CentraleSupélec

About: CentraleSupélec is a based out in . It is known for research contribution in the topics: MIMO & Wireless network. The organization has 1330 authors who have published 2344 publications receiving 30533 citations. The organization is also known as: CentraleSupelec & CentraleSupelec of the Paris-Saclay University.

Generative Collaborative Networks for Single Image Super-Resolution

Abstract: A common issue of deep neural networks-based methods for the problem of Single Image Super-Resolution (SISR), is the recovery of finer texture details when super-resolving at large upscaling factors. This issue is particularly related to the choice of the objective loss function. In particular, recent works proposed the use of a VGG loss which consists in minimizing the error between the generated high resolution images and ground-truth in the feature space of a Convolutional Neural Network (VGG19), pre-trained on the very "large" ImageNet dataset. When considering the problem of super-resolving images with a distribution "far" from the ImageNet images distribution (\textit{e.g.,} satellite images), their proposed \textit{fixed} VGG loss is no longer relevant. In this paper, we present a general framework named \textit{Generative Collaborative Networks} (GCN), where the idea consists in optimizing the \textit{generator} (the mapping of interest) in the feature space of a \textit{features extractor} network. The two networks (generator and extractor) are \textit{collaborative} in the sense that the latter "helps" the former, by constructing discriminative and relevant features (not necessarily \textit{fixed} and possibly learned \textit{mutually} with the generator). We evaluate the GCN framework in the context of SISR, and we show that it results in a method that is adapted to super-resolution domains that are "far" from the ImageNet domain.

Axial Ferrite-Magnet-Assisted Synchronous Reluctance Motor

Abstract: This paper presents a noval 18 poles /16 slots Axial Flux Permanent Magnet-Assisted Synchronous Reluctance Motor (AF-PMASynRM) with non-overlapping concentrated winding. At first, the torque ripple and iron losses are analyzed using 3D Finite Element Analysis (3D-FEA). Then, a comparison between 3D-FEA and 2D-FEA based on flux and iron losses is established. In this paper, we propose to design the motor for high torque low speed application using a multiobjective optimization. In this kind of iterative procedure, the use of Finite Element is generally time consuming. Thus, we propose a 2D analytical saturated model that considers the local saturation near the iron bridges and the slot tangential leakage flux. The magnetic model is coupled with an electrical model that computes the power factor and the voltage at the motor terminals. A loss model is also developed to calculate the copper and the iron losses. The proposed analytical model is 5 times faster than the 2D- FEA. The optimal axial structure is compared to a previously optimized radial motor in order to evaluate the design benefits of axial flux machines.

Quantum filtering for multiple measurements driven by fields in single-photon states

Abstract: In this paper, we derive the stochastic master equations for quantum systems driven by a single-photon input state which is contaminated by quantum vacuum noise. To improve estimation performance, quantum filters based on multiple-channel measurements are designed. Two cases, namely diffusive plus Poissonian measurements and two diffusive measurements, are considered.

A unified model for low-cost high-performance AC drives: The Equivalent Flux concept

Abstract: This paper presents a unified modeling approach of alternating current (AC) machines for low-cost high-performance drives. The Equivalent Flux concept is introduced. Using this concept, all AC machines can be seen as a non-salient synchronous machine with modified (equivalent) rotor flux. Therefore, complex salient-rotor machines models are simplified, and unified shaft-sensorless AC drives can be sought. For this purpose, a unified observer-based structure for rotor-flux position and speed estimation is proposed. The equivalent flux concept generalizes the existing concepts, such as the extended back-electromotive force, the fictitious flux and the active flux.

Queueing Stability and CSI Probing of a TDD Wireless Network with Interference Alignment

Abstract: This paper characterizes the performance of interference alignment (IA) technique taking into account the dynamic traffic pattern and the probing/feedback cost. We consider a time-division duplex (TDD) system where transmitters acquire their channel state information (CSI) by decoding the pilot sequences sent by the receivers. Since global CSI knowledge is required for IA, the transmitters have also to exchange their estimated CSIs over a backhaul of limited capacity (i.e. imperfect case). Under this setting, we characterize in this paper the stability region of the system under both the imperfect and perfect (i.e. unlimited backhaul) cases, then we examine the gap between these two resulting regions. Further, under each case, we provide a centralized probing algorithm (policy) that achieves the max stability region. These stability regions and scheduling policies are given for the symmetric system where all the path loss coefficients are equal to each other, as well as for the general system. For the symmetric system, we compare the stability region of IA with the one achieved by a time division multiple access (TDMA) system where each transmitter applies a simple singular value decomposition technique (SVD). We then propose a scheduling policy that consists in switching between these two techniques, leading the system, under some conditions, to achieve a bigger stability region. Under the general system, the adopted scheduling policy is of a high computational complexity for moderate number of pairs, consequently we propose an approximate policy that has a reduced complexity but that achieves only a fraction of the system stability region. A characterization of this fraction is provided.