Teaching a machine to accurately identify human activities from sensor data poses a significant challenge, which is further compounded by considerations of data privacy, resource costs, and responsiveness, particularly within the constraints of devices like smartphones. While current solutions efficiently identify activities, trained models are barely portable in scenarios composed of diverse activities and limited battery life devices, such as smartphones. This paper introduces Transferable and Copyright-Preserving Human Activity Recognition (TCP-HAR), a mobile-based HAR system that integrates digital watermarking, Federated Learning (FL), Transfer Learning (TL), and compression techniques to provide efficient human activity recognition while providing copyright protection of deep neural network models over Android smartphones. Our solution optimizes the utilization of FL, TL, and their combination (FTL) by extensively testing standalone TL models in offline contexts and comparing these results with FL across a network of mobile devices. Our findings highlight the benefits of TCP-HAR for mobile environments in terms of accuracy, F1-score, and training time. In addition, our proposed watermarking mechanism is robust yet computationally efficient, ensuring ownership verification without compromising the scalability of the TFL process.

Federated Learning (FL) enables collaborative model training without sharing raw data, making it attractive for privacy-preserving applications at the wireless edge. However, when executed over real 5G networks, FL performance degrades due to uplink congestion, heterogeneous client capabilities, and intermittent connectivity. Most existing approaches attempt to mitigate these issues indirectly by optimizing clients (through adaptive participation, local training, or selection strategies) or by optimizing models (via pruning, quantization, or compression), but they ignore potential network bottlenecks. This paper introduces FLAG, an FL architecture that embeds innetwork aggregation directly into 5G gNodeBs, transforming the network into an active participant in the learning process. In particular, FLAG performs parameter aggregation at line rate within the 5G Service Data Adaptation Protocol layer and incorporates three mechanisms: Partial-Contribution Correction for loss-tolerant averaging, a timer-driven pipeline for real-time scheduling, and a deadline-based grouping strategy to mitigate stragglers. Experiments with realistic wireless emulation show that FLAG achieves up to 5.1× faster time-to-accuracy and maintains accuracy within 0.8% of a loss-free baseline, while reducing gNB-to-server bandwidth by aggregating pergNB rather than per-client. FLAG requires no modifications to clients or the parameter server, demonstrating how 5G-aware system design can make federated learning scalable, efficient, and resilient under real-world wireless conditions.

Network traffic telemetry plays a crucial role in the management of modern networks. Estimation of the network traffic matrix is a widely recognized problem whose solutions can span a diverse set of applications. Current approaches to traffic matrix inference through statistical methods often rely on assumptions about the matrix structure, which may be invalid in certain scenarios. Data-driven methods, instead, often use detailed information about the network topology that may be unavailable or impractical to collect. To overcome these challenges, we propose a super-resolution technique for traffic matrix inference that leverages coarser measurements to predict fine-grained network traffic. Furthermore, we devise a distributed learning procedure and adapt our model to scenarios of partial network visibility. Our experiments on real network traces demonstrate that the proposed approach can infer fine-grained network traffic with high precision. Moreover, we prove that our distributed approach improves the inference accuracy with respect to its centralized counterpart, significantly lowering the training time, even in scenarios with partial network knowledge.

As our digital presence expands, safeguarding private data and preserving online privacy becomes paramount. Thus, motivating the development of secure DNS systems, such as DNS over TLS or HTTPS. The vulnerability of these protocols against privacy attacks has led to the development of the Oblivious DNS-over-HTTPS (ODoH) protocol. Nevertheless, the extent of ODoH’s effectiveness in protecting clients’ privacy is still unknown. This study investigates ODoH resiliency against website fingerprinting attacks in the open-world setting. We deploy an ODoH testbed on GENI for data collection and employ deep learning techniques such as ensemble learning for data analysis. Our findings reveal that a passive adversary can identify targeted websites using ODoH traces with an accuracy of 94%. Additionally, we analyze the impact of various factors, including clients’ locations, available resolvers, and time stability, on the attack’s success. Finally, we prototype a mitigation strategy and demonstrate its effectiveness in safeguarding clients privacy.

Effective congestion control is critical for wireless networks, where rapidly varying channel conditions and diverse traffic demands can severely degrade performance. Traditional congestion control algorithms rely on static heuristics that are often ill-suited for dynamic wireless environments. In this paper, we introduce Flecto, a Reinforcement Learning (RL)-based congestion control solution integrated into the QUIC protocol that, leveraging cross-layer metrics, including Signal-to-Noise Ratio, Block Error Rates, and Round-Trip Time measurements, can take decisions using a comprehensive view of network conditions. We implemented Flecto on a 5G testbed using OpenAirInterface and ETTUS USRP B210 radios, showing how it adapts transmission rates in real-time to maximize throughput and minimize latency while maintaining stability. Experimental results show that Flecto achieves an average throughput of 4539.5 KB/s approximately 6% higher both than Cubic (4267.2 KB/s) and New Reno (2674.1 KB/s) while reducing the average Round-Trip Time to 21.8 ms, significantly lower than Cubic’s 27.6 ms and New Reno’s 174.9 ms. These performance gains underscore the promise of integrating RL with cross-layer feedback for adaptive, efficient congestion control in next-generation wireless networks. Moreover, the modular design of Flecto facilitates its extension to other transport protocols and multi-user scheduling frameworks, paving the way for broader adoption in future wireless systems.

Microbursts - microsecond-scale congestion events - are a major cause of packet loss and performance degradation in modern datacenter networks. While packet deflection techniques can help manage microbursts, current implementations lead to excessive packet reordering, exacerbated congestion under high load, and head-of-line blocking in switch buffers. In this paper, we design and implement RobinHood, a novel in-network burst-tolerant protocol. At its core, the protocols mechanisms and policies are based on work-stealing, a technique originally designed to reduce job completion times in operating systems. Through extensive trace-driven simulations on leaf-spine and fattree topologies, we show that RobinHood improves flow completion times up to 22% over Equal-Cost Multi-Path (ECMP), and up to 7% over recent solutions, DIBS and Vertigo, under high load scenarios.

In recent decades, applications in environmental sustainability, education, and housekeeping have become increasingly distributed and sophisticated, leveraging a wide range of devices to perform complex tasks. While a large number of agents can reduce computation time, managing these distributed systems presents significant challenges due to resource constraints such as power consumption and storage. To address this, the literature has explored various model compression techniques, such as pruning, to optimize performance in distributed environments. In this paper, we propose DFL-Trim, a solution for trimming models in Decentralized Federated Learning (FL) that meets network constraints while maintaining satisfactory performance. We demonstrate how pruning can be implemented in decentralized settings, analyze its effect on bandwidth usage, and discuss the trade-offs between compression and model accuracy.

De-authentication Denial of Service (DoS) attacks in wireless networks allow adversaries to maliciously disassociate devices, interrupting communication and effectively denying service. The 802.11w protocol was designed to counter this issue using Protected Management Frames (PMF). However, our analysis reveals that during de-authentication DoS attacks, throughput drops significantly, and client disconnections may occur, exposing the limitations of the 802.11w protocol. Extended Berkeley Packet Filter (eBPF) and eXpress Data Path (XDP) technologies, recently adopted in wired networks to enhance packet processing efficiency, remain largely unexplored in wireless networks and their unique challenges, such as those posed by the 802.11 protocol. In this paper, we introduce a novel approach that integrates eBPF/XDP into the mac80211 Linux kernel module to mitigate de-authentication attacks in near real-time with minimal overhead. Our solution partially overcomes the shortcomings of the 802.11w protocol, offering a more robust defense.

As the Internet of Things (IoT) rapidly expands, there is a growing need for advanced routing solutions that accommodate constrained devices and diverse network requirements. Traditional single-metric routing approaches typically focused on parameters such as delay, hop count, or Maximum Transmission Unit (MTU), lack the adaptability required by IoT environments where energy efficiency, bandwidth considerations, and security constraints must be balanced. A more flexible protocol could improve network performance and prolong device lifetimes. To this aim, building on the extensibility of AODVv2, we developed and integrated a multi-metric routing protocol into the ns-3 simulator. Our implementation allows for the prioritization of multiple factors, such as residual battery levels, node reliability, and hop count, through a customizable metric ordering. Using Type-Length-Value (TLV) encoding, the protocol seamlessly incorporates additional metrics without compromising existing functionalities. Experimental results demonstrate that our multi-metric AODVv2 outperforms the original AODVv2 in terms of network performance, reduces energy consumption, and enhances the overall network longevity, thus providing significant benefits for IoT network management. Our released code paves the way for the programmability of other routing metric systems, including battery discharge rates, and the development of dynamic trust adjustment mechanisms for robust and adaptive routing.

Early detection and mitigation of crop pests like the Fall Armyworm is critical to ensuring food security, particularly for pest-sensitive crops such as maize and sorghum. Traditional monitoring approaches based on manual scouting or centralized analytics struggle to meet the high-precision real-time requirements of pest outbreaks, especially in remote and resource-constrained agricultural settings. In this paper, we present the design vision of AgriCane, a cyber-physical system (CPS) architecture that integrates legged robotic platforms, edge-native machine learning pipelines, and low-bandwidth communication to form a closed sensing-compute-actuation loop. Our vision is to have quadruped robots equipped with manipulators for fine-grained inspection under leaves and at soil level, performing lightweight, on-device detection and escalating uncertain cases to cloud-based models via compressed, opportunistic backhaul. This paper outlines the architectural pillars of AgriCane and identifies key open challenges across the communication, computing, and control stack, ranging from uncertainty-aware inference and real-time robotic control to hierarchical decision pipelines and adaptive scouting. Our goal is to foster cross-disciplinary discussion on the design of resilient, autonomous CPS for openfield agriculture.

