IMDEA Networks Institute Digital Repository
Not a member yet
1915 research outputs found
Sort by
Real-Time Encrypted Traffic Classification in Programmable Networks with P4 and Machine Learning
No full textNetwork traffic encryption has been on the rise in recent years, making Encrypted Traffic Classification (ETC) an important area of research. Machine Learning (ML) methods for ETC are widely regarded as the state-of-the-art. However, most existing solutions either rely on offline ETC based on collected network data or on online ETC with models running in the control plane of Software-Defined Networks (SDN), all of which do not run at line rate and would not meet the strict requirements of ultra-low-latency applications in modern networks. This work exploits recent advances in data plane programmability to achieve real-time ETC in programmable switches at line rate, with high throughput and low latency. An extensive analysis is first conducted to show how tree-based models excel in ETC on various datasets. Then, a workflow is proposed for in-switch ETC with tree-based models. The proposed workflow builds on (i) an ETC-aware Random Forest (RF) modelling process where only features based on packet size and packet arrival times are used, and (ii) an encoding of the trained RF model into off-the-shelf P4-programmable switches. The performance of the proposed in-switch ETC solution is evaluated on 3 use cases based on publicly available encrypted traffic datasets. Experiments are then conducted in a real-world testbed with Intel Tofino switches, in the presence of high-speed background traffic. Results show how the solution achieves high classification accuracy of up to 95% in QUIC traffic classification, with sub-microsecond delay, while consuming less than 10% on average of the total hardware resources available on the switch.Project PCI2022-133013 (ECOMOME), funded by MICIU/AEI/10.13039/501100011033 and the European Union "NextGenerationEU"/PRTRProject ORIGAMI, funded by the Smart Networks and Services Joint Undertaking (SNS JU) under the European Union’s Horizon Europe research and innovation programme under grant agreement No. 101139271.TRUEpu
Efficient 5G Mobile Network Management: Network Slicing and Global Roaming Optimization
Full text linkThe rapid transformation of mobile network infrastructures from hardware-based systems to software-defined networks (SDNs) has introduced several novel paradigms that aim to enhance scalability, flexibility, and efficiency. One of the most significant advancements in this shift is the concept of network slicing. In essence, network slicing allows a single physical network to be divided into multiple virtual networks, each tailored to the specific requirements of different use cases. These slices can vary in terms of bandwidth, latency, security, and resource allocation, making them highly adaptable to different industries, such as autonomous vehicles, smart cities, and Internet of Things (IoT) applications. This capability offers unprecedented flexibility for service providers, as they can deploy and manage various services on the same physical infrastructure without the need for expensive and complex hardware upgrades.
However, realizing the full potential of network slicing presents several challenges. One of the key issues is the complex management of multiple network slices that may have conflicting requirements. For example, ensuring ultra-reliable low-latency communication for one slice while maximizing bandwidth for another slice can be difficult, especially when they share the same underlying physical resources. The dynamic nature of network slicing, where slices are created, modified, and terminated on demand, also requires highly efficient orchestration and automation systems. Furthermore, maintaining security and isolation between slices is essential to prevent interference or data breaches. As mobile networks continue to evolve, addressing these challenges will be crucial to fully unlocking the benefits of network slicing and realizing the vision of highly flexible, software-driven networks.
This thesis explores innovative approaches to improve the profitability of Mobile Network Operators (MNO) through advanced network slicing strategies in 5G networks. The research focuses on three key areas: overbooking network slices for maximize financial gains, cost-efficient network slice management, and operational performance of Mobile Network Aggregators (MNA) within Network Slicing as a Service (NSaaS) context under roaming scenarios. These innovations leverage state-of-the-art techniques, including deep learning, classical optimization, and data-driven algorithms, to tackle the complex challenges of resource provisioning, network function lifecycle management, and global service optimization.
