The
interest on networks has dramatically
increased in the last fifteen years, since it
has been recognized that the network formalism
is the best approach for dealing with complex
systems made by discrete, interacting
units like individuals, companies, computers,
etc. Although plenty of methods of analysis
have been developed, and many phenomena that
crucially characterize network structures and
dynamics can now be studied in a rigorous way,
the challenge of the next decade is to
transform the 'networks science' from a set of
elegant theoretical results to a powerful
toolbox for the everyday activity of
scientists and engineers. A specific field
where this development is more urgent is control,
as the control and the 'network science'
communities have put in common their rich
theoretical knowledge only to a very little
extent.
The
Satellite Meeting will be a forum for
discussing recent advances in the
study of network analysis and control.
The workshop is intended to be fully
interdisciplinary, and bring together
both theoretical studies and
applications to any field of science and
engineering.
Particular
attention will be devoted to the
following topics:
Control and
identification of complex
networks
Consensus and
synchronization on networks
Methods of dynamical
systems and control theory applied
to network analysis
Methods of network
analysis applied to control problems
Organization
of adaptive networks as a dynamical system
Real-world
networks are not static, but continuously
change to meet the evolving needs of society.
To manage and control such dynamic networks,
we have to understand the nature of the
adaptively changing networks, in which the
reformation of the networks and the dynamical
processes occurring on the network are
interdependent.
We considered a simple model of co-evolving
network dynamics, combining the dynamics of
random walkers and the dynamics of weighted
connections which are regulated by the traffic
of the walkers, and analyzed the organization
of the network as a dynamical system.
Under suitable conditions, the density of the
walkers and the link weights converged to
stationary power-law distributions at the
macroscopic level. However, they continued to
change with time at the microscopic level,
even though the dynamics of the proposed model
is completely deterministic without any random
processes. We numerically and theoretically
analyzed the equilibrium states from
perspective of the dynamical system and found
that the system has multi-stability including
chaotic states.
Alberto Bemporad IMT Institute for Advanced
Studies, Lucca [web]
Stochastic economic
model predictive control of complex drinking
water networks
Drinking
water networks are a real-life example of
dynamical networks that include a large number
of nodes and are affected by uncertainty. In
this talk we consider the goal of controlling
the operational management of possibly
hundreds of tanks, pumping stations and demand
nodes, under uncertain upcoming demand,
in an economically profitable
and risk-averse way. State-of-the-art
time series analysis methods are used
to produce 24-hours ahead hourly
estimates of water demand. The uncertainty
of these estimates is represented in the
form of scenario trees; the approach
is entirely data driven and no
assumptions are imposed on the
respective probability distribution
functions. We develop stochastic
model predictive control (SMPC)
methods taking into account the volatility
associated with water demand. The SMPC
problem is solved on an hourly basis in a
receding horizon fashion to optimally
decide current pumping actions and valve
positions aiming at a smooth
and economic network operation that
accounts for the constraints on tank
capacities, pumping capabilities, and
pressures.
Certain performance indicators are
chosen to evaluate the closed-loop behavior of
the controlled system. The formulated MPC
problem gives rise to a large-scale
yet well-structured QP problem, which can
be solved efficiently and in a
distributed manner by exploiting
the separable structure of its dual, thus
enabling the real-time application of
such a control system.
Controllability
and observability of complex systems
The
ultimate proof of our understanding of complex
systems is reflected in our ability to control
them. Although control theory offers
mathematical tools for steering engineered
systems towards a desired state, a framework
to control complex systems is lacking. In this
talk I will show that many dynamic properties
of complex systems can be quantitatively
studied, via a combination of tools from
control theory, network science and
statistical physics. In particular, I will
focus on two dual concepts, i.e.
controllability and observability, of general
complex systems. Controllability concerns our
ability to drive the system from any initial
state to any final state within finite time,
while observability concerns the possibility
to deduce the system’s internal state from
observing its input-output behavior. I will
show that by exploring the underlying network
structure of complex systems one can determine
the driver (or sensor) nodes that with
time-dependent inputs (or measurements) will
enable us to fully control (or observe) the
whole system.
Roberto Tempo CNR-IEIIT, Politecnico di
Torino [web]
Distributed
randomized algorithms in social and sensor
networks
We study a
new model of opinion dynamics in social
networks which has two main features. First,
agents asynchronously interact in pairs, and
these pairs are chosen according to a random
process. Following recent literature, we refer
to this communication model as "gossiping''.