The escalating complexity and length of online privacy policies pose a substantial obstacle to user comprehension, thereby undermining informed consent. While manual annotation efforts have improved transparency, they are inherently limited in scalability. To address this challenge, we introduce COAT (Comprehensive Online Agreement Transparency), a novel framework for the automated analysis and risk scoring of privacy policies using Large Language Models (LLMs). Within the COAT system, this paper presents a comparative study evaluating several LLMs, including OpenAI’s GPT series and open-source models. Our methodology benchmarks LLM performance against humanannotated privacy risk scores using a set of specific policy clauses. This study validates the feasibility of using LLMs for scalable, automated privacy policy evaluation and highlights the performance disparities among current models. To foster further research and evaluation, the resulting dataset of LLM-generated scores is made publicly available.

Geospatial artificial intelligence (GeoAI) and data processing techniques have significantly advanced object detection, prediction, and classification tasks. However, the availability of machine learning-ready, labeled data for specific applications such as plant disease detection remains the major challenge for the broader adoption of GeoAI. For instance, collecting temporal unmanned aerial vehicle (UAV) imagery of agricultural crops to track disease emergence and progress requires substantial human labor and resources, which is often limited to a small spatial scale. Recognizing the pivotal role of temporal data in pattern recognition, object detection, and scene reconstruction, we introduce an innovative approach to augment multispectral temporal datasets: the geospatial time machine (GTM). Our proposed methodology combines graph neural network (GNN) and generative adversarial network (GAN) architectures to generate comprehensive synthetic temporal data encompassing multivariate time series. The results demonstrate that imagery generated through backcasting can enhance the accuracy of downstream classification tasks by up to 53% in plant disease detection, particularly in the initial stages of analyzing a crop growth using multispectral and multitemporal datasets.

AI/ML has enhanced computer networking, aiding administrators in decision-making and automating tasks for optimized performance. Despite such advances in network automation, there remains limited trust in these uninterpretable models due to their inherent complexity. To this aim, eXplainable AI (XAI) has emerged as a critical area to demystify (deep) neural network models and to provide more transparent decision-making processes. While other fields have embraced XAI more prominently, the use of these techniques in computer network management remains largely unexplored. In this paper, we shed some light by presenting, an XAI-based approach designed to clarify the opaque nature of data-driven traffic engineering solutions in general, and efficient network telemetry, in particular. It does so by examining the intrinsic behavior of the adopted models, thereby reducing the volume of data needed for effective learning. Our extensive evaluation revealed how our approach not only reduces training time and overhead in network telemetry models but also maintains or improves model accuracy, leading, in turn, to more efficient and clear ML models for network management.

In recent years, electromyography (EMG) has become a crucial tool in developing human-machine interfaces (HMIs) and assistive robotics by enabling intent recognition through muscle activity analysis. Analyzing EMG data allows us to develop control algorithms that respond to human intent, enabling context-aware responses and complex interactions. However, accurately classifying motor intent through EMG signals presents challenges. Hence, this study introduces EMG-TransNN-MHA, a transformer-based model to tackle the challenges and accurately classify EMG signals. While classifying the EMG signals, EMG-TransNN-MHA achieved an average training accuracy of 96.88% and a test accuracy of 96.39%, outperforming traditional deep learning and machine learning models. Implementation details and code are available at https://github.com/madibabaiasl/EMGIntentPaper.

Machine learning (ML) has emerged as a compelling approach to identify attacks in network traffic security. Existing malware detection strategies often concentrate on specific facets, such as efficient data collection, particular types of malware, or handling data scarcity. While valid, these strategies typically overlook the potential for minimizing sample size, focusing instead on data augmentation. This work introduces a novel method to determine the minimum sample size necessary to achieve a specified accuracy level, measured by the F1 score derived from the confusion matrix. We focus on TCP header traffic data transformed into images through flow-splitting techniques for multi-class traffic classification. In addition, we introduce a diffusion model to generate new synthetic traffic images and show that our method outperforms existing techniques in terms of stability and predictability. This study also compares the effectiveness of synthetic image augmentation using Generative Adversarial Networks (GANs) and Denoising Diffusion Probabilistic Models (DDPM) in improving image recognition and classification accuracy.

The adoption of Extended Reality (XR) technology has been hindered by the need for high-bandwidth and low-latency networks to provide immersive experiences. Head-mounted devices used in XR are still heavy and not portable, limiting the potential of XR applications in various contexts. In this paper, we propose EdgeVerse, a multi-user XR system that leverages edge computing and extended Berkeley Packet Filter (eBPF) to offload computation, thereby reducing the dependency on high-bandwidth networks in support of XR applications. Our approach enables lightweight clients, such as devices with an edge browser, making XR more accessible to users. The key design of EdgeVerse focuses on offloading the connection and network synchronization of multiple XR users at the edge. We used an XDP bidirectional router that processes XR traffic faster to enhance user interaction, responsiveness, and immersive experience. To establish the practicality of our approach, we evaluate our results on a prototype that indicates improved response times and reduced latency with respect to baseline solutions.

In the context of NextG Wireless Networks, addressing the challenges of wireless communication link reliability is paramount to ensure efficient Distributed Learning systems. However, many recent solutions have overlooked key challenges, such as packet-level losses and the impact of TCP retransmissions, which are crucial for the robustness of these systems. In this paper, we propose the integration of fountain codes into the distributed learning process to offer a robust mechanism to counteract packet loss. Specifically, we propose a cumulative strategy logic based on fountain codes specifically tailored for packet exchanges in Distributed Learning applications. Our evaluation shows that fountain codes significantly enhance the efficiency and reliability of distributed learning model updates under severe packet loss conditions, e.g., a packet reduction of ≈ 84% (≈ 60%) at the UE (gNB) side compared to traditional TCP methods when packet loss probability reaches 0.9 in Federated Learning context. However, under low packet loss scenarios, fountain codes computational overhead becomes non-negligible. These results highlight the potential of fountain codes to serve as a robust alternative to conventional communication protocols in distributed learning systems, particularly in environments characterized by unstable network conditions.

Network measurement and telemetry techniques are central to the management of modern computer networks. Internet traffic matrix estimation is a popular technique employed for network management and telemetry to reconstruct missing information. Existing approaches use statistical methods, which often make impractical assumptions about the structure of the Internet traffic matrix. Data-driven methods, instead, heavily rely on the assumption of full knowledge of network topology data, that may be unavailable or impractical to collect. In this work, we propose ResCue, a deep residual networks technique to infer fine-grained Internet network traffic starting from spatial coarse-grained measurements. To address scenarios with network visibility constraints, we design a federated learning approach for fine-grained traffic prediction with partial network knowledge. Our evaluation across real-world traffic data shows that our proposed approach outperforms existing interpolation techniques and that our federated learning design achieves similar accuracy with respect to its centralized counterpart while requiring only partial knowledge of the network.

Recent advances in Machine Learning (ML) brought several advantages also within computer network management. For programmable data planes, however, it is more challenging to benefit from these advantages, given their limited resource capabilities colliding with the complexity of ML models. In this paper, we propose ART, an attempt to simplify ML-based solutions for routing, so that they can "fit", i.e., be executed, on P4 switches. To provide such model simplification, ART relies on efficient knowledge distillation techniques, converting, in particular, Deep Reinforcement Learning (DRL) models into a simpler Decision Tree (DT). Our evaluation results validate the accuracy of the extracted model and the application of the model logic directly into switches with little impact, paving the way for a more reactive data plane programmability via machine learning integration.

Crop disease recognition is a fundamental keystone in enabling disease control, limiting disease spread, and mitigating farmers’ losses. Recently, advanced image processing techniques for crop disease detection, based on deep learning, have gained significant popularity. However, the practical deployment of these models in real farms remains challenging. This is mostly due to the lack of Internet connectivity which prevents the transmission of the acquired images to sufficiently powerful edge/cloud servers to execute such complex models. LoRa has emerged as a promising network solution for rural areas, thanks to its extensive communication range and cost-efficient deployment. However, the low data rate of this technology prevents its effective application for the transmission of large images for crop disease detection. In this paper, we propose a LoRa-based framework called iCrop. iCrop enables high disease classification accuracy while exploiting the cost-effectiveness of LoRa transmission technologies. Specifically, iCrop is based on a LoRa Node, which captures crop leaf images and preprocesses them through image segmentation. The node selects and transmits the most informative segments over LoRa to the LoRa Edge Server. The server, in turn, runs the disease classification using a Convolutional Nerual Network (CNN) deep learning model empowered with majority voting among segments. To prevent data losses, typical of LoRa transmission, we develop a reliable transmission protocol on top of LoRa, which takes care of retransmissions and efficient communication. Extensive experiments on a real LoRa testbed show the advantages over two comparison approaches with respect to several performance metrics.

NextG wireless will heavily rely on Federated Learning (FL) applications to learn context-aware AI solutions from the massive amount of generated data. Ensuring the reliability of wireless links for such applications is paramount, especially for FL where packet loss can severely hamper performance and efficiency. Traditional approaches fall short under the high packet loss characteristics of wireless networks. This demo shows how the integration of Fountain Codes (FC) into the FL process can bring notable improvements in packet transmission efficiency, especially under high packet loss conditions.