This document delves into these advancements, structured across several chapters that explore these key areas in depth. In the first contribution, we explore the NSaaS concept and introduce slice overbooking as a promising strategy to maximize resource utilization and boost net profit in cloud-native mobile networks. Network slicing enables the creation of virtualized, custom-tailored network slices, but managing these slices efficiently remains a challenge. Overbooking allows network operators to admit more slices than available physical resources by capitalizing on the fact that tenants seldom use their total reserved capacity simultaneously. The chapter presents a complete NSaaS management solution, overbooKing-Aware Network Slicing as a Service (kaNSaaS), which integrates deep learning with classical optimization to address the dual problems of admission control and resource allocation. Through extensive experimentation with large-scale real-world data on tenant demands, the results show that kaNSaaS boosts operator profits potentially multiplying it by four compared to non-overbooking strategies under real-world conditions.
In the second contribution, the focus shifts to zero-touch management systems, which promise autonomous network operation with minimal human intervention. As modern network infrastructures have evolved into systems with numerous virtualized network functions, traditional manual approaches to lifecycle management are increasingly inadequate. In response, this chapter introduces AZTEC+, a data-driven solution for anticipatory resource provisioning in network slicing environments. Using a hybrid and modular deep learning architecture, AZTEC+ forecasts future service demands and determines optimal trade-offs between resource provisioning, instantiation, and reconfiguration costs and performance requirements. Tested on a large-scale network, AZTEC+ outperformed existing state-of-the-art management strategies by up to 5.85 times, proving its effectiveness in reducing network costs and addressing the complexity of virtualized mobile networks. It effectively balances costs associated with resource instantiation and reconfiguration, making it a highly efficient solution for managing dynamic network slices autonomously. This chapter emphasizes how zero-touch management, paired with anticipatory resource provisioning, offers a scalable approach to future network management.
In the third contribution, we explore the added complexity introduced by MNAs, which represent a new frontier in global mobile telecommunications in NSaaS. MNAs, such as Google Fi, Twilio, and Truphone, operate by leveraging multiple MNOs to provide mobile communication services across different regions. Unlike traditional MNOs, which are limited by geographic boundaries, MNAs dynamically connect to the MNO that offers the best performance based on location and time, ensuring optimal service quality for users, especially those frequently crossing borders. However, the dynamic nature of MNAs introduces a new layer of complexity in meeting network slicing’s Quality of Service (QoS) guarantees as the isolation and management of slices become more intricate with the involvement of multiple, regionally dispersed MNOs. To address this, we quantify and compare the performance of MNA-driven models against traditional MNOs, offering insights into the challenges and trade-offs in achieving reliable QoS in these
advanced operator frameworks. This section presents a detailed performance analysis of the three
aforementioned MNAs, comparing their performance for key applications like web browsing and video streaming within NSaaS across diverse geographical regions, namely the USA and Spain. While MNAs may introduce slight delays compared to local MNOs in certain regions, they significantly outperform the traditional home-routed roaming model in terms of service quality. Moreover, emulation studies using open-source 5G implementations deployed across Amazon Web Services (AWS) locations illustrate the performance gains MNAs can achieve through advanced network function virtualization. This chapter highlights the potential of the MNA model to reshape global mobile services, offering more flexible, efficient, and seamless experiences for end-users.