Second, agents are not completely open-minded,
but instead they take into account their
initial opinions, which may be thought of as
their "prejudices''. In the literature, such
agents are often called "stubborn". We show
that the opinions of the agents fail to
converge, but persistently undergo ergodic
oscillations, which asymptotically concentrate
around a mean distribution of opinions. This
mean value is exactly the limit of the
synchronous dynamics of the expected opinions.
In the second part of the talk, we demonstrate
how a modified version of this approach is
useful in sensor networks. In particular, we
consider two problems: clock synchronization
and optimal coverage. In both cases, the
solution is obtained by means of a distributed
least-squares randomized algorithm, provided
that a suitable time-averaging operation is
performed.
Controllability
metrics, limitations and algorithms for complex
networks
This
presentation will consider the problem of
controlling complex networks, i.e., the joint
problem of selecting a set of control nodes
and of designing a control input to steer a
network to a target state. For this problem,
1) we propose a metric to quantify the
difficulty of the control problem as a
function of the required control energy, 2) we
derive bounds based on the system dynamics
(network topology and weights) to characterize
the tradeoff between the control energy and
the number of control nodes, and 3) we propose
an open-loop control strategy with performance
guarantees. In our strategy, we select control
nodes by relying on network partitioning, and
we design the control input by leveraging
optimal and distributed control techniques.
Our findings show several control limitations
and properties. For instance, for Schur stable
and symmetric networks: 1) if the number of
control nodes is constant, then the control
energy increases exponentially with the number
of network nodes; 2) if the number of control
nodes is a fixed fraction of the network
nodes, then certain networks can be controlled
with constant energy independently of the
network dimension; and 3) clustered networks
may be easier to control because, for
sufficiently many control nodes, the control
energy depends only on the controllability
properties of the clusters and on their
coupling strength. We validate our results
with examples from power networks, social
networks and epidemics spreading.
Takaaki
Aoki, Kagawa
University [web] Dario
Bauso, University of
Palermo [web]
Alberto
Bemporad, IMT Institute for Advanced Studies, Lucca
[web]
Guanrong Chen, City University
of Hong Kong [web]
Mario Di Bernardo,
University
of Bristol and University of Naples "Federico II" [web] Hideaki
Ishii, Tokio Institute of
Technology [web]
Yang-Yu Liu, Harvard University
[web]
Antonis Papachristodoulou, University of Oxford [web]
Carlo Piccardi, Politecnico di Milano [web] Roberto Tempo, CNR-IEIIT, Politecnico di Torino [web]
Please notice that time
scheduling might be modified. Check
frequently for updates.
The abstracts of the invited talks are
above in this page.
The abstract of the contributed talks
are available here.
Thursday, September 25,
2014, Villa Bottini
(invited talks = 45min, contributed
talks = 15min)
8:50 - 9:00
opening of
the meeting
morning
session, chair: Carlo Piccardi
9:00 - 9:50
invited
talk: Controllability
and observability of complex systems Yang-Yu Liu
9:50 - 10:10
Local and global dynamics
of complex systems: when does the network
structure matter? Jean-Charles Delvenne, Renaud
Lambiotte, Luis E.C. Rocha
10:10 -
11:00
invited
talk: Controllability
metrics, limitations and algorithms for
complex networks Sandro Zampieri
coffee break
11:30 -
11:50
Bayesian estimation of
simple models for networked oscillators
based on experimental data Kaiichiro Ota, Toshio Aoyagi
11:50 -
12:40
invited
talk: Distributed
randomized algorithms in social and sensor
networks Roberto Tempo
12:40 -
13:00
Dynamics and control of
game interactions on complex networks Dario Madeo, Chiara
Mocenni
lunch
afternoon
session, chair: Mario Di Bernardo
14:30 -
15:20
invited
talk: Organization
of adaptive networks as a dynamical system Takaaki Aoki
15:20 -
15:40
Speed-gradient control of
cluster synchronization by adaptation of
network topology Philipp Hövel, Judith Lehnert,
Anton Selivanov, Alexander Fradkov, Eckehard
Schöll
15:40 -
16:00
Motion control of Vicsek's
agents in an inhomogeneous space Arturo Buscarino, Luigi
Fortuna, Mattia Frasca, Alessandro Rizzo
16:00 -
16:20
Event-triggered pinning
control of switching networks Antonio Adaldo, Francesco
Alderisio, Davide Liuzza, Guodong Shi,
Dimos V. Dimarogonas, Mario di
Bernardo, Karl Henrik Johansson
16:20 -
17:10
invited
talk: Stochastic
economic model predictive control of complex
drinking water networks Alberto Bemporad, Ajay
Kumar Sampathirao, Pantelis Sopasakis