skip to main content

AOE Home / About / AOE seminars / Archive / 2016 

Design and security of autonomous constrained control systems.

March 30, 2016

  • Dr. Abhishek Dutta
  • University of Illinois at Urbana-Champaign
  • Torg 1060
  • 4:00 p.m.
  • Faculty Host: Dr. Farhood

Abstract:  Most cyber-physical systems are constrained in their inputs and outputs. Model predictive control provided a systematic means of handling all forms of constraints leading to tremendous impact on industrial control practice. MPC transcribes the control objective into a cost function and solves the constrained optimization online. The improvement in efficiency of the online optimization led to its adoption in autonomous mechatronic systems, however stability and non-linearity posed challenges. In the first part, I shall certify stability by guaranteeing that the problem remains infinitely feasible and prove convergence of iterations in nonlinear MPC with demonstrations on simplified flight control.


Consequently, as more and more critical infrastructures such as aerospace, power and ocean systems are being embedded with sensing and control and linked to the internet, the resulting security vulnerability can be exploited to inflict systematic damage to the connected physical systems. The class of false-data injection attacks is of particular interest as it only requires the ability to compromise the measurements. In this concluding part, I shall construct such attacks, that are stealthy to set-membership based anomaly detectors over widely used control systems with bounded disturbances. The attacker uses constrained optimization based techniques to first estimate the unknown disturbance set to masquerade itself as disturbance and at the same time hijacks the system by maximizing the state estimation error or by driving it to another feasible target. These methods are used to demonstrate the vulnerability and resilience of a test autonomous aerospace system.


Bio: Abhishek Dutta is an aerospace postdoc at the University of Illinois at Urbana-Champaign with Cedric Langbort. He holds a PhD in electromechanical engineering (model predictive control) under the supervision of Robin De Keyser at Ghent University and under the advise of Jan Maciejowski as a junior member of Wolfson College Cambridge. He has a MSc with distinction from the University of Edinburgh with an informatics prize for outstanding thesis and an European masters from the University of Trento with final mark 110/110. He has held research positions at Technical University Munich (electrical engineering) and Nanyang Technological University (mechanical and aerospace)

 

 

Tags

    Tags