Nous vous rappelons que, afin de garantir l'accès de tous les inscrits aux salles de réunion, l'inscription aux réunions est gratuite mais obligatoire.
Inscriptions closes à cette réunion.
11 personnes membres du GdR ISIS, et 27 personnes non membres du GdR, sont inscrits à cette réunion.
Capacité de la salle : 134 personnes.
Network information theory (NIT) studies the fundamental limits of data storage and data transmission in networks. At the same time, it aims at developing coding schemes that approach these limits as close as possible. The birth of NIT traces back to the seminal contributions of Claude E. Shannon during the second half of the 20-th century. Since then, many elegant results have been developed for a variety of network models with potential applications in many real-world scenarios. Nonetheless, today this theory remains far from being complete. This GDR-ISIS meeting aims at reuniting the community and providing the opportunity to young and senior researchers to share recent results in network information theory. In particular, the meeting welcomes contributions in the areas of, but is not limited to caching and data storage, index coding, Gaussian channels and secrecy, energy harvesting in communications, source-channel coding, finite length analysis and polar codes. To this end, researchers and students from communications, electrical engineering, computer sciences, statistics and applied mathematics are very welcome to attend. Industry also has an important place in this meeting and thus, researchers from different companies are also invited to attend.
The talks will be made in a coherent and simplified manner that should make them accessible to a large audience. The meeting is sponsored in part by INRIA and it takes place the 24th of November at the campus of the Universite de Lyon.
Bibliothéque Marie Curie de l'INSA de Lyon,
Amphithéâtre Emilie du Châtelet,
31 Avenue Jean Capelle, 69100 Villeurbanne,
Campus de la Doua - Université de Lyon
Access by train or by plane
From Lyon-St Exupéry airport:
From Part-Dieu train station:
From Perrache train station:
Access from the highway (GPS: Latitude : 45.78264 | Longitude : 4.878073)
Poster Session (7 posters)
Meryem Benammar (Huawei, France)
One of the most challenging multi-terminal source coding problems is the Gray-Wyner problem where two sources need to be conveyed, each to a different receiver, and where the encoder is linked to both decoders through a common communication link while sharing a private link with each of the two. We impose in our setting that each source be reconstructed at its intended receiver to within a given distortion level whilst maintained at a minimum level of equivocation at the opposite receiver. We derive the optimal rate-distortion-equivocation region and give insights on how to construct optimal source codes when requiring equivocation constraints which need to balance the amount of information transmitted with the resulting leakage of information.
Asma Ghorbel (CentraleSupelec, France)
We consider a cache-enabled K-user broadcast erasure packet channel in which a server with a library of N files wishes to deliver a requested file to each user who is equipped with a cache of a finite memory. Assuming that the transmitter has state feedback and user caches can be filled during off-peak hours reliably by decentralized cache placement, we characterize the optimal rate region as a function of the memory size, the erasure probability. The proposed delivery scheme, based on the scheme proposed by Gatzianas et al., exploits the receiver side information established during the placement phase. Our results enable us to quantify the net benefits of decentralized coded caching in the presence of erasure. The role of state feedback is found useful especially when the erasure probability is large and/or the normalized memory size is small.
Guiseppe Durisi (Chalmers University of Technology, Sweden)
Future wireless systems are expected to support a much wider range of use-case scenarios than current wireless standards. Emerging applications, such as industrial automation and traffic safety, may require the exchange of short packets, sometimes under stringent latency and reliability constraints. In this talk, I will illustrate how to use finite-blocklength information theory to optimally design such emerging communication systems. Specifically, I will present finite-blocklength bounds and approximations on the maximum coding rate achievable over multiple-antenna fading channels. Through these bounds, one can characterize accurately the tradeoff between the rate gain obtainable by spreading each codeword over all available time-frequency-spatial degrees of freedom, and the rate loss caused by the need of estimating the fading coefficients over these degrees of freedom. In particular, the bounds allow one to determine the optimal number of transmit antennas and the optimal number of time-frequency diversity branches that maximize the rate.
The bounds also indicate whether the available transmit antennas should be used to provide spatial diversity or spatial multiplexing. I will also demonstrate that classic information-theoretic performance metrics such as the ergodic and the outage capacity are inadequate performance predictors for short packet sizes. This indicates the necessity of accurate finite-blocklength analyses when studying short-packet wireless communications. In the talk, I will provide an accessible introduction to finite-blocklength information theory, with specific focus on AWGN and fading channels. I will also outline further information-theoretic challenges related to the design of low-latency ultra-reliable wireless communication systems.
Inaki Esnaola (University of Sheffield, UK)
The smart grid paradigm is founded on the integration of existing power grids with advanced sensing and communication infrastructures. While the benefits provided by this setting are crucial for the future development of power grids, it also increases the dependency on data acquisition and system monitoring procedures. Central to the control and optimization of power systems is the state estimation problem in electricity grids. One of the main contingencies faced by state estimation procedures is intentionally corrupted data by a malicious attacker. In this talk data injection attacks are studied in a game theoretic setting. First, it is assumed that the network operator acquires the state variables via generalized least squares estimation and different attack performance metrics are proposed. The scenarios defined by the performance metrics are then analyzed. In the second part of the talk, attack constructions are studied assuming minimum-mean-square-error state estimation in centralized and decentralized scenarios. In both scenarios closed form expressions for best response functions and Nash equilibria are given. The analytical results are evaluated in practical state estimation settings by numerically evaluating the performance of different attack strategies in IEEE test systems.