Network slicing in next-generation wireless networks is a mechanism that ensures isolation and optimal resource distribution among users under constraints imposed by limited information availability at Base Stations (BS). While several strategies to optimize wireless slices have been proposed, this poster introduces an approach to resource allocation in 5G networks that integrates detailed end-to-end flow-level data at the transport layer to refine the NextG resource block allocation process. In particular, we present an architecture that leverages the host’s network stack and a novel in-network processing and scheduling component to optimize the dynamic resource allocation process in slicing. The proposed system design, illustrated through a comprehensive system architecture, aims to distribute Resource Blocks (RBs) effectively, accounting for the unique transport-level metrics of each flow. We present some initial evaluation results with an event-driven simulator, reflecting diverse traffic types and leveraging the Rayleigh fading channel model to simulate dynamic user movement. Our findings demonstrate a promising improvement in flow completion times and a reduction in packet loss, compared to traditional allocation methods such as Round Robin and Proportional Fair schemes, typically deployed in 5G production networks.

A large fraction of recent network management tasks rely on Internet traffic matrices, ranging from planning and troubleshooting to routing and anomaly detection. Despite extensive research efforts over the years, acquiring a comprehensive overview of network traffic remains a difficult and error-prone task. While the literature has mostly proposed increasingly accurate and complex Machine Learning (ML) models to reconstruct missing information, in this paper we propose an alternative approach to further enhance this process: combining the ML model with eXplainable AI (XAI) to analyze the model behavior, detect most significant features, and limit the reconstruction process to such reduced input. With this methodology, not only we simplify the problem, but the entire solution finds greater deployability as the data acquisition phase is also simplified. Numerical results demonstrate that, with our solution on a Convolution Neural Network model, the error during completion can be lowered by 80% for a network telemetry traffic reduction of 75%.

Monitoring plantations is crucial for crop management and producing healthy harvests. Unmanned Aerial Vehicles (UAVs) have been used to collect multispectral images that aid in this monitoring. However, given the number of hectares to be monitored and the limitations of flight, plant disease signals become visually clear only in the later stages of plant growth and only if the disease has spread throughout a significant portion of the plantation. This limited amount of relevant data hampers the prediction models, as the algorithms struggle to generalize patterns with unbalanced or unrealistic augmented datasets effectively. To address this issue, we propose PlantPlotGAN, a physics-informed generative model capable of creating synthetic multispectral plot images with realistic vegetation indices. These indices served as a proxy for disease detection and were used to evaluate if our model could help increase the accuracy of prediction models. The results demonstrate that the synthetic imagery generated from PlantPlotGAN outperforms state-of-the-art methods regarding the Frichet inception distance. Moreover, prediction models achieve higher accuracy metrics when trained with synthetic and original imagery for earlier plant disease detection compared to the training processes based solely on real imagery.

Data center traffic management challenges increase with the complexity and variety of new Internet and Web applications. Efficient network management systems are often needed to thwart delays and minimize failures. In this regard, it seems helpful to identify in advance the different classes of flows that (co)exist in the network, characterizing them into different types based on different latency/bandwidth requirements. In this paper, we propose Howdah, a traffic identification and profiling mechanism that uses Machine Learning and a load-aware forwarding strategy to offer adaptation to different classes of traffic with the support of programmable data planes. With Howdah, the sender and gateway elements inject in-band traffic information obtained by a supervised learning algorithm. When a switch or router receives a packet, it exploits this host-based traffic classification to adapt to a desirable traffic profile, for example, to balance the traffic load. We compare our solution against recent traffic engineering proposals and demonstrate the effectiveness of the cooperation between host traffic classification and P4-based switch forwarding policies, reducing packet transmission time in data center scenarios.

Network measurement and telemetry techniques are central to the management of modern computer networks. Traffic matrix estimation is a popular technique that supports several applications. Existing approaches use statistical methods, which often make invalid assumptions about the structure of the traffic matrix. Data-driven methods, instead, leverage detailed information about the network topology that may be unavailable or impractical to collect. In this work, we propose a super-resolution technique for traffic matrix estimation that can infer fine-grained network traffic. In our experiment, we demonstrate that the proposed approach with high precision outperforms existing data interpolation techniques. We also expand our design by employing a federated learning model to address scalability and improve performance. We find that our model increases the accuracy of the inference with respect to its centralized counterpart.

Traffic matrices are used for many network management operations, from planning to repairing. Despite years of research on the topic, their estimation and inference on the Internet are still challenging and error-prone. For example, missing values are unavoidable due to flaws in the measurement systems and possible failure in data collection systems. It is thus helpful for many network operators to recover the missing data from the partial direct measurements. Some existing matrix completion methods do not fully consider network traffic behavior and hidden traffic characteristics, showing the inability to adapt to multiple scenarios. Others instead make assumptions on the matrix structure that may be invalid or impractical, curtailing the applicability. In this paper, we propose Hide & Seek, a novel matrix completion and prediction algorithm based on a combination of generative autoencoders and Hidden Markov Models. We demonstrate with an extensive experimental evaluation on real-world datasets how our algorithm can accurately reconstruct missing values while predicting their short-term evolution.

Years of research on congestion controls have highlighted how end-to-end and in-network protocols might perform poorly in some contexts. Recent advances in data plane network programmability could also bring advantages in transport protocols, enabling mining and processing in-network congestion signals. However, the new machine learning-based congestion control class has only partially used data from the network, favoring a more sophisticated model design but neglecting possibly precious pieces of data. In this paper, we present HINT, an in-band network telemetry architecture designed to provide insights into network congestion to the end-host TCP algorithm during the learning process. In particular, the key idea is to adapt switches’ behavior via P4 and instruct them to insert simple device information, such as processing delay and queue occupancy, directly into transferred packets. Initial experimental results show that this approach comes with a little network overhead but can improve the visibility and, consequently, the accuracy of TCP decisions of the end-host. At the same time, the programmability of both switches and hosts also enables customization of the default behavior as the user’s needs change.

Every day, large amounts of sensitive data are distributed across mobile phones, wearable devices, and other sensors. Traditionally, these enormous datasets have been processed on a single system, with complex models being trained to make valuable predictions. Distributed machine learning techniques such as Federated and Split Learning have recently been developed to protect user data and privacy better while ensuring high performance. Both of these distributed learning architectures have advantages and disadvantages. In this article, we examine these tradeoffs and suggest a new hybrid Federated Split Learning architecture that combines the efficiency and privacy benefits of both. Our evaluation demonstrates how our hybrid Federated Split Learning approach can lower the amount of processing power required by each client running a distributed learning system, and reduce training and inference time while keeping a similar accuracy. We also discuss the resiliency of our approach to deep learning privacy inference attacks and compare our solution to other recently proposed benchmarks.

The computer networking community has been steadily increasing investigations into machine learning to help solve tasks such as routing, traffic prediction, and resource management. The traditional best-effort nature of Internet connections allows a single link to be shared among multiple flows competing for network resources, often without consideration of in-network states. In particular, due to the recent successes in other applications, Reinforcement Learning has seen steady growth in network management and, more recently, routing. However, if there are changes in the network topology, retraining is often required to avoid significant performance losses. This restriction has chiefly prevented the deployment of Reinforcement Learning-based routing in real environments. In this paper, we approach routing as a reinforcement learning problem with two novel twists: minimize flow set collisions, and construct a reinforcement learning policy capable of routing in dynamic network conditions without retraining. We compare this approach to other routing protocols, including multi-agent learning, with respect to various Quality-of-Service metrics, and we report our lesson learned.

Internet traffic matrices are used nowadays for a variety of network management operations, from planning to repairing. Despite years of research on the topic, obtaining a global view of traffic is still challenging and error-prone. Due to flaws in the measurement systems and possible failure in data collection tools, missing values are unavoidable. It is thus helpful for many network operators to recover the missing data from the partial direct measurements. While some existing matrix completion methods allowed this reconstruction, they do not fully consider network traffic behavior and hidden traffic characteristics, showing the inability to adapt to multiple scenarios. Others instead make assumptions about the matrix structure that may be invalid or impractical, curtailing the applicability. In this paper, we propose Hide & Seek, a novel matrix completion and prediction algorithm based on a combination of generative autoencoders and Hidden Markov Models. After an extensive experimental evaluation based on both real-world datasets and on a testbed, we demonstrated how our algorithm can accurately reconstruct missing values while also predicting their short-term evolution.

Despite years of research on transport protocols, the tussle between in-network and end-to-end congestion control has not been solved. This debate is due to the variance of conditions and assumptions in different network scenarios, e.g., cellular versus data center networks. Recently, the community has proposed a few transport protocols driven by machine learning, nonetheless limited to end-to-end approaches. In this paper, we present Owl, a transport protocol based on reinforcement learning, whose goal is to select the proper congestion window learning from end-to-end features and network signals, when available. We show that our solution converges to a fair resource allocation after the learning overhead. Our kernel implementation, deployed over emulated and large scale virtual network testbeds, outperforms all benchmark solutions based on end-to-end or in-network congestion control.

The challenges of managing datacenter traffic increase with the complexity and variety of new Internet and Web applications. Efficient network management systems are often required to thwart delays and minimize failures. In this regard, it appears helpful to identify in advance the different classes of flows that (co)exist in the network, characterizing them into different types according to the different latency/bandwidth requirements. In this paper, we propose Howdah, a traffic identification and profiling mechanism that uses Machine Learning and a congestion-aware forwarding strategy to offer adaptation to different traffic classes with the support of programmable data-planes. With Howdah, sender and gateway elements inject in-band traffic information obtained using supervised learning. When a switch or a router receives a packet, it exploits such host-based traffic classification to adapt to a desirable traffic profile, for example, balancing the load. We compare our solutions against recent traffic engineering solutions and show the efficacy of cooperation between host traffic classification and P4-based switch forwarding policies, reducing packet transmission time in datacenter scenarios.

