IMDEA Networks Institute Digital Repository
Not a member yet
1915 research outputs found
Sort by
Gen-TWIN: Generative-AI-Enabled Digital Twin for Open Radio Access Networks
Get PDFThe realization of efficient Artificial Intelligence (AI) solutions for the optimization of next-generation Radio Access Network (RAN) relies on the availability of expansive, high-quality datasets that accurately capture nuanced, site-specific conditions. However, obtaining such abundant, domain-specific measurements poses a significant challenge, especially as network complexity and energy efficiency demand surge toward 6G. In response, we introduce GenerativeAI-enabled Digital Twin (Gen-TWIN), a synthetic data generation framework underpinned by a soft- attention LSTM-based generative adversarial network (soft-GAN). Our model augments realistic transmitter and receiver-focused RF datasets by supplementing scarce empirical samples and providing the synthetic data volumes essential for training advanced AI models on RAN. Accuracy results show that soft-GAN provided 19% performance improvement compared to baseline models.TRUEpu
Demonstrating Deep Learning-based Spatial Diffusion
Get PDFMetadata geolocation, i.e., mapping information collected at a cellular Base Station (BS) to the geographical area it covers, is a central operation in producing statistics from mobile network measurements. This task requires modeling the probability that a device attached to a BS is at a specific location, and it is currently accomplished via simplistic approximations based on Voronoi tessellations. However, Voronoi cells exhibit poor accuracy compared to real-world geolocation data, which can reduce the reliability of downstream research pipelines. To overcome this limitation, DEEPMEND proposes a new data-driven approach relying on a teacher-student paradigm that combines probabilistic inference and deep learning. Similarly to other benchmarks, DEEPMEND can produce geolocation maps using only the BS positions, yielding a 56% accuracy gain compared to Voronoi tessellations. Our demonstrator will show visual and qualitative comparisons between DEEPMEND and several competitor approaches, allowing users to explore BS deployments from different geographical regions and operators.Comunidad de MadridEuropean UnionTRUEinpres
HELIX: High-speed Real-Time Experimentation Platform for 6G Wireless Networks
Get PDFMobile networks are evolving rapidly, with 6G promising unprecedented capabilities in terms of data rates and ultra-low latencies. However, the development of testbed platforms for wireless experimentation has not kept pace. Existing platforms typically offer either end-to-end capabilities with low bandwidth or high band width with limited or no real-time functionality. In this paper, we introduce HELIX , an experimentation platform with 6G scalable real-time capabilities. HELIX integrates a comprehensive physical layer subsystem with multi-numerology support alongside an advanced mixed software-hardware control unit responsible for interacting with the fronthaul network and dynamically configuring
the functional split in real time. On the server side, we implement the necessary drivers and routines to enable seamless integration with O-RAN systems, thus facilitating open and end-to-end experimentation. We demonstrate the capabilities of HELIX through a variety of experiments at sub-6 GHz, 28 GHz, and 60 GHz frequencies. Notably, HELIX achieves data rates of up to 1200 Mbps
using 256-QAM modulation with over 417 MHz of bandwidth, and end-to-end bidirectional latencies of 500 ��s. We show advanced features, including the implementation of Integrated Sensing And
Communication ( ISAC), and discuss how the platform could be extended to support bandwidths of up to 1670 MHz.TRUEinpres
A Scalable DNN Training Framework for Traffic Forecasting in Mobile Networks
Get PDFThe exponential growth of mobile data traffic
demands efficient and scalable forecasting methods to optimize
network performance. Traditional approaches, like training
individual models for each Base Station ( BS) are computationally
prohibitive for large-scale production deployments. In this paper,
we propose a scalable Deep Neural Networks (DNN) training
framework for mobile network traffic forecasting that reduces
input redundancy and computational overhead. We minimize
the number of input probes (traffic monitors at Base Stations
(BSs)) by grouping BS s with temporal similarity using K-means
clustering with Dynamic Time Warping (DTW ) as the distance
metric. Within each cluster, we train a DNN model, selecting
a subset of BSs as inputs to predict future traffic demand for
all BSs in that cluster. To further optimize input selection, we
leverage the well-known EXplainable Artificial Intelligence ( XAI)
technique, LayeR-wise backPropagation ( LRP) to identify the
most influential BS s within each cluster. This makes it possible
to reduce the number of required probes while maintaining high
prediction accuracy. To validate our newly proposed framework,
we conduct experiments on two real-world mobile traffic datasets.
