IMDEA Networks Institute Digital Repository
Not a member yet
1915 research outputs found
Sort by
DISC: A Dataset for Integrated Sensing and Communications in mmWave Systems
Full text linkThis article presents DISC, a dataset of millimeter-wave channel impulse response measurements for integrated human activity sensing and communication. This is the first dataset collected with a software-defined radio testbed that transmits 60 GHz IEEE 802-11ay-compliant packets and estimates the channel response, including scattered signals off the moving body parts of subjects moving in an indoor environment. The provided data consists of three parts for more than two hours of channel measurements with high temporal resolution (0.27 ms inter-packet time). DISC contains the contribution of seven subjects performing five different activities and includes data collected from two distinct environments. Unlike available radar-based millimeter-wave sensing datasets, our measurements are collected using uniform
packet transmission times and sparse traffic patterns from real Wi-Fi deployments. We develop, train, and release open-source baseline algorithms based on DISC to perform human sensing tasks. Our results demonstrate that DISC can serve as a multi-purpose benchmarking tool for machine learning-based human activity recognition, radio frequency gait analysis, and sparse sensing algorithms for next generation integrated sensing and communications.TRUEinpres
An Urban Geography of Mobile Application Usage: Connecting Demand Dynamics and Urban Fabrics
Full text linkThe surge in usage of mobile applications generates a massive volume of traffic data exhibiting unique dynamics that are hard to unravel. In this work, we leverage factor analysis to pin down recurrent patterns of mobile traffic over the three dimensions of space, time and services in multi-city measurements of unprecedented resolution. We link the revealed structures of real-world mobile demands to urban fabrics, i.e., the combination of infrastructures and social characteristics that determine the functionality of urban territory, hence establishing connections between specific city landscapes and the mobile application consumption they create. Our study provides a new understanding about the diversity of mobile service dynamics in metropolitan areas, including insights on how economic status drives the adoption of specific applications, how residential versus commercial areas create a dichotomy in applications usage, how private and public transport drive surges in the prevalence of different sets of applications or
how nightlife or university studies stimulate the utilization of specific classes of services.European UnionFrench National Research AgencyFonds de Recherche du Qu´ebecTRUEinpres
MUDGUARD: Taming Malicious Majorities in Federated Learning using Privacy-Preserving Byzantine-Robust Clustering
Full text linkByzantine-robust Federated Learning (FL) aims to counter malicious clients and train an accurate global model while maintaining an extremely low attack success rate.
Most existing systems, however, are only robust when most of the clients are honest.
\texttt{FLTrust} (NDSS '21) and \texttt{Zeno++} (ICML '20) do not make such an honest majority assumption but can only be applied to scenarios where the server is provided with an auxiliary dataset used to filter malicious updates.
\texttt{FLAME} (USENIX '22) and \texttt{EIFFeL} (CCS '22) maintain the semi-honest majority assumption to guarantee robustness and the confidentiality of updates.
It is, therefore, currently impossible to ensure Byzantine robustness and confidentiality of updates without assuming a semi-honest majority.
To tackle this problem, we propose a novel Byzantine-robust and privacy-preserving FL system, called \texttt{MUDGUARD}, to capture malicious minority and majority for server and client sides, respectively. Our experimental results demonstrate that the accuracy of \texttt{MUDGUARD} is practically close to the FL baseline using FedAvg without attacks (0.8\% gap on average). Meanwhile, the attack success rate is around 0\%-5\% even under an adaptive attack tailored to \texttt{MUDGUARD}.
We further optimize our design by using binary secret sharing and polynomial transformation, leading to communication overhead and runtime decreases of 67\%-89.17\% and 66.05\%-68.75\%, respectively.MLEDGETRUEinpres
PCP-YOLO: an approach integrating non-deep feature enhancement module and polarized self-attention for small object detection of multiscale defects
Full text linkThe detection of small objects within multiscale defects amidst complex background interference presents a formidable challenge in industrial defect detection. To address this issue and achieve precise and expeditious identification in industrial defect detection, this study proposes PCP-YOLO, a novel network that incorporates a non-deep feature extraction module and a polarized filtering feature fusion module for small object defect detection. Initially, YOLOv8 is employed as the foundational model. Subsequently, a lightweight, non-deep feature extraction module, PotentNet, is designed and integrated into the backbone network. In the neck network, a feature fusion module incorporating polarized self-attention, C2f_ParallelPolarized, has been developed. Finally, CARAFE is utilized to substitute the original upsampling module in the neck network. The efficacy of this approach has been rigorously evaluated using three datasets: the publicly available NEU-DET and PKU-PCB datasets, and the real-world industrial dataset GC10-DET. The [email protected] values achieved are 79.4%, 96.1%, and 77.6%, significantly outperforming other detection methods. The method also has a fast inference speed. These results demonstrate that PCP-YOLO exhibits substantial potential for rapid and accurate defect detection.TRUEpu
Beneath the surface: An analysis of OEM customizations on the Android TLS protocol stack
Full text linkThe open-source nature of the Android Open Source Project (AOSP) allows Original Equipment Manufacturers (OEMs) to customize the Android operating system, contributing to what is known as Android fragmentation. Google has implemented the Compatibility Definition Document (CDD) and the Compatibility Test Suite (CTS) to
ensure the integrity and security of the Android ecosystem. However, the effectiveness of these policies and measures in warranting OEM compliance remains uncertain. This paper empirically studies for the first time the nature of OEM customizations in the Android TLS protocol stack, and their security implications on user-installed mobile apps across thousands of Android models. We find that approximately 80% of the analyzed Android models deviate from the standard AOSP TLS codebase and that OEM customizations often involve code changes in functions used by app developers for enhancing TLS security, like end-point and certificate verification. Our analysis suggests that these customizations are likely influenced by factors such as manufacturers’ supply chain dynamics and patching prioritization tactics, including the need to support legacy components. We conclude by identifying potential root causes and emphasizing the need for stricter policy enforcement, better supply chain controls, and improved patching processes across the ecosystem.Spanish National Cybersecurity Institute (INCIBE)Comunidad de MadridTRUEinpres
NLP-Driven Approaches to Measuring Online Polarization and Radicalization
Full text linkThe growing popularity of social media has coincided with a massive number of real-world issues and crises that are controversial and polarizing. Recent issues such as Russo-Ukrainian and Israeli-Palestinian conflicts, alongside classic issues such as abortion-ban and gun-control, have raised heated debates offline and online. Throughout the past two decades, Computational Social Scientists have been introducing methods of modeling and measuring online polarization and radicalization. Yet, most of the proposed methods rely on traditional tools such as graph analysis and classic NLP models. These tools are accompanied by limitations in terms of scalability, granularity, and availability of data (e.g., follow network is no longer publicly available on Twitter).
