2025Activity reportProject-Team‌WIDE

Browser-based​​​‌ fog computing

The dramatic‌ increase in the amount‌​‌ of data being produced​​ and processed by connected​​​‌ devices has led to‌ paradigms that seek to‌​‌ decentralize the traditional cloud​​ model. In 2011 Cisco​​​‌  48 introduced the vision‌ of fog computing that‌​‌ combines the cloud with​​ resources located at the​​​‌ edge of the network‌ and in between. More‌​‌ generally, the term edge​​ computing has been associated​​​‌ with the idea of‌ adding edge-of-the network storage‌​‌ and computation to traditional​​ cloud infrastructures  42.​​​‌

A number of efforts‌ in this directions focus‌​‌ on specific hardware, e.g.​​ fog nodes that are​​​‌ responsible for connected IoT‌ devices  49. However,‌​‌ many of today's applications​​ run within web browsers​​​‌ or mobile phones. In‌ this context, the recent‌​‌ introduction of the WebRTC​​ API, makes it possible​​​‌ for browsers and smartphones‌ to exchange directly between‌​‌ each other, enabling mobile,​​ or browser-based decentralized applications.​​​‌

Maygh  77, for‌ example, uses the WebRTC‌​‌ API to build a​​ decentralized Content Delivery Network​​​‌ that runs solely on‌ web browsers. The fact‌​‌ that the application is​​ hosted completely on a​​​‌ web server and downloaded‌ with enabled websites means‌​‌ that webmasters can adopt​​ the Content Delivery Network​​​‌ (CDN) without requiring users‌ to install any specific‌​‌ software.

For us, the​​ ability of browsers to​​​‌ communicate with each other‌ using the WebRTC paradigm‌​‌ provides a novel playground​​ for new programming models,​​​‌ and for a browser-based‌ fog architecture combining both‌​‌ a centralized, cloud-based part,​​ and a decentralized, browser-supported​​​‌ part.

This model offers‌ tremendous potential by making‌​‌ edge-of-the-network resources available through​​ the interconnection of web-browsers,​​​‌ and offers new opportunities‌ for the protection of‌​‌ the personal data of​​ end users. But consistently​​​‌ engineering browser-based components requires‌ novel tools and methodologies.‌​‌

In particular, WebRTC was​​ primarily designed for exchanging​​​‌ media and data between‌ two browsers in the‌​‌ presence of a coordinating​​ server. Its complex mechanisms​​​‌ for connection establishment make‌ many of the existing‌​‌ peer-to-peer protocols inefficient. To​​ address this challenge, we​​​‌ plan to consider two‌ angles of attack. First,‌​‌ we plan to design​​ novel protocols that take​​​‌ into account the specific‌ requirements set by this‌​‌ new technology. Second, we​​ envisage to investigate variants​​​‌ of the current WebRTC‌ model with cheaper connection-establishment‌​‌ protocols, in order to​​ provide lower delays and​​​‌ bandwidth consumption in large-scale‌ browser-based applications.

We also‌​‌ plan to address the​​ trade-offs associated with hybrid​​​‌ browser-cloud models. For example,‌ when should computation be‌​‌ delegated to browsers and​​ when should it be​​​‌ executed on the cloud‌ in order to maximize‌​‌ the quality of service?​​ Or, how can a​​​‌ decentralized analytics algorithms operating‌ on browser-based data complement‌​‌ or exploit the knowledge​​ built by cloud-based data​​​‌ analytics solutions?

Emergent micro-service‌ deployment and management

Micro-services‌​‌ tend to produce fine-grained​​​‌ applications in which many​ small services interact in​‌ a loosely coupled manner​​ to produce a wide​​​‌ range of services within​ an organization. Individual services​‌ need to evolve independently​​ of each other over​​​‌ time without compromising the​ availability of the overall​‌ application. Lightweight isolation solutions​​ such as containers (Docker,​​​‌ ...), and their associated​ tooling ecosystem (e.g. Google's​‌ Borg  76, Kubernetes​​  46) have emerged​​​‌ to facilitate the deployment​ of large-scale micro-service-based applications,​‌ but only provide preliminary​​ solutions for key concerns​​​‌ in these systems, which​ we would like to​‌ investigate and extend.

Most​​ of today's on-line computer​​​‌ systems are now too​ large to evolve in​‌ monolithic, entirely pre-planned ways.​​ This applies to very​​​‌ large data centres, for​ example, where the placement​‌ of virtual machines to​​ reduce heating and power​​​‌ consumption can no longer​ be treated using top-down​‌ exhaustive optimisation approaches beyond​​ a critical size. This​​​‌ is also true of​ social networking applications, where​‌ different mechanisms—e.g. to spread​​ news notifications, or to​​​‌ recommend new contacts—must be​ adapted to the different​‌ sub-communities present in the​​ system.

To cope with​​​‌ the inherent complexity of​ building complex loosely-coupled distributed​‌ systems while fostering and​​ increasing efficiency, maintainability, and​​​‌ scalability, we plan to​ study how novel programming​‌ techniques based on declarative​​ programming, components and epidemic​​​‌ protocols can help design,​ deploy, and maintain self-adaptive​‌ structures (e.g. placement of​​ VM) and mechanisms (e.g.​​​‌ contact recommendations) that are​ optimized to the local​‌ context of very large​​ distributed systems. To fulfill​​​‌ this vision, we plan​ to explore a three-pronged​‌ strategy to raise the​​ level of programming abstraction​​​‌ offered to developers.

• First,​ we plan to explore​‌ the use of high-level​​ domain-specific languages (DSL) to​​​‌ declare how large-scale topologies​ should be achieved, deployed,​‌ and maintained. Our vision​​ is a declarative approach​​​‌ to describe how to​ combine, deploy and orchestrate​‌ micro-services in an abstract​​ manner thus abstracting away​​​‌ developers from the underlying​ cloud infrastructures, and from​‌ the intricacies involved in​​ writing low-level code to​​​‌ build a large-scale distributed​ application that scales. With​‌ this effort, we plan​​ notably to directly support​​​‌ the twin properties of​ emergence (the adaptation “from​‌ within”) and differentiation (the​​ possibility from parts of​​​‌ the system to diverge​ while still forming a​‌ whole). Our central objective​​ is to search for​​​‌ principled programming constructs to​ support these two capabilities​‌ using a modular and​​ incremental software development approach.​​​‌
• On a second strand​ of work, we plan​‌ to investigate how unikernels​​ enable smaller footprints, more​​​‌ optimization options, and faster​ boot times for micro-services.​‌ Isolating micro-services into VMs​​ is not the most​​​‌ adequate approach as it​ requires the use of​‌ hypervisors, or virtual machine​​ monitors (VMMs), to virtualize​​​‌ hardware resources. VMMs are​ well known to be​‌ heavyweight with both boot​​ and run time overheads​​​‌ that may have a​ strong impact on performances.​‌ Unikernels seem to offer​​ the right balance between​​​‌ performance and flexibility to​ address this challenge. One​‌ of the key underlying​​ challenges is to compile​​ directly the aforementioned provided​​​‌ DSL to a dedicated‌ and customized machine image,‌​‌ ready to be deployed​​ directly on top of​​​‌ a large set of‌ bare metal servers.
• Depending‌​‌ on the workload it​​ is subjected to, and​​​‌ the state of its‌ execution environment (network, VMs),‌​‌ a large-scale distributed application​​ may present erratic or​​​‌ degraded performance that is‌ hard to anticipate and‌​‌ plan for. There is​​ therefore a strong need​​​‌ to adapt dynamically the‌ way resources are allocated‌​‌ to a running application.​​ We would like to​​​‌ study how the DSL‌ approach we envisage can‌​‌ be extended to enable​​ developers to express orchestration​​​‌ algorithms based on machine‌ learning algorithms.

Privacy-preserving decentralized learning

Personalization‌ and recommendation can be‌​‌ seen as a specific​​ case of general machine​​​‌ learning. Production-grade recommenders and‌ personalizers typically centralize and‌​‌ process the available data​​ in one location (a​​​‌ data-center, a cloud service).‌ This is highly problematic,‌​‌ as it endangers the​​ privacy of users, while​​​‌ hampering the analysis of‌ datasets subject to privacy‌​‌ constraints that are held​​ by multiple independent organizations​​​‌ (such as health records).‌ A decentralized approach to‌​‌ machine learning appears as​​ a promising candidate to​​​‌ overcome these weaknesses: if‌ each user or participating‌​‌ organization keeps its data,​​ while only exchanging gradient​​​‌ or model information, privacy‌ leaks seem less likely‌​‌ to occur.

In some​​ cases, decentralized learning may​​​‌ be achieved through relatively‌ simple adaptations of existing‌​‌ centralized models, for instance​​ by defining alternative learning​​​‌ models that may be‌ more easily decentralized. But‌​‌ in all cases, processing​​​‌ growing amounts of information​ calls for high-performance algorithms​‌ and middleware that can​​ handle diverse storage and​​​‌ computation resources, in the​ presence of dynamic and​‌ privacy-sensitive data. To reach​​ this objective, we will​​​‌ therefore leverage our work​ in distributed and privacy-preserving​‌ algorithms and middleware  50​​, 52, 53​​​‌ as well as the​ results of our work​‌ on large-scale hybrid architectures​​ in Objective 1.

Personalization​​​‌ in user-oriented applications

As​ a first application perspective,​‌ we plan to design​​ tools that exploit decentralized​​​‌ analytics to enhance user-centric​ personalized applications. As we​‌ observed above, such applications​​ exhibit an inherent trade-off​​​‌ between personalization quality and​ privacy preservation. The most​‌ obvious goal in this​​ direction consists in designing​​​‌ algorithms that can achieve​ high levels of personalization​‌ while protecting sensitive user​​ information. But an equally​​​‌ important one consists in​ personalizing the trade-off itself​‌ by adapting the quality​​ of the personalization provided​​​‌ to a user to​ his/her willingness to expose​‌ information. This, like other​​ desirable behaviors, appears at​​​‌ odds with the way​ current systems work. For​‌ example, a user of​​ a recommender system that​​​‌ does not reveal his/her​ profile information penalizes other​‌ users causing them to​​ receive less accurate recommendations.​​​‌ We would like to​ mitigate this situation by​‌ means of protocols that​​ reward users for sharing​​​‌ information. On the one​ hand, we plan to​‌ take inspiration from protocols​​ for free-riding avoidance in​​​‌ peer-to-peer systems  54,​ 60. On the​‌ other hand, we will​​ consider blockchains as a​​​‌ tool for tracking and​ rewarding data contributions. Ultimately,​‌ we aim at enabling​​ users to configure the​​​‌ level of privacy and​ personalization they wish to​‌ experience.

Privacy preserving decentralized​​ aggregation

As a second​​​‌ setting we would like​ to consider target applications​‌ running on constrained devices​​ like in the Internet-of-Things​​​‌ (IoT). This setting makes​ it particularly important to​‌ operate on decentralized data​​ in a light-weight privacy-preserving​​​‌ manner, and further highlights​ the synergy between this​‌ objective and Objective 1.​​ For example, we plan​​​‌ to provide data subjects​ with the possibility to​‌ store and manage their​​ data locally on their​​​‌ own devices, without having​ to rely on third-party​‌ managers or aggregators, but​​ possibly storing less private​​​‌ information or results in​ the cloud. Using this​‌ strategy, we intend to​​ design protocols that enable​​​‌ users themselves, or third-party​ companies to query distributed​‌ data in aggregate form,​​ or to run data​​​‌ analytics processes on a​ distributed set of data​‌ repositories, thereby gathering knowledge​​ without violating the privacy​​​‌ of other users. For​ example, we have started​‌ working on the problem​​ of computing an aggregate​​​‌ function over a subset​ of the data in​‌ a distributed setting. This​​ involves two major steps:​​​‌ selection and aggregation. With​ respect to selection, we​‌ envision defining a decentralized​​ data-selection operation that can​​​‌ apply a selection predicate​ without violating privacy constraints.​‌ With respect to aggregation,​​ we will continue our​​​‌ investigation of lightweight protocols​ that can provide privacy​‌ with limited computational complexity​​  43.