Overall, this research provides valuable insights into NSaaS overbooking management, zero-touch resource provisioning, and the global reach of MNAs. Together, these innovations represent significant strides toward realizing the next generation of mobile networks, where resource efficiency, automation, and global service quality are paramount. The research highlights the economic benefits of slice overbooking, demonstrating how operators can significantly increase profitability by intelligently managing their resources. Furthermore, by integrating hybrid models of AI and optimization, it lays a strong foundation for future developments in network slicing and cost optimization, offering practical implications for both MNO and application developers. As 5G networks continue to evolve, this work sheds light on the shifting landscape of global network operators and provides a roadmap for addressing the growing complexities of resource allocation and service management in a cloud-native, AI-driven environment. Through a blend of deep learning, anticipatory algorithms, and network virtualization, the telecommunications industry is better positioned to meet the ever-evolving demands of users while optimizing network performance across a wide range of dynamic environments.Horizon 2020Telematics EngineeringUniversidad Carlos III de Madrid, Spai
DeExp: Revealing Model Vulnerabilities for Spatio-Temporal Mobile Traffic Forecasting with Explainable AI
Full text linkThe ability to perform mobile traffic forecasting effectively with Deep Neural Networks (DNN) is instrumental to optimize
resource management in 5G and beyond generation mobile networks. However, despite their capabilities, these DNNs often act as
complex opaque-boxes with decisions that are difficult to interpret. Even worse, they have proven vulnerable to adversarial attacks which
undermine their applicability in production networks. Unfortunately, although existing state-of-the-art EXplainable Artificial Intelligence
(XAI) techniques are often demonstrated in computer vision and Natural Language Processing (NLP), they may not fully address the
unique challenges posed by spatio-temporal time-series forecasting models. To address these challenges, we introduce DEEXP in this
paper, a tool that flexibly builds upon legacy XAI techniques to synthesize compact explanations by making it possible to understand
which Base Stations (BSs) are more influential for forecasting from a spatio-temporal perspective. Armed with such knowledge, we run
state-of-the-art Adversarial Machine Learning (AML) techniques on those BSs to measure the accuracy degradation of the predictors
under adversarial attacks. Our comprehensive evaluation uses real-world mobile traffic datasets and demonstrates that legacy XAI
techniques spot different types of vulnerabilities. While Gradient-weighted Class Activation Mapping (GC) is suitable to spot BSs sensitive
to moderate/low traffic injection, LayeR-wise backPropagation (LRP) is suitable to identify BSs sensitive to high traffic injection. Under
moderate adversarial attacks, the prediction error of the BSs identified as vulnerable can increase by more than 250%.Ministerio de Ciencia, Innovación y UniversidadesMinisterio de Asusntos Económicos y Transformación DigitalRamón y Cajal RYC2022-036375-ITRUEinpres
AIChronoLens: AI/ML Explainability for Time Series Forecasting in Mobile Networks
Full text linkForecasting is increasingly considered a fundamental enabler for the management of next-generation mobile networks. While deep neural networks excel at short- and long-term forecasting, their complexity hinders interpretability, a crucial factor for production deployment. The existing EXplainable Artificial Intelligence (XAI) techniques, primarily designed for computer vision and natural language processing, struggle with time series data due to their lack of understanding of temporal characteristics of the input data. In this paper, we take the research on XAI for time series forecasting one step further by proposing AIChronoLens, a new tool that links legacy XAI explanations with the temporal properties of the input. AIChronoLens allows diving deep into the behavior of time series predictors and spotting, among other aspects, the hidden causes of forecast errors. We show that AIChronoLens’s output can be utilized for meta-learning to predict when the original time series forecasting model makes errors and fix them in advance, thereby improving the accuracy of the predictors. Extensive evaluations with real-world mobile traffic traces pinpoint model behaviors that would not be possible to identify otherwise and show how model performance can be improved by 32% upon re-training and by up to 39% with meta-learning.TRUEpu
Demystifying Resource Allocation Policies in Operational 5G mmWave Networks
