----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Mardi 15/11/2106    14h45-17h15
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
14h45-14h50         Introduction
14h50-15h15         Enrico Bini -  Optimal Sampling for Linear Control Systems
15h15-15h40         Sophie Tarbouriech - Stability of Discrete-time Control Systems with Uniform and Logarithmic Quantizers
15h40-16h05         Antoine Girard - Timing contracts for embedded control systems
16h05-16h30         Alexandre Seuret - Robust control under weakened real-time constraints
16h30-17h00         Stefan Valentin - Anticipatory video streaming for mobile users
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Mercredi 16/11/2016   14h15-17h05
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
14h15-14h45        Jean-Pierre Richard  - Networked Control problems: issues, models, solutions
14h50-15h20        Luigi Palopoli  -  Real-time issues in Robotics  
15h25-15h55        Thibault Hilaire - From controller to code: an overview of the finite-precision implementation flow
16h00-16h30        Dorine Petit, Jean-Philippe Georges, Thierry Divoux  - Analyse de la fraîcheur de séquences fonctionnelles dans les lanceurs spatiaux

16h30-17h05        Discussions

 

--------------------------------------------------------------------------------------------------------------------------------------------
* Enrico Bini - Optimal Sampling for Linear Control Systems

In embedded control systems, reducing the number of sampling instants per time unit without degrading the control performance can
lead to several benefits such as energy savings, reduced communication bandwidth, etc.

In this talk, the problem of optimal sampling is discussed. Such a problem is solved explicitly for first-order systems. The explicit
solution of the optimal sampling reveals a surprising connection with the problem of quantization of the optimal continuous-time control input.

-----------------------------------------------------------------------


* Sophie Tarbouriech - Stability of Discrete-time Control Systems with Uniform and Logarithmic Quantizers

The talk deals with stability analysis of discrete-time linear systems involving finite quantizers on the input of the controlled plant. Two kinds of quantization are analyzed: uniform and logarithmic. Through LMI-based conditions, an attractor of the state trajectories and a set of admissible initial conditions are determined. A method is proposed to compare the performances of the two kinds of quantization in terms of the dimensions of the attractor, considering a scenario of Networked Control Systems (NCS). Computational issues are discussed and a numerical example is presented to validate the work.

-----------------------------------------------------------------------

* Antoine Girard - Timing contracts for embedded control systems

Physical systems equipped with embedded controllers are becoming pervasive (smart buildings, intelligent cars, drones, robots, etc.), thus increasing the need for high-confidence analysis and design tools that are able to handle  tight interactions between the physical and digital worlds. In this context, contract-based approaches have been identified as a promising direction for cyber-physical systems design.
For instance, for embedded controller implementation, timing contract specify the constraints on the time instants at which certain operations are performed such as sampling, actuation or computation. Under such contracts, the control engineers are responsible for designing a control law that is robust to all possible timing variation specified in the contract while the software engineers can focus on implementing the proposed control law so as to satisfy the timing contract. In this talk, we propose techniques that are useful within this framework.

In the first part of the talk, we will consider the problem of stability verification: given models of the physical plant and of the controller and a timing contract, verify that the resulting dynamical system is stable. We propose an approach based on the notion of reachable set, and which builds on efficient over-approximation algorithms developed over the past decade.  
We then tackle the problem of timing contract synthesis:  given models of the physical plant and of the controller, determine a set of timing contracts that guarantee stability of the resulting system.

In the second part of the talk, will consider the problem of schedulability: given several controllers, each with a timing contract, to be implemented on a common computational platform, synthesize a dynamic scheduler allocating the shared computational resources, which guarantees that the timing contracts of all controllers are satisfied. We propose an approach based on timed games, whose solution provides a suitable schedule.

-----------------------------------------------------------------------

* Alexandre Seuret -  Robust control under weakened real-time constraints

A weakened implementation scheme for real-time feedback controllers is proposed to reduce the conservatism
due to traditional worst-cases considerations, while preserving the stability and control performance. Based on recent results
to assess stability of linear systems with delayed and sampled-data inputs, this paper takes into account both the effects of
deadline misses of control tasks and uncertainties in the plant. The methodology is applied to the pitch control of an aircraft,
showing that weakening the real-time constraints allows for saving computing power while preserving the system’s stability.

