This seminar is held online approximately bi-weekly. The seminars cover the broad areas of communication and information sciences.

If you are interested in virtually participating in the seminar, as a speaker or attendant, send an email to samir.perlaza@inria.fr and/or padakand@eurecom.fr.

## Talk Schedule

Date | Title | Speaker | Video |

Nov 24, 2023 | Towards Practical Massive Random Access | Maxime Guillaud | Not Recorded |

Dec 14,2023 | Loosing control and interfering | Laurent Clavier | Public |

Jan 12, 2024 | Topological Algebra for Wireless Systems | Laurent Decreusefond | Public |

Jan 26, 2024 | Distributed Computation over Networks | Derya Malak | Public |

Feb 08, 2024 | Copulas and Compression | Malcolm Egan | Public |

Mar 01, 2024 | On the computation of the rate-distortion-perception function | Fotios Stavrou | Public |

Mar 29, 2024 | Communication Models for Intelligent Metasurfaces Using Multiport Network Theory | Marco Di Renzo | Public |

Apr 4, 2024 | Information-Theoretic Proofs based on Change of Measure Arguments | Michèle Wigger | Public |

May 17, 2024 | Spatial Network Calculus and Performance Guarantees in Wireless Networks | Ke Feng | |

May 31, 2024 | TBA | Mireille Sarkiss | |

## Talk Abstracts

- May 31. 2024:
**Mireille SARKISS****(**Telecom SudParis**)****Title:**TBA**Abstract:**TBA

- May 17. 2024 at 14H: Ke Feng
**(**LINCS**)****Title:**Spatial Network Calculus and Performance Guarantees in Wireless Networks**Abstract:**Network calculus is a methodology allowing one to provide performance guarantees in queuing networks subject to regulated traffic arrivals and service guarantees. It is a key design tool for latency-critical wireline communication networks where it allows one to guarantee bounds on the end-to-end latency of all transmitted packets. In wireless networks, service guarantees are more intricate as electromagnetic signals propagate in a heterogeneous medium and interfere with each other. In this talk, we present a novel approach toward performance guarantees for all links in arbitrarily large wireless networks. We introduce spatial regulation properties for stationary spatial point processes, which model transmitter and receiver locations, and develop the first steps of a calculus for this type of regulation. This can be seen as an extension to space of the (classical) network calculus developed with respect to time. Using this approach, we derive reliability, rate, and latency guarantees for all links in spatially regulated wireless networks. Such guarantees do not exist in networks without spatial regulations, e.g., Poisson networks.

- April 4, 2024 :
**Michèle Wigger**(Telecom Paris) — See Video**Title: Information-Theoretic Proofs based on Change of Measure Arguments****Abstract :**We will be presenting information-theoretic proofs based on change of measure arguments for basic source coding and hypothesis testing problems, and if time permits also for channel coding. For the basic source and channel coding setups, the proofs only use a change of measure argument on the strongly typical set and are established solely by analyzing the asymptotic behaviour of the new measure. This proof method is also extended to a source coding setup under the relaxed expected-rate constraint, in which case the minimum compression rate depends on the allowed probability of error $\epsilon$ and an $\epsilon$-dependent converse is required. In the second part of the talk we present converse proof methods for hypothesis testing setups, where in addition asymptotic Markov Chains need to be established.

- Mar 29, 2024
**Marco Di Renzo**(Paris-Saclay University – CNRS and CentraleSupelec, Paris, France) — See Video**Title**:**Communication Models for Intelligent Metasurfaces Using Multiport Network Theory****Abstract**: Intelligent metasurfaces are an emerging technology that enhances the reliability of data transmission by appropriately shaping the propagation of electromagnetic waves in the wave domain, by turning radio propagation environments into smart radio propagation environments. Despite the potential performance gains and applications that this technology may provide in future wireless networks, a major limiting factor preventing information, communication, and signal processing theorists from realizing its full potential and unveiling its ultimate performance limits lies in understanding the electromagnetic and physical properties and limitations underpinning it. In this talk, we present an electromagnetically consistent approach for modeling and optimizing metasurface-aided systems based on multiport network theory.

- Mar 01, 2024 :
**Fotios Stavrou**(EURECOM) — See Video**Title: On the computation of the rate-distortion-perception function**