Full text linkFive years after the initial 5G rollout, several research works have analyzed the performance of operational 5G mmWave networks. However, these measurement studies primarily focus on single-user performance, leaving the sharing and resource allocation policies largely unexplored. In this paper, we fill this gap by conducting the first systematic study, to our best knowledge, of resource allocation policies of current 5G mmWave mobile network deployments through an extensive measurement campaign across four major US cities and two major mobile operators. Our study reveals that resource allocation among multiple flows is strictly governed by the cellular operators and flows are not allowed to compete with each other in a shared queue. Operators employ simple threshold-based policies and often over-allocate resources to new flows with low traffic demands or reserve some capacity for future usage. Interestingly, these policies vary not only among operators but also for a single operator in different cities. We also discuss a number of anomalous behaviors we observe in our experiments across different cities and operators.TRUEpu
How mature is 5G deployment? A cross-sectional, year-long study of 5G uplink performance
Full text linkAfter a rapid deployment worldwide over the past few years, 5G is expected to have reached a mature deployment stage to provide measurable improvement of network performance and user experience over its predecessors. In this study, we aim to assess 5G deployment maturity via three conditions: (1) Does 5G performance remain stable over a long time span (1 year)? (2) Does 5G provide better performance than its predecessor Long-Term Evolution (LTE)? (3) Does the technology offer similar performance across diverse geographic areas and cellular operators? We answer this important question by conducting two year-long measurement campaigns of 5G uplink performance leveraging a custom Android app: one crowd-sourced, cross-sectional campaign spanning 8 major cities in 7 countries and two different continents (Europe and North America), and one controlled campaign focusing on mmWave deployment at a fixed location in the downtown area of Boston, MA. Our datasets show that 5G deployment in major cities appears to have matured, with no major performance improvements observed over a one-year period, but 5G does not provide consistent, superior measurable performance over LTE, especially in terms of latency, and further there exists clear uneven 5G performance across the 8 cities. Our study suggests that, while 5G deployment appears to have stagnated, it is short of delivering its promised performance and user experience gain over its predecessor.TRUEpu
6G Standardization Potential of the ORIGAMI Novel Architectures and Use Cases
Full text linkThe transition to 6G presents many barriers to be
overcome, as well as opportunities for innovation. The integration
of Network Intelligence (NI) is pivotal in optimizing network
performance, enhancing security, and improving resource allo-
cation. The ORIGAMI project identifies 8 critical barriers to 6G
deployment and proposes both architectural and NI innovations
to overcome them. This paper discusses the standardization
potential of such innovations, which respond to 10 different use
cases, each with diverse necessity and impact on the associated
components, and across multiple network domains. We present in
detail three key architectural innovations, namely the Compute
Continuum Layer (CCL), the Zero Trust Layer (ZTL), and
the Global Service-Based Architecture (GSBA), are leveraged to
ensure dynamic adaptation and zero-trust business models; we
then discuss two representative NI innovations targeting energy
efficiency and infrastructure management. Overall, this paper
shows how ORIGAMI’s comprehensive approach to innovation
aligns with and impacts ongoing standardization efforts.Smart Networks and Services Joint Undertaking (SNS JU) under the European Union’s Horizon Europe research and innovation program under Grant Agreement No. 101139270MCIU/AEI/10.13039/501100011033 and the European Union “NextGenerationEU”/PRTRTRUEinpres
Systematic literature review on quantum applications in nanotechnology
Full text linkThe review of progress in quantum computing (QC) is very pertinent nowadays. There is a remarkable challenge in terms of the contributions that this field can provide at the level of improvements in computing time, but perhaps more importantly, in terms of how to rethink the way in which many of the current problems can be approached. Thus, the objective of this work is a systematic literature review that basically revolves around two questions: How does nanoassembled technology affect quantum computing? And what advantages does quantum computing offer to the advancement of nanotechnology? Therefore, this work analyzes how the advance of quantum computing has been influenced by nanotechnology and vice versa, and how quantum computing affects nanotechnology itself. In this way, this article clarifies the paths at which nanotechnology and quantum computing are connected on the route to future technologies in society. In conclusion, we found out that nanotechnology is crucial for the advancement of QC due to the quantumness stands in the nanometric size and the QC-based industry relies on the solid physics state nanoassembly, while on the other hand, QC significantly increases the performance of nanotransistors, imprint better sensibility features on nanosensors, among other things.TRUEpu