Spread maximization​​​‌

Our goal is in‌ particular to study spread‌​‌ maximization in a broader​​ class of diffusion processes​​​‌ than the basic independent‌ cascade (IC) and linear‌​‌ threshold (LT) models of​​ influence  68, 66​​​‌, 67 that have‌ been studied in this‌​‌ context so far. This​​ includes the randomized rumor​​​‌ spreading (RS) model for‌ information dissemination  57,‌​‌ biased versions of the​​​‌ voter model  62 modelling​ influence, and the (graph-based)​‌ Moran processes  70 modelling​​ the spread of mutations.​​​‌ We would like to​ consider several natural versions​‌ of the spread maximization​​ problem, and the relationships​​​‌ between them. For these​ problems we will use​‌ the greedy algorithm and​​ the submodularity-based analytical framework​​​‌ of  68, and​ will also explore new​‌ approaches.

Immunization optimization

Conversely​​ we would also like​​​‌ to explore immunization optimization​ problems. Existing works on​‌ these types of problem​​ assume a perfect-contagion model,​​​‌ i.e., once a node​ gets infected, it deterministically​‌ infects all its non-immunized​​ neighbors. We plan to​​​‌ consider various diffusion processes,​ including the standard susceptible–infected​‌ (SI), susceptible–infected–recovered (SIR) and​​ susceptible–infected–susceptible (SIS) epidemic models,​​​‌ and explore the extent​ to which results and​‌ techniques for the perfect-contagion​​ model carry over to​​​‌ these probabilistic models. We​ will also investigate whether​‌ techniques for spread maximization​​ could be applied to​​​‌ immunization problems.

Some immunization​ problems are known to​‌ be hard to approximate​​ in general graphs, even​​​‌ for the perfect-contagion model,​ e.g., the fixed-budget version​‌ of the fire-fighter problem​​ cannot be approximated to​​​‌ any $n^{1-ϵ}$ factor  45.​‌ This strand of work​​ will consider restricted graph​​​‌ families, such as trees​ or graphs of small​‌ treewidth, for such problems.​​ In addition, for some​​​‌ immunization problems, there is​ a large gap between​‌ the best known approximation​​ algorithm and the best​​​‌ known inaproximability result, and​ we would like to​‌ make progress in reducing​​ these gaps.

Randomized algorithm for mutual​‌ exclusion and coordination

To​​ exploit the power of​​​‌ massive distributed applications and​ systems (such as those​‌ envisaged in Objectives 1​​ and 2) or multiple​​​‌ processors, algorithms must cope​ with the scale and​‌ asynchrony of these systems,​​ and their inherent instability,​​​‌ e.g., due to node,​ link, or processor failures.​‌ Our goal is to​​ explore the power and​​​‌ limits of randomized algorithms​ for large-scale networks of​‌ distributed systems, and for​​ shared memory multi-processor systems,​​​‌ in effect providing fundamental​ building blocks to the​‌ work envisioned in Objectives​​ 1 and 2.

For​​​‌ shared memory systems, randomized​ algorithms have notably proved​‌ extremely useful to deal​​ with asynchrony and failures.​​​‌ Sometimes probabilistic algorithms provide​ the only solution to​‌ a problem; sometimes they​​ are more efficient; sometimes​​ they are simply easier​​​‌ to implement. We plan‌ to devise efficient algorithms‌​‌ for some of the​​ fundamental problems of shared​​​‌ memory computing, such as‌ mutual exclusion, renaming, and‌​‌ consensus.

In particular, looking​​ at the problem of​​​‌ mutual exclusion, it‌ is desirable that mutual‌​‌ exclusion algorithms be abortable​​. This means that​​​‌ a process that is‌ trying to lock the‌​‌ resource can abort its​​ attempt in case it​​​‌ has to wait too‌ long. Abortability is difficult‌​‌ to achieve for mutual​​ exclusion algorithms. We will​​​‌ try to extend our‌ algorithms for the cache-coherent‌​‌ (CC) and the distributed​​ shared memory (DSM) model​​​‌ in order to make‌ them abortable, while maintaining‌​‌ expected constant Remote Memory​​ References (RMRs) complexity, under​​​‌ optimistic system assumptions. In‌ order to achieve this,‌​‌ the algorithm will use​​ strong synchronization primitives, called​​​‌ compare-and-swap objects. As part‌ of our collaboration with‌​‌ the University of Calgary,​​ we will work on​​​‌ implementing those objects from‌ registers in such a‌​‌ way that they also​​ allow aborts. Our goal​​​‌ is to build on‌ existing non-abortable implementations 59‌​‌. We plan then​​ later to use these​​​‌ objects as building blocks‌ in our mutual exclusion‌​‌ algorithm, in order to​​ make them work even​​​‌ if the system does‌ not readily provide such‌​‌ primitives.

We have also​​ started working on blockchains,​​​‌ as these represent a‌ new and interesting trade-off‌​‌ between probabilistic guarantees, scalability,​​ and system dynamics, while​​​‌ revisiting some of the‌ fundamental questions and limitations‌​‌ of consensus in fault-prone​​ asynchronous systems.

Modular theory​​​‌ of distributed computing

Practitioners‌ and engineers have proposed‌​‌ a number of reusable​​ frameworks and services to​​​‌ implement specific distributed services‌ (from Remote Procedure Calls‌​‌ with Java RMI or​​ SOAP-RPC, to JGroups for​​​‌ group communication, and Apache‌ Zookeeper for state machine‌​‌ replication). In spite of​​ the high conceptual and​​​‌ practical interest of such‌ frameworks, many of these‌​‌ efforts lack a sound​​ grounding in distributed computation​​​‌ theory (with the notable‌ exceptions of JGroups and‌​‌ Zookeeper), and often provide​​ punctual and partial solutions​​​‌ for a narrow range‌ of services. We argue‌​‌ that this is because​​ we still lack a​​​‌ generic framework that unifies‌ the large body of‌​‌ fundamental knowledge on distributed​​ computation that has been​​​‌ acquired over the last‌ 40 years.

To overcome‌​‌ this gap we would​​ like to develop a​​​‌ systematic model of distributed‌ computation that organizes the‌​‌ functionalities of a distributed​​ computing system into reusable​​​‌ modular constructs assembled via‌ well-defined mechanisms that maintain‌​‌ sound theoretical guarantees on​​ the resulting system. This​​​‌ research vision arises from‌ the strong belief that‌​‌ distributed computing is now​​ mature enough to resolve​​​‌ the tension between the‌ social needs for distributed‌​‌ computing systems, and the​​ lack of a fundamentally​​​‌ sound and systematic way‌ to realize these systems.‌​‌

To progress on this​​ vision, we plan in​​​‌ the near future to‌ investigate, from a distributed‌​‌ software point of view,​​ the impact due to​​​‌ failures and asynchrony on‌ the layered architecture of‌​‌ distributed computing systems. A​​​‌ first step in this​ direction will address the​‌ notions of message adversaries​​ (introduced a long time​​​‌ ago in  75)​ and process adversaries (investigated​‌ in several papers, e.g.​​  74, 56,​​​‌ 64, 65,​ 69). The aim​‌ of these notions is​​ to consider failures, not​​​‌ as “bad events”, but​ as part of the​‌ normal behavior of a​​ system. As an example,​​​‌ when considering round-based algorithms,​ a message adversary is​‌ a daemon which, at​​ every round, is allowed​​​‌ to suppress some messages.​ The aim is then,​‌ given a problem $\mathrm{P}$, to find the​​​‌ strongest adversary under which​ $P$ can be solved​‌ (“strongest” means here that​​ giving more power to​​​‌ the adversary makes the​ problem impossible to solve).​‌ This work will allow​​ us to progress in​​​‌ terms of general layered​ theory of distributed computing,​‌ and allow us to​​ better map distributed computing​​​‌ models and their relations,​ in the steps of​‌ noticeable early efforts in​​ this direction  74,​​​‌ 41.

Objective​‌ 1: Large-scale Trustless Sybil-Resistant​​ Systems

We plan to​​​‌ contribute to the theoretical​ understanding of Blockchain-based and​‌ Byzantine-tolerant systems by exploring​​ reusable abstractions that can​​​‌ allow programmers to develop​ Byzantine-tolerant applications more easily.​‌ We plan for example​​ to extend existing work​​​‌ on weak consistency to​ a BFT setting, building​‌ for instance on recent​​ proposals on Byzantine Fault-Tolerant​​​‌ CRDTs 63. To​ address scale, we plan​‌ to explore novel scalable​​ Byzantine fault-tolerant algorithms, both​​​‌ in the context of​ closed systems, and then​‌ in the more challenging​​ case of open (aka​​​‌ permissionless) systems. Our line​ of attack is to​‌ focus on lightweight BFT​​ primitives that can enable​​​‌ faster and more resource-efficient​ algorithms 55, 61​‌. In the case​​ of open systems, we​​​‌ will leverage the expertise​ of our team in​‌ theoretical distributed algorithms and​​ randomized algorithms to address​​​‌ Sybil attacks through novel​ countermeasures providing (hopefully) cheaper​‌ and more equitable alternatives​​ to proof-of-work of proof-of-stake​​ algorithms. One open, yet​​​‌ enticing, questions is whether‌ anonymous computing models could‌​‌ provide a path to​​ address this issue. We​​​‌ would also like to‌ investigate how storage can‌​‌ be improved in Blockchains​​ and BFT large-scale systems.​​​‌ Most of these systems‌ are fully replicated, incurring‌​‌ formidable costs (up to​​ 2.6PB of distributed storage​​​‌ in the case of‌ Bitcoin). Coding techniques, that‌​‌ we have used in​​ the past, and adaptable​​​‌ redundancy based on Byzantine‌ quorums 71 are some‌​‌ avenues we would like​​ to explore to address​​​‌ this challenge.

Objective 2:‌ Robustness and Security at‌​‌ Scale

Although WIDE did​​ not focus initially on​​​‌ security issues per se,‌ our historical interest in‌​‌ privacy concerns and Byzantine​​ fault-tolerance has progressively led​​​‌ us to consider a‌ broader range of security‌​‌ properties in distributed and​​ decentralized systems, ranging from​​​‌ anonymity (in anonymity networks,‌ explored in the PhD‌​‌ of Quentin Dufour) to​​ malware protection through large-scale​​​‌ computations.

In terms of‌ malware protection, we would‌​‌ like to harness the​​ power of distribution and​​​‌ collaborative data gathering to‌ help antivirus designers improve‌​‌ and optimize malware detection.​​ We plan in particular​​​‌ to work on the‌ automatic creation of test‌​‌ datasets for antivirus software​​ using automated mutation techniques,​​​‌ building upon our preliminary‌ work in this area.‌​‌ Such a tool is​​ of primary importance in​​​‌ both the academic and‌ industrial fields to be‌​‌ able to quantify the​​ effectiveness of new countermeasures.​​​‌

On the front of‌ privacy, we plan to‌​‌ investigate the design of​​ a distributed digital data​​​‌ vault able to securely‌ store personal data, leveraging‌​‌ our experience on privacy-preserving​​ decentralized systems 43,​​​‌ and on trusted-execution environments‌ (e.g. SGX). We have‌​‌ started collaborating with the​​ CIDRE team at Inria​​​‌ Rennes, with colleagues at‌ KTH (Sweden), and with‌​‌ the company AriadNext (H2020​​ Soteria project) on these​​​‌ topics.

At an infrastructure‌ level, and following the‌​‌ recruitment of Djob Mvondo,​​ we plan to explore​​​‌ how progress in virtualization‌ can help advance the‌​‌ team's agenda in terms​​ of large-scale robustness, in​​​‌ particular in a cloud-computing‌ setting 72, 73‌​‌. Specifically we would​​ like to investigate how​​​‌ novel heterogeneous architectures that‌ embed a range of‌​‌ ASICs and specialized units​​ (GPU, FPGA, SMARTNIC, PIM-devices)​​​‌ can be leveraged to‌ provide more robust and‌​‌ more efficient virtualized services.​​

Objective 3: Fundamentals of​​​‌ distributed randomized algorithms

We‌ plan to continue our‌​‌ theoretical exploration of simple​​ randomized distributed algorithms, where​​​‌ individual entities (nodes or‌ mobile agents) have limited‌​‌ computation and communication power,​​ and are often unreliable.​​​‌ These distributed randomized algorithms‌ are closely related to‌​‌ the mechanisms we plan​​ to explore for Sybil​​​‌ attack protection (Objective 1),‌ and privacy protection (Objective‌​‌ 2).