Michèle Wigger (Télécom ParisTech, France)
We consider two extensions of the celebrated Maddah-Ali & Niesen cache-aided one-to-many broadcast scenario, where communication takes place in two phases: In the first placement phase the single transmitter pre-stores data in the various receivers' caches, even before knowing which files are requested by the receivers. In the second delivery phase the transmitter conveys the requested files over a common bit-pipe to the various receivers, which can also exploit the contents pre-stored in their caches. Our first extension is a rate-distortion version of the Maddah-Ali & Niesen scenario where the various files can be correlated and the receivers need only approximate reconstructions of their desired files. This extension applies in particular to the compression of multi-view videos where the frames corresponding to the various views are correlated and where approximate reconstructions suffice. For this first extended setup, we present new feasibility (coding scheme) and infeasibility (converse) results, and we show that they coincide in some cases. We also show connections to the Gray & Wyner source coding system and to the Wyner and the Gacs & Korner common informations. In our second extension, the delivery phase takes place over a noisy broadcast channel (e.g., an erasure broadcast channel). This extended setup models communication that takes place over highly congested networks. We present new feasibility (coding scheme) and infeasibility (converse) results that match in some special cases. In particular, we show that separation does not hold in cache-aided networks, but distributed joint cache-channel coding (similar to distributed joint source-channel coding) is necessary, such as our recently introduced piggyback coding. The talk is based on joint work with Bernhard Geiger and Roy Timo from TU Munich, and with Shirin Saeedi Bidokhti from Stanford University.
Albert Guillén i Fabregas (Universitat Pompeu Fabra, Spain)
Two alternative exact characterizations of the minimal error probability of Bayesian M-ary hypothesis testing are reviewed. The first expression corresponds the error probability in an induced binary hypothesis test and implies the tightness of the meta-converse bound by Polyanskiy, Poor and Verdú; the second expression implies the tightness of a generalized Verdú-Han lower bound. The expressions help to characterize the minimal error probability of several problems in information theory and to identify the steps where existing converse bounds are loose.
Selma Belhadj-Amor (Inria, France)
In this talk, the fundamental limits of simultaneous information and energy transmission in the two-user Gaussian multiple access channel with and without feedback are fully characterized. All the achievable information and energy transmission rates (in bits per channel use and Joules per channel use respectively) are identified. Thus, the capacity-energy region is defined in both cases.
Deniz Gunduz (Imperial College, UK)
For efficient and reliable electrical energy generation and distribution, advanced control mechanisms have been introduced into the grid. These mechanisms help us monitor the grid in order to rapidly diagnose potential problems, and dynamically react to variations in demand and supply. Smart meters (SMs) are a key component of smart grid control mechanisms. However, SM readings also reveal sensitive private information about consumer behavior since every electric appliance has its own detectable power consumption signature. In this talk, we will introduce novel "physical techniques" to provide privacy to SMs, instead of resorting to purely "cyber" approaches which focus on reporting distorted or combined SM readings. Our physical approach, achieved thanks to the availability of an alternative energy source (e.g., a renewable energy source), retains the critical monitoring and control capabilities SMs provide to the smart grid. We are particularly interested in information theoretic privacy measures in order to provide technology-independent privacy guarantees to consumers. We consider average as well as instantaneous energy constraints on the alternative energy source, and provide single-letter information theoretic performance bounds on the privacy in certain scenarios.
Adriano Pastore (EPFL, Switzerland )
On multiple-access channels with imperfect channel-state information at the receiver, it is demonstrated how a rate-splitting and successive-decoding approach can enhance an inner bound on the rate region achieved by Gaussian codes. Under mild conditions this improved inner bound is shown to be asymptotically tight at high SNR if the channel-state information becomes exact as the SNR tends to infinity. Extensions of the rate-splitting approach to fading channels with no channel-state information are presented. In the context of mismatched decoding, we will explain how this improved bound can be achieved through the use of a successive nearest-neighbor decoding scheme
Ligong Wang (CNRS, France)
We introduce a new quantization problem for point processes: to describe the points so as to allow the reconstructor to generate a covering set that is small in its Lebesgue measure and that is yet guaranteed to contain all the points. The asymptotic trade-off between the description length and the size of the covering set is determined for Poisson processes in various settings. We also present several results for other random point processes and for arbitrarily chosen point patterns. This is joint work with Amos Lapidoth and Andreas Malär.
