| Alessandro Abate |
Research | |||||||||
Main Bio Research
Publications Teaching Contacts Links Barnum |
Broadly, my research interests lie in the field of System Theory and Control Engineering.
Main Affiliations:
Current External Affiliations: Computer Science Lab, SRI International, Menlo Park, CA; Department of Information Engineering, University of Padova, Italy.
My main research focus is in the theory of hybrid systems, i.e. nonlinear systems that are generalizations of dynamical systems
in the sense that they involve hierarchical and mixed models of computation, and in particular in the novel introduction and
study of concepts of stochastic hybrid systems within this setting. The formulation of hybrid systems is a recently developed
modeling framework which has proved to be modeling accurately engineering systems which exhibit complex behaviors, like air
traffic control systems, infrastructure networks, and natural systems like biological networks. Hybrid systems theory presents
technically challenging and original problematics related to the intrinsic interplay between continuous and discrete components
and their dynamics. The necessity of modeling uncertain, noisy or stochastic systems has motivated the subsequent introduction
of probabilistic concepts in the framework. This, in turn, requires the development of new concepts and techniques from
probability theory that are quite sophisticated. Stochastic hybrid systems have recently been established as the novel and
strongest modeling choice for their broad generality and wide potential applications. Systems and Computational Biology
Box Invariance for biologically-inspired Systems:
Simulation of Genetic, Protein and Metabolic Networks: Control of Wireless Communication Networks
The problem of congestion control and packet exchange on wireless networks is currently of great interest,
both for its underlying theoretical aspects, as well as for its clear practical implications and applications. In collaboration
with Minghua Chen, we considered a mathematical model for the fluid flow approximation of
the real Transmission Control Protocol (TCP) for wired networks and focused on the problem of extending this scheme to
the wireless scenario. The new proposed scheme was rigorously analyzed, its stability derived and its robustness properties
studied. The necessity of introducing a specific wireless model is motivated by the presence of channel error (due to
intrinsic noise or channel corruption), which often is not known exactly. This leads to further modification of the model
by approximating parts of its structure with binary functions. These new discontinuous elements, while greatly
simplifying in practice the structure of the algorithm, complicated the theoretical analysis of its dynamical properties.
They were therefore approximated with continuous functions with limiting convergence. We then
investigated the important issues of existence and uniqueness of the equilibrium for the new dynamical system, and
of local asymptotic stability. Furthermore, it was shown that this equilibrium solves a concave net utility optimization
problem, of which the classical one for wired networks is a special case. The scheme, proposed to handle the traffic
on a wireless network, appears not only to be theoretically meaningful, but has also the potential to be translated into a
practical application layer implementation.
The entire analysis of the model, which contained strong nonlinearities, posed interesting difficulties that was
possible to overcome by using results form nonlinear systems theory. In particular, the idea of referring to results from the
theory of 'singular perturbations' was key in enabling the study of the properties of systems with dynamics
that happen at different time scales (in the particular instance it was the case of the packets sent through a network and
the number of users on the network). The work appears to be of potentially high applicability and impact.
Robust MPC:
Non linear Control of a Motorcycle: Prof. Claire Tomlin, my postdoc advisor at Stanford Prof. Shankar Sastry, my PhD research advisor at Berkeley Prof. Laurent El Ghaoui, my MS co-advisor at Berkeley Ling Shi, at CDS, Caltech Prof. Slobodan Simic, at San Jose' State University Dr. Minghua Chen, now at CUHK, previously at MSR and EECS, UC Berkeley Prof. Avideh Zakhor, at EECS, UC Berkeley Prof. Ruggero Frezza, at DEI, Padova Dr. Alessandro Saccon, at DEI, Padova Dr. Aaron Ames, now post-doc at CDS, Caltech Prof. Gert Lanckriet, CS, UCSD Saurabh Amin, CE, UC Berkeley Dr. Alessandro D'Innocenzo, at the University of L'Aquila/ GRASP Lab, UPenn Dr. Giordano Pola, now at EE, UCLA/EE L'Aquila Prof. John Lygeros, Automatic Control Lab, ETH Zurich Prof. Maria Prandini, Politecnco di Milano, Italy Dr. Ashish Tiwari, at the CS Lab, SRI International Dr. Fabrizio Dal Moro, Department of Urology, School of Medicine, University of Padova |
|||||||||