More concretely, we​​ will investigate three settings:​​​‌ in the first setting,‌ agents perform independent or‌​‌ mildly dependent random walks​​ on a graph, and​​​‌ interact when they meet.‌ In the second (more‌​‌ traditional) setting, the interacting​​ entities are the nodes​​​‌ of graph. Finally, in‌ a third setting, nodes‌​‌ are the computing entities​​​‌ and the goal is​ to modify the graph​‌ edges to achieve certain​​ desirable graph properties (an​​​‌ expander graph 44,​ or a k-nearest neighbor​‌ graph), by means of​​ local decentralized operations (typically​​​‌ adjacent nodes interact by​ exchanging some of their​‌ incident edges). In all​​ three cases, we will​​​‌ strive to derive time-​ and space- optimal algorithms,​‌ with strong robustness guarantees.​​

7.1.1​ DecentralizedFlower

• Name:
  DecentralizedFlower
• Keyword:​‌
  Decentralized Learning
• Functional Description:​​
  DecentralizedFlower is a framework​​​‌ to test decentralized machine​ learning algorithms in a​‌ cluster environment, in a​​ production environment, and in​​​‌ a combination of the​ two. The framework enables​‌ developers to test algorithms​​ on a testing environment​​​‌ and then seamlessly deploy​ them into a production​‌ setting. The software is​​ based on the Flower​​​‌ federated-learning library developed by​ the University of Cambridge​‌ and the German Company​​ Adap.
• Contact:
  Davide Frey​​​‌

7.1.2 nodemanager

• Keywords:
  Peer-to-peer,​ Peer-sampling, Distributed, Distributed Applications​‌
• Functional Description:
  Nodemanager is​​ a solution for setting​​​‌ up peer-to-peer applications. It​ is essentially written in​‌ Rust, but provides interfaces​​ for use in Python.​​​‌
• Contact:
  Davide Frey

7.1.3​ DecentralizedDeclearn

• Keyword:
  Decentralized Learning​‌
• Functional Description:
  DecentralizedDeclearn is​​ a Python library for​​ testing decentralized machine learning​​​‌ algorithms. This library provides‌ developers with a simulation‌​‌ framework for testing their​​ applications before deploying them.​​​‌ This library is based‌ on the Declearn library,‌​‌ which is a Python​​ package providing a framework​​​‌ for federated learning. It‌ was developed by the‌​‌ Magnet team at Inria.​​
• Contact:
  Davide Frey

7.1.4​​​‌ decentralised-data-wallet

• Name:
  SOTERIA Data‌ Wallet Prototype
• Keywords:
  Privacy,‌​‌ Data management, Data analytics,​​ Distributed systems, Cryptography, Decentralized​​​‌ Learning
• Functional Description:

  data-wallet-prototype‌ is a Rust library‌​‌ that provides the basic​​ functionality of a digital​​​‌ data wallet, with the‌ particular constraint that distributed‌​‌ computations and interaction with​​ outside parties occur in​​​‌ a fully decentralised manner.‌ This contrasts with existing‌​‌ solutions that rely on​​ centralised systems, such as​​​‌ cloud providers, which are‌ typically used to store‌​‌ encrypted personal information and​​ carry out computations.

  In​​​‌ short, this allows: -‌ Users to securely store‌​‌ their own personal data​​ on their own devices​​​‌ without relying on external‌ service providers (using suitable‌​‌ encryption and hardware security​​ measures) - Third parties,​​​‌ such as research bodies,‌ to perform computations across‌​‌ a network of digital​​ wallets in a decentralized​​​‌ manner without compromising user‌ privacy (using protocols from‌​‌ the literature designed to​​ make this possible).

  This​​​‌ software is designed to‌ be flexible with respect‌​‌ to the hardware limitations​​ of its environment, enabling​​​‌ it to run on‌ a range of personal‌​‌ devices, including Android systems.​​

  This was funded as​​​‌ part of the SOTERIA‌ Digital Security and Privacy‌​‌ project.

• Contact:
  Davide Frey​​

7.1.5 CAC

• Name:
  Context​​​‌ Adaptative Cooperation
• Keyword:
  Distributed‌ systems
• Functional Description:
  Context-Adaptive‌​‌ Cooperation (CAC) is a​​ novel copperation abstraction that​​​‌ allows an arbitrary set‌ of processes to propose‌​‌ values while multiple value​​ acceptances are triggered. Furthermore,​​​‌ each acceptance comes with‌ information about other acceptances‌​‌ that can possibly occur.​​ This code simulates an​​​‌ instance of CAC. Let‌ n be the number‌​‌ of processes, t the​​ number of Byzantine processes​​​‌ whose behaviour may diverge‌ from the initial one.‌​‌ Let m be the​​ total number of proposals​​​‌ made by m different‌ processes. At the end‌​‌ of the simulation, each​​ of the n-t non-Byzantine​​​‌ processes will have accepted‌ one or more of‌​‌ the initial proposals. But​​ if we look at​​​‌ the intersection of all‌ these proposals, only one‌​‌ remains.
• Contact:
  Davide Frey​​

7.1.6 QAAT

• Name:
  Quasi-Anonymous​​​‌ Asset Transfer
• Keywords:
  Asset‌ transfer, Distributed computing
• Functional‌​‌ Description:
  QAAT is the​​ first asset transfer system​​​‌ that achieves anonymity, and‌ consensus-freedom while incurring as-low-as-possible‌​‌ storage and communication costs.​​ QAAT provides the following​​​‌ three properties. - Quasi-anonymity:‌ QAAT hides the amount‌​‌ and the receiver's identity​​ of every asset transfer.​​​‌ - Lightness: QAAT uses‌ only succinct cryptographic schemes,‌​‌ i.e. with at most​​ polylogarithmic proof size and​​​‌ verification time. Moreover, the‌ storage cost incurred by‌​‌ each process is linear​​ in its number of​​​‌ transfers for a fixed‌ security parameter, and the‌​‌ associated communication cost remains​​ as low as possible.​​​‌ - Consensus-Freedom: QAAT is‌ a deterministic algorithm that‌​‌ can operate in an​​​‌ asynchronous setting prone to​ failures, thereby supporting responsive​‌ applications. This software artifact,​​ currently under development, provides​​​‌ the first working implementation​ of the QAAT algorithm​‌
• Contact:
  Davide Frey

7.1.7​​ Splitchain

• Name:
  Splitchain Protocol​​​‌
• Keywords:
  Blockchain, Rust, Distributed​ systems
• Functional Description:
  This​‌ software is a node​​ in the distributed Splitchain​​​‌ system. It allows to​ join the system, submit​‌ new transactions, participate in​​ consensus to create new​​​‌ blocks, and manages automatically​ the split and merge​‌ of the shards, and​​ the routing of data.​​​‌
• Contact:
  Davide Frey

7.1.8​ oversim-ipfs

• Name:
  OverSim for​‌ the InterPlanetary File System​​ (IPFS)
• Keywords:
  C++, Distributed​​​‌ systems, Distributed Storage Systems,​ IPFS, Network simulator
• Functional​‌ Description:

  A fork of​​ the event-driven simulation framework​​​‌ OverSim (2007, Baumgart et​ al)*.

  This project aims​‌ to provide a realistic​​ model of IPFS nodes​​​‌ interacting over a network.​ This includes generating file​‌ chunks, publishing provider records,​​ querying the overlay and​​​‌ providing methods to analyse​ file availability over time.​‌

  It also aims to​​ study a novel re-publishment​​​‌ algorithm under development in​ conjunction with the COAST​‌ team and Hivenet.

  *​​ http://www.oversim.org/

• Contact:
  Davide Frey​​​‌
• Partner:
  Hivenet

7.1.9 PPFDIMS​

• Name:
  Privacy-Preserving and fully-Distributed​‌ Identity Management System
• Keywords:​​
  Distributed systems, Privacy, Identity​​​‌ management
• Functional Description:

  The​ system consists of certificate​‌ issuers, verifiers, and users.​​

  - A user requests​​​‌ a certificate to an​ issuer in order to​‌ certify personal data (identity,​​ age, education level, medical​​​‌ information, etc.). - The​ issuer sends the user​‌ a certificate, proving that​​ the user's personal data​​​‌ is true. - The​ user sends their certificate​‌ to a verifier to​​ access a service (authorisation,​​​‌ purchase, entry, etc.). -​ The verifier grants or​‌ denies the user access​​ to the service based​​​‌ on the certificate's validity.​ Verifiers and issuers perform​‌ these operations without knowing​​ each other's identities, thus​​​‌ preserving everyone's anonymity. Verifiers​ and issuers can also​‌ manage certificate revocation via​​ a pseudo-consensus mechanism not​​​‌ based on a blockchain.​

• URL:
  https://­gitlab.­inria.­fr/­WIDE/­dims/
• Contact:
  Davide​‌ Frey

8.1.1 Contention-Aware Cooperation​

Participants: Michel Raynal,​‌ Davide Frey, François​​ Taïani.

As shown​​​‌ by Reliable Broadcast and​ Consensus, cooperation among a​‌ set of independent computing​​ entities (sequential processes) is​​​‌ crucial in fault-tolerant distributed​ computing. Considering $n$-process​‌ asynchronous message-passing systems where​​ some processes may be​​​‌ Byzantine, this work 21​ introduces a novel cooperation​‌ abstraction, Contention-Aware Cooperation (CAC).​​ While Reliable Broadcast is​​​‌ a one-to-$n$ cooperation​ abstraction and Consensus is​‌ an $n$-to-$\mathrm{n}$ cooperation abstraction, CAC is​​​‌ a $d$-to-$\mathrm{n}$ cooperation abstraction where $\mathrm{d‌}$ ($1\le \mathrm{d}\le n$) varies​​​‌ with each run and​ remains unknown to the​‌ processes. Correct processes accept​​ the same set of​​​‌ $\ell$ pairs $\langle \mathrm{v},i\rangle$ (​‌$v$ is the value​​ proposed by $p_{\mathrm{i‌}}$) from the $\mathrm{d}$ proposer processes, where $\mathrm{1‌}\le \ell \le \mathrm{d}$ and (as $d$)​​​‌ $\ell$ remains unknown to​ the processes (except in​‌ specific cases). Those $\mathrm{\ell }$ values are accepted one​​ at a time, potentially​​​‌ in different orders at‌ each process. In addition,‌​‌ CAC provides each process​​ with an imperfect oracle​​​‌ that provides insights into‌ the values that they‌​‌ may accept in the​​ future. Interestingly, the CAC​​​‌ abstraction is particularly efficient‌ in favorable circumstances, when‌​‌ the oracle becomes accurate,​​ which processes can detect.​​​‌ To illustrate its practical‌ utility, the work details‌​‌ two applications leveraging CAC:​​ a fast consensus implementation​​​‌ optimized for low contention‌ (named Cascading Consensus), and‌​‌ a novel naming problem​​ that can be solved​​​‌ under full asynchrony. All‌ algorithms presented require signatures.‌​‌

8.1.2 Ethical Risk Analysis​​ of L2 Rollups

Participants:​​​‌ Davide Frey, François‌ Taïani.

Layer 2‌​‌ rollups improve throughput and​​ fees, but can reintroduce​​​‌ risk through operator discretion‌ and information asymmetry. In‌​‌ this work 18,​​ we ask which operator​​​‌ and governance designs produce‌ ethically problematic user risk.‌​‌ We adapt Ethical Risk​​ Analysis to rollup architectures,​​​‌ build a role-based taxonomy‌ of decision authority and‌​‌ exposure, and pair the​​ framework with two empirical​​​‌ signals, a cross sectional‌ snapshot of 129 projects‌​‌ from L2BEAT and a​​ hand curated incident set​​​‌ covering 2022 to 2025.‌ We analyze mechanisms that‌​‌ affect risks to users’​​ funds, including upgrade timing​​​‌ and exit windows, proposer‌ liveness and whitelisting, forced‌​‌ inclusion usability, and data​​ availability choices. We find​​​‌ that ethical hazards rooted‌ in L2 components control‌​‌ arrangements are widespread: instant​​ upgrades without exit windows​​​‌ appear in about 86‌ percent of projects, and‌​‌ proposer controls that can​​ freeze withdrawals in about​​​‌ 50 percent. Reported incidents‌ concentrate in sequencer liveness‌​‌ and inclusion, consistent with​​ these dependencies. We translate​​​‌ these findings into ethically‌ grounded suggestions on mitigation‌​‌ strategies including technical components​​ and governance mechanisms.