-----------------------------------------------------------------------

* Stefan Valentin - Anticipatory Video Streaming for Mobile Users

Mobile video streaming is a rising business but poses significant challenges for network and service operators. In mobile networks, time-variant wireless channels often violate the content's rate requirement, which leads to poor streaming quality and wasted channel capacity. This tutorial will cover new techniques for anticipatory adaptation, which drastically improve this situation. By proactively adapting load and resource allocation to a prediction of the wireless channel state, the network prevents service outages even before they occur. Drive tests and simulation consistently show impressive gains in video quality at no significant cost for the operators. After presenting powerful heuristics, we will survey optimal formulations based on linear programming, quadratically constrained programming, and Markov decision processes. Channel prediction will be covered from the perspective of applying support vector machines and Bayesian spatio-temporal inference to radio maps. We will conclude with a vision for context and user-centric adaptation that could play a key role for providing truly seamless services in 5G.

-----------------------------------------------------------------------

* Jean-Pierre Richard - Networked Control problems: issues, models, solutions

Internet and its wired/wireless implementations present great advantages for flexible and low-cost networking. However, for control applications, they are not as reliable as dedicated CANs: integrating them while preserving some performance, constitutes an interesting challenge.
It will be briefly explained how using few dynamical models (delay/hybrid systems) allows for catching many of the effects introduced by the presence of such unreliable networks in the control loops, including non-uniform scheduling and sampling, lost packets, and variable transmission times.
Various theoretical techniques allow for analyzing the resulting models. Some of them need the delays to be constant, which can be obtained by simplification or by using waiting strategies involving buffers and leading to time-driven solutions. Some other introduce prediction so to tolerate fast varying delays and allow for event-driven, faster solutions. Some experimental cases will illustrate the talk.

-----------------------------------------------------------------------

* Luigi Palopoli - Real-time issues in Robotics

In this talk we will report some of the experiences of the author with some real—life robotic application of industrial complexity.
We will restrict the focus on a few issues that come up when the the robot interacts with the environment and collects data through complex
sensors (e.g., RADAR, cameras), and execute complex operations that require accurate planning.
In cases like this, the complexity of the plan largely depends on the complexity of the scene, which generates time variable execution profiles.
Hard real—time design is ineffective in these cases for it leads to unnecessary system complexity in the face of the robots known resilience
to occasional timing failures. We will describe the real—time design problems and discuss possible avenues to their effective solution.

-----------------------------------------------------------------------

* Thibault Hilaire - From controller to code: an overview of the finite-precision implementation flow

La transformation d'un filtre ou d'un régulateur en du code s'exécutant sur une plateforme numérique (ou un circuit matériel le réalisant) est une tâche compliquée et nécessitant une large expertise (control/signal, architecture des ordinateurs, arithmétique des ordinateurs).
Classiquement, cette tâche est réalisée avec des outils d'assistance tels que Matlab/simulink permettant d'évaluer, par simulation l'impact du passage en précision finie (en arithmétique virgule fixe). Mais cette approche ne permet pas de prendre en compte toutes les possibilités de transformation : réorganisation de l'algorithme (changer la structure des calculs), réorganisations des calculs (leur ordre), optimisation des largeurs (se permettre de calculer qu'avec le nombre minimum de bit requis), etc. Nous présenterons donc un flot permettant de prendre en compte tous ses aspects lors de l'automatisation de l'implémentation en virgule fixe de filtres et régulateurs linéaires, et, pour chacun, de quantifier et manière fiable leur impact sur la précision.

-----------------------------------------------------------------------

* Dorine Petit, Jean-Philippe Georges, Thierry Divoux  - Analyse de la fraîcheur de séquences fonctionnelles dans les lanceurs spatiaux  

Afin de réduire les coûts financiers, les problématiques de propriété intellectuelle, de masse et de passage à l’échelle, les nouvelles applications spatiales évoluent aujourd’hui vers des composants sur étagère (COTS) et des solutions standardisées comme les réseaux Ethernet commutés. Pour accompagner cette évolution, il est nécessaire pour les lanceurs d’être capable de vérifier la sûreté de fonctionnement et le bon pilotage via une télémesure bord/sol. Le problème consiste ici à recouvrer, à partir d’un fichier de capture des trames sur un réseau, la satisfaction d’une exigence de fraîcheur (d’informations de mesure et/ou de commande). Après avoir formalisé cette fraîcheur, nous montrons les conditions dans lesquelles il est possible de statuer sur la conformité d’un vol. Des résultats expérimentaux illustrent la démarche dans différentes conditions de défaillances.