Specifically, our approach achieves competitive accuracy while
reducing the total number of input probes by approximately 81%
compared to state-of-the-art predictors.TRUEinpres
User Empowerment in Adaptive Video Streaming over Best-Effort Networks
Get PDFVideo streaming is not only the largest source of Internet traffic but also one of the most economically significant industries, particularly in its adaptive form, where content is delivered in multiple quality levels via the hypertext transfer protocol (HTTP). Users purchase streaming services from streaming platforms (SPs) and network access from Internet service providers (ISPs), both of which strive to enhance user satisfaction, known as quality of experience (QoE), through contrasting methods. QoE is crucial for the ecosystem’s economy and technological advancements, but its subjective nature complicates measurement and application. Consequently, easier-to-use yet less accurate QoE proxies, termed QoE models, become prevalent.
Service providers, particularly SPs, increasingly incorporate active user involvement to boost QoE. This trend, known as user empowerment, offers users active opportunities to enhance their streaming experience. It targets increasing QoE for those who are willing to invest effort, while not impacting those who prefer a passive role. Because user participation requires balancing effort and reward, empowerment strategies must manage this trade-off effectively while ensuring simplicity and privacy. From the providers’ viewpoint, these strategies serve as extensions of core services, necessitating cost-effectiveness and easy integration. This approach enhances QoE and gives engaged users more control, fostering trust, loyalty, and a competitive advantage for the platform.
Currently, user empowerment techniques in video streaming are still nascent. This thesis aims to bridge the gap in user empowerment within adaptive video streaming, focusing on enhancing QoE through active engagement with SPs and ISPs. The work presents four key contributions: 1) A holistic exploration of the video streaming landscape, emphasizing adaptive streaming of long-form videos over current Internet architecture, while reviewing and classifying state-of-the-art methods and identifying promising development directions. 2) A method for creating personalized QoE models by engaging users in brief assessment sessions, resulting in significant QoE improvements through active learning. 3) A novel mechanism for in-band QoE communication from users to ISPs, utilizing the SP’s client interface for QoE estimation and transmission, supported by a prototype demonstrating its feasibility. 4) An evaluation of shortcomings in QoE practices and models, providing guidelines for improvement.TelematicsUniversidad Carlos III de Madrid, Spai
'Hey mum, I dropped my phone down the toilet': Investigating Hi Mum and Dad SMS Scams in the United Kingdom
Get PDFSMS fraud has surged in recent years. Detection techniques have improved along with the fraud, necessitating harder-to- detect fraud techniques. We study one of these where scammers send an SMS to the victim addressing mum or dad, pretend to be their child, and ask for financial help. Unlike pre- vious SMS phishing techniques, successful scammers interact with victims, rather than sending only one message which contains a URL. This recent impersonation technique has proven to be more effective worldwide and has been coined the ‘hi mum and dad’ scam. In this paper, we collaborate with a UK-based mobile network operator to access the initial ‘hi mum and dad’ scam messages and related user spam reports. We then interact with suspicious scammers pretending to be potential victims. We collect 582 unique mule accounts from 711 scammer interactions where scammers ask us to pay more than £577k over three months. We find that scammers deceive their victims mainly by using kindness and distraction principles followed by the time principle. The paper presents how they abuse the services provided by mobile network operators and financial institutions to conduct this scam. We then provide suggestions to mitigate this cybercriminal operation.TRUEinpres
ALPHAS: Adaptive Bitrate Ladder Optimization for Multi-Live Video Streaming
Get PDFLive streaming routinely relies on the Hypertext Transfer Protocol (HTTP) and content delivery networks (CDNs) to scalably disseminate videos to diverse clients. A bitrate ladder refers to a list of bitrate-resolution pairs, or representations, used for encoding a video. A promising trend in HTTP-based video streaming is to adapt not only the client's representation choice but also the bitrate ladder during the streaming session. This paper examines the problem of multi-live streaming, where an encoding service coordinates CDN-assisted bitrate ladder adaptation for multiple live streams delivered to heterogeneous clients in different zones via CDN edge servers. We design ALPHAS, a practical and scalable system for multi-live streaming that accounts for CDNs' bandwidth constraints and encoders' computational capabilities and also supports stream prioritization. ALPHAS, aware of both video content and streaming context, seamlessly integrates with the end-to-end streaming pipeline and operates in real time transparently to clients and encoding algorithms. We develop a cloud-based ALPHAS implementation and evaluate it through extensive real-world and trace-driven experiments against four prominent baseline approaches that encode each stream independently. The evaluation shows that ALPHAS outperforms the baselines, improving quality of experience, end-to-end latency, and per-stream processing by up to 23%, 21%, and 49%, respectively.MICIU/AEI/10.13039/501100011033 and ERDF, EUAustrian Federal Ministry for Digital and Economic AffairsNational Foundation for Research, Technology and Development, AustriaChristian Doppler Research AssociationTRUEpu