RISENSE: Long-Range In-Band Wireless Control of Passive Reconfigurable Intelligent Surfaces
Full text linkReconfigurable Intelligent Surfaces (RIS) are a promising technology for creating smart radio environments by controlling wireless propagation. However, several factors hinder the integration of RIS technology into existing cellular networks, including the incompatibility of RIS control interfaces with 5G PHY/MAC procedures for synchronizing radio scheduling decisions and RIS operation, and the cost and energy limitations of passive RIS technology. This paper presents RISENSE, a system for practical RIS integration in cellular networks. First, we propose a novel, low-cost, and low-power RIS design capable of decoding control messages without complex baseband operations or additional RF chains, utilizing a power sensor and a network of microstrip lines and couplers. Second, we design an effective in-band wireless RIS control interface, compatible with 5G PHY/MAC procedures, that embeds amplitude-modulated (AM) RIS control commands directly into standard OFDM-modulated 5G data channels. Finally, we propose a low-overhead protocol that supports swift on-demand RIS re-con gurability, making it adaptable to varying channel conditions and user mobility, while minimizing the wastage of 5G OFDM symbols. Our experiments validate the design of RISENSE and our evaluation shows that our system can reconfigure a RIS at the same pace as users move, boosting 5G coverage where static or slow RIS controllers cannot.This work received funding from the EU’s Horizon Europe program under Grant Agreements No. 101139270 (ORIGAMI) and 101192521 (MultiX), the Spanish Ministry of Economic A airs and Digital Transformation through UNICO 5G I+D (OPEN6G), the CERCA Programme, the Spanish MCIN/AEI/FEDER, EU (PID2022-136769NB-I00), and the Madrid Regional Government through Project TUCAN6-CM (TEC-2024/COM-460). Syed Waqas Haider Shah is a Juan de la Cierva awardee (JDC2023-050996-I), funded by MCIU/AEI/10.13039/501100011033 and the EU (ESF+).TRUEinpres
Sharing Heterogeneous Computing Resources in Virtualized Open Radio Access Networks
Full text linkVirtualization has recently become a fundamental paradigm in the implementation of 5G Networks, specifically in the context of Radio Access Networks (RANs). RAN virtualization enables baseband processing on general-purpose computing platforms, thus overcoming the coupling of RAN functions with dedicated hardware of traditional hardwired RANs. This approach allows RAN operators to disrupt traditional hardware vendor lock-ins and enable sharing and multiplexing of the available computing resources, as the RAN Base Station (BS) is disaggregated into minimal Radio Unit (RU) hardware connected to cloud-oriented computing platforms that run in software virtual signal processing tasks of the Distributed Unit (DU) and Centralized Unit (CU). However, the execution of such tasks in a timely manner at high probabilities (reliably) is challenging due to their computationally intensive nature, especially at DU level. For this purpose, carrier-grade virtualized RANs (vRANs) today rely on general-purpose computing platforms equipped with Hardware Accelerators (HAs), which are generally energy-hungry and monetarily expensive but can guarantee DU processing reliability. Traditional HAs include Application-Specific Integrated Circuits (ASICs) and Field Programmable Gate Arrays (FPGAs). Recently, Graphics Processing Units (GPUs) have also been considered as HAs, hinging on their unique capability to be easily programmable in software and to efficiently process Machine Learning (ML) workloads, whose algorithms can be used to automate and optimize DU operations. HAs also significantly increase the energy and monetary costs of vRANs, which puts at stake the environmental and economic sustainability of next-generation mobile networks. Therefore, deploying reliable hardware accelerated vRANs with the lowest possible energy toll while reducing deployment costs has become a challenging problem for operators. Indeed, current industrial solutions fail to provide energy and cost efficiency in vRAN, as respectively (i) more energy-efficient processors are shunned for DU