  

10/28/13  

−600 −400 −200 0 200 400 600

−600

−400

−200

0

200

400

600

x

y

Arr fixesDep fixesAirport

Centralized zone

Handover zone

Distributed zone

Park  et  al.,  IEEE  Trans.  on  Intelligent  Transp.  Sys.  2013  

Page  5  

High  Confidence  Networked  Control  

10/28/13  

A2A communication

ADS-B ground station

A2I communication

Ground radar systemGround infrastructure system

Airborne system

Satellite system

Safety msg Cryptographic materialADS-Bavionics

GPS receiver

INS

*  Secure,  fault-­‐tolerant  control  in  the  presence  of  adversaries  *  Distributed  control  using  onboard  

threat  detection  *  GPS  and  inertial  sensor  data  fusion  *  Verification  using  Doppler  effect  and  RSS  of  ADS-­‐B  messages  from  neighboring  aircraft  *  Control  objectives  *  Conflict  avoidance,  maintaining  

separation  in  the  presence  of  uncertainty  *  Minimizing  flight  times  *  Fault  detection  

INS (IV-B)

Doppler (IV-C)

GPS (IV-B)

KF (V)

TX

RXRSS (IV-C)

EKF (VI)

MMSE (VI-B)

Control (IX)+-

+-

Verification (VIII)

Verification (VIII)

Distance

Position

Receive ADS-B msg

Transmit ADS-B msg

GPS/INS System

Doppler/RSS System

Position

KF

RSS D

etection (VII)

Park  et  al.,  under  review,  2013  

Page  6  

*  Sequenced  to  land  (takeoff)  on  a  runway,  and  determine  their  landing  (takeoff)  times  *  Separation  requirements  (safety)  *  Limited  flexibility  afforded  to  air  traffic  

controllers  *  Operational  constraints  (including  arrival/

departure  time  windows)  *  Precedence  constraints  

*  Objectives:  Throughput,  robustness,  equity  *  Results  *  Solution  space  can  be  represented  as  a  

network  whose  size  is  linear  in  the  number  of  aircraft    

*  Various  interesting  extensions  can  be  solved  in  (pseudo-­‐)polynomial  time  as  shortest-­‐path  problems  on  variations  of  this  network  

*  Can  evaluate  tradeoffs  between  multiple  objectives  

Safe,  Efficient  and  Robust  Scheduling  

10/28/13  

0

5

10

15

20

25

29 30 31 32 33 34 35 36 37 38 39

Throughput (# of aircraft / makespan)

Reliab

ilit

y

FCFS

k=1

k=2

FCFS with buffering

max.  number  of  position  shifts  from  FCFS  

Baseline  

Reliability  =  robustness  of  schedule    robustness  of  baseline  

Chandran  &  Balakrishnan,  ACC  2008  Balakrishnan  &  Chandran,  Oper.  Res.,  2010  Lee  &  Balakrishnan,  Proc.  of  IEEE,  2008  

Page  7  

*  Initial  allocation  of  resources  typically  adopt  an  optimistic  view  of  capacity  *  Algorithms  for  reallocating  resources  given  stochastic  capacity  forecasts  *  Exchange  mechanisms  *  Pareto-­‐efficiency  (no  other  allocation  preferable  to  all  airlines)  *  Voluntary  participation  (incentive  to  participate)  *  Incentive  compatibility  (incentive  to  report  true  preferences)  *  Core  allocation  (no  incentive  for  airlines  to  deviate  by  forming  coalitions)  

*  Stochastic  optimization  algorithms  given  scenario-­‐tree  forecasts  *  Evaluation  of  incentives  to  participate,  using  realistic  aircraft  delay  costs  *  Evaluation  of  tradeoffs  between  adaptability  (extent  of  dynamic  replanning)  and  

flexibility  available  to  airlines  

*  Mechanisms  that  combine  optimization  with  (monetary)  transfers    

Resource  Reallocation  in  the  Presence  of  Uncertainty  

10/28/13  

Balakrishnan,  CDC  2007  Ramanujam  &  Balakrishnan,  under  review,  2013  

Page  8  

*  New  technologies  present  opportunities  for  robust  control  algorithms  *  New  challenges  pertaining  to  *  Safety    *  Security  *  Information-­‐sharing  *  Interactions  between  new  and  legacy  infrastructure  systems  *  Integration  and  co-­‐design  of  Economic  Incentives  (EI)  and  Robust  

Control  (RC)  algorithms  for  better  system  performance  

Summary  

10/28/13