8.1.3​​​‌ Communication abstractions in systems‌ prone to malicious attacks‌​‌

Participants: Achour Mostefaoui.​​

Abstractions play a central​​​‌ role in distributed computing,‌ as they capture essential‌​‌ synchronization properties. Equivalence results​​ among abstractions clarify their​​​‌ relative computational power. While‌ many such equivalences are‌​‌ well established in crash-prone​​ systems, far less is​​​‌ known about Byzantine-prone environments,‌ where faulty processes may‌​‌ behave arbitrarily.

This year,​​ we revisited the equivalence​​​‌ landscape in Byzantine systems,‌ with a particular focus‌​‌ on communications between processes,​​ namely, shared registers and​​​‌ broadcast abstractions. We establish‌ three new reductions. Specifically,‌​‌ we prove that the​​ broadcast abstraction called Byzantine​​​‌ Set-Constrained Delivery Broadcast (BSCD-Broadcast)‌ can implement a Snapshot/Append‌​‌ object and vice versa​​ (i.e., a two-way reduction),​​​‌ and that the FIFO‌ variant of the renowned‌​‌ Byzantine Reliable Broadcast (BRB-Broadcast)​​ can implement BSCD-Broadcast under​​​‌ a majority of correct‌ processes. Interestingly, this assumption‌​‌ mirrors the one required​​ in crash-prone systems for​​​‌ a similar transformation.

On‌ antoher axis, we worked‌​‌ on the construction of​​ a privacy-preserving single-writer multi-reader​​​‌ (SWMR) atomic register in‌ a Byzantine-prone distributed model.‌​‌ Specifically, we consider a​​ closed model, in which​​​‌ one process can write‌ values in the register‌​‌ and only a subset​​ of the other processes​​​‌ are allowed to read‌ the value. The aim‌​‌ is to ensure that​​​‌ processes that do not​ have the requisite reading​‌ right are unable to​​ read the content of​​​‌ the register, even when​ they are Byzantine. This​‌ makes the content of​​ the register private. We​​​‌ ensure this privacy by​ encoding the value written​‌ by the writer, using​​ secret sharing, into multiple​​​‌ shards and disseminating them​ among the participating reader​‌ processes. The technical challenge​​ is then to organize​​​‌ the coordination between the​ correct reading processes to​‌ achieve Byzantine linearizability, without​​ knowing the content of​​​‌ the register. The main​ contribution of this work​‌ is a linearizable read-write​​ (R/W) privacy-preserving register for​​​‌ $t<\frac{n}{\mathrm{7}}$, where $t$ is​‌ the number of Byzantine​​ processes and $n$ denotes​​​‌ the total number of​ processes in the system.​‌

8.1.4 Discreet: Distributed delivery​​ service with context-aware cooperation​​​‌

Participants: Davide Frey.​

End-to-end encrypted messaging applications​‌ such as Signal became​​ widely popular thanks to​​​‌ their capability to ensure​ the confidentiality and integrity​‌ of online communication. While​​ the highest security guarantees​​​‌ were long reserved to​ two-party communication, solutions for​‌ n-party communication remained either​​ inefficient or less secure​​​‌ until the standardization of​ the MLS Protocol (Messaging​‌ Layer Security). This new​​ protocol offers an efficient​​​‌ way to provide end-to-end​ secure communication with the​‌ same guarantees originally offered​​ by the Signal Protocol​​​‌ for two-party communication. However,​ both solutions still rely​‌ on a centralized component​​ for message delivery, called​​​‌ the Delivery Service in​ the MLS Protocol. The​‌ centralization of the Delivery​​ Service makes it an​​​‌ ideal target for attackers​ and threatens the availability​‌ of any protocol relying​​ on MLS. In order​​​‌ to overcome this issue,​ we proposed DiSCreet (Distributed​‌ delIvery Service with Context-awaRE​​ coopEraTion), a design that​​​‌ allows clients to exchange​ protocol messages efficiently and​‌ without any intermediary. It​​ uses a Probabilistic Reliable-Broadcast​​​‌ mechanism to efficiently deliver​ messages and the Cascade​‌ Consensus Protocol to handle​​ messages requiring an agreement.​​​‌ Our solution strengthens the​ availability of the MLS​‌ Protocol without compromising its​​ security. We compare the​​​‌ theoretical performance of DiSCreet​ with another distributed solution,​‌ the DCGKA protocol, and​​ detail the implementation of​​​‌ our solution. We published​ this work in the​‌ Annals of Telecommunications 19​​. This work was​​​‌ done in the context​ of the Alvearium Inria​‌ Challenge and involved a​​ collaboration with Ludovic Paillat​​​‌ and Amine Ismail from​ Hive Computing as well​‌ as with Claudia Lavigna​​ Ignat from the LORELEY​​​‌ Inria team and Mathieu​ Turuani from the PESTO​‌ Inria team.

8.1.5 Luby's​​ MIS algorithms made self-stabilizing​​​‌

Participants: George Giakkoupis.​

In 17, we​‌ reconsider two well-known distributed​​ randomized algorithms computing a​​​‌ maximal independent set, proposed​ in the seminal work​‌ of Luby (1986). We​​ enhance these algorithms such​​​‌ that they become self-stabilizing​ without sacrificing their run-time,​‌ i.e., both stabilize in​​ $O(log\mathrm{n‌})$ synchronous rounds with​ high probability on any​‌ $n$-node graph. The​​ first algorithm gets along​​​‌ with three states, but​ needs to know an​‌ upper bound on the​​ maximum degree. The second​​ does not need any​​​‌ information about the graph,‌ but uses a number‌​‌ of states that is​​ linear in the node​​​‌ degree. Both algorithms use‌ messages of logarithmic size.‌​‌

This work was done​​ in collaboration with Volker​​​‌ Turau (Hamburg University of‌ Technology), and Isabella Ziccardi‌​‌ (IRIF, Paris).

8.1.6 On​​ the h-majority dynamics with​​​‌ many opinions

Participants: George‌ Giakkoupis.

In 34‌​‌, we present the​​ first upper bound on​​​‌ the convergence time to‌ consensus of the well-known‌​‌ $h$-majority dynamics with​​ $k$ opinions, in the​​​‌ synchronous setting, for $\mathrm{h‌}$ and $k$ that are‌​‌ both non-constant values. We​​ suppose that, at the​​​‌ beginning of the process,‌ there is some initial‌​‌ additive bias towards some​​ plurality opinion, that is,​​​‌ there is an opinion‌ that is supported by‌​‌ $x$ nodes while any​​ other opinion is supported​​​‌ by strictly fewer nodes.‌ We prove that, with‌​‌ high probability, if the​​ bias is $\omega (‌sqr\mathrm{t‌}x)$ and the‌​‌ initial plurality opinion is​​ supported by at least​​​‌ $x=\omega (‌logn)$ nodes,‌​‌ then the process converges​​ to plurality consensus in​​​‌ $O(log\mathrm{n‌})$rounds whenever $\mathrm{h‌‌}=\omega (\mathrm{n}logn/\mathrm{x‌})$. A main‌ corollary is the following:‌​‌ if $k=\mathrm{o}(n/log‌n)$ and the‌ process starts from an‌​‌ almost-balanced configuration with an​​ initial bias of magnitude​​​‌ $\omega \left(\sqrt{n/‌k}\right)$ towards the‌​‌ initial plurality opinion, then​​ any function $h=‌\omega (klog‌n)$ suffices to‌​‌ guarantee convergence to consensus​​ in $Ol\mathrm{o‌}gn)$ rounds,‌ with high probability. Our‌​‌ upper bound shows that​​ the lower bound of​​​‌ $\Omega (k/‌h^2)$ rounds‌​‌ to reach consensus by​​ Becchetti et al. (2017)​​​‌ cannot be pushed further‌ than $\tilde{\Omega }(‌‌k/h)$. Moreover, the bias​​​‌ we require is asymptotically‌ smaller than the $\mathrm{\Omega ‌‌}\left(\sqrt{nlog\mathrm{n}}\right)$ bias that guarantees​​​‌ plurality consensus in the‌ 3 -majority dynamics: in‌​‌ our case, the required​​ bias is at most​​​‌ any (arbitrarily small) function‌ in $\omega \left(\sqrt{\mathrm{x‌‌}}\right)$ for any $\mathrm{k}\ge 2$.

This​​​‌ work was done in‌ collaboration with Francesco d'Amore‌​‌ (Gran Sasso Science Institute,​​ Italy) , Niccolò d'Archivio​​​‌ (COATI Inria team, Sophia‌ Antipolis), and Emanuele Natale‌​‌ (CNRS, COATI Inria team,​​ Sophia Antipolis).

8.2.1‌ Off-the shelf network traffic‌​‌ analysis

Participants: Barbe Mvondo​​ Djob, Yerom David​​​‌ Bromberg.

Offloading malware‌ detection to large scale‌​‌ platforms poses serious threats​​ in terms on information​​​‌ flow control for mobile‌ devices. Especially for stalkerwares‌​‌ that require the analysis​​ of users network activity.​​​‌ We show in 30‌ that an on-device approach‌​‌ running at the kernel-level​​ can achieve the same​​​‌ level of protection without‌ leaking any user network‌​‌ traffic data and with​​ a minimal overhead on​​​‌ applications performance. Furthermore, our‌ approach is more secure‌​‌ and performant than on-device​​​‌ VPNs which were the​ standard approaches to dealing​‌ with these issues.

8.2.2​​ Towards efficient kernel network​​​‌ processing for VPNs

Participants:​ Honore Cesaire Mounah,​‌ Barbe Mvondo Djob,​​ Yerom David Bromberg.​​​‌

VPNs are software programs​ essential for both several​‌ use cases such as​​ securing remote accesses or​​​‌ providing privacy online. However,​ at scale, in our​‌ work 29, we​​ uncover that VPNs suffer​​​‌ from several bottlenecks due​ to their underlying designs.​‌ Concretely, their performance can​​ drop by more than​​​‌ 65% when hitting 80%​ of the network capacity​‌ of the network card,​​ with different behaviours in​​​‌ terms of input network​ traffic. We propose several​‌ design changes leveraging existing​​ routines in the Linux​​​‌ kernel, to improve performance​ and reach optimal performance.​‌ Concretely, the best open-source​​ VPNs, Wireguard, benefits from​​​‌ our design almost reaching​ the full duplex capacity​‌ of the underlying network​​ card capacity, independently of​​​‌ the input traffic variations.​

This work was done​‌ in collaboration with Julia​​ Lawall from Inria Paris​​​‌ (Whisper team).

8.2.3 Efficient​ Load balancing for multi-tier​‌ applications

Participants: Brice Ekane​​ Apah, Barbe Mvondo​​​‌ Djob, Yerom David​ Bromberg.

We introduce​‌ DISC 25, a​​ system that tackles the​​​‌ backpressure problem in multi-tier​ and microservice applications, where​‌ large response payloads (often​​ produced by backend services​​​‌ like databases) are redundantly​ relayed through intermediate and​‌ frontend tiers, hurting scalability.​​ DISC lets multiple tiers​​​‌ safely share the same​ TCP connection so that​‌ final response data can​​ bypass unnecessary tiers, while​​​‌ lightweight metadata (headers/footers) still​ follows the normal path.​‌ Unlike prior approaches, DISC​​ works with arbitrary multi-tier​​​‌ depths, heterogeneous protocols (e.g.,​ HTTP, IMAP), and TLS,​‌ and requires only modest,​​ localized application changes. Evaluations​​​‌ on both classic benchmarks​ and modern microservices show​‌ substantial gains: up to​​ 41.5% lower cumulative CPU​​​‌ usage, up to 45%​ higher throughput, and dramatic​‌ tail-latency reductions (up to​​ 5.7×), effectively restoring performance​​​‌ independence between tiers and​ making systems scale where​‌ it actually matters—at the​​ backend.

This work was​​​‌ done in collaboration with​ Alain Tchana and Renaud​‌ Lachaize from the University​​ of Grenoble Alpes (Krakos​​​‌ team), and Daniel Hagimont​ from University of Toulouse​‌ III (Sepia team).

8.2.4​​ Containers as alternatives for​​​‌ Cloud gaming

Participants: Adrien​ Gegout, Barbe Mvondo​‌ Djob, Davide Frey​​.

