This thesis presents an approach to perform formal verification of artificial pancreas system using S-Taliro, a tool for testing for metric temporal logic MTL specification on Matlab tm based models.
Violations from the spline based approach can be used to find such scenarios in actual data and the information can be used to correct for such failures in the future controller designs. We present conditions under which a given change of basis transformation for a non-linear system can define an abstraction.
The approaches are evaluated on a suite of Sriram sankaranarayanan thesis examples drawn from the literature and automatically synthesized benchmarks.
This is achieved by first setting up a non-linear program that encodes the invariance conditions for the vector field at each facet of the current polyhedron. The techniques presented here allow us to discover, given a non-linear system, if a change of bases transformation involving degree-bounded polynomials yielding an algebraic abstraction exists.
We compare our approaches to synthesizing Lyapunov functions with approaches based on SOS programming relaxations. In a broader context, it examines the need for enhanced formal verification techniques applied to more sophisticated artificial pancreas controllers that remain under development.
The search yields tunable parameter values at which the performance requirements are satisfied with a high probability, despite variations in the stochastic parameters. Next, we examine a promising simulation-based falsification approach based on robustness semantics of temporal logics.
Our iterative procedure attempts to prove at each stage that the given polyhedron is a positive invariant. We compare both types of relaxations by examining the class of polynomials that can be shown to be positive in each case. We illustrate the use of S-Taliro for finding interesting property violations in a PID-based hybrid closed loop control system.
Next, we present a progression of increasingly more powerful LP relaxations based on expressing the given polynomial in its Bernstein form, as a linear combination of Bernstein polynomials.
We present a simulation-based approach that provides a piecewise approximation of a certain quantile function for the responses of interest. Our approach is applied to three benchmarks: Violations are visualized and discussed and remedies are suggested. Malfunction of these devices can cause death or serious injury to the people using them.
In particular, the LP approach is shown to be as fast as the SOS programming approach, but less prone to numerical problems.
Ideas about checking this model is presented. If so, our technique yields the resulting abstract system, as well. We first examine two classes of relaxations for proving polynomial positivity: We argue the need for offline and online runtime verification for these devices, and discuss challenges that make verification hard.
Current clinical trials test these devices only for a few patients and do not test for the wide range of failure scenarios that could occur during daily use. Violations are found for both the model based approach as well as the spline based approach.
Subsequently, we encode the positive-definiteness of the function, and the negative-definiteness of its derivative over the domain of interest.
These can be naturally cast as chance-constrained optimization problems, which however, are hard to solve precisely.
Nevertheless, we demonstrate that the new approach allows our procedure to recover from a poor choice of templates initially to yield better invariants.
Automated formal verification can test for corner case failure conditions that could be addressed before production. Such problems are common in robust system design, including feedback controllers, biomedical devices, and many others.
Forlenza, Sriram Sankaranarayanan, and David M. The violations can be used by the control system designer to find the root cause of the failure and to design a new controller that satisfies the specified properties.
The goal is to find values of the tunable parameters that ensure the satisfaction of given performance requirements with a high probability. The well-known bounds on Bernstein polynomials over the unit box help us formulate increasingly precise LP relaxations that help us establish the positive definiteness of a polynomial over a bounded domain.Response Threshold Based Task Allocation in Multi-Agent Systems Performing Concurrent Prof.
Sriram Sankaranarayanan Date The nal copy of this thesis has been examined by the signatories, and we nd that both. Sriram Sankaranarayanan 3 of 17 TimePass (): Veriﬁcation of hybrid control systems using time trajectory approximations for ordinary differential equations through positive invariant computation as well as guaranteed.
This thesis presents progress in both areas. studies, the group included Sriram Sankaranarayanan, Henny Sipma, C esar S anchez, Matteo Slanina, Calogero Zarba, and Ting Zhang.
Bernd Finkbeiner tolerated my undergraduate antics but graduated before I became a Ph.D. student. Outside of the STeP group, Damon Mosk.
•Sriram Sankaranarayanan, AleksandarChakarov, and Sumit Gulwani,Static Analysis for Probabilistic Programs: Inferring Whole Program Properties from Finitely Many PathsIn PLDI AleksandarChakarov, PhD Thesis, University of Colorado, Boulder, August Co-Authors AleksandarChakarov Univ.
Colorado, Boulder now at Phase Change. Thesis Committees: Member of the following thesis committees Hyojung Han, University of Colorado, Boulder, PhD, Hyondeuk Kim, University of Colorado, Boulder, PhD, Sriram Sankaranarayanan, and Naveen Sharma, Object model construction for inheritance in c++ and its applications to program analysis, Compiler Construction (CC.
This thesis presents an approach to perform formal verification of artificial pancreas system using S-Taliro, a tool for testing for metric temporal logic (MTL) specification on Matlab(tm) based models. Sriram Sankaranarayanan.
Second Advisor. David Maahs.
Third Advisor. Pavol Cerny. Abstract.Download