Deep Neural Networks (DNNs) process big datasets achieving high accuracy on incredibly complex tasks. However, this progress has led to a scalability impasse, as DNNs require massive amounts of processing power and local memory to be trained, making them impossible or impractical to be used on a single device. This situation has led to the design of distributed training architectures, where the DNN and the training data can be split among multiple processors. How to choose the appropriate distributed training architecture, however, remains an open question. To help bring insights into this debate, in this work we design a Distributed Training Simulator (DTS) that estimates the training time of a DNN in a distributed architecture through a mathematical model of the distributed architecture and resource-allocation heuristics. We illustrate the power of the proposed DTS through the implementation of five different distributed architectures, Pipeline Learning, Federated Learning, Split Learning, Parallel Split Learning, and Federated Split Learning, and we validate the accuracy of the training estimates using three different datasets of varying complexity and two different DNNs. Finally, we present a trade-off analysis to demonstrate the coherence of DTS estimates for diverse high-performance computing scenarios by comparing these estimates with the behaviors of a real computer cluster.

Localization has achieved great attention in 5G networks, pushed by standardization. However, experimentation in 5G networks lacks the integration of network function modules designed for localization. We present our implementation of the 5G Localization Management Function. It complies with the 3GPP standard and OpenAirInterface, the most advanced framework that implements a full 5G-New Radio stack. We show that we are able to extend the functionality of OpenAirInterface, enabling location services. Finally, we demonstrate that the tool's performance satisfies the 5G Key Performance Indicators required by 3GPP for localization.

Translating high-level policies to lower-level network rules is one of the main goals of control or data plane network programmability. To further abstract requirements and propel automation in networking, several industries have proposed the paradigm of “network intent”. However, the translation from intents to low-level policies is considered critical to program data planes and other network elements, especially when dealing with P4-enabled switches. In this paper, we present NLP4, an architecture that helps translate intents, in the form of human language, into data-plane programs, in the form of P4 rules. In particular, NLP4 uses Natural Language Processing (NLP) techniques to translate high-level human-language intents, a MultiLayer Perceptron (MLP) model for processing the NLP output and converting it into mid-level policy. An API then uses this information, which separates the intent from the network to generate commands readable by P4-enabled switches. Our initial prototype on a network emulator validates our architecture for a specific case: load profiling, demonstrating how even users with limited P4 expertise may customize their networks by merely specifying intents.

The Internet of Things (IoT), coupled with the edge computing paradigm, is enabling several pervasive networked applications with stringent real-time requirements, such as telemedicine and haptic telecommunications. Recent advances in network virtualization and artificial intelligence are helping solve network latency and capacity problems, learning from several states of the network stack. However, despite such advances, a network architecture able to meet the demands of next-generation networked applications with stringent real-time requirements still has untackled challenges. In this paper, we argue that only using network (or transport) layer information to predict traffic evolution and other network states may be insufficient, and a more holistic approach that considers predictions of application-layer states is needed to repair the inefficiencies of the TCP/IP architecture. Based on this intuition, we present the design and implementation of Reparo. At its core, the design of our solution is based on the detection of a packet loss and its restoration using a Hidden Markov Model (HMM) empowered with adversarial autoencoders. In our evaluation, we considered a telemedicine use case, specifically a telepathology session, in which a microscope is controlled remotely in real-time to assess histological imagery. Our results confirm that the use of adversarial autoencoders enhances the accuracy of the prediction method satisfying our telemedicine application’s requirements with a notable improvement in terms of throughput and latency perceived by the user.

Mobile phones, wearable devices, and other sensors produce every day a large amount of distributed and sensitive data. Classical machine learning approaches process these large datasets usually on a single machine, training complex models to obtain useful predictions. To better preserve user and data privacy and at the same time guarantee high performance, distributed machine learning techniques such as Federated and Split Learning have been recently proposed. Both of these distributed learning architectures have merits but also drawbacks. In this work, we analyze such tradeoffs and propose a new hybrid Federated Split Learning architecture, to combine the benefits of both in terms of efficiency and privacy. Our evaluation shows how Federated Split Learning may reduce the computational power required for each client running a Federated Learning and enable Split Learning parallelization while maintaining a high prediction accuracy with unbalanced datasets during training. Furthermore, FSL provides a better accuracy-privacy tradeoff in specific privacy approaches compared to Parallel Split Learning.

Handling High-Performance Computing (HPC) workflows often requires the orchestration of a collection of parallel flows. Traditional techniques to optimize flow-level metrics do not perform well in optimizing such collections because the network is usually agnostic to application requirements. A Coflow is a recently proposed abstraction that created new opportunities in network scheduling for datacenter networks. However, recent work on coflow scheduling has focused on merely two objectives: decreasing communication time of data-intensive jobs and guaranteeing predictable communication time. In this paper, we take a step further and propose some initial results towards the design of heuristics that optimize also the energy consumption of a data center that hosts HPC jobs. To this aim, we built and released an energy-aware coflow scheduling simulator to the community that helps analyze the tradeoff between energy efficiency and coflow completion time. We also propose two scheduling algorithms that consider coflow completion time, CPU utilization, and energy consumption efficiency. Our initial results using the simulator clarify how each policy should be tuned to the application needs and the computational resources available.

Edge computing technologies have improved delays and privacy of several applications, including in medical imaging and eHealth. In this paper, we consider ultrasound technology and echocardiology (echo) and empower it with edge computing. Despite the many advances that ultrasound technology has seen recently, e.g., it is possible to perform echo scans using wireless ultrasound probes, the use of Artificial Intelligence (AI) techniques is becoming a necessity, for faster and more accurate echo diagnosis (not limited to heart diseases). While a few proprietary solutions exist that embed AI within echo devices, none of them uses resource-intensive tasks on handheld devices, and none of them is open-source. To this end, we propose EdgeEcho, an architecture that captures ultrasound data originated from handheld ultrasound probes and tags it using semantic segmentation performed on edge cloud. Our prototype focuses on optimizing the management of edge resources to address the specific requirements of echocardiology and the challenges of serving AI algorithms responsively. As a use case, we focus on a ventricular volume detection operation. Our performance evaluation results show that EdgeEcho can support multiple parallel medical video processing streaming sessions for continuing medical education, demonstrating a promising edge computing application with life-saving potential.

Defending a cyber asset from a targeted attack based on port scanning is a challenging task because attackers exploit protocol behavior essential for productive use of applications. For instance, TCP or UDP ports opened for applications such as file transfer or video can be exploited to launch denial of service attacks. There is a need for proactive methods that can detect port scanning based attacks at their initial stage so that countermeasures can be initiated before the attack impact is disruptive on a cyber asset. This paper presents methods to counteract the initial stages of network attacks involving TCP and UDP port scanning. Our methods analyze outgoing traffic to identify ICMP 3.3 and TCP RST response packets that indicate the beginning of an attack launch. We specifically describe two countermeasures based on software-defined networking controller (at the network level) and Linux utility (at the host level) modules we developed. To validate the effectiveness of our proactive detection based methodology, we set up a testbed with a scheme of a polygon and conducted experiments related to distortion of the port status of attacks. Our results demonstrate that our approach is effective and the accuracy of determining open TCP ports did not exceed 15%, and it did not reach 2% for the remaining ports (closed TCP, UDP of any type).

The soaring complexity of networks has led to more and more complex methods to manage and orchestrate efficiently the multitude of network environments. Several solutions exist, such as OpenFlow, NetConf, P4, DPDK, etc., that allow net-work programmability at both control and data plane level, driving innovation in many focused high-performance networked applications. However, with the increase of strict requirements in critical applications, also the networking architecture and its operations should be redesigned. In particular, recent advances in machine learning have opened new opportunities to the automation of network management, exploiting existing advances in software-defined infrastructures. We argue that the design of effective data-driven network management solutions needs to collect, merge, and process states from both data and control planes. This paper sheds light upon the benefits of utilizing such an approach to support feature extraction and data collection for network automation.

Years of research on transport protocols have not solved the tussle between in-network and end-to-end congestion control. This debate is due to the variance of conditions and assumptions in different network scenarios, e.g., cellular versus data center networks. Recently, the community has proposed a few transport protocols driven by machine learning, nonetheless limited to end-to-end approaches.In this paper, we present Owl, a transport protocol based on reinforcement learning, whose goal is to select the proper congestion window learning from end-to-end features and network signals, when available. We show that our solution converges to a fair resource allocation after the learning overhead. Our kernel implementation, deployed over emulated and large scale virtual network testbeds, outperforms all benchmark solutions based on end-to-end or in-network congestion control.

The most significant challenge of our time is global warming, it impacts every area of our lives. This study was motivated by the observation that to train Artificial Intelligence and Machine learning (AI/ML) algorithms result in staggering carbon footprints. Moreover, centralized implementations are becoming a bottleneck of several AI/ML applications that needs frequent retraining and low latency responses. To overcome the limitations of a centralized ML research community has proposed Federated Learning, a technique used to train AI/ML algorithms in a distributed fashion. There has been significant previous work to reduce power consumption by adopting efficient hardware techniques; while such techniques yield large savings, they are not focusing on distributed learning. We propose an Energy-efficient Consensus Protocol (EECP) for sustainable Federated Learning. Our protocol iterates over the bidding phase and agreement (or consensus) phase by only exchanging bids and a few other policy-driven information with neighbor workers. Our simulations show significant energy savings of up to 22.7% with respect to our benchmark.