VMs are the​​​‌ standard isolation environments for​ large scale cloud gaming​‌ infrastructures. In 27,​​ we explore an alternative,​​​‌ containers. Concretely, we show​ why VMs are rigid​‌ for the fast changing​​ and flexible cloud gaming​​​‌ environments and perform early​ evaluations to show that​‌ containers can not only​​ be on par in​​​‌ terms of gaming experience​ but can lead to​‌ energy savings opportunities. We​​ present an early design​​​‌ showing how VMs can​ be replaced and discussed​‌ several potential issues especially​​ regarding GPU sharing.

This​​​‌ work was done with​ Pascal Manchon from Blacknut.​‌

8.2.5 Disconnecting Users from​​ Virtual Worlds with a​​​‌ Single Packet: an Unreal​ Untold Story

Participants: Hugo​‌ Bertin, Yerom David​​ Bromberg.

Online worlds,​​ with online games at​​​‌ the forefront, have become‌ ubiquitous in our lives.‌​‌ In 2024, the gaming​​ industry's worldwide revenue was​​​‌ estimated at US$455 billion.‌ Yet, this industry is‌​‌ facing a growing number​​ of cheating actors and​​​‌ techniques. In this paper‌ 22, we introduce‌​‌ new attacks targeting multiplayer​​ games based on Unreal​​​‌ Engine (UE), such as‌ Fortnite, PUBG, and Valorant.‌​‌ These attacks disconnect players​​ from ongoing game sessions​​​‌ against their will. Cheaters‌ can launch them as‌​‌ a Denial-of-Service against opponents​​ with very few packets​​​‌ (sometimes only one) to‌ steal the victory from‌​‌ the target without exposing​​ themselves as fraudsters. This​​​‌ paper shows how such‌ issues present in a‌​‌ single game engine can​​ spread widely across several​​​‌ games produced by different‌ editors, focusing on Unreal‌​‌ Engine, whose source code​​ is publicly available. UE​​​‌ is also commonly found‌ in digital twins, virtual‌​‌ reality, and other Metaverse​​ solutions. We present our​​​‌ analysis of the design‌ and implementation choices made‌​‌ within Unreal Engine. We​​ cover how to exploit​​​‌ UE networking protocols and‌ discuss how to defeat‌​‌ some common countermeasures used​​ on the Internet against​​​‌ IP spoofing. We propose‌ some mitigation strategies for‌​‌ video game developers.

This​​ work was done in​​​‌ collaboration with Ilies Benhabbour‌ (KAUST) and Marc Dacier‌​‌ (KAUST).

8.3.1​​​‌ Strengthening malware analysis against‌ obfuscation and packing

Participants:‌​‌ Victoire Nganfang, Barbe​​ Mvondo Djob, Yerom​​​‌ David Bromberg.

Existing‌ malware detection techniques fall‌​‌ short against obfuscation and​​ packing techniques. In this​​​‌ work 32, we‌ introduce a new vision-based‌​‌ malware detector designed to​​ remain effective and robust​​​‌ against the aforementioned techniques.‌ Concretely, we introduce DroidHunter‌​‌ that disassembles apps to​​ extract low-level Smali instructions,​​​‌ encodes each instruction as‌ a single RGB pixel,‌​‌ and produces semantic-rich images​​ that preserve opcode and​​​‌ operand information. Our large‌ scale evaluation on approximately‌​‌ 500K APKs, including obfuscated​​ malwares, show up to​​​‌ 99.9% detection accuracy, and‌ outclass nine state-of-the-art detectors.‌​‌ Furthermore, our work achieves​​ stronger resistance to concept​​​‌ drift which is essential‌ for robustness over time.‌​‌

This work was done​​ with Simon Queyrut, Valerio​​​‌ Schiovani (University of Neuchatel),‌ and Vianney Kengne Tchendji‌​‌ (University of Dschang).

8.3.2​​ Low-Cost Privacy-Preserving Decentralized Learning​​​‌

Participants: Dimitri Lereverend,‌ Davide Frey, François‌​‌ Taiani.

Decentralized learning​​ (DL) is an emerging​​​‌ paradigm of collaborative machine‌ learning that enables nodes‌​‌ in a network to​​ train models collectively without​​​‌ sharing their raw data‌ or relying on a‌​‌ central server. In this​​ work, we introduced Zip-DL​​​‌ 23, a privacy-aware‌ DL algorithm that leverages‌​‌ correlated noise to achieve​​ robust privacy against local​​​‌ adversaries while ensuring efficient‌ convergence at low communication‌​‌ costs. By progressively neutralizing​​ the noise added during​​​‌ distributed averaging, Zip-DL combines‌ strong privacy guarantees with‌​‌ high model accuracy. Its​​ design requires only one​​​‌ communication round per gradient‌ descent iteration, significantly reducing‌​‌ communication overhead compared to​​ competitors. Our work established​​​‌ theoretical bounds on both‌ convergence speed and privacy‌​‌ guarantees. Moreover, it demonstrated​​​‌ Zip-DL's practical applicability with​ extensive experiments that show​‌ it outperforms state-of-the-art methods​​ in the accuracy vs.​​​‌ vulnerability trade-off. Specifically, Zip-DL​ (i) reduces membership-inference attack​‌ success rates by up​​ to 35% compared to​​​‌ baseline DL, (ii) decreases​ attack efficacy by up​‌ to 13% compared to​​ competitors offering similar utility,​​​‌ and (iii) achieves up​ to 59% higher accuracy​‌ to completely nullify a​​ basic attack scenario, compared​​​‌ to a state-of-the-art privacy-preserving​ approach under the same​‌ threat model. These results​​ position Zip-DL as a​​​‌ practical and efficient solution​ for privacy-preserving decentralized learning​‌ in real-world applications.

This​​ work was done in​​​‌ collaboration with Romaric Gaudel​ from the MALTE Inria​‌ team, and with Sayan​​ Biswas, Anne-Marie Kermarrec, Rafael​​​‌ Pires, and Rishi Sharma​ from EPFL, Lausanne.

8.4.1 An​​ asymptotically optimal algorithm for​​​‌ generating bin cardinalities

Participants:​ Dimitrios Los.

In​‌ the balls-into-bins setting, $\mathrm{n}$ balls are thrown uniformly​​​‌ at random into $\mathrm{n}$ bins. The naïve way​‌ to generate the final​​ load vector takes $\mathrm{\Theta ‌}\left(n\right)$ time.​ However, it is well-known​‌ that this load vector​​ has with high probability​​​‌ bin cardinalities of size​ $\Theta (log\mathrm{n‌}/loglog\mathrm{n})$. In 15​​​‌, we present an​ algorithm in the RAM​‌ model that generates the​​ bin cardinalities of the​​​‌ final load vector in​ the optimal $O(‌logn/loglogn)$ time​​​‌ in expectation and with​ high probability. Further, the​‌ algorithm that we present​​ is still optimal for​​​‌ any $m\in [n,nlog‌n]$ balls and​​ can also be used​​​‌ as a building block​ to efficiently simulate more​‌ involved load balancing algorithms.​​ In particular, for the​​​‌ Two-Choice algorithm, which samples​ two bins in each​‌ step and allocates to​​ the least-loaded of the​​​‌ two, we obtain roughly​ a quadratic speed-up over​‌ the naïve simulation.

This​​ work was done in​​​‌ collaboration with Luc Devroye​ (McGill University, Canada).

9.1.1​ CIFRE with Broadpeak

Participants:​‌ Yerom David Bromberg,​​ Barbe Mvondo Djob,​​​‌ Alexandre Duvivier.

The​ goal of this thesis​‌ is to design and​​ implement mechanisms that improve​​​‌ the performance of cache​ servers and, consequently, improving​‌ services that rely on​​ the latter, such as​​​‌ streaming services provided by​ BroadPeak. This thesis is​‌ supervised by Yerom David​​ Bromberg, Barbe Mvondo Djob,​​​‌ and Nicolas Le Scouarnec​ (Broadpeak). The currently deployed​‌ systems at Broadpeak achieve​​ up to 60Gbps and​​​‌ can even reach 150Gbps​ regarding network throughput. The​‌ goal is to achieve​​ 400Gbps on the existing​​​‌ hardware with novel software​ designs while reducing energy​‌ consumption. The thesis will​​ explore ideas that revolve​​​‌ around improving the interaction​ of user-space applications with​‌ kernel network stack subsystems.​​

9.1.2 CIFRE with Blacknut:​​​‌ Efficient Containerized Cloud-Gaming Platforms​

Participants: Davide Frey,​‌ Barbe Mvondo Djob,​​ Adrien Gegout.

Cloud​​​‌ gaming enables users without​ high-end consoles or computers​‌ to play video games​​ online on any device​​ with a compatible Internet​​​‌ connection. Users send their‌ commands via a gamepad‌​‌ to a remote server,​​ which applies them and​​​‌ transmits a video stream‌ with game images. Although‌​‌ this paradigm requires few​​ resources on the part​​​‌ of users, it generates‌ a high consumption of‌​‌ resources and energy in​​ the cloud to provide​​​‌ a good quality of‌ service to users with‌​‌ games that perform well,​​ even at start-up. This​​​‌ thesis, supervised by Davide‌ Frey, Barbe Mvondo Djob,‌​‌ Pascal Manchon (Blacknut), and​​ Eric L’Hostis (Blacknut) aims​​​‌ to reduce this resource‌ consumption while improving performance‌​‌ as perceived by users.​​ In particular, we aim​​​‌ on the one hand‌ to enable games to‌​‌ run on containers instead​​ of virtual machines as​​​‌ they do today, and‌ on the other, to‌​‌ predict user demands by​​ pre-allocating resources where it​​​‌ is really useful and‌ necessary.

9.1.3 Flexible Virtualization‌​‌ for Processing In Memory​​ YUMPIM (ANR - PRCE)​​​‌

Participants: Brice Ekane Apah‌, Yerom David Bromberg‌​‌.

Processing In-Memory (PIM)​​ follows near-data processing principles​​​‌ by moving computation closer‌ to where data resides.‌​‌ This project focuses on​​ UPMEM because it is​​​‌ currently the only commercial‌ PIM technology that can‌​‌ be deployed on off-the-shelf​​ servers using standard DDR4/DDR5​​​‌ memory protocols, it comes‌ with an SDK that‌​‌ eases PIM application development,​​ and it is widely​​​‌ studied in recent research.‌ Prior work shows UPMEM‌​‌ can be a viable​​ alternative to CPUs and​​​‌ GPUs across various workloads‌ (e.g., machine learning, bioinformatics,‌​‌ cryptography), offering high flexibility​​ thanks to its general-purpose​​​‌ compute cores and often‌ delivering better energy efficiency‌​‌ when it outperforms CPUs/GPUs​​ (up to 23.2x vs​​​‌ CPUs and 2.54x on‌ average vs GPUs).

The‌​‌ core objective of the​​ project is UPMEM virtualization​​​‌ for cloud environments. Since‌ virtualization is fundamental to‌​‌ cloud computing for resource​​ sharing and efficient utilization,​​​‌ enabling broad PIM adoption‌ requires supporting time-sharing on‌​‌ UPMEM so that dynamically​​ arriving jobs from uncoordinated​​​‌ users can run. Achieving‌ such time-sharing remains challenging‌​‌ on UPMEM and other​​ current PIM technologies.

In​​​‌ collaboration with Alain Tchana‌ from Krakos Inria team‌​‌ and VATES

10.1.1 Associate Teams‌ in the framework of‌​‌ an Inria International Lab​​ or in the framework​​​‌ of an Inria International‌ Program

10.1.2 Inria​​​‌ associate team not involved​ in an IIL or​‌ an international program

ANR JCJC Project sGOV​​​‌ (2023-2027)

Participants: Barbe Mvondo‌ Djob, Yerom David‌​‌ Bromberg.

In this​​ project, we propose to​​​‌ design smart governors (sGOV)‌ to tackle the sub-optimal‌​‌ energy management of idle​​ VMs in the Cloud.​​​‌ In a nutshell, the‌ main objective of sGOV‌​‌ is to identify VMs​​ idle periods, and not​​​‌ account the idle period‌ in the computing of‌​‌ the next CPU state​​ to switch. sGOV design​​​‌ goals are (i) genericity:‌ should be generic enough‌​‌ to be applied to​​ mainstream virtualization systems, and​​​‌ (ii) non-intrusiveness: should not‌ require legacy code to‌​‌ run in user VMs​​ to favor adoption by​​​‌ Cloud providers.