Poster Session (6 posters)
German Bassi (CentraleSupelec, France)
It is a well-known fact that feedback does not increase the capacity of point-to-point memoryless channels, however, its effect in secure communications is not fully understood yet. In this work, two achievable schemes for the wiretap channel with generalized feedback are presented. One of the schemes, which is based on joint source-channel coding, correlates the codewords to be transmitted with the feedback signal in the hope of ``hiding'' them from the eavesdropper's observations. The other scheme, which uses the feedback signal to generate a shared secret key between the legitimate users, encrypts the message to be sent at the bit level. These schemes recover previous known results from the literature, thus they can be seen as a generalization and hence unification of several results in the field. Additionally, we present new capacity results for a class of channels and, for the Gaussian wiretap channel with noisy feedback, the schemes are shown to achieve positive secrecy rates even in unfavorable scenarios where the eavesdropper experiences much better channel conditions than the legitimate user.
Inaki Estella (Huawei, France)
We consider the uplink transmission over a cloud radio access network, in which the signal transmitted by multiple users (UE) is received at multiple base stations (BS). The BSs are connected to a central processor (CP) via finite-capacity backhaul links, where the UE's messages have to be recovered. Focusing on maximizing the achievable sum-rate, we propose a lattice based coding scheme, in which linear combinations of the UE's messages are decoded at the BS. Each BS recovers the channel input associated to the decoded linear combination, jointly compresses it with the received signal, and forwards it to the CP. This scheme exploits the correlation between received signals and reconstructed equations, as well as the partial denoising achieved by locally decoding linear combinations of messages. The proposed scheme is shown to outperform the classical schemes for this scenario, namely compute-and-forward and compress-and-forward. The results are illustrated through some numerical examples.
Victor Exposito (Mitsubishi Electric R&D Centre Europe, CentraleSupelec, France)
We consider the problem of transmitting a common message from a transmitter to two receivers over a broadcast channel, which is also called multicast channel in this case. The two receivers are allowed to cooperate with each other in full-duplex over non-orthogonal channels. We investigate the information-theoretic upper and lower bounds on the achievable rate of such channels. In particular, we propose a two-round cooperation scheme in which the receivers interactively perform compress-forward (CF) and then decode-forward (DF) to improve the achievable rate. Numerical results comparing the proposed scheme to existing schemes and the cutset upper bound are provided. We show that the proposed scheme outperforms the non-interactive DF and CF schemes as well as the noisy network coding. The gain over the DF scheme becomes larger when the channel becomes symmetric, while the gain over the CF scheme becomes larger when the channel becomes asymmetric.
Abdallah Fayed (Qatar University, Qatar)
We tackle a two-user Interference channel where every transmitter-receiver pair communicates two messages, one of them is public while the other is hidden from the receiver in the other transmitter- receiver pair. Every receiver is a legitimate receiver for two messages, an eavesdropper for a third one, antipathetic for a fourth one, which could or could not be used in reception by the receiver. The secrecy level is measured by the equivocation rate, we characterize an achievable rate region and an outer bound.
Alexandre Marquet (GIPSA-lab, France)
Un signal multiporteuses peut être vu comme une expansion sur une famille de Gabor dont les coefficients correspondent aux symboles à transmettre. En notant F0 l'espace entre sous-porteuses et T0 la durée symbole multiporteuses, il est alors nécessaire d'avoir F0.T0 > 1 pour pouvoir retrouver ces symboles sans erreur à l'aide d'un récepteur linéaire (en l'absence de bruit). Dans ce poster nous nous intéressons au cas où F0.T0 < 1, ce qui permet d'augmenter la densité de signalisation au prix d'une apparition d'interférence entre impulsions de mise en forme. On montre que cette interférence est minimisée par l'emploi de frames de Gabor étroites en émission et en réception.
Roy Timo (TUM, Germany)
We present a converse bound for lossy source coding in the finite blocklength regime. The bound is based on $\dif$-tilted information, and it combines ideas from two different converse techniques by Kostina and Verdú. When particularised to the binary and Gaussian memoryless sources, the new bound gives slightly tighter results in certain blocklength regimes. Joint work with Lars Palzer from TUM.
Victor Quintero (Inria, France)
Perfect channel-output feedback enlarges the capacity region of the two-user linear deterministic interference channel in most of the cases. However, in the case in which feedback links are impaired by noise, the conditions over which implementing feedback makes a significant difference are not yet well identified. This presentation describes the exact conditions for which implementing feedback enlarges the capacity region. From an engineering point of view, this presentation provides an answer to two particular questions: (a) If feedback can be implemented in only one of the transmitter-receiver pairs, what is the most convenient choice if the objective is to increase the sum rate or one of the two individual point-to-point rates?; (b) What are the conditions over which implementing feedback is absolutely useless from the point of view of enlarging the capacity region? The presentation concludes with some examples of particular study cases.
Date : 2015-11-24
Lieu : Bibliotheque Marie Curie de lINSA Lyon, Amphi Emilie du Chatelet
Thèmes scientifiques :
D - Télécommunications : compression, protection, transmission
Inscriptions closes à cette réunion.
Accéder au compte-rendu de cette réunion.
(c) GdR 720 ISIS - CNRS - 2011-2022.