Network Function Virtualization (NFV) replaces physical middleboxes with elastic Virtual Network Functions (VNFs). Those VNFs need to be instantiated, and their resources dynamically scaled to meet application and traffic fluctuation requirements. Despite recent extensive research, deciding how to map virtual resources optimally to the underlying infrastructure remains practically a challenge. Existing approaches mostly assign fixed resources to each VNF instance, and transfer virtual flows using a single physical path, without prior knowledge of traffic patterns and available bandwidth. Such resource binding strategies lead to suboptimal physical link utilization. We advance the state of the art in this regard by presenting OctoMap, a system designed to support with learning theory any chain embedding algorithm. OctoMap utilizes a Convolution Neural Network for traffic prediction and provisioning, and a contextual multi-armed bandit algorithm to solve the online VNF chain embedding problem. We show the performance benefits of OctoMap with a trace-driven simulation campaign using publicly available datasets. In particular, we show how OctoMap reduces the costs of provisioning network services under node and link constraints, comparing different predictors and different multi-armed bandit policies.

Due to ubiquitous increase of mobile services and powerful Internet-of-Things (IoT) devices, the interest for mobile-edge computing (MEC) solutions has grown both in industry and academia. One of the fundamental mechanisms of MEC is offloading, i.e., delegation of a computation from the user to a server (set) placed near to the edge. Edge servers may have poor incentives to run a delegated service, for example, for the temporary limited resources. In this article, we dissect the incentive mechanisms within a MEC ecosystem with the aim of ensuring a fault-tolerant edge service under unreliable scenarios. In particular, we design an auction mechanism to model the interaction between the MEC players, and model the edge users’ probability of successful offloading, assuming that the cost of executing each offloading request is private. Scrutinizing the demand uncertainty of edge users, the main motive of our auction method is to optimize the offloading cost to engage more edge users in this process, while imposing probabilistic guarantees of offloading service execution. Our offloading cost minimization problem is considered to be an NP-hard. For the solution, we use a heuristic methodology to get the optimal approximation ratio and provide economical fairness. We provide exhaustive simulation results to show the excellent performance of our scheme.

Next-generation networks are expected to provide higher data rates and ultra-low latency in support of demanding applications, such as virtual and augmented reality, robots and drones, etc. To meet these stringent requirements of applications, edge computing constitutes a central piece of the solution architecture wherein functional components of an application can be deployed over the edge network to reduce bandwidth demand over the core network while providing ultra-low latency communication to users. In this article, we provide solutions to resource orchestration and management for applications over a virtualized client-edge-server infrastructure. We investigate the problem of optimal placement of pipelines of application functions (virtual service chains) and the steering of traffic through them, over a multi-technology edge network model consisting of both wired and wireless millimeter-wave (mmWave) links. This problem is NP-hard. We provide a comprehensive “microscopic” binary integer program to model the system, along with a heuristic that is one order of magnitude faster than optimally solving the problem. Extensive evaluations demonstrate the benefits of orchestrating virtual service chains (by distributing them over the edge network) compared to a baseline “middlebox” approach in terms of overall admissible virtual capacity. Moreover, we observe significant gains when deploying a small number of mmWave links that complement the Wire physical infrastructure in high node density networks.

The service function chaining paradigm links ordered service functions via network virtualization, in support of applications with severe network constraints. To provide wide-area (federated) virtual network services, a distributed architecture should orchestrate cooperating or competing processes to generate and maintain virtual paths hosting service function chains while, guaranteeing performance and fast asynchronous consensus even in the presence of failures. To this end, we propose a prototype of an architecture for robust service function chain instantiation with convergence and performance guarantees. To instantiate a service chain, our system uses a fully distributed asynchronous consensus mechanism that has bounds on convergence time and leads to a (1 - 1/e)-approximation ratio with respect to the Pareto optimal chain instantiation, even in the presence of (non-byzantine) failures. Moreover, we show that a better optimal chain approximation cannot exist. To establish the practicality of our approach, we evaluate the system performance, policy tradeoffs, and overhead via simulations and through a prototype implementation. We then describe our extensible management object model and compare our asynchronous consensus's overhead against Raft, a recent decentralized consensus protocol, showing superior performance. We furthermore discuss a new management object model for distributed service function chain instantiation.

The abiding attempt of automation has also permeated the networks, with the ability to measure, analyze, and control themselves in an automated manner, by reacting to changes in the environment (e.g., demand). When provided with these features, networks are often labeled as “self-driving” or “autonomous”. In this regard, the provision and orchestration of physical or virtual resources are crucial for both Quality of Service (QoS) guarantees and cost management in the edge/cloud computing environment. To effectively manage the lifecycle of these resources, an auto-scaling mechanism is essential. However, traditional threshold-based and recent Machine Learning (ML)-based policies are often unable to address the soaring complexity of networks due to their centralized approach. By relying on multi-agent reinforcement learning, we propose Mystique, a solution that learns from the load on links to establish the minimal set of active network resources. As traffic demands ebb and flow, our adaptive and self-driving solution can scale up and down and also react to failures in a fully automated, flexible, and efficient manner. Our results demonstrate that the presented solution can reduce network energy consumption while providing an adequate service level, outperforming other benchmark auto-scaling approaches.

Current efforts to address and alleviate the suffering and harms associated with gambling disorder (GD) predominately involve in-person psychotherapy or static web-based interventions, while dynamic, proactive approaches to the treatment of GD are needed. Currently, several medical health (mHealth) apps for GD exist, but none are empirically validated and, even though they could be potentially useful, these interventions are static, reactive approaches and they are not based on the geo-fencing principle. Advances in smartphone technology now allow for a paradigm shift. Just-in-time adaptive interventions (JITAI) are technology-based dynamic interventions that proactively respond to time-varying information to provide intervention at critical moments. In this paper, a novel JITAI mHealth app for GD called “Don't Go There” (DGT) is presented. It capitalizes on smartphones' global positioning software (GPS) or other zero-permission embedded sensors to recognizes a user's location. When the patient turns off the GPS capabilities, we infer the geolocation solving a probabilistic route matching problem. The purpose of DGT is to construct a geofence around a gambler's favored gambling establishment, to discourage participation. We hypothesize that the deployment of this app could lead to a reduction in gambling behavior and problems and improve psychological functioning.

Network measurement and telemetry techniques are central to the management of today's computer networks. One popular technique with several applications is the estimation of traffic matrices. Existing traffic matrix inference approaches that use statistical methods, often make assumptions on the structure of the matrix that may be invalid. Data-driven methods, instead, often use detailed information about the network topology that may be unavailable or impractical to collect.

The abiding attempt of automation has also pervaded computer networks, with the ability to measure, analyze, and control themselves in an automated manner, by reacting to changes in the environment (e.g., demand) while exploiting existing flexibilities. When provided with these features, networks are often referred to as "self-driving". Network virtualization and machine learning are the drivers. In this regard, the provision and orchestration of physical or virtual resources are crucial for both Quality of Service guarantees and cost management in the edge/cloud computing ecosystem. Auto-scaling mechanisms are hence essential to effectively manage the lifecycle of network resources. In this poster, we propose Relevant, a distributed reinforcement learning approach to enable distributed automation for network orchestrators. Our solution aims at solving the congestion control problem within Software-Defined Network infrastructures, while being mindful of the energy consumption, helping resources to scale up and down as traffic demands fluctuate and energy optimization opportunities arise.

Distributed learning systems are increasingly being adopted for a variety of applications as centralized training becomes unfeasible. A few architectures have emerged to divide and conquer the computational load, or to run privacy-aware deep learning models, using split or federated learning. Each architecture has benefits and drawbacks. In this work, we compare the efficiency and privacy performance of two distributed learning architectures that combine the principles of split and federated learning, trying to get the best of both. In particular, our design goal is to reduce the computational power required by each client in Federated Learning and to parallelize Split Learning. We share some initial lessons learned from our implementation that leverages the PySyft and PyGrid libraries.

We consider a Multiaccess Edge Computing (MEC) network where distributed servers have energy harvesting (e.g., solar) and storage (e.g., batteries) capabilities. Energy from a connected power grid is also available, in case that harvested from ambient sources is scarce or absent. Network processors are deployed according to a given network topology, across two tiers, and computing tasks are flexibly allocated depending on considerations related to load balancing, energy consumption (for communication and computing) and energy purchases from the power grid. Specifically, an on-line optimization problem, exploiting a predictive control approach, is formulated to minimize the monetary cost incurred in the energy purchases from the power grid, by dispatching the computation jobs to those servers that have enough energy and computation resources. Our proposed framework uses forecasts of exogenous processes, such as the amount of energy harvested and job arrivals, which are estimated on the fly to steer the allocation of computation jobs to the servers.

Multi-homed mobile devices are capable of aggregating traffic transmissions over heterogeneous networks. MultiPath TCP (MPTCP) is an evolution of TCP that allows the simultaneous use of multiple interfaces for a single connection. Despite the success of MPTCP, its deployment can be enhanced by controlling which network interface to be used as an initial path during the connectivity setup. In this paper, we proposed an online MPTCP path manager based on the contextual bandit algorithm to help choose the optimal primary path connection that maximizes throughput and minimizes delay and packet loss. The contextual bandit path manager deals with the rapid changes of multiple transmission paths in heterogeneous networks. The output of this algorithm introduces an adaptive policy to the path manager whenever the MPTCP connection is attempted based on the last hop wireless signals characteristics. Our experiments run over a real dataset of WiFi/LTE networks using NS3 implementation of MPTCP, enhanced to better support MPTCP path management control. We analyzed MPTCP's throughput and latency metrics in various network conditions and found that the performance of the contextual bandit MPTCP path manager improved compared to the baselines used in our evaluation experiments. Utilizing edge computing technology, this model can be implemented in a mobile edge computing server to dodge MPTCP path management issues by communicating to the mobile equipment the best path for the given radio conditions. Our evaluation demonstrates that leveraging adaptive context-awareness improves the utilization of multiple network interfaces.