**Abstract :**In this talk, we study methods for the computation of a generalization of the rate-distortion function called the rate-distortion-perception function (RDPF). This generalization to the rate-distortion problem was defined in the machine learning community in recent years, to give a mathematical intuition to the observation that “low distortion” is not a synonym for “high perceptual quality”, and, from an optimization standpoint, there is a tradeoff between the distortion and the perception. The RDPF is also a relevant semantic based metric in compression. For discrete sources, with single-letter distortion and a perception constraint that belongs to the family of f-divergences, we compute the RDPF by proposing an approximate alternating minimization approach a la Blahut-Arimoto algorithm that converges to a globally optimal point. For scalar Gaussian sources with squared error distortion and a perception constraint that can be either the KL divergence, the squared Wasserstein distance, the squared Hellinger distance or the geometric Jensen-Shannon divergence, we derive closed-form expressions under the assumption of jointly Gaussian processes. Our closed-form solutions are also supported by their corresponding test-channel forward realizations, the first of their kind. For vector Gaussian sources, we propose a generic alternating minimization algorithm using a block-coordinated descent method that can compute optimally the vector Gaussian RDPF. For the extreme case of the perfect-realism vector Gaussian RDPF, that is to say, when the Gaussian distribution of the source and that of the reconstruction are the same, we derive in closed-form solution that reveals a new adaptive reverse-water-filling solution. We corroborate our findings with various simulation studies.

- February 08, 2024 :
**Malcolm Egan**(INRIA, Lyon) — See Video**Title**:**Copulas and Compression****Abstract :**A key property of Gaussian models is the exponential decay of the probability density function. In many applications, large values occur with much higher probabilities, leading to heavy-tailed probability distributions. In this talk, we consider the family of regularly varying distributions, which captures many popular heavy-tailed models such as alpha-stable and student-t distributions. This family also includes shot noise models which can arise as interference models in wireless communications. By exploiting copula models of statistical dependence, we study (distributed) compression of such non-Gaussian data, highlighting the role of extremal dependence measures and their connection to information measures.

- January 26, 2024 :
**Derya Malak**(EURECOM) — See Video**Title : Distributed Computation over Networks****Abstract :**Large-scale distributed computing systems, such as MapReduce, Spark, or distributed deep networks, are critical for parallelizing the execution of computational tasks. Nevertheless, a struggle between computation and communication complexity lies at the heart of distributed computing. There has been recently a substantial effort to address this problem for a class of functions, such as distributed matrix multiplication, distributed gradient coding, linearly separable functions, etc. The optimal cost has been achieved under some constraints, based mainly on ideas of linear separability of the tasks and linear space intersections. Motivated by the same challenge, we propose a novel distributed computing framework where a master seeks to compute an arbitrary function of distributed datasets in an asymptotically lossless manner. Our approach exploits the notion of characteristic graphs, which have been widely utilized by Shannon, Korner, and Witsenhausen to derive the rate lower bounds for computation, and later by Alon-Orlitsky, Orlitsky-Roche, Doshi-Shah-Medard, and Feizi-Medard, to resolve some well-known distributed coding and communication problems, allowing for lowered communication complexity and even for a) correlated data, b) a broad class of functions, and c) well-known topologies. The novelty of our approach lies in accurately capturing the communication-computation cost tradeoff by melding the notions of characteristic graphs and distributed placement, to provide a natural generalization of distributed linear function computation, thus elevating distributed gradient coding and distributed linear transform to the realm of distributed computing of any function.

- January 12, 2024:
**Laurent Decreusefond**(TelecomParis) — See Video**Title**:**Topological Algebra for Wireless Systems****Abstract**: We show how the mathematical theory called topological algebra can be used to assess and optimize the performance of cellular networks.

- December 14, 2023:
**Laurent Clavier**(IMT Atlantique) — See Video**Title**:**Loosing control and interfering**

**Abstract**: In dense networks, with objects that are expected to operate for several years without maintenance, access to resources can no longer be provided in a globally coordinated way. Nor is it possible or efficient to completely orthogonalize communications. Releasing control, however, will mean an increase in interference, at least some of which will have to be treated as noise. We are interested in the statistical properties of this noise. A Gaussian model is not suitable. Instead, we propose heavy-tailed distributions, which are particularly well-suited to representing the rare but highly penalizing cases of strong interference. The dependence between the different elements of an interference vector is then a key characteristic that needs to be modelled, copulas being a possible way forward.

- November 24, 2023:
**Maxime Guillaud**(Inria)**Title**:**Towards Practical Massive Random Access****Abstract**: In multi-user wireless communications, random access departs from the classical set-up of multiple access in the fact that the number and identity of the active transmitters are unknown to the receiver (typically because the sporadic nature of the traffic does not allow for coordination between transmitters). This is a significant departure from the well-understood multiple access schemes (including non-orthogonal multiple access, NOMA). Random access arises e.g. in massive internet-of-things, ultra-reliable and low-latency communications, and non-terrestrial networks applications. This talk will outline why, compared to multiple access, multi-user decoding in random access scenarios is markedly more difficult, and requires to revise some of the basic assumptions that underpin modern multi-user communications systems, such as the pre-existence of synchronization and timing offset compensation, or centralized assignment of pilot sequences. We will focus on massive random access, which explicitly considers the high-contention regime, i.e. the case where the number of simultaneously active transmitters can be very large, and discuss some of the practical waveforms and coding approaches that have been proposed in practice to solve this problem.