Papers by Montassar Ezzine
2021 IEEE Global Communications Conference (GLOBECOM), 2021
Packet routing is a fundamental problem in communication networks that decides how the packets ar... more Packet routing is a fundamental problem in communication networks that decides how the packets are directed from their source nodes to their destination nodes through some intermediate nodes. With the increasing complexity of network topology and highly dynamic traffic demand, conventional modelbased and rule-based routing schemes show significant limitations, due to the simplified and unrealistic model assumptions, and lack of flexibility and adaption. Adding intelligence to the network control is becoming a trend and the key to achieving high-efficiency network operation. In this paper, we develop a model-free and data-driven routing strategy by leveraging reinforcement learning (RL), where routers interact with the network and learn from the experience to make some good routing configurations for the future. Considering the graph nature of the network topology, we design a multi-agent RL framework in combination with Graph Neural Network (GNN), tailored to the routing problem. Three deployment paradigms, centralized, federated, and cooperated learning, are explored respectively. Simulation results demonstrate that our algorithm outperforms some existing benchmark algorithms in terms of packet transmission delay and affordable load.
Section de rattachement : 61 Secteur : Secondaire RÉSUMÉ : Nous appliquons, dans cet article, deu... more Section de rattachement : 61 Secteur : Secondaire RÉSUMÉ : Nous appliquons, dans cet article, deux méthodes de synthèse d'observateurs d'ordre réduità entrées inconnues pour estimer lesétats d'uneéolienne. La première méthode est une approche temporelle où après avoir donné les conditions d'existence et de stabilité de l'observateur, nous donnons un algorithme de synthèse. Puis nous donnons l'algorithme concernant l'approche fréquentielle avant de montrer l'efficacité de ces deux méthodes sur un modèle d'éolienne.
Section de rattachement: 61
2021 International Conference on Control, Automation and Diagnosis (ICCAD), 2021
The problem of positive filters design for the class of positive generalized systems, known as de... more The problem of positive filters design for the class of positive generalized systems, known as descriptor or singular systems is investigated. In fact, we propose a new positive full order filter for positive descriptor systems affected by disturbance vector. Positivity conditions for such systems is firstly given. Then, the proposed procedure is based on the unbiasedness of the filter using a Sylvester equation, where all positive filter matrices are parameterized by an unique gain matrix Z; These matrices are designed, such that the dynamics of the estimation error is positive and asymptotically stable.
2021 International Conference on Control, Automation and Diagnosis (ICCAD), 2021
Dynamin (DNM) is a family of large GTPases possessing a unique mechanical ability to "pinch&... more Dynamin (DNM) is a family of large GTPases possessing a unique mechanical ability to "pinch" off vesicles entering cells. DNM2 is the most ubiquitously expressed member of the DNM family. We developed a novel tool based on elastin-like polypeptide (ELP) technology to quickly, precisely, and reversibly modulate the structure of DNM2. ELPs are temperature-sensitive biopolymers that self-assemble into microdomains above sharp transition temperatures. When linked together, DNM2 and a temperature-sensitive ELP fusion organize into a range of distinct temperature-dependent structures above a sharp transition temperature, which were not observed with wild-type DNM2 or a temperature-insensitive ELP fusion control. The structures comprised three different morphologies, which were prevalent at different temperature ranges. The size of these structures was influenced by an inhibitor of the DNM2 GTPase activity, dynasore; furthermore, they appear to entrap co-expressed cytosolic ELPs. Having demonstrated an unexpected diversity of morphologically distinct structures, DNM2-ELP fusions may have applications in the exploration of dynamin-dependent biology.
2020 American Control Conference (ACC), 2020
This paper concerns the design of new positive filters for positive linear systems. In fact, we p... more This paper concerns the design of new positive filters for positive linear systems. In fact, we propose a new positive full order filter for positive linear systems subject to unknown inputs and bounded disturbances. The designed filter is always nonnegative at any time and converges asymptotically to the real state vector. This paper is among first attempts to design positive unknown input filters (PUIF) for positive linear systems. The proposed approach is based on the unbiasedness of the estimation error and by imposing the positivity of the design parameters; then a new method to avoid the derivative of the disturbance vector in the filtering error dynamics is proposed. Based on the extended strictly positive real (ESPR) design, this problem is solved by applying the Linear Matrix Inequalities (LMI)-based ESPR Lemma. Note that all structural constraints on filter matrices are addressed in terms of LMI formulation. An algorithm that summarizes the different steps of the proposed positive filter design is given. A numerical example is finally given to illustrate the effectiveness of the proposed method.
Int J Control Autom Syst, 2011
International Review of Automatic Control (IREACO), 2016
This paper investigates the problem of the full order H∞ unknown input filter designs in time and... more This paper investigates the problem of the full order H∞ unknown input filter designs in time and frequency domain. The considered systems are linear singular ones with delayed states, delayed input, unknown input and bounded disturbance. Firstly, an Unknown Input Observer (UIO) and a H∞ Unknown Input Filter (H∞ UIF) for such systems are proposed respectively in the time domain. This approach is based on Lyapunov-Krasovskii stability theory, where the gain implemented in the design is an optimal solution of linear matrix inequalities (LMIs). Secondly, a frequency-domain approach is derived from time-domain results by using the factorization approach. The proposed methods show their effectiveness on an illustrative example.