Our core‌ idea with sGOV is‌​‌ that VMs idle periods​​ have specific signatures regarding​​​‌ the interaction between the‌ VM and virtualization system.‌​‌ For example, when a​​ process in a VM​​​‌ stalls waiting for an‌ I/O event (e.g., the‌​‌ arrival of a network​​ packet), no processing is​​​‌ performed on its I/O‌ device interface until the‌​‌ event arises. However, a​​ VM waiting for a​​​‌ hardware event such as‌ the network packet will‌​‌ not behave similarly as​​ a VM waiting for​​​‌ a software interrupt or‌ signal from a process‌​‌ (e.g., SIGALARM signal). Additionally,​​ these behaviors can differ​​​‌ depending on the hardware‌ architecture — a sleep()‌​‌ instruction will not follow​​ the same pattern on​​​‌ an Intel CPU as‌ on AMD or ARM‌​‌ for example.

Partners: IRISA​​ (coordinator, U. Rennes). Budget:​​​‌ 286 814.5€

ANR Second‌ Chance (2023-2027, PRCE)

Participants:‌​‌ Yerom David Bromberg,​​ Barbe Mvondo Djob.​​​‌

Virtualization is a key‌ technology for datacenters and‌​‌ cloud computing, enabling flexible​​ resource allocation through virtual​​​‌ machines (VMs). Running multiple‌ VMs on the same‌​‌ physical host reduces hardware​​ and management costs while​​​‌ minimizing environmental impact. Central‌ to this process is‌​‌ the hypervisor, a software​​ layer that abstracts physical​​​‌ resources into virtual ones‌ for VMs, each running‌​‌ its own guest operating​​ system to support high-performance​​​‌ applications like web services,‌ databases, and AI tasks.‌​‌ While containers, such as​​ those managed by Docker​​​‌ or Podman, are widely‌ used, they complement rather‌​‌ than replace hypervisors, which​​ offer advanced features like​​​‌ security, performance isolation, persistent‌ storage, and snapshot management.‌​‌ Public cloud platforms often​​ encapsulate containers from different​​​‌ tenants within separate VMs.‌ A critical hypervisor capability‌​‌ is live VM migration,​​ a mature technique that​​​‌ moves a running VM‌ between physical machines without‌​‌ disrupting operations or degrading​​ performance. This feature is​​​‌ essential for cloud and‌ datacenter platforms, supporting administrative‌​‌ tasks while ensuring application​​ availability and performance, with​​​‌ providers like Google performing‌ millions of such migrations‌​‌ monthly.

Given that live​​ migration is commonly used​​​‌ for applications with stringent‌ availability and performance requirements,‌​‌ addressing the problem involves​​​‌ several challenges: determining migration​ safety without being overly​‌ conservative, ensuring acceptable application​​ performance during and after​​​‌ migration, developing extensible techniques​ to handle new types​‌ of CPU feature heterogeneity​​ and emerging application workloads,​​​‌ and maintaining transparency for​ application developers by avoiding​‌ modifications or recompilation of​​ guest code.

Partners: IRISA​​​‌ (coordinator, U. Rennes), LIG​ (Grenoble INP), Orange Business​‌ Services (Eolas).

ANR Project​​ ByBloS (2021-2025)

Participants: George​​​‌ Giakkoupis, Michel Raynal​, Davide Frey,​‌ Yerom David Bromberg,​​ François Taïani, Timothé​​​‌ Albouy.

Blockchain-based systems​ have over the last​‌ 10 years profoundly impacted​​ society and research. They​​​‌ come however with many​ inefficiencies, that are inherent​‌ to the problem they​​ attempt to solve, Byzantine​​​‌ Tolerant Agreement, one of​ the most difficult problems​‌ of distributed computing. Many​​ Blockchain-based applications do not​​​‌ require the strong guarantees​ that an agreement provides.​‌ Building on this insight,​​ Byblos seeks to explore​​​‌ the design, analysis, and​ implementation of lightweight Byzantine​‌ decentralized mechanisms for the​​ systematic construction of large-scale​​​‌ Byzantine-tolerant Privacy-Preserving distributed systems.​

Partners: IRISA (coordinator, U.​‌ Rennes) in Rennes, LIRIS​​ (INSA Lyon) in Lyon,​​​‌ and LS2N (Université de​ Nantes) in Nantes. Budget:​‌ 252 220€

Inria Challenge​​ Project FedMalin

Participants: François​​​‌ Taiani, Davide Frey​, Cyrille Kenfack.​‌

FedMalin (project.inria.fr/fedmalin/)​​ is a research project​​​‌ that spans 11 Inria​ research teams and aims​‌ to push FL research​​ and concrete use-cases through​​​‌ a multidisciplinary consortium involving​ expertise in ML, distributed​‌ systems, privacy and security,​​ networks, and medicine. We​​​‌ propose to address a​ number of challenges that​‌ arise when FL is​​ deployed over the Internet,​​​‌ including privacy and fairness,​ energy consumption, personalization, and​‌ location/time dependencies.

FedMalin will​​ also contribute to the​​​‌ development of open-source tools​ for FL experimentation and​‌ real-world deployments, and use​​ them for concrete applications​​​‌ in medicine and crowdsensing.​

The FedMalin Inria Challenge​‌ is supported by Groupe​​ La Poste, sponsor of​​​‌ the Inria Foundation.

Within​ Fedmalin, Davide Frey and​‌ François Taïani co-supervised the​​ PhD thesis of Rémy​​​‌ Raes, together with Lionel​ Seinturier and Romain Rouvoy​‌ from the Spirals team​​ form Inria Lille. Davide​​​‌ Frey also supervises the​ work of Cyril Kenfack​‌ (Engineer) in order to​​ contribute to a benchmarking​​​‌ environment for the experimentation​ with federated and decentralized​‌ learning platforms and algorithms.​​

Inria Challenge Project Alvearium​​​‌

Participants: François Taiani,​ Davide Frey.

The​‌ Alvearium project (project.inria.fr/alvearium/​​) aims to provide​​​‌ a sovereign alternative peer-to-peer​ cloud that provides both​‌ compute and data storage​​ through a peer-to-peer network​​​‌ rather than from a​ centralized set of data​‌ centers. The company Hive​​ (www.hivenet.com) proposes​​​‌ to exploit the unused​ capacity of computers and​‌ to incentivize users to​​ contribute their computer resources​​​‌ to the network in​ exchange for similar capacity​‌ from the network and/or​​ monetary compensation. By exchanging​​​‌ similar computing resources and​ network capacity, users can​‌ benefit from all cloud​​ services while ensuring the​​​‌ confidentiality of their data​ as it is fragmented,​‌ encrypted and spread across​​ the peer-to-peer network.

The​​ Inria COAST, COATI, MAGELLAN,​​​‌ PESTO and WIDE teams‌ participating in this challenge‌​‌ bring their expertise on​​ aspects of reliable and​​​‌ cost-efficient data placement and‌ repair in the case‌​‌ of node failures, collaboration​​ on shared data, data​​​‌ security and management of‌ malicious nodes in the‌​‌ context of unreliable distributed​​ storage.

Inria Challenge Project​​​‌ Cupseli

Participants: Davide Frey‌, Yerom David Bromberg‌​‌.

The Cupseli challenge​​ aims to demonstrate that​​​‌ it is possible to‌ run complex applications (particularly‌​‌ in the field of​​ machine learning) on heterogeneous,​​​‌ distributed, and volatile resources,‌ while achieving strong parallel‌​‌ efficiency and preserving both​​ accuracy and confidentiality. Building​​​‌ on the combined expertise‌ of hive and Inria‌​‌ in storage technologies illustrated​​ in Alvearium (www.inria.fr/en/alvearium​​​‌), this strategic partnership‌ explores algorithmic and system‌​‌ solutions to optimize computation,​​ memory, and communications, while​​​‌ ensuring security and fault‌ tolerance. The work is‌​‌ organized around three axes:​​ Frugality (adapting training and​​​‌ inference to limited and‌ dynamic resources), Security and‌​‌ Confidentiality (protecting data and​​ models through encryption, secure​​​‌ enclaves, and defenses against‌ attacks), and Volatility (ensuring‌​‌ robustness and performance despite​​ the unpredictable arrival and​​​‌ departure of resources). The‌ shared goal is to‌​‌ offer a green and​​ sovereign alternative to data​​​‌ centers, by leveraging already-existing‌ resources for the benefit‌​‌ of AI and Big​​ Data applications. This collaboration,​​​‌ anchored in this joint‌ challenge, brings together researchers,‌​‌ PhD students, engineers, and​​ postdocs, with large-scale experiments​​​‌ conducted on hive’s infrastructure.‌

Inria Challenge Project OS‌​‌

Participants: Yerom David Bromberg​​, Barbe Mvondo Djob​​​‌.

Data centers are‌ today at the heart‌​‌ of all computing, from​​ providing the computing power​​​‌ that supports machine learning,‌ databases, video streaming, etc.,‌​‌ down to providing tiny​​ sensors with extra computing​​​‌ power and storage. By‌ centralizing computing, data centers‌​‌ have the potential to​​ deliver massive computing resources​​​‌ while adapting the resource‌ consumption efficiently to changing‌​‌ needs. Nevertheless, data centers​​ have not fully realized​​​‌ their potential of optimizing‌ large-scale computing usage. Instead,‌​‌ studies have consistently shown​​ that, even though new​​​‌ data centers continue to‌ be built, existing data‌​‌ centers are massively underused,​​ typically reaching a usage​​​‌ ratio of only 50‌

The essential problem of‌​‌ managing a data center​​ is to allocate hardware​​​‌ resources, in an environment‌ in which application requirements‌​‌ are not known a​​ priori and are constantly​​​‌ changing, and where at‌ the same time hardware‌​‌ capabilities are regularly evolving.​​ The Defi OS will​​​‌ attack the problem of‌ data center underusage at‌​‌ the operating system level​​ and hypervisor level, as​​​‌ these are the software‌ components that interact directly‌​‌ with the hardware. The​​ project OS (project.inria.fr/defios/​​​‌) brings together researchers‌ from the Whisper, WIDE,‌​‌ KrakOS, and Benagil teams​​ and will investigate how​​​‌ virtual machine migration, heterogeneous‌ architectures, rack scale computing,‌​‌ and custom resource management​​ policies can be harnessed​​​‌ to raise the data‌ center usage ratio toward‌​‌ 90

11.1.1 Scientific events: organisation​‌

11.1.2​​ Scientific events: selection

11.1.3 Journal

11.1.4 Invited talks

• George​​ Giakkoupis. Distributed Stochastic Graph​​​‌ Algorithms. 3rd Bertinoro Workshop​ on Distributed Geometric Algorithms​‌ (DiG@BiCi25), Bertinoro, Italy, Sep.​​ 30 2025
• George Giakkoupis.​​​‌ Naively sorting evolving data.​ ADYN Seminar: Algorithms, Dynamics,​‌ and Information Flow in​​ Networks, Virtual, Germany, Mar.​​​‌ 31 2025
• Davide Frey​ gave a talk at​‌ the Workshop on "Distributed​​ Computing: Past, Present, Future"​​​‌ in honor of Maurice​ Herlihy at LIP6, Paris.​‌

11.1.5 Leadership within the​​ scientific community

• François Taïani​​​‌ serves as co-chair of​ the scientific committee of​‌ GDR RSD (Groupement de​​ Recherche Réseaux & Systèmes​​​‌ Distribués) since 2024.
• George​ Giakkoupis was a member​‌ of the steering committee​​ (member-at-large) of the ACM​​​‌ Symposium on Principles of​ Distributed Computing (PODC) from​‌ 2023-2025.