tasks processing and (ii) dedicated HAs are assigned to individual DUs following an overprovisioned approach. This thesis investigates the deployment of energy- and cost-effective yet reliable vRANs in order to close the gap with respect to the aforementioned standard solutions. The vision set forth by this thesis is twofold: (i) increasing the energy-efficiency of traditional hardware accelerated vRANs by opportunistically complementing HAs with less energy-hungry computing processors like Central Processing Units (CPUs) to process DU tasks; (ii) improving the cost-efficiency by means of DU centralization, that allows to amortize the cost of each expensive HA by sharing the same HA across multiple DUs. In line with this vision, the first solution proposed in this thesis is ECORAN, an efficient multi-agent contextual bandit ML algorithm operating in the O-RAN Near Real-Time RAN Intelligent Controller (Near-RT-RIC). ECORAN configures policies to
opportunistically offload DU workloads to either a GPU-based HA or CPUs in the O-Cloud to save energy while preserving the reliability of the vRAN. To address cost-efficiency, instead, ECORAN applies concepts from mean field theory to be fully scalable and thus deal with an arbitrarily large and dynamic number of DUs, that are centralized in the same shared and HA-powered computing platform. Using traffic traces from a production mobile network, ECORAN can provide up to 40% energy savings and roughly up to 60x cost gains with respect to the standard approach used today by the industry. As with many other studies on RAN control in the literature, the offloading policy used in ECORAN is determined by an ML model that requires GPU resources for efficient training and execution in production. However, dedicating a GPU to each ML model that
automates a specific RAN control function and reserving a GPU-based HA solely for DU
processing is not a cost-efficient approach. Conversely, indiscriminately co-locating DU
workloads and multiple ML services can compromise the processing reliability of the
former and the throughput performance of the latter. Under this perspective, this thesis explores reliable multiplexing opportunities of the resources of a single GPU to further
squeeze cost-efficiency in vRANs. To this end, this thesis proposes YinYangRAN, an innovative system operating in the Non-Real-Time-RIC that supervises the multiplexing of the computing resources of a GPU-based HA as to ensure reliability in processing DU tasks while maximizing the throughput of a concurrent ML service running in the same GPU. Experiments performed with workloads collected in real RANs show that YinYangRAN can potentially reduce the deployment cost by a factor of N compared to the solution using N dedicated GPUs per process, and improve vRAN reliability by over 50% compared to hardware-accelerated vRANs using conventional GPU multiplexing
methods, with minimal impact on co-located ML workloads. Based on insights from tracking workload dynamics in real-world cells, it is observed that traffic exhibits burstiness at the Transmission Time Interval (TTI) level, i.e. with a timescale of 1 ms. However, YinYangRAN operates in Non-Real-Time (timescale≥ 1 s), and ECORAN operates in Near-Real-Time (timescale ∼ 10 − 100 ms). Consequently, to ensure reliability in the vRAN, both solutions adopt a conservative approach by configuring resources for the highest expected peak over the entire decision period, that is significantly longer than 1 ms. This approach leads to wasting resources for most of the decision period due to overprovisioning. Thus, additional gains in energy and cost efficiency can potentially be achieved by exploiting a heterogeneous O-Cloud infrastructure with both HAs and CPUs through a real-time controller capable of responding to TTI-level traffic fluctuations. To address this, this thesis introduces CloudRIC, a real-time brokering system powered by lightweight data-driven models that jointly coordinates a centralized access for multiple DUs to a heterogeneous pool of computing processors, including HAs and CPUs, and assists DUs with compute-aware radio policies while meeting vRAN-specific reliability targets. Extensive experimental evaluations on GPU-accelerated vRANs demonstrate that CloudRIC can achieve, respectively, 3x and 15x average gains in energy and cost efficiency under real and even dense RAN workloads compared to the industry-standard solution that assigns dedicated HAs to individual DUs, while maintaining the same 99.999% target reliability. At the time of writing this thesis, the proposed solutions are, to the best of our knowledge, the only approaches aimed at reliably deploying energy- and cost-efficient hardware-accelerated vRANs through both DU centralization and the combined use of HAs and CPUs for DU processing.Telematics EngineeringIMDEA Network