The network management community has explored and exploited light, copper, and several wireless spectra (including acoustics) as a media to transfer control or data traffic. Meanwhile, haptic technologies are being explored in end-user (wearable) devices, and Tactile Internet is being used merely as a metaphor. However, with rare exceptions and for smaller scoped projects, to our knowledge, vibration has been largely untouched as networking communication media.In this paper, we share the lessons learned while creating and optimizing a pilot testbed that serves as an inexpensive starting point for the exploration of vibration-defined networking. We demonstrate the feasibility of vibrations as a tool for resiliency, physical layer security, and an innovative method of teaching networking concepts to the Visually Impaired (VI) community.

The edge computing paradigm allows computationally intensive tasks to be offloaded from small devices to nearby (more) powerful servers, via an edge network. The intersection between such edge computing paradigm and Machine Learning (ML), in general, and deep learning in particular, has brought to light several advantages for network operators: from automating management tasks, to gain additional insights on their networks. Most of the existing approaches that use ML to drive routing and traffic control decisions are valuable but rarely focus on challenged networks, that are characterized by continually varying network conditions and the high volume of traffic generated by edge devices. In particular, recently proposed distributed ML-based architectures require either a long synchronization phase or a training phase that is unsustainable for challenged networks. In this paper, we fill this knowledge gap with Blaster, a federated architecture for routing packets within a distributed edge network, to improve the application's performance and allow scalability of data-intensive applications. We also propose a novel path selection model that uses Long Short Term Memory (LSTM) to predict the optimal route. Finally, we present some initial results obtained by testing our approach via simulations and with a prototype deployed over the GENI testbed. By leveraging a Federated Learning (FL) model, our approach shows that we can optimize the communication between SDN controllers, preserving bandwidth for the data traffic.

Telepathology aims to replace the pathology operations performed on-site, but current systems are limited by their prohibitive cost, or by the adopted underlying technologies. In this work, we contribute to overcoming these limitations by bringing the recent advances of edge computing to reduce latency and increase local computation abilities to the pathology ecosystem. In particular, this paper presents LiveMicro, a system whose benefit is twofold: on one hand, it enables edge computing driven digital pathology computations, such as data-driven image processing on a live capture of the microscope. On the other hand, our system allows remote pathologists to diagnosis in collaboration in a single virtual microscope session, facilitating continuous medical education and remote consultation, crucial for under-served and remote hospital or private practice. Our results show the benefits and the principles underpinning our solution, with particular emphasis on how the pathologists interact with our application. Additionally, we developed simple yet effective diagnosis-aided algorithms to demonstrate the practicality of our approach.

The demand of low latency applications has fostered interest in edge computing, a recent paradigm in which data is processed locally, at the edge of the network. The challenge of delivering services with low-latency and high bandwidth requirements has seen the flourishing of Software-Defined Networking (SDN) solutions that utilize ad-hoc data-driven statistical learning solutions to dynamically steer edge computing resources. In this paper, we propose RoPE, an architecture that adapts the routing strategy of the underlying edge network based on future available bandwidth. The bandwidth prediction method is a policy that we adjust dynamically based on the required time-to-solution and on the available data. An SDN controller keeps track of past link loads and takes a new route if the current path is predicted to be congested. We tested RoPE on different use case applications comparing different well-known prediction policies. Our evaluation results demonstrate that our adaptive solution outperforms other ad-hoc routing solutions and edge-based applications, in turn, benefit from adaptive routing, as long as the prediction is accurate and easy to obtain.

With edge computing, it is possible to offload computationally intensive tasks to closer and more powerful servers, passing through an edge network. This practice aims to reduce both response time and energy consumption of data-intensive applications, crucial constraints in mobile and IoT devices. In challenged networked scenarios, such as those deployed by first responders after a natural or human-made disaster, it is particularly challenging to achieve high levels of throughput due to scarce network conditions.In this paper, we present an algorithm for traffic management that takes advantage of a deep learning model to implement the forwarding mechanism during task offloading in these challenging scenarios. In particular, our work explores if and when it is worth using deep learning on a switch to route traffic generated by microservices and offloading requests. Our approach differs from classical ones in the design: we do not train centralized routing decisions. Instead, we let each router learn how to adapt to a lossy path without coordination, by merely using signals from standard performance-unaware protocols such as OSPF. Our results, obtained with a prototype and with simulations are encouraging, and uncover a few surprising results.

Network virtualization has enabled new business models by allowing infrastructure providers to lease or share their physical infrastructure. A fundamental network management problem that infrastructure providers face to support customized virtual network services is the Virtual Network Embedding (VNE). This requires solving the (NP-hard) problem of matching constrained virtual networks onto the physical network. In this paper, we propose EVA, an architecture that solves the virtual network embedding problem with evolutionary algorithms. By tuning the fitness function and several other policies and parameters, EVA adapts to different types of network topologies and virtualization environments. We compared a few representative policies of EVA with recent virtual network embedding and virtual network function chain allocation solutions; our findings show how with EVA we obtain higher acceptance ratio performance as well as quicker convergence time. We release our implementation code to allow researchers to experiment with evolutionary policy programmability.

Edge Computing brings flexibility and scalability of virtualization technologies at the edge of the network, enabling service providers to deploy new applications over a richer network infrastructure. However, the coexistence of such variety of applications on the same infrastructure exacerbates the already challenging problem of coordinating resource allocation while preserving the resource assignment optimality. In fact, (i) each application can potentially require different optimization criteria due to their heterogeneous requirements, and (ii) we may not count on a centralized orchestrator due to the highly dynamic nature of edge networks. To solve this problem, we present DRAGON, a Distributed Resource AssiGnment and OrchestratioN algorithm that seeks optimal partitioning of shared resources between different applications running over a common edge infrastructure. We designed DRAGON to guarantee both a bound on convergence time and an optimal (1-1/e)-approximation with respect to the Pareto optimal resource assignment. We evaluate convergence and performance of DRAGON on a prototype implementation, assessing the benefits compared to traditional orchestration approaches.

Traditionally, Network Function Virtualization uses Service Function Chaining (SFC) to place service functions and chain them with corresponding flows allocation. With the advent of Edge computing and IoT, a retiable composition of latency-sensitive SFCs is needed to support applications in geo-distributed cloud infrastructures. However, the optimal SFC composition in this case becomes the NP-hard integer multi-commodity-chain flow (MCCF) problem that has no known approximation guarantees. In this paper, we present a novel practical and near optimal SFC composition approach for geo-distributed cloud infrastructures that also admits end-to-end network QoS constraints such as latency, packet loss, etc. Specifically, we propose a novel metapath composite variable approach that reaches 99% optimality on average and takes seconds for practically sized integer MCCF problems of US Tier-1 (~300 nodes) and regional (~600 nodes) infrastructure providers' topologies. To ensure reliability, we compose SFCs with capacity chance-constraints and backup policies. Using trace-driven simulations comprising of challenging disaster-incident conditions, we show that our solution composes twice as many SFCs than the state-of-the-art network virtualization methods.

In this demonstration we illustrate how haptic communication can occur via a mobile application. In particular, we present an Android-based platform that allows connectivity among haptic devices and smartphones. Our app was designed specifically for educational applications, and we used the Hapkit as our first exemplary hardware platform. Our open-source implementation offers several haptic emulations designed for K-12 Physics courses; furthermore, our Android application provides instrumented metrics capable of analyzing latency and other network parameters among communicating hardware and software processes.

Edge computing paradigm aims at offloading microservices from mobile devices to the edge of network, reducing latency and increasing computational ability. Differently from the monolithic architecture, in a microservice-based system, an application is developed as a suite of microservices, each running independently on containers. To optimize the offloading decision, existing schemes often require a priori knowledge of the service type, latency-requirement, or both. Such limitations, in turn, hinders the quality-of-service (QoS) for real-time (mobile) applications. This paper presents FLAVOUR, a novel distributed and latency-optimal microservice scheduling mechanism for edge computing platform to achieve minimal service latency, while providing guaranteed transmission rates to minimize microservice completion times (MSCT). To design FLAVOUR, we first formulate a stochastic service delay minimization problem with constraints on MSCT and network stability. By solving this problem, we derive an optimal latency-optimal heuristic scheduling problem, which establishes the theoretical foundation for FLAVOUR. We have implemented a FLAVOUR prototype and evaluated FLAVOUR through both the testbed experiments and CloudSim simulations. Our preliminary results show that FLAVOUR holds great promise in terms of latency and throughput under different traffic dynamics.

Network Function Virtualization (NFV) is opening new opportunities for both the business and the research community. As the need to softwarize functions grows, managing the underlying hosting infrastructure faces new challenges. In this paper, we focus on one of these challenges: the ability to provision enough virtualizable infrastructure resources to guarantee smooth and responsive network and application operations. To this aim, we learn from observed patterns of requests to an infrastructure hosting virtual network functions, and we model the problem of appropriately scaling resources to provision them. Using months of real Internet traffic requests, we train and compare the performance of several (classical and more recent) learning algorithms. Our goal is to predict future NFV requests to proactively provision our infrastructure using constraint optimization. Our results, obtained with simulations and with a prototype that deploys Linux containers, show both expected and surprising results, and aims at fostering debates on when and if the juice of (supervised) deep learning techniques is worth the squeeze.