11.1.6 Scientific expertise​​

• Achour Mostefaoui served as​​ a member of the​​​‌ evaluation committee CE25 of‌ the ANR (French research‌​‌ funding agency) on Software​​ platforms, systems and networks​​​‌ from 2022 to 2025.‌
• Davide Frey was an‌​‌ expert reviewer for the​​ National Science Center, Poland.​​​‌
• George Giakkoupis is a‌ member of the Working‌​‌ Group GT CoA: Complexité​​ et Algorithmes since 2023.​​​‌

11.1.7 Research administration

• François‌ Taïani served as a‌​‌ member of the thesis​​ committee of the Doctoral​​​‌ School Matisse (ED N.‌ 601).
• François Taïani served‌​‌ as a member of​​ the Bureau du Comité​​​‌ des Projets (BCP) of‌ the Inria Centre at‌​‌ Rennes University.
• François Taïani​​ served as a member​​​‌ of the local Inria‌ secondment committee (Comission des‌​‌ délégations Inria) in Rennes.​​
• François Taïani served as​​​‌ Career Advice Person, (Référent‌ conseil-parcours professionel chercheurs) for‌​‌ IRISA/Inria Centre at Rennes​​ University since 2019.
• François​​​‌ Taïani served as vice-chairman‌ of the recruitment committee‌​‌ for a Professorship on​​ "Cybersécurité, réseaux, systèmes" of​​​‌ ESIR (University of Rennes).‌
• François Taïani served on‌​‌ the recruitment committee for​​ an MCF "Systèmes, IA,​​​‌ HPC" of INSA Lyon.‌
• François Taïani served on‌​‌ the recruitment committee for​​ a Professorship "Computer Science"​​​‌ of Nantes University.
• Davide‌ Frey served on the‌​‌ recruitment committee for Maitre​​ de Conference at IDMC,​​​‌ Université de Lorraine, Nancy.‌
• George Giakkoupis is a‌​‌ local correspondent of the​​ Inria Centre at Rennes​​​‌ University for the preparation‌ of the Annual Activity‌​‌ Reports by the project​​ teams since 2023.

11.2.1‌ Teaching

• Engineering School: François‌​‌ Taïani, Operating Systems, 28h,​​ 2nd year of Engineering​​​‌ School (M1), ESIR /‌ U. Rennes, France.
• Engineering‌​‌ School: François Taïani, Distributed​​ Systems, 12h, 3rd year​​​‌ of Engineering School (M2),‌ ESIR / U. Rennes,‌​‌ France.
• Engineering School: François​​ Taïani, Introduction to Operating​​​‌ Systems, 72h, 1st year‌ of Engineering School (L3),‌​‌ ESIR / U. Rennes,​​ France.
• Engineering School: Barbe​​​‌ Mvondo Djob, Network and‌ Security for IOT, 45h,‌​‌ ESIR M1, Rennes, France​​
• Engineering School: Barbe Mvondo​​​‌ Djob, Cloud Computing for‌ IOT, 45h, ESIR M2,‌​‌ Rennes, France
• Engineering School:​​ Barbe Mvondo Djob, Cybersecurity​​​‌ and Hacking, 30h, Polytechnic‌ 2nd year, Paris, France‌​‌
• Computer Science Dpt.: Achour​​ Mostefaoui, Graph Theory, 42h,​​​‌ 3rd year of Bachelor‌ of Science (L3), ISTIC‌​‌ / U. Rennes, France.​​
• Master: Davide Frey, Scalable​​​‌ Distributed Systems, 10h, M1,‌ EIT/ICT Labs Master School,‌​‌ U. Rennes, France.
• ENS​​ L3 : Davide Frey,​​​‌ Distributed Algorithms, 11h, L3‌ parcours SI, ISTIC, ENS‌​‌ Rennes, France.
• ENS L3:​​ George Giakkoupis, Distributed Algorithms,​​​‌ 9h, L3 parcours SI,‌ ISTIC, ENS Rennes, France.‌​‌
• Master: Davide Frey, Cloud​​ Computing, 12h, M2-MIAGE, U.​​​‌ Rennes, France.
• Computer Science‌ Dpt: Brice Ekane Apah,‌​‌ Networking, 45h, L2, U.​​ Rennes, France.
• Computer Science​​​‌ Dpt: Brice Ekane Apah,‌ Entreprise network architectures, 36h,‌​‌ M2, U. Rennes, France.​​
• Computer Science Dpt: Brice​​​‌ Ekane Apah, Next-Generation Network‌ Architecture, 23h, M2, U.‌​‌ Rennes, France.
• Computer Science​​​‌ Dpt: Brice Ekane Apah,​ Software Techniques for Cloud​‌ Computing, 33h, M2, U.​​ Rennes, France.
• Computer Science​​​‌ Dpt: Brice Ekane Apah,​ Learning Progress Monitoring, 13.5h,​‌ M2, U. Rennes, France.​​

11.2.2 Supervision

• PhD (defended​​​‌ in December 2025): Vincent​ Kowalski, Useful byzantine abstraction​‌ at computability level of​​ shared memory, supervised by​​​‌ Achour Mostefaoui and Matthieu​ Perrin (Univ. Nantes)
• PhD​‌ (defended in December 2024):​​ Timothé Albouy, Towards Lightweight​​​‌ Scalable and Open Byzantine-Fault-Tolerant​ Distributed Objects, U. Rennes,​‌ supervised by François Taïani​​ and Davide Frey.
• PhD​​​‌ in progress: Dimitri Lerévérend,​ Privacy-Preserving Decentralized Learning Through​‌ Model Fragmentation and Private​​ Aggregation, started in September​​​‌ 2023, supervised by Davide​ Frey, Romaric Gaudel (MALT​‌ team) and François Taïani.​​
• PhD in progress: Manon​​​‌ Sourisseau, Byzantine-Tolerant Netcodes For​ Tomorrow's Metaverse, started in​‌ October 2023, supervised by​​ François Taïani, Yerom David​​​‌ Bromberg, and Jérémie Découchant​ (TU Delft).
• PhD in​‌ progress: Rémy Raes, Distributed​​ Machine Learning in Ubiquitous​​​‌ Environments using Location-dependent Models,​ started in 2023, supervised​‌ by Davide Frey, François​​ Taiani, Romain Rouvoy and​​​‌ Lionel Seinturier (Spirals team,​ Inria Lille).
• PhD in​‌ progress: Augustin Godinot, Auditing​​ the mutations of AI-models,​​​‌ started in November 2022,​ supervised by Erwan Le​‌ Merrer, Gilles Trédan (LAAS/CNRS),​​ François Taïani, and Camilla​​​‌ Penzo (PEReN).
• PhD in​ progress: Hua Junrui, Advanced​‌ Techniques for Efficient Distributed​​ Hash Tables Management with​​​‌ Fault Tolerance against Byzantine​ Faults in Large-Scale Distributed​‌ Systems, started in November​​ 2024, supervised by François​​​‌ Taiani, Gérald Oster (LORELEY​ team, Inria Nancy), and​‌ Alexandru Dobrila (Hive).
• PhD​​ in progress: Ludovic Paillat,​​​‌ Security for peer-to-peer cloud​ storage without central authority,​‌ started in 2023, supervised​​ by Davide Frey, Claudia​​​‌ Ignat (LORELEY team, Inria​ Nancy), Alexandru Dobrila (HIVE),​‌ Mathieu Turiani (PESTO Team,​​ Inria Nancy).
• PhD in​​​‌ progress: Adrien Gegout, Efficient​ containerized Cloud Gaming, started​‌ in October 2023, supervised​​ by Davide Frey, Barbe​​​‌ Mvondo Djob, Pascal Manchon​ (Blacknut).
• PhD in progress:​‌ Cesaire Honoré, Scheduling in​​ heterogeneous architectures, started in​​​‌ December 2022, supervised by​ Yerom David Bromberg and​‌ Barbe Mvondo Djob
• PhD​​ in progress: Victoire Nganfang,​​​‌ Resisting to Massive proliferation​ of new Android malware​‌ threats, started in November​​ 2024, supervised by Yerom​​​‌ Yerom David Bromberg and​ Valério Schiavoni (University of​‌ Neuchâtel, Switzerland)
• PhD in​​ progress: Stella Tchoutcha, Energy-Efficient​​​‌ Function-as-a-Service (FaaS) for Low-Power​ Edge and IoT Deployments,​‌ started in December 2025,​​ supervised by Barbe Mvondo​​​‌ Djob and Nikos Parlavantzas​ (Magellan team)
• PhD in​‌ progress: Caleb Fonyuy-Asheri, Heterogeneous​​ VM migration , started​​​‌ in February 2024, supervised​ by Yerom David Bromberg,​‌ Alain Tchana (Grenoble INP),​​ Barbe Mvondo Djob, Renaud​​​‌ Lachaise (UGA)
• PhD in​ progress: Alexandre Duvivier, CDN​‌ performance optimization, started in​​ October 2023, supervised by​​​‌ Yerom David Bromberg, Barbe​ Mvondo Djob, Nicolas Le​‌ Scouarnec (Broadpeak)
• PhD in​​ progress: Amelie Gonzalez, Linux​​​‌ network stack optimization, started​ in September 2023, supervised​‌ by Yerom David Bromberg,​​ Barbe Mvondo Djob, Julia​​​‌ Lawal (Whisper team, Inria​ Paris)
• PhD in progress​‌ : Hugo Bertin, Attacking​​ Games to Strengthen their​​​‌ Defenses, started in February​ 2025, Supervised by Yerom​‌ David Bromberg and Marc​​ Dacier (KAUST)
• PhD in​​ progress : Elie Raspaud,​​​‌ Enhancing privacy in distributed‌ machine learning using homomorphic‌​‌ cryptography, started in September​​ 2025, supervised by Davide​​​‌ Frey, Philippe Chartier and‌ Mohammed Lemou (CNRS)
• PostDoc:‌​‌ Georgy Ishmaev, Byzantine Fault-Tolerant​​ Distributed Ledgers, May 2024-October​​​‌ 2025, supervised by François‌ Taïani and Davide Frey,‌​‌ funded by ANR project​​ ByBloS n. ANR-20-CE25-0002.
• PostDoc:​​​‌ Dimitrios Los, Distributed Algorithms‌ on Evolving Data, until‌​‌ Feb 2024, supervised by​​ George Giakkoupis, funded by​​​‌ Action Exploratoire DisEvo: Distributed‌ Algorithms on Evolving Data‌​‌

11.2.3 Juries

• François Taïani​​ was a reviewer for​​​‌ Yacine Belal's PhD thesis:‌ Trustworthy Collaborative Learning: Personalization,‌​‌ Privacy, and Robustness at​​ the Edge, INSA Lyon​​​‌ (France), 10 June 2025.‌
• François Taïani was a‌​‌ reviewer for Sara Tucci-Piergiovanni's​​ HDR thesis: Blockchains: from​​​‌ The Wild to Distributed‌ Computing, Université Paris-Saclay (France),‌​‌ 29 September 2025.
• Achour​​ Mostefaoui was a reviewer​​​‌ for Alejandro Naser's PhD‌ thesis: Fault-Tolerant Compu<ng with‌​‌ Unreliable Channels, IMDEA Research​​ Intitute (Madrid, Spain), 10​​​‌ December 2025.
• Davide Frey‌ was a reviewer for‌​‌ Eugenio Lomurno's PhD thesis:​​ Adversarial and Generative Deep​​​‌ Learning for Data Privacy‌ in Human-Centered Artificial Intelligence,‌​‌ Politecnico di Milano (Milano,​​ Italy), 9 May 2025.​​​‌
• Davide Frey was a‌ reviewer for Hassan Nazeer‌​‌ Chaudhry's PhD thesis: Efficient​​ Processing of Graph-Based Data​​​‌ Streams, Politecnico di Milano‌ (Milano, Italy), 29 August,‌​‌ 2025.
• Davide Frey was​​ a reviewer for Bulat​​​‌ Aydarovich NASRULIN's PhD thesis:‌ Trustworthy Foundations for Web3,‌​‌ Delft University of Technology​​ (Delft, The Netherlands), 26​​​‌ September 2025.
• Brice Ekane‌ Apah was an examiner‌​‌ for PAPA Asane FALL​​ PhD thesis: Linux as​​​‌ a micro-kernel : the‌ memory management's case, University‌​‌ of Grenoble Alpes (Grenoble,​​ France), 05 December 2025.​​​‌
• Barbe Mvondo Djob served‌ as one of the‌​‌ French Baccaleureate jury presidents​​ (président du jury du​​​‌ baccalauréats) for 2025.