Multimodal packaging waste brand identification approach for extended producer responsibility traceability
No full textExtended Producer Responsibility (EPR) policies in packaging wastes are challenging due to waste traceability in their post-consumer stage. Tracking packages after disposal involves identifying their producers under extreme conditions. Several Computer Vision (CV) approaches for waste material recognition have been successfully tested. However, the identification of waste producers remains unexplored mainly due to difficult conditions for brand recognition and the requirement of large datasets that vary from place to place and over time. We propose a multimodal approach for waste brand identification that utilizes only one ”real” image per product for each brand, achieving a macro F1-score of 0.75 with 23 brands and 38 products. The approach leverages package texts and visual features extracted with pre-trained models and predicts the brand using a KNN model with a custom distance based on the Levenshtein distance. Our method employs data augmentation and random word sampling to create synthetic samples from each product image. The KNN model uses random words and a vector of visual features extracted with a previously trained CNN model for prediction. During prediction, the distance of the nearest neighbors is computed as the weighted sum of the visual features distance and the sum of the minimum words Levenshtein distances. This study demonstrates the feasibility of brand identification on packaging waste for EPR traceability without the burden of large dataset acquisition.TRUEpu
Voice App Developer Experiences with Alexa and Google Assistant: Juggling Risks, Liability, and Security
Get PDFVoice applications (voice apps) are a key element in Voice
Assistant ecosystems such as Amazon Alexa and Google
Assistant, as they provide assistants with a wide range of
capabilities that users can invoke with a voice command. Most
voice apps, however, are developed by third parties—i.e., not
by Amazon/Google—and they are included in the ecosystem
through marketplaces akin to smartphone app stores but with
crucial differences, e.g., the voice app code is not hosted by
the marketplace and is not run on the local device. Previous
research has studied the security and privacy issues of voice
apps in the wild, finding evidence of bad practices by voice
app developers. However, developers’ perspectives are yet to
be explored.
In this paper, we report a qualitative study of the experiences of voice app developers and the challenges they face.
Our findings suggest that: 1) developers face several risks
due to liability pushed on to them by the more powerful
voice assistant platforms, which are linked to negative privacy
and security outcomes on voice assistant platforms; and 2)
there are key issues around monetization, privacy, design, and
testing rooted in problems with the voice app certification
process. We discuss the implications of our results for voice
app developers, platforms, regulators, and research on voice
app development and certification.EU-NextGenerationEU, MCIN/AEI /10.13039/501100011033 and the EU-NextGenerationEUTRUEpu
Roaming across the European Union in the 5G Era: Performance, Challenges, and Opportunities
Get PDFRoaming provides users with voice and data connectivity when traveling abroad. This is particularly the case in Europe where the “Roam like Home” policy established by the European Union in 2017 has made roaming affordable. Nonetheless, due to various policies employed by operators, roaming can incur considerable performance penalty as shown in past studies of 3G/4G networks. As 5G provides significantly higher bandwidth, how does roaming affect user-perceived performance? We present, to the best of our knowledge, the first comprehensive and comparative measurement study of commercial 5G in four European countries.
Our measurement study is unique in the way it makes it possible to link key 5G mid-band channels and configuration parameters (“policies”) used by various operators in these countries with their effect on the observed 5G performance from the network (in particular, the physical and MAC layer) and applications
perspectives. Our measurement study not only portrays the observed quality of experience of users when roaming, but also provides guidance to optimize the network configuration as well as to users and application developers in choosing mobile operators. Moreover, our contribution provides the research community with, to our knowledge, the largest cross-country roaming 5G dataset to stimulate further research.Ministerio de Ciencia e InnovaciónTRUEpu