Software-Defined Networking (SDN) is a paradigm that enables easier network programmability based on separation between network control plane and data plane. Network Function Virtualization (NFV) is another recent technology that has enabled design, deploy, and management of softwarized networking services. The vast majority of SDN and NFV based architectures, whether they use Virtual machines (VMs) or Lightweight Virtual Machines (LVMs), are designed to program forwarding, probably the most fundamental among all network mechanisms. In this paper instead we demonstrated that there are other (as important) networking mechanisms that need programmability. In particular, we designed, implemented and extensively tested an architecture that enables policy-programmability of (live) migration of LVMs. Migration is used for maintenance, load balancing, or as a security mechanism in what is called Moving Target Defence (a virtual host migrates to hide from an attacker). Our architecture is based on Docker and it is implemented within a Software-Defined Infrastructure. Migration mechanism can be set easily by means of configuration file, to make a novel policy-based architecture. We evaluated the performance of our system in several scenarios, over a local Mininet-based testbed. We analyzed the tradeoff between several Load Balancing policies as well as several Moving Target Defense solutions inspired by network coding.

Securing Internet of Things (IoT) devices and protecting their applications from privacy leaks is a challenge, due to their weak (computational and storage) capabilities, and their proximity with sensitive data. Considering the resource-constrains of such devices, their long lifetime, and the intermittent connections, classical security approaches are often too difficult or impractical to apply. Moving target defense is an established technique whose goal is to lower the attack surface to malicious users by constantly modifying device footprint. Changing the address to an IoT device without privacy leaks is, however, a nontrivial task. In this paper, we propose a novel method to perform a network-wide (Internet protocol and medium access control) address shuffling procedure, called address shuffling algorithm with HMAC (AShA), which is simple to implement, and whose network overhead is minimal. To demonstrate its effectiveness, we analyze our approach via theoretical analysis and simulations. Our analysis shows how AShA parameters can be adapted to various network sizes while our simulations results show how AShA can be used to successfully perform a global collision-free address renewal on networks of more than 2000 nodes using 16-bit addresses.

Edge Computing is moving resources toward the network borders, thus enabling the deployment of a pool of new applications that benefit from the new distributed infrastructure. However, due to the heterogeneity of such applications, specific orchestration strategies need to be adopted for each deployment request. Each application can potentially require different optimization criteria and may prefer particular reactions upon the occurrence of the same event. This paper presents a Service-Defined approach for orchestrating cloud/edge services in a distributed fashion, where each application can define its own orchestration strategy by means of declarative statements, which are parsed into a Service-Defined Orchestrator (SDO). Moreover, to coordinate the coexistence of a variety of SDOs on the same infrastructure while preserving the resource assignment optimality, we present DRAGON, a Distributed Resource AssiGnment and OrchestratioN algorithm that seeks optimal partitioning of shared resources between different actors. We evaluate the advantages of our novel Service-Defined orchestration approach over some representative edge use cases, as well as measure convergence and performance of DRAGON on a prototype implementation, assessing the benefits compared to conventional orchestration approaches.

Virtual network services that span multiple data centers are important to support emerging data-intensive applications in fields such as bioinformatics and retail analytics. Successful virtual network service composition and maintenance requires flexible and scalable “constrained shortest path management” both in the management plane for virtual network embedding (VNE) or network function virtualization service chaining (NFV-SC), as well as in the data plane for traffic engineering (TE). In this paper, we show analytically and empirically that leveraging constrained shortest paths within recent VNE, NFV-SC and TE algorithms can lead to network utilization gains (of up to 50%) and higher energy efficiency. The management of complex VNE, NFV-SC and TE algorithms can be, however, intractable for large scale substrate networks due to the NP-hardness of the constrained shortest path problem. To address such scalability challenges, we propose a novel, exact constrained shortest path algorithm viz., neighborhoods method (NM). Our NM uses novel search space reduction techniques and has a theoretical quadratic speed-up making it practically faster (by an order of magnitude) than recent branch-and-bound exhaustive search solutions. Finally, we detail our NM-based SDN controller implementation in a real-world testbed to further validate practical NM benefits for virtual network services.

Discovering "closely related entities" in any network, a.k.a. communities, is a key goal for various network analytic applications. In particular, cliques and bicliques are two community structures that have influenced several tools and techniques in Big Data and social networking. A clique is a complete subgraph of an undirected graph and similarly, a biclique is a complete bipartite subgraph. A maximal clique (or biclique) is a clique that is not subset of any other clique (or biclique). Algorithms to list all maximal cliques in general graphs or bicliques in bipartite graphs have been previously studied. In this paper, we enhance these solutions explaining how a novel structure, that we call clique-biclique can be used to unravel richer communities with respect to different problems in a wide variety of networks. We then give two algorithms for efficiently listing maximal labelled-cliques and maximal clique-bicliques. The first algorithm is an extension of the maximal clique enumeration and the second cleverly combines enumeration of maximal cliques and maximal bicliques. We conduct an experimental analysis over different synthetic and real datasets to evaluate performance of the algorithms, and we found that even richer communities compared to cliques and bicliques can be efficiently found in most networks of interest.

For several years researchers have used the term "network orchestration" as a metaphor. In this paper, we make the metaphor reality; we describe a novel approach to network orchestration that leverages sounds to augment or replace various network management operations. We test our Music-Defined Networking approach with both a real and a virtual network testbed, on several mechanisms and applications: from datacenter server fan failure detection to authentication, from load balancing to explicit congestion notification and detection of heavy hitter flows. Our approach can be used with and without a Software-Defined Network controller. Despite its limitations, we believe that sound-based network management has potential to be further explored as an effective and inexpensive out-of-band orchestration technique.

Children constitute a vulnerable population and special considerations are necessary in order to provide proper care for them during disasters. After disasters such as Hurricane Katrina, the rapid identification and protection of separated children and their reunification with legal guardians is necessary to minimize secondary injuries (i:e, physical and sexual abuse, neglect and abduction). At Camp Gruber, an Oklahoma shelter for Louisianan's displaced by Hurricane Katrina, of the 254 children at the camp, 36 ((i.e, 14.2%) were separated from their legal guardians. It took 6 months to reunify the last children; 70% of the children were with their legal guardian after 2 weeks. Imagine not knowing for 2 weeks (or 6 months) if your children are dead or alive. To exacerbate these natural challenges, during a disaster Internet connectivity is scarse or unreliable.

The Cloud computing paradigm is enabling innovative and disruptive services by allowing enterprises to lease computing, storage and network resources from physical infrastructure owners. This shift in infrastructure management responsibility has brought new revenue models and new challenges to Cloud providers. One of those challenges is to efficiently migrate multiple virtual machines (VMs) within the hosting infrastructure with minimum service interruptions. In this paper we first present a live-migration performance testing, captured on a production-level Linux-based virtualization platform, that motivates the need for a better multi-VM migration strategy. We then propose a geometric programming model whose goal is to optimize the bit rate allocation for the live-migration of multiple VMs and minimize the total migration time, defined as a tradeoff cost function between user-perceived downtime and resource utilization time. By solving our geometric program we gained qualitative and quantitative insights on the design of more efficient solutions for multi-VM live migrations. We found that merely few transferring rounds of dirty memory pages are enough to significantly lower the total migration time. We also demonstrated that, under realistic settings, the proposed method converges sharply to an optimal bit rate assignment, making our approach a viable solution for improving current live-migration implementations.

In recent studies, researchers have developed various computation offloading frameworks for bringing cloud services closer to the user via edge networks. Specifically, an edge device needs to offload computationally intensive tasks because of energy and processing constraints. These constraints present the challenge of identifying which edge nodes should receive tasks to reduce overall resource consumption. We propose a unique solution to this problem which incorporates elements from Knowledge-Defined Networking (KDN) to make intelligent predictions about offloading costs based on historical data. Each server instance can be represented in a multidimensional feature space where each dimension corresponds to a predicted metric. We compute features for a "hyperprofile" and position nodes based on the predicted costs of offloading a particular task. We then perform a k-Nearest Neighbor (kNN) query within the hyperprofile to select nodes for offloading computation. This paper formalizes our hyperprofile-based solution and explores the viability of using machine learning (ML) techniques to predict metrics useful for computation offloading. We also investigate the effects of using different distance metrics for the queries. Our results show various network metrics can be modeled accurately with regression, and there are circumstances where kNN queries using Euclidean distance as opposed to rectilinear distance is more favorable.

One of the main goals of mobile edge computing is to support new generation latency-sensitive networked applications. To manage such demanding applications, a fine-grained control of end-to-end paths is imperative. End-to-end delay estimation and forecast techniques were essential traffic engineering tools even before the mobile edge computing paradigm pushed the cloud closer to the end user. In this paper, we model the path selection problem for edge traffic engineering using a risk minimization technique inspired by portfolio theory in economics, and we use machine learning to estimate the risk of a path. In particular, using real latency time series measurements, collected with and without the GENI testbed, we compare four short-horizon latency estimation techniques, commonly used by the finance community to estimate prices of volatile financial instruments. Our initial results suggest that a Bayesian Network approach may lead to good latency estimation performance and open a few research questions that we are currently exploring.