22nd Mediterranean Conference on Control and Automation, 2014
This paper investigates a discrete-time domain design of unknown input functional observer (UIFO)... more This paper investigates a discrete-time domain design of unknown input functional observer (UIFO) for discrete-time linear singular delayed systems. Note that both constant and variable delays are considered. The main procedure is based on Lyapunov-Krasovskii stability theory, where the optimal gain in the functional observer is a solution of linear matrix inequalities (LMIs). Sufficient conditions for observer stability are proposed. The effectiveness of the proposed approach is illustrated by a numerical example.
ABSTRACT The paper deals with a frequency domain solution to the standard Hinfini filtering probl... more ABSTRACT The paper deals with a frequency domain solution to the standard Hinfini filtering problem for linear time-invariant multivariable systems where all measurements are affected by disturbances using a polynomial approach. The design procedure is first obtained by considering time domain solution which is related to a Riccati equation , and then due to the connecting relationship that parameterizes the dynamics behavior between time and frequency domain, given by Hippe, the full order filter representation in the frequency domain is obtained. The filter is easy to calculate as it requires the computation of a single gain and it is easily implementable also.
ABSTRACT Tunisie [email protected]; [email protected]; [email protected]... more ABSTRACT Tunisie [email protected]; [email protected]; [email protected]; [email protected] Section de rattachement : 61 Secteur : Secondaire ESU E : This paper proposes an easier frequency domain design of observers for linear systems, where all measurements are not corrupted by disturbances, with the aid of a factor-ization approach . The design procedure in the frequency domain is first obtained from the time domain solution where we use the full order observer of type Luenberger , and then by the use of the factorization approach where we propose to define some useful coprime factor-ization, the representation in the frequency domain is derived. The frequency designed filter is easy to calculate as it requires only computation of a suitably coprime factorization. MOTS C ES : Time domain, frequency domain, linear system, observer-based controller, Linear Matrix inequalities (LMIs), Matrix fractions descriptions (MFDs).
Cet article propose une synthese temporelle puis frequentielle de filtres Hinfini d'ordre ple... more Cet article propose une synthese temporelle puis frequentielle de filtres Hinfini d'ordre plein pour les systemes singuliers sujets a des perturbations a energie bornee. La synthese temporelle est basee sur le non biais du filtre, la resolution d'une equation de Sylvester, et afin d'eviter d'avoir a deriver les perturbations, le systeme d'erreur est transforme en un systeme singulier. Puis la methode LMI (Inegalites Matricielles Lineaires) est utilisee pour trouver le gain du filtre satisfaisant en plus de la stabilite, une attenuation des perturbations. La synthese frequentielle est deduite a partir du filtre obtenu dans le domaine temporel par l'utilisation de descriptions matricielles fractionnaires (MFD) et de l'approche factorisation.
International Journal of Intelligent Computing and Cybernetics, 2011
ABSTRACT The purpose of this paper is to propose solutions for both discrete-time and frequency-d... more ABSTRACT The purpose of this paper is to propose solutions for both discrete-time and frequency-domain designs of unbiased H ∞ functional filters for discrete-time linear systems affected by bounded norm energy disturbances. The discrete-time procedure design is based on the unbiasedness of the functional filter using a Sylvester equation; then the problem is expressed in a singular system one and is solved in terms of Linear Matrix Inequalities (LMIs). The frequency procedure design is derived from discrete-time domain results by defining some useful matrix fraction descriptions and mainly, establishing the useful and equivalent form of the connecting relationship that parameterizes the dynamics behavior between discrete-time and z-domain. The performance of the proposed approach is illustrated with the aid of a practical example. The proposed methods are easily implementable and concern a more general class of systems, as the transformation of the system in a singular one permits to treat the problem of perturbance advanced. First, the order of this filter is equal to the dimension of the vector to be estimated, which is benefit in case of control purpose (reduction of time calculation comparing to the full order one). Second, all recent works on the functional filtering consider systems which permit to avoid to have advanced perturbation term in the error dynamics; the authors propose here an approach which resolves the H ∞ filtering problem even when the term is present. In addition, it permit to consider more general class of discrete-time systems. Furthermore, the LMI approaching the discrete-time case permits to handle with more general problem (H ∞ ,L 2 -H ∞ ) than the classical Riccati one.
Proceedings of the 2010 American Control Conference, 2010
The paper deals with a frequency domain solution to the standard H∞ filtering problem for linear ... more The paper deals with a frequency domain solution to the standard H∞ filtering problem for linear time-invariant multivariable systems where all measurements are affected by disturbances using a polynomial approach. The design procedure is first obtained by considering time domain solution which is related to a Riccati equation , and then due to the connecting relationship that parameterizes the dynamics
Proceedings of the 2010 American Control Conference, 2010
This paper proposes both time and frequency domain design of functional filters for linear time-i... more This paper proposes both time and frequency domain design of functional filters for linear time-invariant multivariable systems where all measurements are affected by disturbances. The order of this filter is equal to the dimension of the vector to be estimated. The time procedure design is based on the unbiasedness of the filter using a Sylvester equation; then the problem is expressed in a singular system one and is solved via Linear Matrix Inequalities (LMI) to find the optimal gain implemented in the observer design. The frequency procedure design is derived from time domain results by defining some useful Matrix Fractions Descriptions (MFDs) and mainly, establishing the useful and equivalent form of the connecting relationship that parameterizes the dynamic behavior between time and frequency domain, given by Hippe in the reduced-order case. A numerical example is given to illustrate our approach.
Uploads
Papers by Montassar Ezzine