Fortunately, in the past few years, thanks to the invention of the transformers architecture, the world has witnessed massive breakthroughs in the field of Natural Language Processing (NLP). Especially, Large Language Models (LLMs) have grasped the attention of both public and scientific communities. These breakthroughs have also created unprecedented opportunities for advancing classic techniques in various domains of Computational Social Sciences, including polarization detection and opinion mining.
This thesis aims to propose novel approaches using state-of-the-art NLP techniques to model and track polarization on social media. It introduces a scalable method for quantifying echo chambers with sentence transformers, revealing asymmetries in discourse diversity across political ideologies. Furthermore, it applies LLMs to analyze the content of cross-partisan interactions, showing that cross-party engagement does not necessarily lead to productive discourse. The thesis also investigates radicalization in gender-based communities and compares the spread of radical content across platforms like Reddit and Discord. Lastly, it addresses the limitations of existing language models in detecting stance polarity by fine-tuning a sentence transformer to become stance-aware, enabling more accurate detection of opposing viewpoints on similar topics. Together, these contributions offer Computational Social Scientists new tools for understanding polarization, radicalization, and bias in online environments.Telematics EngineeringUniversidad Carlos III de Madrid, Spai
DUNE: Distributed Inference in the User Plane
Full text linkThe deployment of Machine Learning (ML) models in the user plane enables line-rate in-network inference, significantly reducing latency and improving the scalability of functions like traffic monitoring. Yet, integrating ML models into programmable network devices requires meeting stringent constraints in terms of memory resources and computing capabilities. Previous solutions have focused on implementing monolithic ML models within individual programmable network devices, which are limited by hardware constraints, especially while executing challenging classification use cases. In this paper, we propose DUNE, a novel framework that realizes for the first time a user plane inference that is distributed across the multiple devices that compose the programmable network. DUNE adopts fully automated approaches to (i) breaking large ML models into simpler sub-models that preserve inference accuracy while minimizing resource usage, (ii) designing the sub-models and their sequencing so as to enable an efficient distributed execution of joint packet- and flow-level inference. We implement DUNE using P4, deploy it in an experimental network with multiple industry-grade programmable switches, and run tests with real-world traffic measurements for two complex classification use cases. Our results demonstrate that DUNE not only reduces per-switch resource utilization with respect to legacy monolithic ML designs but also improves their inference accuracy by up to 7.5%.TRUEinpres
CHRONOPROF: Profiling Time Series Forecasters and Classifiers in Mobile Networks with Explainable AI
Full text linkThe next-generation of mobile networks will increasingly rely on Artificial Intelligence (AI)/Machine Learning (ML) for effective network automation, resource orchestration and management. This translates into performing classification and regression tasks on time series data. Unfortunately, the existing AI/ML models are inherently complex and hard to interpret, which hinders their deployment in production networks. Further, the vast majority of the existing EXplainable Artificial Intelligence (XAI) techniques are either primarily conceived for computer vision and natural language processing and thus fail to provide useful insights.
In this paper, we take the research on XAI for time series classification and regression tasks one step further proposing ChronoProf, a new tool that builds on legacy XAI techniques. By creating a linearized version of the original model for different observations, ChronoProf provides insights about the dynamic changes in the model decision-making process across observations and is agnostic to the influence of feature magnitude, which is a key limitation of legacy explainers. Thus, ChronoProf highlights the real influence of model parameters on the output. Our extensive evaluation with real-world mobile traffic traces shows that ChronoProf is able to measure the feature importance, especially in classification tasks where linearized explanations across observations show high consistency.TRUEinpres
User-Plane Algorithms for Stateless and Stateful Inference in Programmable Networks
Full text linkIn the last decade, the complexity of networks has increased significantly to accommodate the rise of innovative applications. This growing complexity has rendered traditional human-in-the-loop network management approaches inadequate, necessitating greater automation and flexibility in managing these networks. The introduction of Software-Defined Networking (SDN) with a programmable control plane marked a major advancement in this direction, enabling a wide range of network automation applications to be executed within the SDN control plane.FALSEpu