11.3.1 Productions‌ (articles, videos, podcasts, serious‌​‌ games, ...)

• François Taïani,​​ Manon Sourisseau, Timothé Albouy,​​​‌ and Davide Frey took‌ part in the filming‌​‌ of the television programme​​ Esprit Sorcier: les surprises​​​‌ de la recherche (Wizard‌ Spirit: the surprises of‌​‌ research), filmed in Rennes​​ in April 2025 with​​​‌ the participation of 340‌ secondary school pupils from‌​‌ the Rennes region. They​​ presented their current research​​​‌ on Byzantine fault-tolerant protocols‌ as part of the‌​‌ ByBloS project. The programme​​ is now available on​​​‌ Internet streaming services and‌ YouTube.

International journals‌

• 15 articleL.Luc‌​‌ Devroye and D.Dimitrios​​ Los. An asymptotically​​​‌ optimal algorithm for generating‌ bin cardinalities.Mathematics‌​‌ and Computers in Simulation​​228February 2025,​​​‌ 147-155HALDOIback‌ to text
• 16 article‌​‌R.Romaric Duvignau,​​ M.Michel Raynal and​​​‌ E. M.Elad Michael‌ Schiller. Self-stabilizing multivalued‌​‌ consensus in the presence​​ of Byzantine faults and​​​‌ asynchrony.Theoretical Computer‌ Science1039June 2025‌​‌, 115184HALDOI​​
• 17 articleG.George​​​‌ Giakkoupis, V.Volker‌ Turau and I.Isabella‌​‌ Ziccardi. Luby's MIS​​ Algorithms Made Self-Stabilizing.​​​‌Information Processing Letters188‌August 2025, 106531‌​‌HALDOIback to​​ text
• 18 articleG.​​​‌Georgy Ishmaev, E.‌Emmanuelle Anceaume, D.‌​‌Davide Frey and F.​​François Taïani. Ethical​​​‌ Risk Analysis of L2‌ Rollups.ACM Transactions‌​‌ on the WebDecember​​ 2025HALDOIback​​​‌ to text
• 19 article‌L.Ludovic Paillat,‌​‌ C.-L.Claudia-Lavinia Ignat,​​ D.Davide Frey,​​​‌ M.Mathieu Turuani and‌ A.Amine Ismail.‌​‌ Discreet: distributed delivery service​​ with context-aware cooperation.​​​‌Annals of Telecommunications -‌ annales des télécommunications80‌​‌3-4April 2025,​​ 357-374HALDOIback​​​‌ to text

International peer-reviewed‌ conferences

• 20 inproceedingsT.‌​‌Timothé Albouy, A.​​Antonio Fernández Anta,​​​‌ C.Chryssis Georgiou,‌ M.Mathieu Gestin,‌​‌ N.Nicolas Nicolaou and​​ J.Junlang Wang.​​​‌ AMECOS: A Modular Event-based‌ Framework for Concurrent Object‌​‌ Specification.OPODIS 2024​​ - 28th International Conference​​​‌ on Principles of Distributed‌ SystemsProceedings of the‌​‌ 28th International Conference on​​ Principles of Distributed Systems​​​‌ (OPODIS 2024)Lucques, Italy‌Schloss Dagstuhl – Leibniz-Zentrum‌​‌ für Informatik2025HAL​​DOI
• 21 inproceedingsT.​​​‌Timothé Albouy, D.‌Davide Frey, M.‌​‌Mathieu Gestin, M.​​Michel Raynal and F.​​​‌François Taïani. Contention-Aware‌ Cooperation.OPODIS 2025‌​‌ - 29th International Conference​​ On Principles Of Distributed​​​‌ Systems9Iasi, Romania‌Schloss Dagstuhl – Leibniz-Zentrum‌​‌ für InformatikDecember 2025​​, 9:1–9:20HALDOI​​​‌back to text
• 22‌ inproceedingsH.Hugo Bertin‌​‌, I.Ilies Benhabbour​​, M.Marc Dacier​​​‌ and Y.-D.Yérom-David Bromberg‌. Disconnecting Users from‌​‌ Virtual Worlds with a​​ Single Packet: an Unreal​​​‌ Untold Story.iMETA‌ 2025 - 3rd International‌​‌ Conference on Intelligent Metaverse​​ Technologies & ApplicationsDubrovnik,​​​‌ CroatiaOctober 2025,‌ 67-74HALDOIback‌​‌ to text
• 23 inproceedings​​S.Sayan Biswas,​​​‌ D.Davide Frey,‌ R.Romaric Gaudel,‌​‌ A.-M.Anne-Marie Kermarrec,​​ D.Dimitri Lerévérend,​​​‌ R.Rafael Pires,‌ R.Rishi Sharma and‌​‌ F.François Taïani.​​ Low-Cost Privacy-Preserving Decentralized Learning​​​‌.Proceedings on Privacy‌ Enhancing Technologies SymposiumPrivacy‌​‌ Enhancing Technologies Symposium2025​​3Washinghton DC, United​​​‌ StatesJuly 2025,‌ 451–474HALDOIback‌​‌ to text
• 24 inproceedings​​​‌Y.-D.Yérom-David Bromberg,​ J.Jérémie Decouchant,​‌ M.Manon Sourisseau and​​ F.François Taïani.​​​‌ Formalizing Rollback Netcodes for​ Robust and Real-Time Client-Server​‌ Architectures.Proceedings of​​ the 29th International Conference​​​‌ on Principles of Distributed​ SystemsOPODIS 2025 -​‌ 29th International Conference on​​ Principles of Distributed Systems​​​‌Iasi, RomaniaSchloss Dagstuhl​ – Leibniz-Zentrum für Informatik​‌2026, 1-17HAL​​DOI
• 25 inproceedingsB.​​​‌Brice Ekane, D.​Djob Mvondo, R.​‌Renaud Lachaize, Y.-D.​​Yérom-David Bromberg, A.​​​‌Alain Tchana and D.​Daniel Hagimont. DISC:​‌ Backpressure Mitigation In Multi-tier​​ Applications With Distributed Shared​​​‌ Connection.Proceedings of​ the 22nd USENIX Symposium​‌ on Networked Systems Design​​ and Implementation (NSDI 25)​​​‌NSDI 2025 - 22nd​ USENIX Symposium on Networked​‌ Systems Design and Implementation​​Philadelphia (Pennsylvania), United States​​​‌April 2025, 55-70​HALback to text​‌
• 26 inproceedingsE.Eli​​ Gafni, G.Giuliano​​​‌ Losa, M.Michel​ Raynal and G.Gadi​‌ Taubenfeld. Brief Announcement:​​ Stranger-Free Tasks.PODC​​​‌ 2025 - 44th ACM​ Symposium on Principles of​‌ Distributed ComputingMexico, Mexico​​ACMJune 2025,​​​‌ 203-206HALDOI
• 27​ inproceedingsA.Adrien Gegout​‌, D.Djob Mvondo​​, D.Davide Frey​​​‌ and P.Pascal Monchon​. Towards an Efficient​‌ Containerized Cloud Gaming Platform​​.ASHES - IPDPS​​​‌ Workshops2025 IEEE International​ Parallel and Distributed Processing​‌ Symposium Workshops (IPDPSW)Milano,​​ FranceJune 2025,​​​‌ 78-86HALDOIback​ to text
• 28 inproceedings​‌A.Augustin Godinot,​​ E. L.Erwan Le​​​‌ Merrer, C.Camilla​ Penzo, F.François​‌ Taïani and G.Gilles​​ Tredan. Queries, Representation​​​‌ &amp; Detection: The Next​ 100 Model Fingerprinting Schemes​‌.Proceedings of the​​ AAAI Conference on Artificial​​​‌ IntelligenceAAAI 2025 -​ 39th Annual AAAI Conference​‌ on Artificial Intelligence39​​16Philadelphia (Pennsylvania), United​​​‌ StatesApril 2025,​ 16817-16825HALDOI
• 29​‌ inproceedingsH. C.Honore​​ Cesaire Mounah, D.​​​‌Djob Mvondo, J.​Julia Lawall and Y.-D.​‌Yérom-David Bromberg. The​​ Impact of Kernel Asynchronous​​​‌ APIs on the Performance​ of a Kernel VPN​‌.SYSTOR 2025 -​​ 18th ACM International System​​​‌ and Storage ConferenceVirtual​ conference, Israel2025,​‌ 167-173HALDOIback​​ to text
• 30 inproceedings​​​‌D. B.Djob Barbe​ Thystere Mvondo Djob and​‌ Y.-D.Yérom-David Bromberg.​​ Secure access to network​​​‌ data for mobile network​ traffic analysis applications.​‌55th Annual IEEE/IFIP International​​ Conference on Dependable Systems​​​‌ and Networks (DSN 2025)​Naples, ItalyJune 2025​‌HALDOIback to​​ text
• 31 inproceedingsD.​​​‌Djob Mvondo, B.​Boris Djomgwe Teabe and​‌ N.Nikos Parlavantzas.​​ Efficient Memory Usage For​​​‌ Edge FaaS Platforms.​PerCom 2026 - 24th​‌ IEEE International Conference on​​ Pervasive Computing and Communications​​​‌Pise, Italy2026HAL​
• 32 inproceedingsV.Victoire​‌ Nganfang, S.Simon​​ Queyrut, D.David​​​‌ Bromberg, V.Valerio​ Schiavoni, D.Djob​‌ Mvondo and V.Vianney​​ Kengne Tchendji. DroidHunter:​​​‌ A Robust Vision-Based Detection​ Against Hidden Android Malware​‌.ASIACCS 2026 -​​ 21st ACM ASIA Conference​​ on Computer and Communications​​​‌ SecurityBangalore, IndiaACM‌June 2026HALDOI‌​‌back to text
• 33​​ inproceedingsP.Pasquale de​​​‌ Rosa, S.Simon‌ Queyrut, Y.-D.Yérom-David‌​‌ Bromberg, P.Pascal​​ Felber and V.Valerio​​​‌ Schiavoni. PhishingHook: Catching‌ Phishing Ethereum Smart Contracts‌​‌ leveraging EVM Opcodes.​​DSN 2025 - 55th​​​‌ Annual IEEE/IFIP International Conference‌ on Dependable Systems and‌​‌ NetworksNaples, FranceIEEE​​2025, 265-266HAL​​​‌DOI
• 34 inproceedingsF.‌Francesco d'Amore, N.‌​‌Niccolò d'Archivio, G.​​George Giakkoupis and E.​​​‌Emanuele Natale. On‌ the h-majority dynamics with‌​‌ many opinions.39th​​ International Symposium on Distributed​​​‌ Computing (DISC 2025)Berlin,‌ GermanySchloss Dagstuhl –‌​‌ Leibniz-Zentrum für Informatik2025​​HALDOIback to​​​‌ text

Edition (books, proceedings,‌ special issue of a‌​‌ journal)

• 35 proceedingsPaPoC​​ '25: 12th Workshop on​​​‌ Principles and Practice of‌ Consistency for Distributed Data‌​‌.PaPoC '25: 12th​​ Workshop on Principles and​​​‌ Practice of Consistency for‌ Distributed DataWorld Trade‌​‌ Center Rotterdam Netherlands, France​​ACMMarch 2025HAL​​​‌DOIback to text‌

Reports & preprints

• 36‌​‌ miscT.Timothé Albouy​​, E.Emmanuelle Anceaume​​​‌, D.Davide Frey‌, M.Mathieu Gestin‌​‌, A.Arthur Rauch​​, M.Michel Raynal​​​‌ and F.François Taïani‌. Asynchronous BFT Asset‌​‌ Transfer: Quasi-Anonymous, Light, and​​ Consensus-Free.February 2025​​​‌HAL
• 37 miscA.‌Aurélien Bellet, E.‌​‌Edwige Cyffers, D.​​Davide Frey, R.​​​‌Romaric Gaudel, D.‌Dimitri Lerévérend and F.‌​‌François Taïani. Unified​​ Privacy Guarantees for Decentralized​​​‌ Learning via Matrix Factorization‌.October 2025HAL‌​‌
• 38 miscT.Timothee​​ Chauvin, E.Erwan​​​‌ Le Merrer, F.‌François Taïani and G.‌​‌Gilles Tredan. Log​​ Probability Tracking of LLM​​​‌ APIs.December 2025‌HAL
• 39 miscG.‌​‌Georgy Ishmaev, E.​​Emmanuelle Anceaume, D.​​​‌Davide Frey and F.‌François Taïani. Ethical‌​‌ Risk Analysis of L2​​ Rollups.December 2025​​​‌HAL
• 40 miscG.‌Georgy Ishmaev. Ethics‌​‌ of Blockchain Technologies.​​2025HALDOI