Traffic engineering at network edges is challenging given the latency-sensitive nature of all applications that need to be supported. End-to-end delay estimation and forecasts were essential traffic engineering tools even before the mobile edge computing paradigm pushed the cloud closer to the end user. In this paper, we model the path selection problem for edge traffic engineering using a risk minimization technique inspired by portfolio theory in economics, and we use machine learning to estimate path selection risks. In particular, using real latency time series measurements, both existing and collected with and without the GENI testbed, we compare four short-horizon latency estimation techniques, commonly used by the finance community to estimate prices of volatile financial instruments. Our results suggest that a Bayesian Network approach may lead to good latency (peak) estimation performance, as long as there are dependencies among the time series path latency measurements.

The service function chain paradigm links ordered service functions via network virtualization, in support of applications with severe network constraints. This paradigm is particularly interesting in (federated) scenarios where is beneficial to decouple heavy processing from the core cloud and distributed it closer to end-users, such as in edge clouds. To provide such wide-area (federated) virtual network services, a distributed architecture should orchestrate processes to allow instantiation and maintenance of virtual paths hosting service function chains, while guaranteeing performance and fast convergence, even in presence of failures. To this end, we propose Catena, an architecture for resilient distributed service function chain instantiation. To instantiate a service chain, Catena uses a fully distributed asynchronous consensus mechanism that has bounds on convergence time and guarantees an optimal (1-1/e)-approximation with respect to the Pareto optimal centralized chain instantiation, even in presence of (non-byzantine) failures. We leverage stochastic optimization theory to design Catena, and we evaluate its performance and policy tradeoffs with simulations and on a (released) prototype implementation, finding surprising results and demonstrating policy programmability for the resilient distributed chain instantiation problem.

Edge computing, which is a fundamental component of emerging 5G architectures, involves onloading or offloading multiple virtual network functions from mobile devices to an edge network substrate. In this paper, we present a model for the complete edge function onloading problem, which consists of three main phases: (1) Cyber foraging, which involves discovery of resources monitoring the state of edge resources, (2) edge function mapping, which involves matching requests to available resources, and (3) allocation, which involves assigning resources to mappings. Using optimization theory, we show how these three phases are tightly connected, and how the wide spectrum of existing solutions that either solve a particular phase, or jointly solve two of the phases (along with their interactions), are incomplete and may lead to inefficiencies. Moreover, with extensive simulation experiments we demonstrate that joint optimization of all three phases enables the edge network to host a larger set of constrained edge function requests.

Network virtualization has become a fundamental technology to deliver services for emerging data-intensive applications in fields such as bioinformatics and retail analytics hosted at multi-data center scale. To create and maintain a successful virtual network service, the problem of generating a constrained path manifests both in the management plane with a physical path creation -chains of virtual network functions or virtual link embedding - and in the data plane with on-demand path adaptation - traffic steering with Service Level Objective (SLO) guarantees. In this paper, we define the virtual path embedding problem to subsume the virtual link embedding and the constrained traffic steering problems, and propose a new scheme to solve it optimally. Specifically, we introduce a novel algorithm viz., `Neighborhood Method' (NM) which provides an on-demand path with SLO guarantees while reducing expensive over provisioning. We show that by solving the Virtual Path Embedding problem in a set of diverse topology scenarios we gain up to 20% in network utilization, and up to 150% in energy efficiency, compared to the existing path embedding solutions.

To provide wide-area network services, resources from different infrastructure providers are needed. Leveraging the consensus-based resource allocation literature, we propose a general distributed auction mechanism for the (NP-hard) virtual network (VNET) embedding problem. Under reasonable assumptions on the bidding scheme, the proposed mechanism is proven to converge, and it is shown that the solutions guarantee a worst-case efficiency of (1-(1/e)) relative to the optimal node embedding, or VNET embedding if virtual links are mapped to exactly one physical link. This bound is optimal, that is, no better polynomial-time approximation algorithm exists, unless P=NP. Using extensive simulations, we confirm superior convergence properties and resource utilization when compared to existing distributed VNET embedding solutions, and we show how by appropriate policy design, our mechanism can be instantiated to accommodate the embedding goals of different service and infrastructure providers, resulting in an attractive and flexible resource allocation solution.

Network virtualization has enabled new business models by allowing infrastructure providers to lease or share their physical network. A fundamental management problem that cloud providers face to support customized virtual network (VN) services is the virtual network embedding. This requires solving the (NP-hard) problem of matching constrained virtual networks onto the physical network. In this paper we present VINEA, a policy-based virtual network embedding architecture, and its system implementation. VINEA leverages our previous results on VN embedding optimality and convergence guarantees, and it is based on a network utility maximization approach that separates policies (i.e., high-level goals) from underlying embedding mechanisms: resource discovery, virtual network mapping, and allocation on the physical infrastructure. We show how VINEA can subsume existing embedding approaches, and how it can be used to design novel solutions that adapt to different scenarios, by merely instantiating different policies. We describe the VINEA architecture, as well as our object model: our VINO protocol and the API to program the embedding policies; we then analyze key representative tradeoffs among novel and existing VN embedding policy configurations, via event-driven simulations, and with our prototype implementation. Among our findings, our evaluation shows how, in contrast to existing solutions, simultaneously embedding nodes and links may lead to lower providers' revenue. We release our implementation on a testbed that uses a Linux system architecture to reserve virtual node and link capacities. Our prototype can be also used to augment existing open-source “Networking as a Service” architectures such as OpenStack Neutron, that currently lacks a VN embedding protocol, and as a policy-programmable solution to the “slice stitching” problem within wide-area virtual network testbeds.

Content Distribution Networks (CDNs) could benefit from peer-to-peer (P2P) network techniques to improve content delivery time by leveraging the upload capacity of the entire network. In most available solutions, peers first select a set of partners, and later select the pieces of content to exchange. It is fundamental instead that both peer and piece selection algorithms are performed in a consistent and integrated way. To this aim, we propose a content distribution protocol that unifies piece and peer selection in a single algorithm which optimizes the swarming effects. Our approach leverages Flajolet-Martin sketches, a technique based on Bloom filters, to estimate the number of distinct pieces in the CDN, and then adopts a fractional knapsack problem approach to effectively utilize the entire upload capacity of each peer. We tested our solution with both simulations and Planetlab experiments, showing how the piece estimation at every peer is a good approximation of the global rarest piece across the network, and not just across the first hop neighborhood. Moreover, we show how our solution improves the average downloading time by up to 20%. If we consider only the fastest 50% of peers, the downloading time is improved by 100%. Furthermore, our solution decreases the average first content uploading time by 80% with respect to standard P2P protocols which use a local rarest first piece selection, and tit-for-tat as peer selection algorithm.

Software-Defined Networks (SDNs) based approaches represent an opportunity for easing the deployment and the management of wide-area wireless network services. In this paper, we focus on a particular SDN management mechanism that wireless infrastructure providers need to adopt to support such services: the wireless virtual network (VN) embedding problem. We formulate the problem leveraging on optimization theory, analyzing its complexity, and proposing a general distributed auction mechanism. This mechanism leverages on the max-consensus literature to guarantee bounds on the embedding time and on the performance with respect to a Pareto optimal solution. Using extensive simulations, we confirm superior resource utilization when compared with existing distributed VN embedding solutions, proving to be an attractive and flexible resource allocation approach for wireless SDNs.

We consider a Delay Tolerant Network (DTN) whose users (nodes) are connected by an underlying Mobile Ad hoc Network (MANET) substrate. Users can declaratively express high-level policy constraints on how "content" should be routed. For example, content can be directed through an intermediary DTN node for the purposes of preprocessing, authentication, etc., or content from a malicious MANET node can be dropped. To support such content routing at the DTN level, we implement Predicate Routing where high-level constraints of DTN nodes are mapped into low-level routing predicates within the MANET nodes. Our testbed uses a Linux system architecture with User Mode Linux to emulate every DTN node with a DTN Reference Implementation code. In our initial architecture prototype, we use the On Demand Distance Vector (AODV) routing protocol at the MANET level. We use the network simulator ns-2 (nsemulation version) to simulate the wireless connectivity of both DTN and MANET nodes. Preliminary results show the efficient and correct operation of propagating routing predicates. For the application of content re-routing through an intermediary, as a side effect, results demonstrate the performance benefit of content re-routing that dynamically (on-demand) breaks the underlying end-to-end TCP connections into shorter-length TCP connections.

The initial phase in a content distribution (file sharing) scenario is delicate due to the lack of global knowledge and the dynamics of the overlay. An unwise distribution of the pieces in this phase can cause delays in reaching steady state, thus increasing file download times. We devise a scheduling algorithm at the seed (source peer with full content), based on a proportional fair approach, and we implement it on a real file sharing client. In dynamic overlays, our solution improves by up to 25% the average downloading time of a standard protocol ala BitTorrent.

In this paper, a middleware architecture that enables transparent inter-operability between two different wireless networks is presented. In particular, access to the communication channel is maintained by adaptively switching between GPRS (General Packet Radio Service) and Bluetooth interfaces, both established automatically with the assistance of an RFID tag. The access channels are continuously monitored, allowing the device to switch to the Bluetooth interface, whenever this lower-cost alternative is available. A preliminary field trial has been set up, which has demonstrated the effectiveness of the approach in supporting seamless handover procedures within a heterogeneous network as envisioned for future wireless communications systems.

In this paper, a middleware architecture allowing the transparent mobility among two different wireless networks is presented. The proposed scheme is basically based on adaptively maintaining the access to the communication channel for a mobile client enabled with GPRS (General Packet Radio Service) and Bluetooth interfaces. In particular, both the GPRS and the Bluetooth connections are automatically established by means of an RFID tag. To validate this architecture, a preliminary field trial have been set up pointing out the effectiveness of the proposed approach in supporting seamless handover procedures within a heterogeneous network.